Flexible Automation in Developing Countries: The Impact on Scale and Scope and the Implications for Location of Production (Unu Intech Studies in New Technology and Development, 6)

FLEXIBLE AUTOMATION IN DEVELOPING COUNTRIES

The diffusion of computer-based flexible automation (FA) has been seen as a key to the future industrialization of developing countries. If, as was thought, the impact of FA on economies of scope and optimal scale were to reduce optimal scales and therefore firm sizes, industrial production would be ideally located in developing countries. However, Flexible Automation in Developing Countries reveals that the diffusion of FA may actually act as an obstacle to industrialization in these countries. Ludovico Alcorta examines the extent of, and motives for, the diffusion of FA at a global level and then turns to the local and firm level, bringing together indepth studies of sixty-two firms in Brazil, India, Mexico, Thailand, Turkey and Venezuela. Research focuses upon the impact of computer-numericallycontrolled machine tools on scale and scope by exploring changes in lot sizes and product variety (product scale and scope), total plant output (plant scale) and total firm output (firm scale). Barriers to setting up FA-based operations are discussed, as are factors which may affect a decision to locate in a developing country. The contributed studies reveal a relatively slow diffusion of FA in developing countries and it is demonstrated that while FA possibly increases scope, it also requires that plant output be increased in order to maintain efficiency. Alcorta concludes that location in developing countries will probably only be viable for a few large domestic firms, multinationals seeking to relocate simple but labourintensive assembly processes and in countries with significant domestic markets. This work is unique in addressing scale and scope issues in developing countries and in the wealth of information regarding machine tools which it provides. The data provided in the appendix includes official United Nations data, previously unpublished. This will be of use for all research into trends in the use of machine tools. Ludovico Alcorta is an economist and research fellow at UNU/INTECH. He has conducted research on economic and industrial development for fifteen years, primarily in Latin America.

UNU/INTECH STUDIES IN NEW TECHNOLOGY AND DEVELOPMENT Series Editors: Charles Cooper and Swasti Mitter The books in this series reflect the research initiatives at the United Nations University Institute for New Technologies (UNU/INTECH) based in Maastricht, the Netherlands. This institute is primarily a research centre within the UN system and evaluates the social, political and economic environment in which new technologies are adopted and adapted in the developing world. The books in the series explore the role that technology policies can play in bridging the economic gap between nations, as well as between groups within nations. The authors and contributors are leading scholars in the field of technology and development; their work focuses on: • • • •

the social and economic implications of new technologies; processes of diffusion of such technologies to the developing world; the impact of such technologies on income, employment and environment; the political dynamics of technology transfer.

The series is a pioneering attempt at placing technology policies at the heart of national and international strategies for development. This is likely to prove crucial in the globalized market, for the competitiveness and sustainable growth of poorer nations. 1 WOMEN ENCOUNTER TECHNOLOGY Changing patterns of employment in the Third World Edited by Swasti Mitter and Sheila Rowbotham 2 IN PURSUIT OF SCIENCE AND TECHNOLOGY IN SUB-SAHARAN AFRICA The impact of structural adjustment programmes John Enos 3 THE POLITICS OF TECHNOLOGY IN LATIN AMERICA Edited by Maria Inês Bastos and Charles M.Cooper 4 EXPORTING AFRICA Technology, trade and industrialization in Sub-Saharan Africa Edited by Samuel M.Wangwe 5 TECHNOLOGY, MARKET STRUCTURE AND INTERNATIONALISATION Issues and policies for developing countries Nagesh Kumar and N.S.Siddharthan 6 FLEXIBLE AUTOMATION IN DEVELOPING COUNTRIES The impact on scale and scope and the implications for location of production Ludovico Alcorta

FLEXIBLE AUTOMATION IN DEVELOPING COUNTRIES The impact on scale and scope and the implications for location of production Ludovico Alcorta in collaboration with Ruy de Quadros Carvalho, Lilia Domínguez, Flor Brown Osvaldo Alonso, Francisco Tamayo, Vanessa Cartaya, Ghayur Alam, Hacer Ansal and Peter Brimble

London and New York

INTECH Institute for New Technologies Published in association with the UNU Press

First published 1998 by Routledge 11 New Fetter Lane, London EC4P 4EE This edition published in the Taylor & Francis e-Library, 2005. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.” Simultaneously published in the USA and Canada by Routledge 29 West 35th Street, New York, NY 10001 © 1998 UNU/INTECH All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging in Publication Data Alcorta, Ludovico. Flexible automation in developing countries/Ludovico Alcorta. p. cm.—(UNU/INTECH studies in new technology and development; 6) Includes bibliographical references. 1. Automation—Economic aspects—Developing countries. 2. Computers—Economic aspects—Developing countries. 3. Industrial location—Developing countries. I. Title. II. Series. HC59.72.A9A43 1998 338′.064–dc21 98–11152 CIP ISBN 0-203-19352-0 Master e-book ISBN

ISBN 0-203-26540-8 (Adobe eReader Format) ISBN 0-415-19153-x (Print Edition)

CONTENTS

List of figures

x

List of tables

xi

List of contributors

xvi

Acknowledgements

xviii

List of abbreviations

xx

Introduction

1

Background

1

Objectives

4

Research questions

4

Organization of this book

6

PART I Concepts, method and synthesis

8

1

Scale and scope: concepts and issues

9

1.

Introduction

9

2.

Origins of economies of scale theory: from ‘division of labour’ to cost curves

10

3.

Scale, economies of scale and optimal scale: definitions, sources and dimensions

12

4.

Scope and economies of scope: origins, definitions and sources

15

5.

Flexible automation and economies of scale and scope: the ‘modern technology’ literature view and its critics

21

6.

Technical change, costs, scale and scope: a graphical representation of the issues

30

2

Methodology: research design and implementation

36

1.

Introduction

36

2.

Approach, method and unit of analysis

37

vi

3.

Sampling: technology, industries and countries

39

4.

Data requirements

45

5.

The model case study: Hydraul

49

6.

Implementation of the study

63

3

The diffusion of flexible automation in developing countries

68

1.

Introduction

68

2.

International diffusion of metal-cutting machine tools

69

3.

Surveyed countries’ CNC machine-tool diffusion in an international perspective

80

4.

Diffusion of flexible automation in selected firms

83

5.

Flexible automation and new organizational techniques

91

6.

Factors underlying the diffusion of CNC machine tools in surveyed firms

97

7.

The process of intra-firm diffusion of CNC machine tools

102

8.

Conclusions

106

4

Impacts on scale and scope

109

1.

Introduction

109

2.

Changes in product scale

110

3.

Changes in product variety or scope

113

4.

Changes in plant and firm scale

119

5.

Factors underlying increases in plant scale

122

6.

Other price and efficiency effects

134

7.

Conclusions

136

5

Flexible automation and location of production in developing countries

141

1.

Introduction

141

2.

Flexible automation and production and cost conditions in engineering

142

3.

Scale and scope and firm entry

147

4.

Transport costs, just-in-time and infrastructure

153

vii

5.

Factor prices and biases

156

6.

Is large local demand still necessary?

158

7.

Location of production in developing countries: an assessment

159

6

Conclusions and policy recommendations

164

PART II The country studies

172

7

The impact of flexible automation on scale and scope in the Brazilian engineering industry RUY DE QUADROS CARVALHO

173

1.

Economic and technical change, 1980–93

173

2.

Technical change in the Brazilian engineering industry: the case-study evidence

180

3.

The impact of flexible automation on scale and scope

194

4.

Flexible automation and costs

203

5.

Conclusion: industrial restructuring, flexible automation, scale 210 of production, and the prospects for the location of production in Brazil

8

The impact of flexible automation on scale and scope in the Mexican engineering industry LILIA DOMÍNGUEZ AND FLOR BROWN

213

1.

Introduction: macroeconomic and industrial performance

213

2.

Technical change in the manufacturing industry: case-study evidence

217

3.

Economies of scope

229

4.

Plant and firm scale

234

5.

Costs and prices

240

6.

Main findings and conclusions

249

9

The impact of flexible automation on scale and scope in the Venezuelan engineering industry OSVALDO ALONSO, FRANCISCO TAMAYO AND VANESSA CARTAYA

254

1.

Introduction: the political and economic context

254

viii

2.

Technical change in the engineering industry: the case-study evidence

262

3.

The impact of flexible automation on scale and scope

272

4.

Changes in unit costs and profits

279

5.

Conclusion: flexible automation, scale and location of production in Venezuela

290

10

The impact of flexible automation on scale and scope in the Indian engineering industry GHAYUR ALAM

295

1.

Introduction: industrial development in India

295

2.

Diffusion of flexible automation in the Indian engineering industry

300

3.

Flexible automation and economies of scope and scale

311

4.

Production costs

316

5.

Conclusions

322

11

The impact of flexible automation on scale and scope in the Turkish engineering industry HACER ANSAL

326

1.

The Turkish economy and the engineering industry

326

2.

The diffusion of flexible automation at firm level

333

3.

The impact of flexible automation on scale

342

4.

Changes in unit costs

352

5.

Conclusions

358

12

The impact of flexible automation on scale and scope in the Thai engineering industry PETER BRIMBLE

361

1.

Introduction: industrial development in Thailand

361

2.

Evidence on diffusion from the case studies

367

3.

Firm-level impacts of flexible automation

379

4.

Plant scale and unit costs

383

5.

Summary and conclusions

386

Appendices

393

ix

Bibliography

426

Index

449

FIGURES

1 Short-run unit costs in flexible and inflexible plants 2 Output space for a two-product firm 3 Economies and diseconomies of scope 4 Optimal scale of production for composite good Y* under alternative technologies

17 19 32 33

TABLES

2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 4.1 4.2 4.3 4.4 4.5

Engineering industry’s value added and output as shares of total manufacturing of surveyed countries, 1963–91 Hydraul: basic economic data Hydraul: batch size composition, before and after Hydraul: product diversity, before and after Hydraul: sources of scale increases under old and new technology Hydraul: costs and unit costs Hydraul: unit cost structure Hydraul: main capital investments Hydraul: employment Hydraul: productivity changes Value of world machine-tool consumption, 1974–94 Value of world metal-cutting machine-tool production, 1973–94 Metal-cutting machine-tool net imports, 1978–94 CNC and non-CNC metal-cutting machine-tool production, 1980– 94, in units CNC and non-CNC metal-cutting machine-tool production, 1980– 94, in value Total CNC machine-tool net imports, 1989–94 CNC machine-tool stocks in selected countries, 1982–94 Diffusion of flexible automation by type of equipment in surveyed countries Metal-cutting CNC machine tools in engineering by firm size, industry and type of ownership in surveyed countries Density of automation by firm size, industry and type of ownership Diffusion of new organizational concepts by density of automation Diffusion of new organizational concepts by firm size, industry and type of ownership Motives for diffusion of flexible automation by firm size, industry and ownership Changes in batch size by density of automation Changes in product scale by firm size, industry and type of ownership Changes in product diversity or scope by density of automation Changes in product diversity or scope by firm size, industry and ownership Changes in scale of production by density of automation

41 50 56 57 59 60 60 61 62 62 71 72 73 76 76 77 81 86 87 89 93 94 99 112 112 115 116 120

xii

4.6 4.7 4.8 4.9 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 7.9 7.10 7.11 7.12 8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 8.9 8.10 8.11 8.12 9.1

Changes in unit costs Unit values of CNC metal-cutting machine tools in selected countries, 1975–94 Relative unit values of CNC and non-CNC lathes Relative unit values of machining centres and non-CNC machine tools (excluding lathes) Brazil: industrial manufacturing output, 1980–93 Brazil: the mechanical engineering capital goods industry, 1980–93 Brazil: sales and exports of the autoparts industry, 1980–93 Brazil: basic characteristics of the sampled firms Brazil: sales, employment, productivity and exports in selected engineering firms, 1985 and 1993 Brazil: diffusion of microelectronics-based automation in selected engineering firms, 1994 Brazil: gains in setting-up and machining cycle times for single parts in selected engineering firms, 1985 and 1994 Brazil: distribution of machining batch sizes in sampled firms, 1985 and 1994 Brazil: increase in product diversity in sampled firms, 1985 and 1994 Brazil: estimates of plant capacity in sampled firms, 1985 and 1993 Brazil: indicators of efficiency gains in total plant setting-up time for sampled firms, 1985 and 1994 Brazil: unit cost structures in sampled firms, 1985 and 1994 Mexico: characteristics of sampled firms, 1994 Mexico: types of machines in sampled firms, 1994 Mexico: organization of production processes in sampled firms before and after flexible automation adoption Mexico: labour, organization and administration in sampled firms Mexico: batch sizes in sampled firms Mexico: changes in product diversity in sampled firms Mexico: volume of production and capacity utilization in sampled firms Mexico: gains in lead-time reduction, labour productivity and quality in sampled firms Mexico: changes in price and total unit cost, in local currency real terms, in sampled firms Mexico: structure of costs in sampled firms, 1989 and 1993 Mexico: changes in employment levels in sampled firms Mexico: changes in sales, price, total cost and profit margin, in local currency real terms, in sampled firms Venezuela: average annual growth of GDP, exports and imports, and average export prices for Venezuelan oil industry, 1950–90

120 132 133 133 175 175 177 181 184 189 195 197 198 200 202 205 218 220 225 227 231 233 235 238 241 243 246 247 255

xiii

9.2

Venezuela: structure of manufacturing industry, 1968, 1974, 1984 and 1991 9.3 Venezuela: the engineering and autoparts industries: evolution of the main variables, 1988–91 9.4 Venezuela: CNC equipment by type, 1992 9.5 Venezuela: sampled firms’ size, product and ownership type 9.6 Venezuela: main economic indicators in sampled firms 9.7 Venezuela: efficiency indexes before automation in sampled firms 9.8 Venezuela: flexible automation and organizational improvements in sampled firms 9.9 Venezuela: reasons for adoption of flexible automation in sampled firms 9.10 Venezuela: setting-up times and average batch sizes in sampled firms 9.11 Venezuela: changes in the scope of production in sampled firms 9.12 Venezuela: changes in production volumes in sampled firms 9.13 Venezuela: utilization and efficiency indicators in sampled firms 9.14 Venezuela: changes in unit cost in sampled firms 9.15 Venezuela: changes in unit cost by source in sampled firms 9.16 Venezuela: share of capital and labour in total unit costs in sampled firms 9.17 Venezuela: changes in reject and waste rates and raw material inventories in sampled firms 9.18 Venezuela: changes in energy, maintenance and repair components of unit cost in sampled firms 9.19 Venezuela: R&D, marketing and administration shares in total unit costs in sampled firms 9.20 Venezuela: prices, profits and delivery times in sampled firms 10.1 India: machine-tool industry, 1980–93 10.2 India: machine-tool industry performance in real terms, 1980, 1984, 1988 and 1992 10.3 India: foreign collaborations agreements, 1980–94 10.4 India: employment and diffusion in sampled firms 10.5 India: reasons for the adoption of flexible automation in sampled firms 10.6 India: changes in setting-up times in sampled firms 10.7 India: delivery requirements and batch sizes in sampled firms 10.8 India: changes in product diversity in sampled firms 10.9 India: changes in machining times in sampled firms 10.10 India: changes in rejection rates in sampled firms 10.11 India: changes in machine availability in sampled firms 10.12 India: numbers of conventional machines replaced by CNC machines in sampled firms 10.13 India: cost of equipment in sampled firms

256 260 261 263 265 267 269 271 274 277 277 280 282 282 283 283 285 287 289 298 299 299 303 304 311 312 314 315 316 316 317 318

xiv

10.14 India: changes in labour requirements of sampled firms 10.15 India: changes in annual labour cost in sampled firms 10.16 India: changes in material inventories in sampled firms 10.17 India: changes in space requirements in sampled firms 11.1 Turkey: economic trends, 1977–92 11.2 Turkey: trends in manufacturing, 1977–92 11.3 Turkey: real wage index in manufacturing, 1980–91 11.4 Turkey: share of the engineering industry in total manufacturing value added, 1980–90 11.5 Turkey: structure of the engineering industry, by subsector, 1983–9 11.6 Turkey: value added per employee in the engineering industry, 1987– 90 11.7 Turkey: production, employment and productivity indexes in engineering industry subsectors, 1987–91 11.8 Turkey: main characteristics of sampled firms 11.9 Turkey: changes in machine-tool adjustment time in sampled firms 11.10 Turkey: batch sizes before and after flexible automation adoption 11.11 Turkey: percentage increases in product diversity 11.12 Turkey: export performance of sampled firms before and after flexible automation 11.13 Turkey: cumulative machining times for typical batch sizes in sampled firms 11.14 Turkey: changes in plant output, production capacity and sales in sampled firms 11.15 Turkey: changes in employment and labour productivity in sampled firms 11.16 Turkey: capacity utilization rate in sampled firms, 1993 11.17 Turkey: changes in unit cost in sampled firms 11.18 Turkey: changes in the shares of manufacturing and overhead costs in unit costs in sampled firms 11.19 Turkey: changes in the shares of capital and labour costs in unit costs in sampled firms 11.20 Turkey: changes in raw materials’ share in unit costs and in scrap rates in sampled firms 11.21 Turkey: changes in the share of energy and repairs in unit costs in sampled firms 12.1 Thailand: summary information on companies approached for interview 12.2 Thailand: characteristics of sampled firms 12.3 Thailand: reasons for introducing flexible automation in sampled firms 12.4 Thailand: effects of flexible automation on set-up times in sampled firms

318 319 320 321 327 329 329 330 331 331 332 335 344 345 346 347 348 349 350 351 353 355 356 357 358 368 371 373 380

xv

12.5 Thailand: effects of flexible automation on batch size and processing time in sampled firms 12.6 Thailand: effects of flexible automation on product diversity in sampled firms 12.7 Thailand: effects of flexible automation on unit costs in sampled firms A.1 Total metal-forming and metal-cutting machine-tool production, 1973–94 A.2 World metal-cutting machine-tool production, 1973–94 A.3 CNC and non-CNC lathe production, 1975–94 (units) A.4 CNC and non-CNC lathe production, 1975–94 (US$m.) A.5 Production of machining centres, 1986–94 (units) A.6 Production of machining centres, 1986–94 (US$m.) A.7 Total machine-tool imports, 1962–94 A.8 Total machine-tool exports, 1962–94 A.9 Total metal-cutting machine-tool exports, 1978–94 A.10 Total metal-cutting machine-tool imports, 1978–94 A.11 Total lathe exports, 1978–94 A.12 Total lathe imports, 1978–94 A.13 Total CNC lathe exports, 1989–94 A.14 Total CNC lathe imports, 1989–94 A.15 Total machining centre exports, 1989–94 A.16 Total machining centre imports, 1989–94 A.17 Other CNC metal-cutting machine-tool exports, 1989–94 A.18 Other CNC metal-cutting machine-tool imports, 1989–94 A.19 World machine-tool consumption, 1973–94

382 384 387 393 394 396 398 400 401 402 406 410 412 414 416 418 419 420 421 422 423 424

CONTRIBUTORS

The following contributors have contributed to the second part of this book: Ghayur Alam is a zoologist and business administrator with a PhD from the University of Manchester. He is director of the Centre for Technology Studies, Gurgaon, Haryana, and is a consultant on issues of technology transfer and capabilities. Osvaldo Alonso is an economist with an MSc in Science and Technology from the Centre for Development Studies (CENDES), Central University of Venezuela (UCV). He is a Research Fellow at FIM-Productividad, Venezuela’s research institute on productivity, and a private consultant in the fields of competitiveness, automation and human resource development. Hacer Ansal is a civil engineer with an MSc in Engineering from Northwestern University and a PhD from Sussex University. She is an associate professor in the Faculty of Management, Department of Economics, Istanbul Technical University. Peter Brimble is an economist with MAs in Economics from Georgetown and Sussex Universities and a PhD from Johns Hopkins University. He is currently president of The Brooker Group Ltd, an investment advisory, consultancy and business research firm in Bangkok. Flor Brown is an economist with postgraduate studies in Mexico. Currently she is lecturer at the Graduate Program in Economic Sciences, School of Sciences and Humanities, National Autonomous University (UNAM), Mexico. Vanessa Cartaya is a development sociologist with postgraduate and doctoral studies in France, United States and Venezuela. At the time of writing she was director of the Centre for Economic and Social Research (CIES) in Caracas, Venezuela. Lilia Domínguez is an economist with an MA in Economics from the University of East Anglia and doctoral studies in the New School for Social Research, New York. She is lecturer at the Graduate Program in Economic Sciences, School of Sciences and Humanities, and at the Faculty of Economics, National Autonomous University (UNAM), Mexico. Ruy de Quadros Carvalho is a business administrator with an MA in Political Science from the University of Campinas (UNICAMP) and a PhD

xvii

from Sussex University. He was director of Brazil’s Institute for Applied Economic Research (IPEA) and government advisor on technology and is currently a research fellow at UNICAMP. Francisco Tamayo is a chemical engineer and MBA from the Institute for Advanced Business Studies (IESA) in Venezuela. He is advisor to the Logistics manager in Venezuela’s national telephone company and a private consultant and lecturer on productivity and total quality.

ACKNOWLEDGEMENTS

It would not have been possible to write this book without the assistance and support of many colleagues, and my friends and family. Professor Charles Cooper, UNU/INTECH’s Director, provided the opportunity for me to engage in this research. He set standards of rigour and achievement which I hope have been matched. The academic staff at UNU/INTECH were a source of continual inspiration and helpful comments. Larry Westphal’s insights were particularly useful in the design stage of the project, while my lunch discussions with Nagesh Kumar often helped me to make sense of issues. Towards the end Mary-Anne Schenk was an efficient and organized research assistant, and Sen McGlinn significantly enhanced the text of the country studies. I must thank also the consultants involved in the study who, sometimes against their own judgements, closely followed the requests made of them. In addition, Martin Bell, Geske Dijstra, John Humphrey, Staffan Jacobsson, Jorge Katz, John Meyer, J.Oosterling, Wilson Peres, Charles Sabel, Ed Steinmueller and Erol Taymaz contributed their name and knowledge to a lively workshop held in Maastricht, as well as to the improvement of individual papers delivered there. My special gratitude goes to Staffan Jacobsson whose comments throughout have presented a challenge. The study benefited also from the support of the many manufacturers, users and institutions involved in diverse ways with flexible automation. They are too numerous to name individually here, except perhaps for Brian Stilwell, Jan Raemackers and Les Pratt who were especially helpful in generating understanding of the potential and the limitations of the new technologies. Henri Maurer, from the European Machine Tools Committee (CECIMO) in Brussels, supplied valuable information on machine-tool production. Ronald Janssen, from the United Nations Statistical Division in Geneva, introduced me to the UN’s COMTRADE data and replied promptly to my regularly ‘urgent’ demands. The acknowledgements would not be complete without an expression of my thanks to my wife and children—twins were born during the course of the project —who put up with most of the difficult times I had to go through while completing this research.

xix

None of those acknowledged here are, of course, responsible for any errors that may remain in the book. Ludovico Alcorta Research Fellow Institute for New Technologies United Nations University Maastricht

LIST OF ABBREVIATIONS

AMT CAD CAE CAM CECIMO CM CNC ECE EOS FA FMS GDP GNP GT ISIC ISO JIT MES MIR NAFTA NPD OCEI OECD OEM PDVSA PLCs R&D

Association for Manufacturing Technology Computer-aided design Computer-assisted engineering Computer-aided manufacturing European Committee for Co-operation of the Machine-Tool Industries Cellular manufacturing Computer numerical control Economic Commission for Europe Economies of scale Flexible automation Flexible manufacturing systems Gross Domestic Product Gross National Product Group technology International Standard Industrial Classification International Standards Organization Just-In-Time production Minimum efficient scale Minimum investment requirements North America Free Trade Agreement New product development Oficina Central de Estadística e Informática Organisation for Economic Co-operation and Development Original equipment manufacturer Petroleos de Venezuela Programme logic controllers Research and development

xxi

SITC SMEs TQM UNIDO UNU/INTECH

Standard International Trade Classification Small and medium-sized enterprises Total quality management United Nations Industrial Development Organization United Nations University/Institute for New Technologies

INTRODUCTION

Background Ever since Adam Smith gave expression to his now-famous dictum, ‘the division of labour is limited by the extent of the market’, the issue of economies of scale and optimal scale has figured prominently in the economics discussion. Smith pointed out that larger markets permit greater specialization of labour and machinery, which leads to significant unit-cost reductions. Other factors, such as technological relationships, permitting equipment having greater capacity with a less than proportional increase in investment, and indivisibilities, which make it worthwhile to spread the costs of lumpy equipment, initial product development or setting-up machines over a larger output, also have been considered sources of economies of scale. Economic efficiency, since Smith, has been closely connected to increases in scale. Developing countries’ industrialization has always been limited by increasing optimal scales. The small size of their domestic markets has prevented industries being established or meant that, when established, firms would be producing lower volumes and at unit costs far higher than those of efficient plants elsewhere. Producing at suboptimal levels, in turn, would require high domestic protection, with attendant effects on social welfare. Changes in consumer preferences, income levels, macroeconomic conditions and/or the degree of foreign competition would lead to unused capacity—and, therefore, even higher unit costs and more protection—or to the closure of the facility. Exports could provide a way out of the scale problem, but a minimum of efficiency and learning is often necessary prior to entering foreign markets. During the 1980s and early 1990s a substantial economics, management and engineering ‘modern technology’ literature emerged claiming that new technologies, particularly microelectronics-based forms of automation and design and associated organizational modifications, are leading to fundamental changes in economies of scale. It is said that, contrary to previous ‘mass production’ technologies, where increases in scale were crucial to unit-cost reductions, recent changes in product and process technologies are increasing the flexibility of production units, enabling a switch to the manufacture of a wider variety of goods resulting in falls of optimal scales of output or ‘descaling’.

2 INTRODUCTION

At product level, it is argued that, unlike ‘specialized’ old technologies, the capacity of new technologies or flexible automation (FA) to integrate diverse equipment and functions and to be programmed helps reduce minimum batch sizes—i.e. quantities of the same product treated in a certain process or sequence of operations—by reducing setting-up times and costs. In addition, by allowing production facilities to vary their product range with ease, to use their equipment fully and reduce setting-up costs, FA leads to economies of scope—i.e. cutbacks in unit costs due to joint-production. At plant level, it is asserted that FA is substantially shrinking the size of machinery and plants while at the same time making it possible for most capital equipment to be available in a wider spectrum of capacities, which, together with falling semiconductor prices, are drastically cutting the cost of capital equipment and allowing the emergence of smaller efficient production facilities. Critics argue that there is no reason why FA should lead to reduction in plant scale rather than to ‘scaling-up’. It is claimed that physical size of equipment and plant should not be confused with economic size, as ‘smaller’ equipment may have larger production capacity. Furthermore, by reducing setting-up times and expanding variety, FA may allow increase in total plant output, so that scale and scope economies at product level reinforce scale economies at plant level. It does not follow either that because equipment is physically smaller it costs less. Despite falls in the cost of microchips and computers, and irrespective of their increase in power, the cost of the production equipment that uses them may still be higher than the technologies it replaces. Availability of equipment in smaller capacities would reduce plant scales only if equipment cost falls in line with capacity. Otherwise, it means no more than that smaller firms may have access to these technologies but at lower efficiency levels. At firm level, it is contended, FA will increase research and development (R&D) and marketing ‘fixed’ costs. R&D costs could rise because of the considerable backlog of scientific and engineering knowledge required for the new products, the increasing technical complexity and novelty of many products, the wider range of products on which R&D is done, the larger design efforts required to take full advantage of the more flexible, and faster, manufacturing capabilities, and the sizeable investments in software and computer specialists necessary to link design to manufacturing and other functions of the firm. Marketing costs also could rise because of the higher information requirements for selling and the growing advertising expenditures necessary to inform on the availability of new products. The net effect of these factors, it is argued, could be ‘scaling-up’ rather than ‘de-scaling’, as higher levels of output may be needed to amortize increasing R&D and marketing ‘fixed’ costs. Falling optimal scales resulting from the adoption of FA may have important implications for developing countries, according to the ‘de-scaling’ literature. They may increase the efficiency of small-scale production and, insofar as some minimum optimal or efficient scale is a barrier to entry, lead to more entry and competition. They may help to cope with the uncertainties created by variable demand or by a changing supply of inputs. They may result also in the more widespread impact of the various forms of learning associated with experience

INTRODUCTION 3

and of the externalities ensuing from the acquisition and use of new knowledge. They could pave the way to new patterns of decentralized industrialization based on small production units located outside of urban centres. They may also enhance opportunities for developing countries’ industrial firms to compete successfully in international markets. However, even if plant and firm scales increase, there may still be opportunities for localization of production in developing countries, as long as economies of scope can be reaped. Provided there is demand for the wider product range, it may still be economically justifiable to start industrial production, as total plant or firm capacity may be fully used by producing adequate quantities of individual products. Efficient production for the local market may, in turn, be the first step towards successful entry into foreign markets. In short, working either through lower optimal scales or economies of scope, FA may positively contribute to the establishment of industry in developing countries. Conversely, higher optimal scales and lack of economies of scope mean that any advance towards industrialization achieved by developing countries, particularly in the context of a reduction of trade barriers around the world, will be reversed as local firms have to compete with more efficient producers from abroad. A progressive loss of international industrial competitiveness may, in turn, result in primary product or cheap-labour trade specialization, with a potentially deleterious effect on domestic income, employment and the environment. Higher optimal scales and lack of economies of scope may also result in lower potential for industrial learning and for the spreading of externalities, and so lead to ‘forgetting’ what was learned in the past. They could bring about a more centralized pattern of industrialization based on a small number of firms or ‘enclaves’ selling to world markets but with few ‘spill-overs’ to the rest of the domestic economy. Whatever the merits of these arguments and counter-arguments, and despite the apparent widespread diffusion of FA, especially in developed countries, and the potential impact on location of industrial production, there is very little empirical evidence supporting either the ‘de-scaling’ or the ‘scaling-up’ view. Few systematic studies have been done in developed countries on the impact of FA on product scale and scope, and even fewer on the effect of FA on plant and firm scale. Indeed, by comparison to the wealth of industry-specific and comparative empirical research on scale issues done in the 1950s, 1960s and 1970s by authors such as Bain (1954, 1956), Chenery (1949), Pratten (1975) and Scherer et al. (1975), hardly any fresh research on scale has emerged during the 1980s. What little fresh research there is, is based on a few observations, journalistic sources, or reviews or updating of previous work. In developing countries, despite research showing an increased diffusion of FA, the evidence on scale and scope is even more meagre and amounts to a couple of simulation exercises and case studies.

4 INTRODUCTION

Objectives The potential impact of FA on industrial development clearly warrants careful analytical examination of the issues involved and calls for a body of empirical evidence on which to base such analysis. Accordingly, the main purposes of this book are: to develop a simple conceptual framework for the discussion of the impact of FA on scale and scope; to present the results of empirical research carried out in several developing countries on the effects of FA on scale and scope; and to examine the possible implications for the location of industrial production in developing countries. The conceptual framework will be based on economic production and cost theory. There is a well-established tradition in economics, beginning with Adam Smith (1776) and running through Marshall (1890), Viner (1953) and presentday neo-classical economists, which elaborates on concepts such as returns to scale, economies of scale, optimal scale and indivisibilities. There is also a much younger, albeit equally important, body of neo-classical theory, beginning perhaps with Stigler (1939) and followed, after many years of individual and team effort, by seminal work by Baumol, Panzar and Willig (1988), that has introduced issues of flexibility, output variety and economies of scope into economic theory. These two traditions will constitute the backbone of our conceptual framework. Admittedly, economic theory is ill-equipped to deal with issues of technological change because of its convexity assumption, its lack of a proper understanding of the complexities and dynamics involved in technological change and its disregard of a wealth of empirical evidence on how firms actually operate (Baumol 1987; Morroni 1992; Nelson 1994; Nelson and Winter 1982; Romer 1990, 1994). But for the problem at stake, which is essentially one of a comparatively static nature —whether a change in technology results in changes in economies of scale and scope—static economic theory would seem to be a reasonable starting-point for organizing the issues and data. Given that, unlike many developed countries, hardly any developing country has carried out systematic surveys of use of FA by manufacturing firms on which to base a fully fledged statistical analysis and that, in any case, the overall number of firms using FA in developing countries was expected to be low, the main approach of the research reported here was an in-depth analysis of data obtained from several case studies. Some sixty-two firms in six developing countries— Brazil, India, Mexico, Thailand, Turkey and Venezuela—were examined for this research. Combined with secondary data available internationally, this approach should provide an overall picture of the diffusion and the scale and scope of the impact of FA in developing countries. Research questions The research was essentially exploratory: there was no general presumption of which way the main results would go. Although there were some expectations regarding specific results, these were neither in line with the ‘conventional

INTRODUCTION 5

wisdom’ nor with the views of many colleagues and consultants participating in the project. There was an advantage in engaging consultants with different expectations, as this imposed a much tougher standard on data collection and interpretation. The very categorical views expressed in the literature and shared by many colleagues suggested that the overall research approach had to be slightly indeterminate and framed in terms of open-ended research questions rather than precisely defined hypotheses. Three questions guided this research: 1. Has flexible automation diffused to developing countries? To what extent? What factors account for its diffusion? 2. What has been the impact of flexible automation on product, plant and firm scale and scope? Has flexible automation reduced plant and firm optimal scale and increased economies of scope? Or, conversely, has the adoption of flexible automation enticed the production of larger optimal volumes of plant and firm output? What factors account for the changes? 3. What has been the impact of flexible automation on the location of industrial production in developing countries? The first question is contextual and will be addressed at two levels. First, it will be addressed at the level of developing countries in general. On the basis of secondary data gathered from a variety of official and private sources we will attempt to depict the extent of diffusion of FA in developing countries. Second, it will be examined in the context of each of the six countries under study, particularly regarding the issue of the determinants of the diffusion of FA. The second question addresses the main subject matter of this research and will be examined with the help of country studies data based on firm interviews. Detailed interviews were conducted in selected firms to obtain data on their experience in using old and new technologies, as well as on output level, product variety and costs. Qualitative accounts were obtained which constitute a significant part of our data set, but great effort was expended in obtaining quantitative data. As has been pointed out, much of the discussion in this field is based on impressions built on limited qualitative assertions, simple observations or journalistic evidence which, apart from being academically questionable, does not give any sense of the magnitude of the issues under scrutiny. Identifying indicators for the concepts under analysis forced all those involved to be precise about intended meanings and the data to be collected. Seeking detailed quantitative data in the context of open-ended interviews also allowed for an immediate check on the consistency of what was being said by the interviewee. In some firms where it was not feasible to obtain the quantitative data required, it was possible to resort to qualitative discussions. The third question looks into the implications FA would have on the location of production in developing countries. Of particular interest were the impact which FA, working through scale and scope, has had on changes in industrial organization both internationally and domestically and the implications for entry

6 INTRODUCTION

by developing-country firms. However, because FA may have a bearing also on location through its impact on production and costs conditions, factor prices and biases and infrastructure requirements, we will address these issues too. Lack of relevant skills or transport systems could, for instance, act as a powerful deterrent to the successful use of FA in developing countries. For the purposes of this research we will take a rather broad approach to the definition of flexible automation. Initially, the emphasis was mainly on microelectronics-based automation and design hardware and software. After some preliminary interviews and discussions with colleagues and consultants for the research it became apparent that the diffusion of some of the new technologies in the countries under study had been closely, even ‘inseparably’, associated with the diffusion of a number of ‘organizational techniques’ which pertain to the way human and material resources, including the new hardware, are organized. The ‘complementarity’ between the ‘hard’ and ‘soft’ sides of technology was something emphasized by the literature. It was necessary, therefore, to define FA in such a way as to take this fact into account. Thus, for the purposes of this book, FA should be understood as the combination of the new microelectronicsbased forms of automation and associated organizational techniques. Indeed, the research should help to elucidate the nature of the relationship between the new ‘hardware’ and the new ‘organizational techniques’. Organization of this book The book is divided into two parts. Part I focuses on the conceptual and methodological foundations of the research and presents the main aggregate results. Chapter 1 briefly examines the evolution of scale and scope concepts since their inception in economic theory and develops a graphic conceptual framework of the issues at stake. Chapter 2 discusses the methodological approach taken and presents the detailed firm case study which was used as a model for fieldwork in developing countries. Chapter 3 addresses the questions of the diffusion and the motives for the adoption of FA in developing countries by examining world FA production and trade data and on the basis of the consolidation of diffusion data produced in individual country studies. Chapter 4 summarizes the main findings on the impact of FA on scale and scope at three levels—product, plant and firm —also on the basis of consolidated data from individual country studies. It then analyses the main reasons for such impact. Chapter 5 draws implications of the growing diffusion of FA for location of production in developing countries, and Chapter 6 presents the main overall conclusions and policy recommendations. Part II of the book presents individual country studies: Brazil (Chapter 7); Mexico (Chapter 8); Venezuela (Chapter 9); India (Chapter 10); Turkey (Chapter 11); and Thailand (Chapter 12). Most studies are similarly structured, beginning with a brief description of macroeconomic and industrial trends and policies, then moving on to discuss developments in the engineering industry including, where relevant, the manufacture of FA equipment. Country studies then analyse the diffusion of FA in sampled firms, covering the amount and

INTRODUCTION 7

types of equipment, the extent of operations being covered by FA and the process of diffusion. The analysis of diffusion is followed by a discussion of the impact of FA on product scale and scope, and on plant and firm scale. Total unit costs as well as capital, labour, inputs, R&D and marketing unit costs are then examined.

Part I CONCEPTS, METHOD AND SYNTHESIS

1 SCALE AND SCOPE Concepts and issues

1 Introduction The purpose of this chapter is to explain the conceptual framework developed in this research and used to address the impact of flexible automation on scale and scope. The concepts of scale and economies of scale and of scope and economies of scope have all too often been used loosely in discussions about the impact of technical change on costs, firm size, industrial organization and industrialization. Much of the confusion has arisen from the fact that the circumstances and relationships these concepts attempt to describe involve a combination of physical, organizational and economic phenomena, which normally are not easy to disentangle. Thus, the first step in this research was to develop a proper understanding of their meaning, of the key factors accounting for their changes, of their different dimensions and of the possible relationships that emerge between them on the basis of established economic theory; and to relate this understanding to the ensuing debate on whether or not new technologies reduce optimal scale and increase economies of scope. It must be emphasized that standard economic theory, despite its limitations, clearly remains the basis on which empirical research on issues of economies of scale and scope is conducted because it continues to provide a useful and recognizable heuristic instrument to organize the ideas and the data in this field. It also facilitates understanding and comparability across research. It is, therefore, an obligatory starting-point for any research of this kind. In outlining the development of the conceptual framework the section 2 briefly discusses the origins and development of the theory of economies of scale by examining the contributions of Adam Smith, Marshall and Viner. The main focus is on the intra-firm division of labour, not the inter-firm division of labour, as our concerns extend only to the firm level. Section 3 defines the concepts of scale, economies of scale and optimal scale, and reviews their main sources and dimensions. The sources of economies of scale to be addressed are specialization, indivisibilities and dimensional effects, while the dimensions of economies of scale to be discussed are those of product, plant and firm. Section 4 concentrates on the origins, definition and sources of economies of scope. It begins by

10 CONCEPTS, METHOD AND SYNTHESIS

addressing Stigler’s contribution, in particular the concepts of plant adaptability and flexibility in the short run, and then moves on to discuss the development of economies of scope theory by Baumol and associates. Section 5 reviews the debate between, on the one side, those who argue that new technologies will result in reductions in optimal scale and economies of scale and increased economies of scope and, on the other, those who contend that the effect will be the opposite. The final section of the chapter develops a graphical representation of both sides of the argument by way of clarifying the debate and as a means of organizing subsequent work. 2 Origins of economies of scale theory: from ‘division of labour’ to cost curves There are few propositions in economics that have caused so much debate and controversy as Adam Smith’s dictum ‘the division of labour is limited by the extent of the market’ and the pin factory example he used to illustrate his statement. This assertion is seen in both classical political economy and neoclassical economics as one the main foundations of production and economies of scale theory. In essence, Smith (1776) argued that the principle of the division of labour and the ensuing work specialization was the most important determinant of productive efficiency because of the increase in dexterity which results from making a single, relatively simple, operation the sole activity of the worker; the saving of time arising from not having to move from one type of task to another; and the associated use of machines which ease and abridge labour. In Smith’s celebrated pin factory example, one man draws out the wire, another straightens it, a third one cuts it, a fourth one points it and so on. Up to 18 distinct operations were identified which with no more than 10 labourers could result in thousands of pins produced in one day. A craft workman producing pins by himself would not be able to produce more than twenty pins in the same time. The main outcome of the division of labour was that it reduced direct and indirect labour per unit of output (see also Atallah 1966; Corsi 1991; Morroni 1992; Skinner 1974; Stigler 1951; Tayler 1985).1 The division of labour was, for Smith, a process resulting not so much from any human wisdom but from the gradual increase in the ‘propensity to exchange’. Because expansion of trade and output is at the origin of the division of labour, the extent of that division is limited by the extent of the market. Where the market is small there is no incentive to specialize as few will be able to purchase the output of those engaged in factory production. Where possibilities of economically transporting products elsewhere exist, the extent of the market is potentially increased by the ‘inhabitants of the lands’ within the reach of those transporting means. For any one country taken in isolation, the extent of its market is limited by its internal transporting facilities and grows ‘in proportion to the richess and populousness of that country’ (Smith 1776:124).2

SCALE AND SCOPE: CONCEPTS AND ISSUES 11

Classical political economy built on these ideas (Atallah 1966; Corsi 1991; Gold 1981; Morroni 1992; Tayler 1985; Vassilakis 1987). Babbage (1832) added that increasing specialization of labour reduces apprenticeship time and wastage of material. It also helps to separate in a clear way physical and mental labour and, thus, to allocate each job to workers with the appropriate skills and qualifications, resulting in skilled and creative individuals concentrating on activities that require judgement and on the development of new machines and products, and less skilled individuals dedicating to more narrowly defined tasks. Marshall (1890) brought the link to mechanization more explicitly into the discussion by pointing out that increasing division of labour creates the conditions for the identification and replacement of certain tasks by specialized production equipment. Work that is uniform and monotonous can be gradually taken over by machinery until there is nothing to do by hand except to supply machines with inputs and take away the product when finished. Even the function of overseeing the work of machinery can be replaced by devices which stop movement when anything goes wrong. The main effects of mechanization are to lower the cost of work while making it more accurate because as the division of labour progresses work is continuously subdivided and made highly specialized; and to make possible the manufacture of interchangeable parts. Marshall also pointed out that ‘internal economies’ arise within any manufacturing firm out of the increasing scale of output which is both the cause and the result of further work and skill specialization, increasing mechanization and improved production organization. Internal economies enable firms to increase their output in more than the proportionate increase of all of their inputs (increasing returns to scale).3 Despite Marshall’s major influence on neo-classical economics and his key contribution to what was emerging as an economies of scale theory, it was Viner who formulated the neo-classical theory of economies of scale as it is known today (Tayler 1985). In his seminal article, published originally in 1931, he developed the first graphical representation of cost theory which included ‘the usual assumptions of atomistic competition and rational economic behaviour on the part of the producers’ (Viner 1953:198). In it he distinguished between shortand long-run cost theory. In the short-run, plant capacity and some production factors such as capital are ‘fixed’, in the sense that they cannot be altered, while others such as raw materials and labour are ‘variable’. Costs associated with each type of factor are, respectively, ‘fixed’ and ‘variable’ costs. Because by definition ‘fixed’ costs cannot be altered, average fixed costs tend to decline as output increases. Since any increase in output is the result of the combination of constantly priced ‘fixed’ and ‘variable’ factors and given diminishing returns for the ‘variable’ factors, average costs will fall initially while the productivity of the ‘variable’ factors rise but will subsequently increase as the productivity of ‘variable’ factors declines. The outcome is a U-shaped total average cost curve. In the long-run all factors become variable and plant capacities were assumed, on equilibrium grounds, to be available to the exact level of minimum average cost output required by any individual producer. Any producer then has the choice of producing at the short-run level of output with the risk of getting it wrong, thus

12 CONCEPTS, METHOD AND SYNTHESIS

increasing average costs; or of investing in plant capacity believed to match expected demand at any point in the future (hence, the name ‘long-run’ or ‘planning’ cost curve). Because there are limitless options between the short and the long run, and given diminishing returns also in the long run, it is possible to have a U-shaped long-run average cost curve joining all the short- and long-run output and capacity trade-offs—the ‘envelope’ curve.4 Viner added that economies of large-scale production were long-run phenomena that only arise out of the adjusting of plant capacity to each successive output. Sources of economies were either technical or pecuniary, emerging out of reductions in the technical coefficients of production or from prices paid for the factors of production. Technical economies arise out of savings in labour, material or equipment due to improved organization or methods of production. Pecuniary economies arise from the possibility of obtaining quantity discounts due to a larger scale of purchases. 3 Scale, economies of scale and optimal scale: definitions, sources and dimensions Since Viner, textbook scale economies theory has evolved to become a more empirical set of concepts and relationships. It has been influenced also by developments in applied economics and engineering. In textbook production theory scale refers to size of output or capacity of production units, and economies of scale refers to reductions in unit costs due to increases in size of output. Economies of scale are said to exist if total cost rises proportionately less than does output, while diseconomies of scale arise when total cost rises proportionately more than does output; and optimal scale occurs at the point where any increase in output no longer reduces but raises unit costs. The main sources of scale economies are specialization, indivisibilities and dimensional effects (Morroni 1992; Rosseger 1986; Scherer and Ross 1990). Specialization sources can be further subdivided into static and dynamic sources. Static specialization gains arise out of larger output and had already been identified by Adam Smith in his discussions of division of labour and work specialization. In a nutshell, increasing scale allows the separation of tasks and workers to do their individual jobs rapidly and precisely and with the appropriate skill content, while avoiding the expense of time and effort associated with moving from one task to another. They allow also the use of more efficient purpose-specific machinery and mechanized production processes and an improved production organization. A too-large output, however, may be complex to plan, coordinate and control, leading to management and organizational problems and to losses in efficiency rather than gains. Dynamic specialization gains arise out of the learning potential of long production runs or the cumulative volume of output through time. Where intricate operations and complex process adjustments are involved, efficiency increases as workers learn by doing: the production process is ‘demystified’, for workers and management alike, enabling

SCALE AND SCOPE: CONCEPTS AND ISSUES 13

them to correct for mistakes. Hence, unit costs are reduced as a result of accumulating output. Any commodity is indivisible if there is a minimum size below which it is unavailable. Morroni (1992) identifies two kinds of indivisibility: economic and technical. Economic indivisibilities arise out of the fact that many commodities can only be traded in a given unit, e.g. (a length of) cloth or (a bushel of) corn. Technical indivisibilities arise out of the physical impossibility of dividing a particular commodity into amounts usable for production and consumption. This is normally the case with capital equipment, the capacities of which vary in discrete quantities and have a fixed minimum. In a vertical sequence of production stages, indivisibilities may arise out of the imbalances that emerge due to the different capacities of capital equipment at each stage as, if all machines are going to be fully utilized, it is the machine with the largest minimum capacity that determines the size of the others. Morroni (1992) also points to labour-related indivisibilities. Wherever there is the need for team work, as in the case of joint lifting of weight, labour cannot be decomposed into its constituents, as it would not be possible to carry out that particular activity individually. Specific skills are available only in certain individuals or groups of them, and often in discrete and minimum quantities. Capital equipment and labour indivisibilities create minimum outlays that have to be incurred by any producer, i.e. ‘fixed costs’. Unit costs fall as any unit of production initially required to produce a smaller output increases its output without a proportional increase in costs.5 Sources of dimensional effects relate to the geometrical volume-surface area relationship of certain kinds of capital equipment such as vessels, containers and pipelines. The cost of construction of any container increases in line with its surface area size, whereas its capacity increases with volume. Since the area of a sphere or cylinder varies as the two-thirds power of volume, the cost of building some equipment rises roughly as the two-thirds power of their capacity. Therefore, the higher the capacity the lower the unit costs per unit of capacity. This source of scale economies is commonly known in engineering as ‘the 0.6 rule’.6 According to Scherer et al. (1975), Scherer and Ross (1990) and Silberston (1972), scale and economies of scale are better analysed in terms of three dimensions: product, plant or firm. Product scale refers to the volume of any single batch, sometimes referred to as lot size or production run. Product-scale economies emerge from the indivisibility and fixed costs of the operations of preparation of equipment, exchange of jigs and fixtures, machine adjustment and trial runs necessary to begin manufacturing a particular batch or product run (Carlsson 1989a; Kaplinsky 1991; Pratten 1975, 1991a; Scherer et al. 1975; Silberston 1972; Steudel and Desruelle 1992). Batch or lot size is the quantity of identical items or products manufactured in a certain process or sequence of operations between set-ups. Set-ups or setting-up time is the time spent between the production of the last unit of the last batch and the first good item or product of the new batch.

14 CONCEPTS, METHOD AND SYNTHESIS

Setting-up times and associated costs are key factors in determining whether and when a new product is manufactured, particularly in discrete product industries (Ayres 1991; Morroni 1991). The classical example is Ford’s replacement of the model-T car by the model-A car, which necessitated closing down the factory for nine months in 1926 (Abernathy 1978).7 The car industry has always been under immense pressure not to change car models, which explains why some models remain in the market for years. In the printing industry, the initial typesetting costs can be so high that they sometimes make the publication of specialized or academic books or journals economically unfeasible. Very few books have print runs to match those of the Bible or Porter’s Competitive Advantage. In many operations in the metalworking sector the ‘typical’ set-up time is 20–30 per cent of processing time, while in a US textile factory setting up a roller printing press to print cotton or synthetic fibres takes around 20 per cent of production time (Carlsson 1989a; Scherer et al. 1975). In general, when production is not on a strict to-order basis, specific decisions on how large a batch size or how long a production run should be are taken with the help of the economic batch- or lot-size model (Mansfield 1988; Steudel and Desruelle 1992; Scherer et al. 1975). In essence, the model seeks to determine the size of the ‘optimal’ batch or lot on the basis of expected demand and set-up and inventory carrying costs, as inventories necessarily build up when large batches are produced:

where Q=the optimal batch size being sought; D=the level of expected demand (normally annual); s=the cost per set-up; c=the cost of inventory carrying. What the economic batch- or lot-size model says is that the larger the set-up costs relative to inventory carrying costs, the greater the incentive to continue producing the same item or product if unit costs are going to be kept in check. The lower the set-up costs relative to the inventory carrying costs, the lower the required batch size. As set-up costs approach zero, there is no optimal batch size, and individual batch size can be directly determined as a function of customers’ demand in order to keep low the inventory levels and costs. Plant scale, in turn, is normally associated with the total output of an entire plant in continuous process or ‘fluid-flow’ industries, such as oil refining, chemicals, steel and cement, in a given period of time—normally one year. It is in these industries where dimensional effects and the ‘0.6 rule’ have the largest impact on economies of scale. But plant scale relates also to the total output or capacity of discrete-good industries, such as printing, mechanical engineering, the electronic, clothing and shoe-making industries. Because discrete-good production involves a wide range of possible combinations of technologies, forms of production organization, and inputs and outputs, the key sources of economies

SCALE AND SCOPE: CONCEPTS AND ISSUES 15

of scale are specialization of labour and machinery, improved production organization, and individual equipment and production process indivisibilities. Both in continuous process and in discrete-product industries, plant economies of scale may arise also out of minimum requirements or indivisibilities of certain functions and needs, such as local management, security, safety, maintenance and the availability of reserve equipment and spare parts, particularly in cases where functions cannot be subcontracted.8 Firm scale relates to the output or capacity of the whole firm, which may or may not involve several plants, and firm economies of scale emerge from the indivisible and fixed nature of certain ‘intangible’ investments, such as research and development (R&D), marketing and management. Budgets for the development of new products and processes are normally ‘rule of thumb’ amounts, reached on the basis of previous years’ sales, levels of retained profits, the expenditure of potential competitors, minimum threshold considerations, the average allocation of the industry and the emerging technological opportunities (Freeman 1982; Hay and Morris 1991). Marketing and distribution expenditures also are set amounts resulting from similar factors. For advertising to be effective there are certain minimum threshold levels for messages that have to be transmitted. Consumer durables and office automation industries require specialized dealer networks and after-sales service. Operating a sales force requires investment in training and specialized equipment (Scherer and Ross 1990). Management costs are also pre-set and depend on a minimum number of functions and hierarchical levels—and therefore managers—and of specialized equipment that are required for the normal operation of any firm (Koutsoyiannis 1980). Scale gains arise, therefore, from the possibility of spreading all these fixed costs among a larger total volume while reducing unit costs. 4 Scope and economies of scope: origins, definitions and sources One of the major assumptions underlying cost curves and economies of scale theory is that plants and firms are single-product producers. Reality seems, however, to be extensively populated by production facilities and firms manufacturing many different products, i.e. multi-product plants and firms. Producers of garments, textiles, consumer electronic goods, home appliances, engines and machine tools, have constantly to switch models or manufacture according to varying technical specifications to cater for differentiated demand and, hence, have to adapt or modify their production processes.9 Initial preoccupations within economics with the issue of production flexibility and costs can be traced back to Stigler (1939). He set out to address the issue of production and costs in the short run which, in turn, allowed for the conceptualization of multi-product plants and firms. Consistent with received neo-classical theory, Stigler accepted that in the short run some production factors were ‘fixed’ while others were ‘variable’. However, he pointed out that ‘fixed factors’, in addition to being ‘divisible’ or ‘indivisible’, could also be

16 CONCEPTS, METHOD AND SYNTHESIS

‘adaptable’ or ‘not adaptable’ to changing quantities of the variable factor. A plant, machine or piece of equipment is adaptable if it does not lose ‘acceptable’ efficiency when operated by a varying number of labourers or with a changing amount of material. Adaptable plants imply that output may be reduced without resulting in unemployment of the fixed factor. Stigler pointed out that illustrations of perfect adaptability were difficult to find in reality but that one could envisage some kind of agricultural land which could be combined with varying amounts of ploughing, seeds, etc, within fairly wide limits. Adaptability arises from the capacity to reorganize production processes. According to Stigler, there are four possible combinations of divisibility and adaptability for operations that are outside the optimal point of production. The first two possibilities that arise in the utilization of a fixed plant concern the cases of perfect divisibility or complete indivisibility with total adaptability. In these cases, because of diminishing returns of the variable factor, the resulting short-run unit-cost curve would be smoothly U-shaped, much in the neo-classical vein. The only difference between the two cases is that, in the latter, the cost function would not be continuous over the whole range of output. The third possibility is a combination of an unadaptable but divisible plant. This is the ‘unrealistic’ case of, for instance, a plant with several identical machines which can be used only with a fixed amount of labour and of materials. In this case, shortrun unit costs would be a horizontal line until the plant is fully employed, and it would not be possible to increase output beyond this point. Finally, in the case of an indivisible and completely unadaptable plant the short-run unit-cost curve would be a vertical line, and it would be possible to produce only at a single level of output. In between the last two possibilities there are a number of combinations which arise out of partial adaptability and, hence, short-run cost curves with either a small or a large output for which unit costs are significantly decreasing or increasing. Stigler adds that it may be possible to introduce some flexibility into the operation of a plant so that it can be acceptably efficient over a range of probable outputs. Production ‘flexibility’ is defined as the capacity to produce different amounts of a given product with the best possible technology, but at the cost of being unable to use the best-known technology for any amount of output. It always requires some adaptability, in the sense that the greater the adaptability of inputs the easier it is for a plant to be flexible, but it goes beyond adaptability as it involves the capacity to modify an existing plant and significantly switch the level of output with little loss of efficiency. Sources of flexibility are, first, the degree of divisibility of fixed plant which reduces the costs of producing at suboptimal levels; and second, the capacity to transform fixed into variable costs. Morroni (1992), who builds heavily on Stigler’s contribution, adds that the main sources of flexibility include the capacity to change the number of workers or the number of working hours per worker (numerical flexibility), the ability to adjust workers’ skills within the firm (functional flexibility), the potential to hire or lease equipment or keep old equipment for catering for short term upsurges in demand, the possibility of keeping inventories for compensating for variations in demand, and the capacity to employ subcontractors.

SCALE AND SCOPE: CONCEPTS AND ISSUES 17

Figure 1 Short-run unit costs in flexible and inflexible plants

Curves UCI and UCF in Figure 1 are the unit-cost curves of an inflexibleindivisible and a flexible-divisible plant. When output is OA and OD the flexible plant has smaller unit costs than does the inflexible plant, with the latter producing at prohibitively high unit costs. Between outputs OB and OC the inflexible plant has lower unit costs. If output is expected to fluctuate between OB and OC then the inflexible plant is the more desirable, but if output is going to fluctuate beyond these points the flexible plant may become the better option. Hence, the unit-cost curve of a flexible plant is flatter but with a minimum cost higher than that of the unit-cost curve of the less flexible plant.10 Although Stigler remained very much in the neo-classical tradition, and did not address the issue of firms manufacturing more than one product, there is no doubt that his contributions paved the way for further developments in this direction.11 First, he introduced issues of production adaptability and flexibility, hitherto unexplored, into the economics’ discussion. It was just a step away conceptually, although it took some time, to make the link with multi-product output concepts after ideas pertaining to production process adaptability and flexibility had been established. Second, he raised the possibility that firms, certainly in the shortrun, had a far wider choice of relatively efficient points in which to operate than what received neo-classical theory was prepared to accept, and that the selection of any level of efficiency involved significant trade-offs in terms of the capacity to respond to changes in demand. One important implication of this contribution is that there may not be a short-run optimal point of production but only a set of relatively, more or less, efficient options depending on entrepreneurial expectations of the market.

18 CONCEPTS, METHOD AND SYNTHESIS

Producing more than one good implies considering not only the degree of production adaptability and flexibility, as Stigler contended, or the setting-up, change-over and investment costs, as traditional economies of scale theory showed, but looking also into the potential cost effects of joint-production. Baumol et al. (1988; see also Bailey and Friedlaender 1982) have addressed this issue and have developed the concepts of scope and economies of scope, which they consider complementary to the traditional concepts of scale and economies of scale. Scope stands for product range; economies of scope arise when the cost of making goods jointly is less than that of making the same total quantity of the same goods separately. They can be defined formally in the case of two outputs, q1 and q2, as follows: Economies of scope arise from the sharing or joint-utilization of inputs (Baumol et al. 1988; Bailey and Friedlaender 1982). They are the result of using inputs that may have some properties of a public good, so that when purchased for one production process they may be freely available for another. More often, however, they are the outcome of using a given factor or input which is imperfectly divisible, so that the production of a specific product, or number of them, may lead to under-utilization of that factor or input, and which can be shared in the production of another good. Or, in slightly different terms, such economies often arise when a multi-product set of production techniques employing a common input exhibits ‘strong’ cost complementarities, in the sense that the marginal cost of producing any one product decreases with increases in the quantities of all other products. Baumol et al. (1988) and Bailey and Friedlaender (1982) argue that, historically, technical change has been a crucial factor in creating the conditions for the emergence of economies of scope. Until refrigeration and fast transportation developed, the joint exploitation of wool and mutton, for instance, was not possible. More recently, it was only after the development of technologies that allow the switching of tasks and varying the order in which parts are transferred it has become possible to reap economies of scope. Baumol et al. (1988) and Bailey and Friedlaender (1982) point to several illustrative cases where plant economies of scope in the production of goods and services may be at work. The first one is the case of an input used in different stages of the production process. It is generally cheaper to produce both wool and mutton than it is to raise sheep separately for each product. The second refers to the indivisibility of certain factors of production. In the light of a falling demand for cars, the use of a stamping machine for the production of light trucks as well as for cars may provide economies of scope. A third example comes from the use of an input by more than one product, like the case of a firm that produces abstracts from journals. The firm produces three separate indexes of abstracts, comprising a general index which contains all the information, and two specialized indexes on the basis of the general one. Scope economies arise because the general index is freely available for the other two indexes. The final

SCALE AND SCOPE: CONCEPTS AND ISSUES 19

Figure 2 Output space for a two-product firm

case comes from the economies of networking. Routeing different flights using larger planes through the same airport and then combining passengers with the same destination in other large aircraft can lead to cost reductions in the overall journey. Teece (1980) and Carlton and Perloff (1994) added that generally, at firm level, economies of scope emanate from the nature of knowledge as a public good, from exchanging and pooling the information and know-how available from several projects being undertaken simultaneously, and from using marketing, distribution and management facilities for more than just a single product. In order to relate economies of scale and scope concepts, Baumol et al. (1988; see also Bailey and Friedlaender 1982 for a clear and succinct explanation of the new concepts) develop a number of additional cost concepts in the context of the multi-product firm. To begin with, they point out that in the multi-product context there is no definition of average or unit cost since there is no meaningful way to aggregate two different products into a single output measure. Furthermore, since the composition of output affects costs, care must be taken to differentiate between scale and scope effects. In the multi-product context the concept of scale economies can be made meaningful with the help of two additional concepts: ray economies of scale and product-specific economies of scale. Ray economies of scale indicate the behaviour of costs as the total output of a fixed combination of goods rises. Since for any output level proportions do not change, it follows that average costs, or more precisely ray average costs (RAC), are the total costs of a new ‘combined’ or ‘composite’ commodity divided by output. In this way the problem of adding together oranges and apples is avoided. In the output space for commodities Y1

20 CONCEPTS, METHOD AND SYNTHESIS

and Y2 shown in Figure 2, any output vector moving along a ray through the origin, such as OC, represents the RAC of the ‘composite’ commodity Y1Y2. Because RAC describes only the cost behaviour when output expands or contracts along a given ray, it is necessary also to address cost changes as output proportions change. To look into this issue the concept of product-specific economies of scale was developed. Product-specific economies of scale refers to cost changes as the output of one commodity changes. In Figure 2, these cost changes are reflected in lines FB or EB, where output of the other commodity is held constant in either case. It is measured by estimating first the average or unitincremental cost (AIC), which is defined as the addition to total costs associated with producing a given product at a specific level of output, as compared with not producing it at all, divided by the output of that product. For product Y1:

Product-specific returns to scale (S) for product Y1 are then given by:

where: MC=marginal cost. If S1 is greater than, equal to or lower than 1, there are, respectively, increasing, constant or decreasing returns to scale. To continue addressing the effects on costs of changes in the composition of output, Baumol et al. turn their attention to the impact of joint-production and bring into the analysis the concept of economies of scope. As was said, economies of scope are present when firms with diversified product mixes have lower total costs than the total costs of firms independently producing the same mix for the same level of total output. In Figure 2 point B represents an economies of scope situation as C(B)
If Sc is greater than zero there are economies of scope; and if it is lower than zero there are diseconomies of scope. Baumol et al. (1988), like Bailey and Friedlaender (1982), then focus on the analysis of the impact on costs of variations in product-specific scale economies and in economies of scope, which together lead to multi-product scale economies or total plant or firm scale economies. In general, in a two-good environment Y1 and Y2, the extent of multi-product scale economies is equal to total costs divided by the weighted sum of marginal costs, with the weights representing the different levels of output:

SCALE AND SCOPE: CONCEPTS AND ISSUES 21

or, after some algebraic manipulation:

where

The authors add that the multi-product or plant degree of scale economies for products Y1 and Y2 is equal also to the weighted average of the degrees of scale economies pertinent to each product, magnified by the economies of scope factor 1/(1–Sc). This implies that strong economies of scope could result in multi-product or plant economies of scale even if there are no economies of scale in some or all of the individual products. Baumol et al. (1988) and Bailey and Friedlaender (1982) point out also that, if the effect of scope economies is greater than the effect of any product-specific scale economies, the cost function is said to be transray convex. This is the case with line AD in Figure 2. Should the effect be the opposite, i.e. product-specific scale economies that are larger than economies of scope, the cost function is said to be transray concave. 5 Flexible automation and economies of scale and scope: the ‘modern technology’ literature view and its critics During the 1980s and early 1990s a substantial body of literature emerged, sometimes associated with terms such as ‘flexible specialization’, ‘postfordism’, ‘lean production’ and ‘new competition’, claiming that new technologies, particularly microelectronics-based forms of automation and design and associated organizational modifications, are leading to fundamental changes in economies of scale. This literature includes works by authors in economics (Acs and Audretsch 1990a, 1990b, 1991a, 1991b; Acs et al. 1990; Auty 1992; Best 1990; Carlsson 1989a, 1989b; Dosi 1988; Hoffman 1989a; Kaplinsky 1984, 1990a, 1991; Milgrom and Roberts 1990; Morroni 1991; Piore and Sable 1984; Rosenberg 1988); engineering (Benson 1989; Benson and Ponton 1991; Biemans and Vissers 1991; Bolwijn et al. 1986; Talaysum et al. 1987); and management (Ayres 1991; Ayres and Miller 1983; Bessant 1991; De Meyer et al. 1989; Gilder 1988; Jaikumar 1986; Womack et al. 1990). Although there are some differences of viewpoint between individual authors, the underlying argument of the modern technology literature is that the idea that increasing division of labour and machine specialization, and the resulting larger scales of output, lead to higher efficiency and lower unit costs has been a guiding

22 CONCEPTS, METHOD AND SYNTHESIS

principle of the manufacturing industry ever since the Industrial Revolution. A number of product and process innovations during the eighteenth and nineteenth centuries, including the manufacture of flintlocks with completely interchangeable parts, the production of standardized nuts and bolts, the development of precision instruments and machine tools, and the establishment of production lines for the manufacture of pulleys used in sailing ships, gradually paved the way for the continuous application of division of labour principles into manufacturing, which was eventually called ‘mass production’ by historians, economists, engineers and managers.12 Yet it was only at the beginning of the twentieth century that ‘mass production’ began to be applied by major manufacturers. First came Taylor’s systematic efforts to study in great detail the simplest tasks, timing them and designing methods and tools that permitted workers to increase productivity significantly with lower physical effort, which were later complemented by concepts and principles about the organization and supervision of work developed by Fayol. Pioneering its application was the Ford Motor Company, which around 1910 started reorganizing production along ‘taylorist’ lines and introduced, a few years later, a moving-belt conveyor in its fly-wheel magnetos plant. Production lines were carefully designed so that there was an optimal division of labour within and between workstations. Within individual stations, operations were decomposed into specialized tasks comprising simple repetitive motion patterns with minimal human intervention in handling and positioning the piece being worked on. This allowed workers to learn easily and perform actions rapidly, without the need of motion or mental adjustment or the replacement of some of their functions by purpose-built machines. Between stations, there was careful synchronization of timing of simultaneous operations and of the speed of transfer between stations so that bottlenecks would not arise. Thereafter similar approaches were applied in Ford’s car body and engine assembly operations, replacing labour, wherever possible, by specialized high-speed production machine tools. Ford’s ‘moving assembly line’ concept was embraced by the whole of the automobile industry, leading to a number of process innovations, including ‘transfer lines’ which coupled machine tools with rudimentary automatic transfer mechanisms and were capable of producing specific engine parts, such as cylinder blocks, practically without human intervention (Abernathy 1978; McNeil 1990). The success of Ford’s efforts in significantly reducing production time and increasing productivity led to the adoption of these ‘mass production’ methods by most car manufacturers and component suppliers, initially in the US but later worldwide. Outside the automobile industry ‘mass production’, increasingly also called ‘fordism’, spread to other discrete-product industries and to continuous process industries which, likewise, saw the cost-reducing potential of expanding the division of labour, using specialized machinery and producing higher output. Many large-scale capital-intensive factories and firms emerged in such industries as textiles, footwear, equipment, precision instruments, office equipment, petroleum, chemicals, steel and cement. It is said that throughout the first half of the twentieth century, ‘industry after industry came under the domination of

SCALE AND SCOPE: CONCEPTS AND ISSUES 23

giant firms using specialised equipment to turn out previously unimagined numbers of standard goods, at prices that local producers could not meet’ (Piore and Sabel 1984:20). Advances in technology since the 1970s have, according to the modern technology literature, progressively undermined the basic principle of ‘mass production’. Developments in microelectronics and information technologies have led not only to a series of innovations in consumer electronics, telecommunications and computers, but also, and perhaps more importantly, to the fabrication of new—and significant modifications and adaptations of old— tools, equipment and production measuring and sensing instruments. These new technologies or ‘flexible automation’ (FA) are radically different from previous vintages of machinery. FA includes programme logic controllers (PLCs) for industrial use; computer numerical control (CNC) machine tools such as lathes or machining centres, weaving and sewing machines, printing presses, automatic packaging systems, and surface mounting devices for assembly of printed circuit boards; industrial robots; computer aided design/manufacturing (CAD/CAM); production process control computers (PCCs) and necessary software; automated guided vehicles (AGV); automated storage and retrieval (AS/R); flexible manufacturing systems (FMS), combining CNC machine tools, process control computers and automated transport; and computer integrated manufacture (CIM), which integrates all design and manufacturing capabilities and other business operations as a single control and data system. FA technologies have a number of technical and economic properties or characteristics that were unavailable in the pre-microelectronic era. First, as their name indicates, they are flexible because they can rapidly switch response. This is the result of the technologies’ capacity to be programmed in different ways and therefore vary according to new circumstances and instructions (Carlsson 1989a; Morroni 1991). Second, they are capable of integrating different pieces of equipment (Bessant 1991). Microprocessors’ and computers’ capability to handle large amounts of information allows a far more precise and hierarchical, while at the same time adaptable, control of individual pieces of equipment and of the whole production process. Better control capabilities facilitate, in turn, the integration of more mechanical functions into individual machines, and of machines into more complex single production systems. Third, they are much faster and more precise in performing any task (Twiss 1981). Electronic sensing and measuring instruments combined with microprocessor control of power and transmission devices allow faster acceleration and deceleration. In many machines hard-wired circuitry is replaced by microchips, increasing operating speeds. Computers’ ability to deal with a considerable amount of information permits the more accurate definition of time, speed, co-ordinates and angles, and facilitates operation within finer tolerance margins. Fourth, they are physically much smaller (Bessant 1991). Because FA includes a significant proportion of microchips and printed circuit boards which are physically small, FA itself tends to be physically smaller and so occupies less space than that required by older technologies. Finally, FA technologies, individually or grouped, have a high level of complementarity. The introduction of, for example, CAD/CAM technology

24 CONCEPTS, METHOD AND SYNTHESIS

makes it cheaper to improve on products and to design and introduce new ones. However, for the equipment to be fully profitable, CNC machine tools which allow the more efficient production of a wider range of goods may be required, and this in turn may be worthwhile only if a process control computer is introduced (Milgrom and Roberts 1990). Complementarities do not limit themselves to FA ‘hardware’ but also to a number of the new ‘soft’ organizational techniques or organizational concepts which have emerged in recent years (Milgrom and Roberts 1990). These include just-in-time production (JIT) or inventory minimizing approaches; total quality management (TQM), i.e. a concern with quality at all stages of production; cellular manufacturing (CM), the cell organization of the factory; and working practices reorganization, which involves changes in job descriptions, worker responsibility and functionality. These new organizational techniques are said to increase the marginal return of FA by permitting the more effective use of the capabilities of the hardware. The combination of FA and associated organizational techniques is expected to have a major impact on economies of scale and scope, according to the modern technology literature. At product level it is argued that, unlike ‘mass production’ old technologies, FA’s ability to integrate diverse equipment and functions, and its programmability, help reduce minimum batch sizes by reducing setting-up times and costs. These ‘fixed’ costs include the costs of changing and adjusting machines with appropriate tools, resetting equipment and/ or re-adjusting transfer lines from a previous batch to a new one (Hoffman 1989a; Kaplinsky 1984, 1990a). Hence, lower setting-up costs are making it economically feasible to produce smaller batches. Concurrently, by allowing production facilities to vary their product range and to make full use of their equipment, FA leads to economies of scope (Bailey and Friedlaender 1982; Baumol et al. 1988; Morroni 1992). Evidence in support of this view is found in a number of studies in the metalworking, bicycle and automobile industries. Research by Carlsson (1989a) on the metalworking industry shows that factories in Japan have reduced set-up times from 1–1.5 hours to only a few minutes by using an electronic pallet system which minimizes machine downtime. Another study by Hoffman and Kaplinsky (1988) shows that between 1970 and 1980 set-up times in forging and casting in Toyota’s stamping division fell from 100–200 minutes to 10 minutes and from 60 minutes to 4 minutes, respectively. Average lot sizes also fell during the same period from 5,000 to 500. Finally, research by Mody et al. (1991b) on costs in developing countries’ bicycle industries, based on simulations under different technological alternatives, provides similar ‘evidence’. The simulations show that, after introducing ‘advanced manufacturing technologies’ and organizational improvements, a typical firm in the industry would not only see the costs of adding a new product variety increase much less than with ‘old mass production’ fixed automation, but would also face an overall unit-cost fall for each bicycle model. The figures show that the unit cost of producing around 70 models using advanced manufacturing technologies would be slightly lower than the unit cost of producing 5 models with old technologies.

SCALE AND SCOPE: CONCEPTS AND ISSUES 25

At plant level, according to the ‘modern technology’ literature, FA is substantially shrinking the size of machinery and plant while, at the same time, making it possible for most capital equipment to be available in a wider spectrum of capacities, which, together with falling semiconductor prices, is drastically cutting the cost of capital equipment and allowing the emergence of smaller efficient production facilities (Acs et al. 1990; Acs and Audretsch 1990a; Carlsson 1989b; Gilder 1988; Jaikumar 1986; Kaplinsky 1990a; Rosenberg 1988; Talaysum et al. 1987). Ayres (1991) and Ayres and Miller (1983) hold that there is potential for further capital cost reductions through the standardization of programmable machine tools and robots. It was also claimed that, particularly for continuous production, lower equipment costs could be possible from the application of economies of scale principles to the design and manufacture of a very large number of small plants or plant modules rather than having to build fewer larger plants. This would now be possible because of advances in plant design technology (Auty 1992; Benson 1989; Benson and Ponton 1991; Stevenson and Walker 1987). Hence, smaller, more divisible and less costly equipment and plant will result in optimal volumes of output which are lower than was possible with old technologies. In Gilder’s words: Rather than pushing decisions up through the hierarchy, microelectronics pulls them remorselessly down to the individual. This is the law of the microcosm. This is the secret of the new American challenge in the global economy. The new law of the microcosm has emerged, leaving Orwell, von Neumann, and even Charles Ferguson in its wake. With the microprocessor and related chip technologies,the computing industry has replaced its previous economies of scale with the new economies of microscale. (1988:57) Carlsson (1989b) and Acs and Audretsch (1990a) add that there is statistical evidence to support these claims. Using employment-based indicators of the sizes of plant, they show that in metalworking industries in countries such as the US, the UK, the former West Germany, Italy and Japan, the proportion of small firms in the total has increased while at the same time the average size of firms has decreased. As these industries have increasingly used CNC machine tools and robotics, they conclude that FA leads to ‘de-scaling’, i.e. to reductions in optimal scale. Evidence of a different sort is presented by Hoffman (1989a). He points to Nissan’s new engine plant in the UK, which was said would be producing 80,000 units efficiently as opposed to the previous optimal scale of 300,000 units. Critics have argued that there is no reason why FA should lead to reductions in plant scales rather than to ‘scaling-up’ (Alcorta 1992a, 1993a, 1994; Bureau of Industry Economics 1988a, 1988b; Ferguson 1988; Markowsky and Jubb 1989; Mody et al. 1992; Rhys 1992). It has been claimed that physical size of equipment and plant should not be confused with economic size. While it is true

26 CONCEPTS, METHOD AND SYNTHESIS

that semiconductors, computers and some electronic products are undergoing ‘miniaturization’ and that, by integrating, machines are occupying less space and therefore getting ‘smaller’, it does not follow that machines are also falling in capacity (Alcorta 1992a, 1994). FA may be faster, more efficient and reliable, and capable of operating for longer periods, thus expanding capacity. Furthermore, by reducing setting-up times and expanding variety, FA may allow an increase in total plant output due to higher capacity utilization, so that scale and scope economies at product level reinforce economies of scale at plant level (multi-product scale economies). Indeed, the study by Mody et al. (1991b) quoted earlier found that increasing variety from 5 to 70 models also meant an increase in total plant output, prompting the authors to conclude: What modern automated equipment does allow is the production of small lot sizes because the time to change machine settings for a different product becomes smaller with such equipment. However, to use the equipment efficiently, the annual throughput rate of the plant must increase, if only marginally. (Mody et al. 1991b:41) It does not follow either that because equipment is physically smaller it costs less. Despite falls in the cost of microchips and computers, and despite their increase in power, the cost of the production equipment in which they are used may still be higher than the cost of the technologies it replaces (Alcorta 1992a, 1994). Indeed, this could be due to the greater sophistication, complexity and integration of FA, even after some degree of standardization has taken place. Availability of equipment in smaller capacities would reduce plant scales only if equipment cost has fallen in line with capacity. Otherwise, it means only that smaller firms may have access to these technologies but at lower efficiency levels (Alcorta 1992a, 1994). In any case, given that ‘greater divisibility’ is restricted mainly to computers or electronic products and has had hardly any impact on production equipment, there is no a priori justification for expecting any significant ‘descaling’ trend in production. Finally, the capacity to design smaller continuous production plants has not been fully developed and is only a possibility for the future. The evidence presented by the modern technology literature is also suspect. On the one hand doubts have been cast on the degree of heterogeneity of products (Alcorta 1993a, 1994). While existing evidence points in the direction of a reduction of setting-up times and change-over costs, a fall in product scale and some increase in economies of scope, it was not clear from the research whether differences among ‘new’ products were substantial. It could well be the case that products with minor differences or ‘cosmetic changes’ vis-à-vis existing ones are now being labelled as ‘new’ products. A leading proponent of ‘de-scaling’ found that after reorganization and FA adoption at Lucas LFS, a leading British engineering manufacturer, ‘product diversity had increased’ as the number of varieties rose from 7 to 9 in one product, from 2 to 8 in another, while in a third

SCALE AND SCOPE: CONCEPTS AND ISSUES 27

product there were no varieties either before or after. In addition there were different mixes of batch size, so the factory depended much less on two particular varieties, which used to account for a large proportion of total sales (Kaplinsky, 1991). But the plant continued producing the same three products as before. On the other hand, questions have been raised as to the interpretation and soundness of the data on plant scale (Alcorta 1993a, 1995b; Harrison 1994). First, the introduction of labour-saving techniques and the ensuing increase in productivity could well result in smaller plants in terms of employees, but not necessarily in terms of optimal output. Plainly, capital intensity increases. Employment-based indicators are therefore poor indicators of ‘de-scaling’ as far as economic plant scale is concerned. Second, even if they were a good proxy for optimal scale, the evidence does not necessarily show ‘de-scaling’ in terms of plant-size reduction as a result of new technologies. It is not possible to tell from average or aggregate figures whether old large plants have been replaced or whether they coexist with new smaller plants that have been built for specialized markets. Third, the level of aggregation of the data, on which the evidence is based—two-digit industry— implies a wide variety of establishments which may have little in common with each other, and does not help to discern which types of firm are adopting what. Finally, the source of the data on the level of output by Nissan’s UK engine factory was apparently a newspaper article which probably referred to the starting level of output. In the event, the factory was manufacturing 200,000 engines and was planning an expansion to 400,000 engines by the mid-1990s (Womack et al. 1990). What little evidence there is would seem to point in the direction of FA having a ‘scaling-up’ impact on the level of plant output. Research by Edquist and Jacobsson (1988) found that, after retooling the complete crankshaft factory of Volvo Components in Sweden with FA equipment, including CNC lathes, robots, AGV and a central computer, there was an increase in production capacity of 33 per cent in comparison with the old machines. A study by Jacobsson (1989) on the production of CNC machine tools shows that the industry, driven today by electronics both in its production processes and in its product technology, requires very high volumes to remain in business. While earlier producers of numerically controlled lathes were required to manufacture around 100 units a year, today’s producers require approximately 2,000 units per year to be efficient. ‘Scaling-up’ is also taking place in process or fluid-flow industries. Interestingly, it is the ‘small-scale’ steel mini-mill sector, for a long-time used as the model of ‘de-scaling’, where this is occurring. During the 1980s, the sector underwent a period of intense industry-specific and microelectronics-based technical change, which included the introduction of more efficient melting techniques, ladle furnaces, continuous casters, computers and electronic control equipment. As a result, there has been a substantial reduction in material, energy, capital and labour unit costs, and a significant improvement in quality. Also, the sector is now capable of producing some tubular products and may, eventually, manufacture flat-rolled sheets. However, this has meant that the minimum economic scale of output for ‘mini-mills’ also has risen. In the 1990s it is expected that competitive mini-mills will increase their annual capacity fivefold

28 CONCEPTS, METHOD AND SYNTHESIS

or tenfold to around 500,000–1 million tonnes, with the larger mills producing up to 2 million tonnes (Mody et al. 1991a). The modern technology literature had contended also that at firm level FA will increase research and development (R&D) and marketing ‘fixed’ costs. R&D costs could rise and additional learning efforts could be required because of the need to adapt and modify product technology and develop the software to make better use of FA. While product adaptation has always been a central activity of the firm, the adoption of FA requires R&D efforts in some new areas. These efforts consist not only of designing new standard and interchangeable parts, but of completely redesigning all parts and products and most of the productive processes with the help of computers capable of modelling physical processes in such a way that production and assembly of goods are eased. In this way, the increasing number and complexity of components can be partially addressed. ‘Designing for assembly’ and ‘designing for producibility’, however, requires detailed written knowledge of the products and production processes and a capacity to write complex models and programs which, in many cases, are factory-specific, and thus ask for an in-house software writing team (Bolwijn et al. 1986; Goldhar and Schlie 1991a, 1991b; Stalk 1991; Whitney 1991). Specialized software development and technical support are also required where highly diverse combinations of technology have to be linked into a ‘network’ so that pieces of equipment can ‘talk’ intelligibly to each other. Even getting a relatively ‘standard’ FMS to operate correctly, for instance, involves up to two years’ preparation and planning and solving complicated integration and communication problems, tasks possible only in experienced engineering and R&D departments. Although attempts to standardize and make production and design software more ‘user-friendly’ are being made by both outside suppliers and specialist consultants, it is only those firms that can expand their specific capabilities in the areas of system integration and software design which will successfully adapt FA to their individual conditions (Bessant 1991; Haywood et al. 1991).13 Rising product development costs result also from increasing the manufacturing speed and flexibility that have reduced product life-cycles, thus requiring the design and development of more products more often to avoid underutilizing capacity. Rather than taking several months or years to develop, new products will have to emerge from R&D departments much faster than hitherto. Moreover, in a context of increasing flexibility, only firms that can continuously offer new products will enjoy a competitive edge (Stalk 1991). Data in support of the higher R&D expenditure view, and the resulting higher scale economies, are found in OECD (1991b:122) figures which show that between 1974 and 1984 expenditure on ‘intangible investment’, which includes R&D, advertising and software, increased in all countries from 2.6 per cent of GDP in 1974 to 3.7 per cent of GDP in 1984. This trend was present in all seven OECD countries which had comparable figures, with the US, Japan and the former West Germany showing the highest increases. More recent figures (1981– 93) show an increase also in real terms on R&D expenditure of 51.5 per cent for European Union countries, the US, Canada and Japan (European Commission

SCALE AND SCOPE: CONCEPTS AND ISSUES 29

1994). Above-average increases were recorded by Germany, Italy, Japan and Spain. Manufacturing R&D expenditure also increased over manufacturing output between 1980 and 1990 from 1.5 per cent to 2.0 per cent in Germany, France, Great Britain and Italy, from 1.2 per cent to 2.6 per cent in Japan and from 2.3 per cent to 3.2 per cent in the US (European Commission 1994). Investment in R&D by the corporate sector accounts for a significant share of this increase. Between 1967 and 1989 annual expenditure in R&D in the automobile industry doubled in real terms in the case of US firms, trebled in the case of European firms and increased six-fold in the case of Japanese firms (Womack et al 1990).14 Marketing costs also are said to rise because of the competitive advantages that arise from linking sales to production, the higher information requirements for selling and the larger variety of goods and resulting advertising expenditures to communicate the availability of new products (Belussi 1987, 1992). Under FA, establishing a network of marketing and distribution integrated with other functions of the company becomes an imperative from the very early stages of application. Speed of response in manufacturing and the capacity to introduce new products and anticipate the market are said to be crucially dependent today on market intelligence and continuous information, which in turn require purpose-built channels, infrastructure and specially trained personnel. This is particularly important if specific growing market segments or trends are to be identified. Customers are also getting more sophisticated and specific in their demands. Often, sales and distribution networks must extend beyond national borders and, as a result, the amount of information to be processed increases, normally requiring a major upgrading of the equipment necessary to handle and transfer data. In addition, insofar as advertising gives the customer information and communicates the availability of new products, more product variety involves providing details about more products. In the case of Benetton, information provided by the retailing system has been fundamental to the firm’s success in the fashion market (Belussi 1987, 1992). In 1985 Benetton had around 2,400 shops worldwide. Around 200 of these shops, located in four strategic markets, are the ‘antennae’ of Benetton’s information system. They provide data on a daily basis about each item sold, including model, colour, size, and so on. The data are then fed into production and design, enabling Benetton to respond to market changes in about ten days. To do so, however, the company has had to equip its 200 key shops with advanced cashregisters, telecommunications links with a central computer and the necessary software and hardware support, which involved a major investment effort. Benetton also had to advertise an increasing number of products and decided to publicize not individual products but specific characteristics, like colours, types of garment or brand to reduce escalating advertising costs. Despite this approach, advertising costs have been increasing continuously and exceed by far those of R&D, amounting 2.8 per cent of sales in the early 1990s as compared to 0.5 per cent in 1984, which was accounted for in part by the fact that national and international campaigns through the mass media had to be used.

30 CONCEPTS, METHOD AND SYNTHESIS

Summing up, there is agreement in the literature that FA’s impact on scale and scope at firm level, particularly because of increasing R&D and marketing costs, is resulting in ‘scaling-up’, as higher levels of output may be needed to amortize higher ‘fixed’ costs. Authors such as Kaplinsky (1990a) and Kline and Rosenberg (1986), however, point out that by becoming imitators, focusing only on minor innovations, or by networking and sharing some overheads (firm-level economies of scope), it may be possible to minimize the impact of these costs on firm scale.15 There is agreement in the literature also that FA is reducing product scale and increasing economies of scope, although there are some doubts as to the extent of the increase in scope. There is deep disagreement on the impact of FA at plant level, with one side in the argument pointing to reductions in economies of scale and optimal scale due to miniaturization and standardization of capital equipment and the other side arguing for increases in economies of scale and optimal scale due to higher machine sophistication, efficiency and cost. 6 Technical change, costs, scale and scope: a graphical representation of the issues Traditionally, the impact of technical change on production is discussed in terms of shifts in the production function. It may lead to more output being produced with the same resources or to the same output being produced with fewer resources. The impact of technical change on production can be approached also from the cost side. In this case, given a set of factor and/or input prices, technical change may allow the firm to produce a higher level of output at the same cost, or the same level of output at a lower cost. The existence of multi-product firms introduces the possibility of technical change leading also to lower levels of cost, due not only to savings in factors and/ or inputs costs in producing individual goods but also to producing goods jointly.16 Figure 3 shows the total cost surfaces of producing the same two goods, Y1 and Y2, prior to and after the introduction of a new technology. Both cost surfaces reflect the total cost of producing either Y1 or Y2, or a combination of them. If Y1 only or Y2 only is produced, then we have a conventional singleproduct cost curve on each horizontal axis. Cost surface C2EF is the result of using ‘old’ specialized equipment. Producing a combination of goods Y1 and Y2 implies a cost penalty due to resetting and changeover costs and input waste. Thus, there are ‘diseconomies’ of scope, as reflected by the shape of the crosssection connecting points E and F—i.e. the costs of producing Y1 and Y2 separately are lower than those of producing them jointly (transray concavity). Following the introduction of a ‘new’ cost-saving technology, a new lowercost surface C1CD emerges. Total costs have fallen due to a reduction of factors and/or inputs costs for each individual good, as exemplified by the shift downwards of each good’s cost curve. There is also an economies of scope gain. Joint-production of both goods is now cheaper than making them separately, as shown by the shape of the cross-section linking points C and D (transray convexity).

SCALE AND SCOPE: CONCEPTS AND ISSUES 31

The discussion thus far does not consider the impact of technical change on scale. Stevenson (1980) and, more recently, Stiglitz (1987) and Markowski and Jubb (1989), have raised the possibility that technical change may also have a scale ‘bias’. In Stevenson’s view: ‘Such a bias would alter the range over which returns to scale of a given degree could be realized—and thus possibly alter the output level at which minimum average costs could be attained’ (1980:163). Markowski and Jubb (1989), in turn, have extended this discussion to the multiproduct setting and have explored some of the emerging scale and cost relationships. To exemplify how technical change may affect scale, let us select a fixed combination of products Y1 and Y2, as that represented along the ray OR in Figure 3, and consider the cost behaviour as the scale of the resulting output bundle Y* is varied—i.e. a ‘slice’ of the cost surface perpendicular to the Y1, Y2 plane and along ray OR. As was pointed out in section 4, as along any particular ray output proportions do not change, by working with a ‘composite’ commodity Y* one circumvents some of the problems posed by aggregating two different goods. Changes in the output of the composite commodity are of the same proportions as those of its individual components. Average unit costs for the chosen composite commodity Y* can then be estimated at any point along this ray or ‘slice’, in the same way as in the single-product case (ray average costs). Figure 4 shows the average unit-cost curves corresponding to costs of producing the composite commodity Y* under ‘old’ and ‘new’ technologies, with optimal-scale points A and B, respectively. As described above, there has been a reduction in costs but the scale of output has not been affected, as the optimal scale remains on the AY*1 line. Only when the ‘optimal’ scale shifts to any point within the shaded area OC2AY*1’ to the left of line AY*1, i.e. at unit costs lower than or equal to those of the ‘old’ technology but with lower volumes, is there scale reduction or ‘de-scaling’. Point D in curve NT2 represents one such case. A commonly mentioned example is the development of the electric arc furnace, which allowed the emergence of a number of steel minimills producing 100,000tons of steel per annum as opposed to the conventional integrated mills, had to produce several million tons to be efficient (Acs et al. 1990; Auty 1992). In essence, this is what the ‘modern technology’ literature is arguing. Conversely, the introduction of a ‘new’ technology may lead to an increase in the optimal scale, or ‘scaling-up’. In this case, lower average unit costs are achieved at higher levels of output than with the ‘old’ technology—the shaded area to the right of AY*1. Point E in curve NT3 illustrates this case. Stiglitz (1987) mentions the example of the chemical industry, where new and more efficient plants always have larger capacities. Critics of the modern technology literature would argue that the optimal scale with FA should be situated in this area. A few final remarks concerning the applicability of this framework are called for. The first relates to the framework’s relevance to the product, plant and firm dimensions of scale and scope. Insofar as the cost axis includes change-over costs, production costs, intangible investments, or a combination of them; and

32 CONCEPTS, METHOD AND SYNTHESIS

Figure 3 Economies and diseconomies of scope

the output axis considers one or two goods produced in a multi-product plant or the same good manufactured in two different plants, the framework thus far developed would seem to be useful for addressing all dimensions of scale and scope. The second relates to the fact that, in reality, interaction and trade-offs between costs, scale and technology will be much more complex. There may be circumstances where FA will lead to the replacement of a production facility producing one or two goods by another producing the same number of goods. But FA may result also in new production facilities capable of producing a much larger variety of goods. On occasions FA may lead to significant scope economies and product-specific scale economies; in other circumstances it may result in minor product-specific scale economies and overwhelming scope

SCALE AND SCOPE: CONCEPTS AND ISSUES 33

Figure 4 Optimal scale of production for composite good Y* under alternative technologies

economies; while in another situation it may result in both economies of scope and multi-product scale economies. There may be circumstances where all possible effects are present. Precisely disentangling and controlling for each one of these effects may require a database that is way beyond this or any other research, so some choices as to extent of information to be gathered have to be made. The third point concerns the degree of ‘flexibility’ in the application of the framework. As Stigler (1939) pointed out, in reality firms do not normally, if at all, operate in the ‘optimal scale’ point of production, but instead choose a range of possible efficient points that permits them to deal with uncertainty. Thus, it may not be possible to identify the ‘optimal scale’ point of production, but only a level of output and costs that, within reasonable parameters of resource utilization and existing demand, may be as efficient as it is possible to be. Notes 1 According to Morroni (1992) and Tayler (1985) the basic idea of higher efficiency through task specialization was not that of Smith himself, but should be traced back to William Petty and, beyond him, to French Enlightenment philosophers and, ultimately, to Classical Greek philosophy. 2 The other major contributor to classical political economy and issues of division of labour and mechanization was, of course, Marx. Essentially, Marx made a much sharper distinction than Smith between division of labour and mechanisation. In the Marxian analysis there is a progression from handicraft production to, what is termed, the ‘manufacturing system’ of production, based on a fully developed division of labour in the workshop. Yet, in the ‘manufacturing system’ production is still dependent on the skill, strength, quickness and sureness with which each

34 CONCEPTS, METHOD AND SYNTHESIS

3

4 5

6

7

8

9

10 11

individual worker handles his/her tools. Only after ‘factory system’ production is introduced, ie where the machine performs with its tools the operations that previously were done by the worker, that production is no longer limited by the physical and mental capabilities of the worker and a fully scientific organization of production can be developed. Cooper (1971) has reviewed some of the theoretical arguments put forward by Marx. Stigler (1951) argued, from a neo-classical standpoint, that if economies of largescale production were so large how can the existence of small firms be explained. He added that either there was a trend to monopoly in all industries or they were competitive, in which case there were no increasing returns. To be sure, Marshall had already pointed to the existence of diminishing returns to scale in the context of agriculture or due to limits in management (for an elaboration of this point, see Scherer et al. 1975), and had argued that increasing returns were reached only in the very long run and only in manufacturing industry, and that large firms catered for markets where demand was large and small firms for those where demand was low. But these views were still dismissed by neo-classical economists, particularly Stigler (1951), on the grounds that even ‘limited’ managers would soon seize the opportunities for securing monopoly positions, and that they failed to provide the conditions for stable equilibrium and were inconsistent with Marshall’s own views on general equilibrium (see also Gold 1981; Tayler 1985; Vassilakis 1987). The concept of ‘envelope’ curve was not in Viner’s original article but was referred to by him in a supplementary note written in 1950. The existence of indivisibilities is a key factor in the breaking of the convexity conditions required by neo-classical theory for the optimization and uniqueness of solutions (Baumol 1987). Tribe and Alpine (1986) have recently criticized the validity of the 0.6 rule on the grounds that there are ‘significant discontinuities in the value of the coefficient within the range of scales’. An earlier criticism along the same lines was voiced by Gold (1981). According to Bessant (1991), during the period it took to change over, Ford lost US $200 million, was forced to lay off thousands of workers—60,000 in Detroit alone —had to scrap 15,000 machine tools and rebuild another 25,000, and lost its market leadership to General Motors. Cost savings can arise also from the fact that the number of staff required for maintenance and service increases much less proportionately than does output (Scherer and Ross 1990; Pratten 1975, 1991a). Empirical research on the extent of product diversity quoted by Carlton and Perloff (1994) shows that 146 out of the top 200 US manufacturing firms by shipments operate on 11 or more industries, while another study on all US manufacturing firms found that single-plant firms produced on average between 1 and 2 distinct products while multi-plant firms produced on average between 2 and 3 distinct products. Put in slightly different terms, the elasticity of unit costs with respect to output for any level of output is higher for the flexible than for the inflexible plant. Paradoxically, the ‘young’ Stigler, also had some very explicit doubts about the validity of the law of diminishing returns: ‘The historical connection between agricultural land and the law of diminishing returns may explain in part the failure

SCALE AND SCOPE: CONCEPTS AND ISSUES 35

12

13 14 15

16

of economists to recognize the difficulties in short-run application of the law’ (1939: 314). Landes (1969) considered the development of parts’ interchangeability to be a crucial step in the drive towards higher scales as it allowed for the routinization of assembly activities and processes, and their change from a nodal to a linear flow. Rising R&D costs may result also from new health, safety and environmental concerns (Kline and Rosenberg 1986). Indeed, increasing R&D costs would seem to be one of the reasons behind the proliferation of research co-operation agreements between firms. Scherer and Ross (1990) indicate that in a study for the US the R&D costs to imitator firms of unpatented goods were 50 per cent of the original innovation costs in 127 industries covered by the study. Duplicating costs exceeded 75 per cent in 40 per cent of the industries researched. By allowing larger product variety or scope, and therefore wider pricing possibilities, technical change may in addition affect marginal revenue.

2 METHODOLOGY Research design and implementation

1 Introduction In the previous chapter we defined the conceptual framework for this research in relationship to four concepts: technology, scale (either in terms of output or capacity), scope or output variety and costs. It is now time to describe how this framework was turned into a workable method and research design, bearing in mind that the focus of the research was firms—institutions which in many cases are not used to providing detailed data and are often suspicious and unwilling to be ‘sniffed’ at by academic researchers. This chapter will look into the methodological choices, the reasons for them and the trade-offs faced in doing this research. It presents a model case study on which studies performed in selected developing countries’ firms were based. Indeed, the case study provides a precise description of the underlying methodology and of the data requested, so the reader may want to spring straight into it. Section 2 analyses the advantages and disadvantages of the firm-based case study approach that was used. It points out that, because of the microeconomic emphasis and the early state of diffusion of FA in developing countries, a case-study approach was the most appropriate methodological tool. Section 3 discusses the selection of industries and countries. In selecting countries, the key criterion was access to a large number of new technologies in order to make the research viable. This criterion excluded most developing countries as FA has barely reached them, leaving only a few, the more advanced ones, from which to select. Section 4 describes the kinds of data requested. As was pointed out in the Introduction, much of the research done in this field has an extremely weak empirical base, and it was felt necessary to improve on this situation. It was believed that some quantification of the extent of the changes, imperfect as it may be, would provide a sense of the magnitude of the issues at stake. Section 5 presents a very detailed case study designed on the basis of contacts and in-depth interviews in a number of CNC machine-tool producer and user firms in Britain, Belgium, Germany and the Netherlands. The case study was intended to provide a model on which later interviews in developing countries could be based. The case study reflects the experience of a mediumsized British engineering firm, ‘Hydraul’, which since 1986 had replaced most of

METHODOLOGY 37

its old machine tools by one FMS and a number of CNC lathes. It was very fortunate for the research that such a collaborative firm was found so early on in the study. Having such clear guidance on what was expected significantly eased field implementation. The case study is interesting in its own right as it covers all the main issues regarding the diffusion and impact of FA in developed and developing countries’ firms. Section 6 recapitulates on the different phases of the research and the technical, academic and industry inputs obtained while implementing the study. 2 Approach, method and unit of analysis One key methodological decision was whether to base the research on surveys or case studies. Performing a survey, if properly designed and implemented, would have the advantage of providing a significant amount of quantitative data and a much more precise idea of the impact of FA on scale and scope throughout industry. Well-known government surveys on use of technology are the US Census and annual surveys of manufacturing technology, which provide a wealth of information including use of and plans to use specific manufacturing technologies, expenditure in automation, factors affecting adoption and the age of equipment (see US Department of Commerce 1988, 1991). Research in this field has resorted also to randomly and non-randomly selected sample surveys, normally postal, to obtain data on the use and implications of FA. The main problem with survey data is its expense. Census and annual surveys can be carried out only by government statistical offices or large governmentsupported research centres which can exploit economies of scale and scope in survey design and application. Even sample surveys, although lest costly, involve considerable investment in questionnaire design, sample selection and postage, and they always require prior knowledge of the population. Moreover, when they are poorly designed and implemented, little can be done to salvage them. Other general limitations with surveys are that the type and number of questions are limited, that it is difficult to target different individuals or functions within an organization as, normally, a single individual fills in the questionnaire and that response rate is generally low, particularly when returns are not mandatory. Where multi-nation analyses are necessary, survey instruments which do not share the same methodology are limited in terms of the comparability and variable reliability of the data returned. Case studies on the other hand have limited value for generalization, as it is not always easy to distinguish results which are specific to the case under scrutiny from those that reflect the wider collectivity. A case-study approach, however, was appropriate for this research because the diffusion of FA is a recent phenomenon in developing countries, meaning that relevant statistics or surveys are either unreliable or unavailable. Performing a survey, even for a single country, would have far exceeded the resources available for this project and many others. Case studies were considered suitable because certain issues, such as the reasons why a specific firm was adopting a technology, can be

38 CONCEPTS, METHOD AND SYNTHESIS

comprehended most fully through the study of the internal workings of firms, the ‘microcosmic realm’ (Westphal et al. 1990; Yin 1984). This, in turn, required gaining in-depth information from different departments and individuals within the organization. Further, to the extent that in developing countries technical change is restricted to few firms and accurate and detailed data from them are required for sensible conclusions to be reached, opting for a case study approach was particularly pertinent. A case-study approach was favoured also because it would provide the more precise knowledge of the technologies being adopted and the extent of their adoption, as well as some indications on costs, necessary to perform meaningful performance comparisons between firms in different countries. The method selected to address the issues at stake was comparative statics, which posits point in time in order to extract ‘before’ and ‘after’ conclusions. Comparative statics is usually used in economics to analyse the impact of a change in the parameters of a model by comparing the equilibrium positions that exist either side of the change (Kehoe 1987). Its main limitations are that it is unable to offer analysis either of the forces that created equilibrium in the first place or of the process of change from one equilibrium position to the other; there are also doubts as to the reality of equilibrium itself, particularly of its conditions of uniqueness and regularity. As far as this research is concerned, however, the aim is not so much to understand the process of diffusion of FA and its outcome as to examine whether the new technologies have changed the balance of economies of scale and scope. For this much narrower purpose, a comparative statics approach seemed the most appropriate method, and one that can be easily related to. In any case, the research provides a reasonable account of the factors that have led to a change in scale and scope balance. As to whether the ‘before’ and ‘after’ are indeed equilibrium points, the research will make no such claim, as was pointed out, following Stigler in the previous chapter. The study focuses on firm level because this was essentially microeconomic research. However, there were a number of choices here too. The first choice concerned which firms to include in the sample. Since it was necessary to perform ‘before’ and ‘after’ comparisons to find significant differences, it was decided to include firms that had some experience of use of both old and new technologies or, alternatively, dissimilar old-technology firms and newtechnology firms which produce similar products. In the latter case, the number of firms had to be enlarged to keep to a pre-set number of ‘comparisons’. In practice, most of the firms included had experience with both vintages of technology. The second decision concerned the number of firms. This was a difficult decision as there was no clear indication as to the population size of firms using FA in developing countries, except perhaps for some research, summarized in Boon and Mercado (1990), Edquist and Jacobsson (1988), Jacobsson, (1986) and Watanabe (1993a), suggesting that it was very small. Considering this fact, and also the in-depth interviews and extensive data requirements, something we will return to shortly, it was decided that around 60 developing country firms or ‘comparisons’ should be examined, each country represented by 8–10 firms or

METHODOLOGY 39

‘comparisons’, would be sufficient to provide a reasonable picture of the issues under research. This number of case studies is large enough to suggest any significant variations. In the event, 62 firms participated in the study, although not all of them provided the full set of data, thereby precluding statistical analysis. Finally, it was necessary to introduce some control criteria in terms of size and ownership. As it was not possible to have a precise idea of the population it was decided to include at least two small, two medium-sized and two large firms by employment. In line with established research for developing countries, firms with under 100 employees were considered small, firms with over 100 and less than 500 employees were classified as medium-sized and firms with over 500 employees were classed as large. It was decided to include also at least two foreign firms per country. Foreign-owned were defined as firms with a majority of foreign investors. Locally owned firms were those fully owned by local residents. In practice, it was possible to keep to the size criteria but, surprisingly, foreign-owned firms were much harder to find, partially because of statutory limitations in some countries. As a result, a third ownership criterion—foreign participation —had to be included. Foreign participation applied to firms with a minority foreign stakeholding. Before moving on to the next section, one further and final remark on generalization. Clearly, despite the efforts that have gone into sample selection and, as we will see, into ensuring comparability and obtaining accurate data, the number and characteristics of firms and the resources available for the research do not permit any kind of generalization in a statistical sense. This is an inescapable limitation. Nonetheless, what it is believed the research can provide is a consistent set of empirical observations which, few in number as they may be, give a sense of the direction and magnitude of the underlying changes. That is as far as this research can go although, it must be emphasized, it is far more, in empirical terms, than what most of the research in this field has to offer. 3 Sampling: technology, industries and countries Microelectronics applications have spread pervasively (Freeman 1982; Freeman and Perez 1988; OECD 1988) within and across industries, while at the same time leading to the emergence of several new industries. FA has also affected most processes and sub-processes and functions within firms. However, of all such industries, engineering is where the impact of microelectronics-based technologies is argued to be the greatest. FA is said to have rapidly diffused both in the design and production of engineering goods, automating the production process throughout the industry, and leading to major improvements in productivity (Edquist and Jacobsson 1988; Watanabe 1993a). Hence, it seemed fitting to focus on engineering for this research. Focusing on the engineering industry was also important for at least three other reasons. First, the industry is characterized by a diversity of equipment, processes and products, so firms have constantly to face choices between

40 CONCEPTS, METHOD AND SYNTHESIS

specializing or diversifying their product range and the underlying technology (Edquist and Jacobsson 1988; Westphal and Rhee 1983). Hence, concentrating on the engineering industry offers a unique opportunity for appraising the possible combinations and trade-offs between economies of scale and economies of scope. Second, the engineering industry has significant forward and backward linkages, is at the forefront of technological progress due to its high knowledgeand learning-intense nature and provides equipment to all sectors of the economy (ECE 1992c; UNIDO 1991d). It is also a very large sector within manufacturing in developing countries, accounting for up to 33 per cent of manufacturing output and value added. Data on the surveyed countries are given in Table 2.1. The emergence of cost-efficient engineering firms in developing countries may, in turn, add to the building of local technological capabilities and long-term development. Third, developments in the engineering industry, particularly in car manufacturing, have normally been used by historians, economists and engineers both to conceptualize the whole of manufacturing industry and as an indication of what is to come in other industries. The very idea of ‘mass production’ was originated in this industry. Engineering, therefore, seems the most relevant industry in which to test the hypothesis presented by the literature.1 Within engineering there was also some choice as to which specific sectors to consider. Initially two sectors were selected, motor vehicle components and general industrial machinery and equipment, on the grounds that these industries should account for a large proportion of machine tool demand if the experience of developed countries is anything to go by, and because of their large share of engineering industry output in developing countries.2 Autocomponents include manufacturing of products such as front and rear axles, brake discs and systems, shock absorbers, transmissions, piston rings and cylinder linings, fuel systems and engine bodies. The general capital equipment industry includes manufacturing of oil and water valves, pumps and electric motors. As the research progressed it became necessary to include a third industry sector— manufacture of customized products—which during the fieldwork proved to be a significant user of CNC machine tools in developing countries. Manufacturers of customized products consist mainly of producers of moulds, firms providing machining services and some other firms, such as those producing gas or sanitary fittings, that could not be classified elsewhere. The diversity of engineering operations meant, however, that some further choices had to be made on the kinds of process that were to be examined. There are several basic operations in engineering (Groover 1987, 1996). • Processing operations involve the transformation of a product from one stage of completion to another without adding any material or component. Only energy, in any of its forms, is added to change the shape of a part, to remove material from it, alter its physical properties or accomplish some other change required. Processing involves: a basic stage, such as casting or forging, where work materials are given their initial form; a secondary processing stage, such as machining, where the workpart is given its final desired geometry; a

Source: UNIDO Industrial Statistics, 1995.

Table 2.1 Engineering industry’s value added (%) and output (%) as shares of total manufacturing (US$m.) of surveyed countries, 1963–91

METHODOLOGY 41

42 CONCEPTS, METHOD AND SYNTHESIS

• • • •

third stage, such as heat-treating operations, where the physical properties of the workpiece are enhanced; and, a finishing stage, such as coating, polishing or chrome-plating, where the final appearance of the workpiece is determined. Assembly operations require the joining of two or more separate components through mechanical fastening operations with screws or nuts or through joining processes such as welding or soldering. Material handling and storage operations comprise the movement and storage of pieces from processing to assembly through manual or automatic means. Inspection and testing include ensuring that products meet the established standards and specifications and function in the desired way. Process control involves the regulation of individual operations as well as the management of plant-level activities.

Machining involves removing excess material from a workpart, and is seen by most engineers as one of the more important manufacturing processes (Groover 1996), because it can be applied to a variety of materials, including plastics and ceramics; it can generate shapes of almost unlimited complexity; it is more accurate than most processes; and it is capable of creating smooth surface finishes. The commercial and technological importance and specific charateristics of machining operations have meant that over the last years they have been affected, to a larger extent than any other of the engineering operations, by developments in microelectronics. Thus, to be included in the research, firms must, at least, be engaged in some machining operations. The choice of machining operations for inclusion in turn dictated that the equipment and machines available in the sampled firms included metal-cutting machine tools. Metal-cutting machine tools are defined as power-driven machines, but excluding hand-held cutting tools, that shape metal by cutting or using electrical techniques (AMT 1989). Old technology or ‘before’-machine tools included conventional or general purpose machine tools such as lathes, drills, milling and boring machines; also included were fixed or special-purpose machine tools such as transfer lines and automatic screw machines (Amstead et al. 1986; De Garmo et al. 1984; Groover 1987, 1996).3 New technology or FA ‘after’-machine tools included CNC lathes, CNC drills, CNC mills and machining centres, which integrate drilling, milling and boring operations in a single machine. Other FA equipment examined comprised programme logic controllers (PLCs), nowadays digital controllers of inputs and outputs of specific production processes, sub-processes or machines; industrial robots, i.e. automatic position-controlled reprogrammable multifunction manipulators capable of handling materials, parts, tools or specialized devices; computer-aided design/manufacturing (CAD/CAM) technology which allows graphic representation and electronic drawing, and generates engineering data and programs for modelling and manufacturing products; process control computers (PCCs) and necessary software, which alone or in combination with other computers and/or PLCs, can simultaneously gather, process and store

METHODOLOGY 43

production data, monitor and control several, if not all, stages of the production process, interface with the operator and communicate with other devices outside the system; and flexible manufacturing systems (FMS), i.e. the combination of CNC lathes and machining centres, load-unload stations, automatic tool changing, electronically controlled vehicles to transport workpieces between stations, and a central computer for process control. However, because of the need to make comparisons between technologies or production processes which are as equal or homogeneous as possible and to measure the impact of FA on production or capacity and product variety, it was decided that firms had to be using CNC metal-cutting machine tools to be included in the sample. CNC metal-cutting machine tools, therefore, became the core technology for this research. On the organizational side five new techniques, ‘concepts’ or ‘approaches’ were examined. 1 Total quality management (TQM), i.e. addressing quality at all stages of goods’ and services’ production, was operationalized on the basis of several aspects: improving product design so that potential quality problems are detected at the design phase or minimized at the production stage; monitoring quality of inputs and suppliers; quality checks at each workstation; introduction of quality circles (groups of workers from near-by workstations discussing and evaluating quality); and statistical process control: random checking of products at the end of the line. Emphasis was, however, put on the third aspect as firms must have introduced quality control in each work-station to be judged to have have adopted TQM. Quality was understood as freedom from deficiencies. 2 Just-in-time (JIT), the practice of organizing delivery of raw materials, workparts and final products to each stage of production or to customers so that inventories and waiting time are minimized, was operationalized on the basis of attempts to promptly organize deliveries inside, ‘internal JIT’, or outside the factory, ‘external JIT’. 3 Working practices reorganization involves the reduction of hierarchical levels, redefinition of job posts and functions, changes in pay practices, multifunctionality in tasks and enhanced responsibility for shopfloor workers. 4 Cellular manufacturing (CM) involves the physical reorganization of the factory lay-out as cell work centres dealing with parts of similar characteristics or in product-oriented work centres. 5 Finally, quality accreditation: early on in the fieldwork, it became apparent that many firms were using their applications to international quality certificates, such as ISO 9000, Ford’s IQ 101 or Japanese product quality standards, as alternatives to TQM techniques to improve on quality. Since the ‘standards’ approach differs from TQM in that the former requires meeting certain product and process norms while the latter involves

44 CONCEPTS, METHOD AND SYNTHESIS

continuously striving for quality, it was decided to include the application to international certification as another organizational technique. Turning to countries, the research concentrated on six developing countries. Some research was carried out in developed countries, but this was mainly exploratory and preparatory for the work to be done in developing countries (see Alcorta 1993a for some of the results of this research). Within developing countries, efforts were made to identify countries with a significant extent of FA diffusion for developing country levels; at the same time, a wide enough geographical coverage was maintained to account for varying economic conditions and regional differences. In Latin America, Brazil had a fairly well-developed engineering industry for developing country levels, and research by Quadros Carvalho (1992), Ferraz et al. (1991), Fleury (1993), Porteous (1992) and Silva (1992) has shown evidence of Brazil’s increasing and sustained adoption of FA. Brazil also had its own CNC machine tool industry. In addition, from being a protected industry mainly oriented to the domestic market, Brazil’s engineering industry was carefully opening up to foreign competition and hence, increasing the domestic supply of capital goods. Mexico, it was believed, would provide an interesting case study, as the engineering industry had recently received a large influx of foreign and local investment aimed at modernizing the industry and entering the US market. The Mexican engineering industry was already, within Latin America one of the most open to foreign competition, and was expected to be internationally competitive by the time the research began. Some research had already been done and statistics on the adoption of FA in the engineering industry by ownership, firm size and sub-sector, documented by Domínguez (1993), could be used as a starting-point. The third country chosen in Latin America was Venezuela. Helped by a sustained influx of foreign exchange from its oil resources, Venezuela had been quietly building its manufacturing base. Indeed, imports of machine tools had overtaken those of Argentina and had been sustained during critical periods (UNIDO 1991d). At the same time Venezuela’s use of machine tools remained far below that by Brazil or Mexico, and was more typical of a medium-sized developing country. In the Middle East, Turkey’s engineering industry had been growing rapidly in the years preceding the research. According to ECE (1992c), the index of engineering industrial production in Turkey increased by more than 46 per cent between 1985 and 1990. Of all the developed (including East European) countries, only Austria and Ireland had performed better. The industry had been exporting metal-forming machine tools to more advanced countries and was, according to metal-cutting machine tool manufacturers in Europe, one of the more attractive markets in developing countries, having achieved a significant degree of technological capability. Another country selected for the research was India. For years India had protected its engineering and machine-tool industries behind tariff and non-tariff barriers. Since the early 1980s import liberalization had had a profound impact on the industry. Annual imports of machine tools were on average 23 per cent

METHODOLOGY 45

higher in 1986–8 than in 1980–6 and were estimated to have risen even further in more recent years (UNIDO 1991d). According to Jacobsson (1991), liberalization was leading to increased diversity of products and equipment supplied both by the local engineering industry and by imports, but this may have been achieved at the expense of losing procurement, R&D, marketing and finance scale economies and limiting the potential for own-innovation capabilities. Finally, Thailand, in the Far East, was studied. Research by O’Connor (1989), UNIDO (1991d) and Watanabe (1993a) have pointed to an emerging hi-tech engineering industry producing components for automobile, motorcycle, agricultural machinery and electrical appliance manufacturers, both for the local market and for exports. Indeed, as this research was being initiated, Hitachi announced the relocation, from Japan to Thailand, of its production of electric motors for its machine tools. 4 Data requirements The research was based on primary and secondary data. Primary data were obtained directly from in-depth semi-structured interviews performed in firms from the six selected countries and for which a detailed questionnaire was prepared. The length of each interview varied from 3 to 12 hours. In some cases there was more than one interview so the accumulated interview time varied from 3 to 20 hours. The Hydraul case-study interview which follows involved two face-to-face interviews of around 16 hours altogether and lengthy follow-up telephone conversations of around 2–3 hours. Often, introductory visits to explain the research to the general manager and to obtain his backing for further visits had to be arranged. Obtaining the backing of the general manager or owner was crucial as it meant fewer restrictions on access to data or documents. Managers interviewed included the general manager, technical manager, production manager and technical staff. In small firms one manager had all the information but in larger firms this was not the case. It was always better to interview more than one person because more information was normally obtained and it was possible to cross-check the data. To get the complete financial picture of the company it was sometimes necessary to get in touch with the accounting or financial departments. Some firms opted, in addition to interviews, for a written questionnaire which they could fill in at their own leisure, although in these cases a considerable amount of chasing had to be done, and some firms never returned the questionnaire. The final ‘output’ of the interviews was a case study for each firm although in every case it was not possible to develop a fully fledged study. An important additional criterion for selecting firms was whether or not the firm would be willing to submit itself to such intense scrutiny. To reduce the risk of failing to gain access to firms, consultants who had already worked with potential or target firms and/or were known in the industry were chosen. In addition, great efforts were made to ensure that consultants were referred to

46 CONCEPTS, METHOD AND SYNTHESIS

specific firms, or individuals within firms, by their suppliers of machine tools, business associations and/or other academics or consultants who had already worked with targeted firms. This proved to be an invaluable way of accessing firms and of gaining their trust. Indeed, one of the consultants has since returned to some enterprises for even more extensive interviews and research (see Domínguez and Brown 1995, 1996). The questionnaire prepared covered five areas and was divided into 23 topics. The five areas were firm history; production process characteristics and process of adoption of FA; impact on product scale and scope; unit costs; and other factors affecting the use and performance of FA. Both quantitative and qualitative questions were included in each section, some requiring consultation of documents or records and completion afterwards, others prompting complex elaboration, thought and discussion of past and present experiences by the interviewees. A brief statement of each firm’s history was sought in order to have an idea of the ‘paths’, and the ‘dependencies’, that firms had followed, and developed, over time, the factors that had influenced such directions and how these may have influenced their technological decisions. In particular the questionnaire emphasized ownership aspects including the origin of the firm, major ownership changes that had taken place through the years, whether the firm had at some time been involved in any take-over or merger and whether the firm was a member of a domestic conglomerate or multinational corporation. Another set of questions addressed the product and market orientations and the participation of the firm including its product profile: whether the company was specialized in manufacturing and assembly of a single component or several components; whether its products were aimed at the original equipment manufacturer (OEM) market or directly at consumers; whether the firm was export or domestic-market oriented; and what the company’s market share was. There were questions also on the long-term evolution of variables such as sales, employment and assets, which some firms addressed by providing company balance sheets. Questions were raised on any major accomplishment in new-product development (NPD) and in adaptation or modification of products or production processes. The basic characteristics of the production processes were ascertained by requesting a description of the production processes, including the number and kind of operations involved, inputs flow, factory lay-out, machinery in use before and after modernization, and which stages, if any, of the process were subcontracted. A factory visit was often requested, and many times granted, providing the interviewer with a visual image of the production processes. Indeed, sometimes the description was made en-route. There were some very specific questions regarding the equipment and organizational techniques in use ‘before’ and ‘after’ modernization. One issue that arose in this regard was determining the ‘before’ and ‘after’ zones in each case, as modernization in most firms is a process that can take several years. ‘Before’ was defined as the period when FA was not in use or accounted for a minimal share of total production— less than 5 per cent of the total stock of machine tools—or involved only the odd new machine, here and there. Alternatively, a cut-off date of 1985 for ‘before’

METHODOLOGY 47

and the moment of the interview for ‘after’ were used, as firms generally felt more comfortable with specific dates. This second criterion was considered acceptable, as aggregate figures suggested that there had been little diffusion of FA in developing countries prior to that year. It also made sense to the firms interviewed, as in most cases 1985 actually reflected their own experiences of transition. In addition to the previous description and questions, there were queries on the reasons for the adoption of FA and the processes involved in its adoption, including the length of time necessary to install the new equipment and any problems faced before the equipment became fully operative. In general, firms provided most of the information requested in this section of the questionnaire. The section requesting data for measuring overall output and product mix was aimed at obtaining data on scale and scope on the basis of firms’ own definitions. Ideally, the full list of products and components, and their prices for all years within the ‘before’ and ‘after’ period, would have been obtained; then, with the help of the Laysperes method, which uses the prices at the beginning of the period as the base, or the Paasche method, which uses the prices of the previous year as the base, and after correcting for inflation, changes in overall real output could have been calculated (Hayes et al. 1988). Although this approach was considered, it would have required an immense amount of data due to the very large number of intermediate components and final goods produced. Processing the data alone would have taken months, in the unlikely event that the required data were available. Instead, the questionnaire sought to obtain data on setting-up times, batch sizes, product variety, product research and development, plant scale, capacity utilization and machine efficiency, as measured by the firms themselves. Although there was some guidance from the interviewer as to definitions, the emphasis in this part of the questionnaire was on obtaining precise descriptions and measurements from interviewees so that comparisons could be made at a later stage. Often, this meant getting back to an interviewee or cross-checking results with other staff. In the case of product variety, for instance, interviewees were asked to define different products in terms of product families, i.e. as some kind of ‘aggregate’ and what this ‘aggregate’ was, in terms of their functional characteristics, in terms of a combination of functional characteristics and sizes, or of any other specific definition the firm was using. Firms were asked also the reasons for defining product scope in the ways in which they did. In the case of scale, some firms pointed to the difficulty of having a figure of total physical output in engineering firms because of the fact that both intermediate components and final goods were produced. They suggested that the practical solution to this problem was to use the volume of castings consumed in production as a total output indicator (see the case study in section 5). Between 50 per cent and 75 per cent of firms provided some data on product scale and scope. The questionnaire included a section on costs, the purpose of which was to capture the cost variations that underlie changes in scale and scope. It would have been desirable to obtain cost figures for individual products and cost

48 CONCEPTS, METHOD AND SYNTHESIS

variations as scale and scope changed but, despite a number of questions directed to such effect, data of this sort were hard to get. Although some firms provided information of individual product unit costs, others did not. This was not necessarily a matter of secrecy but more often than not was due to the fact that firms just do not keep detailed cost records. Some firms admitted this and added that with modernization and FA they were expecting to improve their cost accounting. However, more than 30 firms in the sample provided some data on unit or total costs prior to and after the adoption of FA, including figures or indicators on variations in capital, labour, inputs and overheads, which together with figures on scale and scope provided a picture of developing country firms’ unit-cost structures. Finally, the questionnaire had a section aimed at identifying any other external factor that may have significantly affected the adoption and performance of FA. Economic factors such as sudden and large changes in relative prices or interest rates, a major import liberalization or a recession; social factors such as strikes; and other factors, including drastic cutbacks in the supply of energy, natural disasters or excessive red tape, were included. Firms were all too pleased to address these issues, particularly those in relation to their governments’ policies. In terms of the conceptual framework discussed in the previous chapter, the interview data should provide an accurate description of ‘old’ and ‘new’ technologies. It should give also a relatively precise idea of changes in product scale and scope. For those firms that also provided cost figures, and given a ‘reasonable’ capacity utilization level, the data should indicate where in Figure 1. 4 they are located and whether a specific firm is operating at any point in area OC2AY1* or anywhere at the right of AY1*, i.e. whether there has been ‘scaling up’or ‘de-scaling’. However, it is important to be aware that because there is simply no information on the full cost curve it is not possible to know whether this point is a minimum cost one. Also, the data will not be able to tell us whether changes result from product scale economies and/or from economies of scope. Turning to secondary data, the research obtained published technical information on the characteristics and use of FA from industry journals and machine-tool manufacturers’ documentation. Data on CNC machine-tool production, trade and consumption are fragmented and dispersed, and an attempt was made in this research to bring together different sources. The main sources of data are the American Machinist; the European Committee for Co-operation of the Machine Tool Industries (CECIMO); the UN’s Economic Commission for Europe Industry and Technology Division (ECE, ITD)/Organization for Economic Cooperation and Development (OECD) databank; the UN’s COMTRADE statistics on trade (United Nations 1996); local associations of machine-tool producers; and the governments of the countries studied. The American Machinist makes an annual survey of around 35–40 countries on production, trade and consumption of metal-cutting and metal-forming machine tools. Its main advantage is that it goes as far back as the 1960s, but it does not provide a breakdown for CNC and non-CNC machine tools. CECIMO collects data on production, trade and consumption from national machine-tool associations and statistical offices. The strength of CECIMO’s data is that they

METHODOLOGY 49

provide a breakdown for, respectively, metal-forming and metal-cutting, and numerically and non-numerically controlled machine tools, particularly for production data. It has a similar database also for consumption but this is built on the basis of sometimes incompatible statistics. CECIMO data’s main weakness is that they cover only a very limited number of countries.4 ECE/OECD’s databank delivers a yearly questionnaire to member countries of each organization with diverse indicators for the engineering industry, including data on production, trade and apparent consumption of metal-cutting and CNC machine tools. Although the geographical coverage is wider than that of CECIMO’s data, it backdates only to the late 1980s and is, by its own admission, less complete than that of other sources (ECE 1994). The UN’s COMTRADE database (United Nations 1996) includes most countries and also goes back as far as the 1960s in its first revision of the Standard International Trade Classification (SITC), but provides only trade data on machine tools. More recent (second and third) revisions provide more detailed breakdown but at the expense of reducing the time-span and number countries covered. National statistics are generally tapped by the previous sources, so it was necessary to resort to them only when a particular datum was unavailable in the other sources. However, some data were obtained from US, Indian, Korean and Japanese domestic machine-tool associations. 5 The model case study: Hydraul5 5.1 The firm Hydraul was established in 1961 as a medium-sized engineering-component manufacturer, especially hydraulic valves and other equipment, for the construction equipment industry. One of its major customers was the familyowned Road Construction Group (RCG), whose requirements were parts for road sweepers, the major product of RCG.6 Over time, Hydraul became also a manufacturer of parts for machine tools, mining and tunnelling equipment and was very diversified. Around 1968 the firm encountered serious financial difficulties and was purchased by RCG with the main objective of serving group demand. The takeover brought some financial respite to the company, but by the early 1980s the company was in financial trouble again, forcing it to take major strategic decisions. The problem was that, for many of the products Hydraul sold, volume was essential to keep prices down, while at the same time there was a need to reduce the range of products to remain competitive. Between 1980 and 1985 the company entered a second major restructuring, initially reducing its product range and turning attention to exports. Since restructuring the company has become very profitable (Table 2.2).

50 CONCEPTS, METHOD AND SYNTHESIS

Table 2.2 Hydraul: basic economic data

Notes * Mainly to Continental Europe. Hydraul’s local market share was small for 1990, the only year for which data were available.

5.2 Technical change The production process The production process for engineering goods at Hydraul involves the transformation of metal castings into diverse cylindrical and cubic parts which are assembled mainly into valves, gear pumps and motors and hydraulic cylinders and gearboxes. The product range includes both general purpose valves and customized products whose main final destiny is the specialized industrial vehicle market—such as tractors, bulldozers, forklift trucks and road sweepers. The process consists of up to eight stages since not all parts go through the same stages and not all stages are carried in-house. The first stage consists of machining metal castings, mainly steel, forged iron or aluminium, provided by Hydraul’s local suppliers. The main operations carried out at this stage are drilling, milling and boring. Lathes are used for cutting cylindrical parts, while conventional drills and milling machines, and more recently machining centres, are used to cut cuboid parts. The second stage consists of heating, as many metal parts require high temperatures to achieve the necessary cohesiveness. This stage is contracted-out as Hydraul does not have heating facilities because its volume of output does not justify such an investment, although it is considering adopting a new heat induction process that seems to require lower volumes. This is one of the major bottlenecks to faster throughput times. The third stage is grinding, in order to smooth the parts and achieve accuracy in tolerances. The fourth stage is deburring to smooth down sharp edges and corners which sometimes is followed by another grinding stage (stage five). The sixth stage is assembly, at which all components are fitted into the final product. Randomly selected products then go through an electronic testing stage (stage seven) and, if required, a packaging stage (stage eight).

METHODOLOGY 51

Old technology Although the acquisition of new technologies started in 1979 with the introduction of the first computer-controlled machining centre, followed in 1983 by the first computer-controlled lathe, it was only in the second half of the 1980s that ‘real’ modernization started. Previously, the bulk of products were machined using conventional or numerically controlled lathes, drilling, milling and grinding machines and the arrival of the computer controlled (CNC) machines made little difference to the process, as they were used for only very specific high value-added applications and at rates far below their full capacity. However, Hydraul became well-acquainted with this equipment and during this period perceived its potential, so increasingly it was used to machine other products. Before modernization, the factory lay-out was organized along functional lines, with lathes followed by drilling machines which, in turn, were followed by milling and then grinding machines. Parts proceeded in sequential order through the factory until assembly. It was a typical job-shop. The equipment was ‘multipurpose’ in the sense that it could be used to cut workpieces of any shape, but it was ‘specialized’ in the sense that most machines could perform only a single operation, i.e. turning, milling, drilling or tapping. Although the first numerically controlled lathe—a punch tape system—was purchased in 1972, most of the equipment was conventional, so could not be programmed for repetitive cutting and required an operator to set up and work the machine. There were around 150 machine tools, some of which are still used today for operations that are not performed in the new machines—grinding, for instance, has proved difficult to integrate with the operation of the new machine tools because of their special shape and the abrasive material of which they are made—or in emergencies. Flexible automation (FA) Since 1986 a process of replacing most of the equipment and restructuring production organization has been in place. The new equipment includes five horizontal machining centres Flexible Manufacturing Systems (FMS), manufactured by Flexible Technologies (FT)—FT 500 HMC—linked with railguided vehicles and controlled by Siemens computers. It includes 3 state-of-theart 8-axis German Traub CNC lathes with an automatic rod-feeding mechanism —Traub TNS 65D. A 4-seat three-dimensional CAD system and other PC-based CAD systems have been purchased. Hydraul has replaced the production scheduling and administration, stocktaking and accounting computer by a more up-to-date and powerful UNIX-based industrial computer, and has increased the number of terminals available. Together with the computer new production scheduling and material resource planning software have been purchased. The company is in the process of starting to link all the existing computing systems, but this is proving extremely difficult and expensive because of the software requirements. Finally, Hydraul has organized itself and its suppliers to deliver just-in-time so that inventories of finished goods and raw materials are reduced.

52 CONCEPTS, METHOD AND SYNTHESIS

Foundries, for instance, now deliver at regular precisely timed daily intervals the exact volume of castings required for manufacturing. This has become possible because Hydraul is buying foundry time, i.e. so many hours a week at specific times, rather than a volume of castings, so that the foundry is exclusively dedicated to producing the castings Hydraul requires when it needs them. Operating in this way allows Hydraul to communicate at very short notice, normally one week but possibly less, its precise castings’ requirements for the following days. Actual requirements are given by Hydraul’s order book and production-planning schedule. The number of hours that are ‘booked’ depends on the average tonnage the company handles in any period of time. Future investments include the commissioning of an induction heating machine for treating parts that require high temperatures. In this way Hydraul may reduce delivery time from weeks to days as products needing heating would no longer be contracted-out, sometimes to a supplier around 100 miles away, with the attendant impact on competitiveness. Although the machine seems to be within the financial reach of the company, and would reduce costs and throughput time substantially, the main problem at the moment is that of establishing proprietary rights clearly with the supplier to avoid competitors making use of the innovation and having Hydraul pay for the development costs. Reasons for adoption of flexible automation In the early 1980s the company confronted a second round of serious financial difficulty, forcing it to take major strategic decisions. The cause of the problem was identified as low individual product volumes and excessive product range. It was the same factor that had led to Hydraul’s first major restructuring twelve years earlier. For many of the products Hydraul sold, volume was essential to keep prices down. Some competitors were focusing on a smaller range of products and could offer them at lower price. In addition the great diversity of products required a large number of engineers providing sales and technical support, as the knowledge required for each individual market was highly technical and detailed. Technical support and back-up knowledge for customers were essential requirements and the company simply was not providing it. The main decision made was to reconsider the product range and to concentrate on fewer products. In the general valves’ area the number of products were substantially reduced so that only already high-volume valves, for which an even bigger market could be sought, were produced. In the customized product range the focus was on hydraulic valves, cylinders and gearboxes for the specialized industrial vehicle market, i.e. tractors, forklifts, excavators, and other large construction equipment. In this area the company had built an expertise as a result of sourcing other group members which manufactured similar products, thus constantly demanding components and spare parts for their vehicles. Having a narrower product scope meant that the sales and technical support department had better focus, although it was decided to invest also in flexible plant just in case, once the company had strengthened financially, it was necessary to return

METHODOLOGY 53

to the manufacturing of a wider range of products.7 The possibility of having plant that was also flexible became a key objective of the restructuring process. As to the technology, Hydraul had been already experimenting with CNC lathes and machining centres and was well aware of their potential. The key considerations were the already observed drastic reductions in manufacturing time due to the possibility of performing several turning, drilling, milling and boring operations using the same machine; the significant reduction in machining cycletimes due to the increased speed of the new equipment; the possibility of running the equipment unmanned for several hours; the prospect of reaching far higher machine utilization rates; and the achievement of significant improvements in quality. An important additional advantage of the new technologies, particularly necessary to convince the board which had to authorize the investment, was that the funds did not have to be disbursed in a short period of time but could be spread over several years. This was so because the FMS could be built on the basis of independent machining centres to be joined only when the last one arrived —which in fact was eighteen months later. Indeed the system grew to some extent according to the availability of funds. The financial success of the first investment made it considerably easier to obtain approval for the purchase of the three Traub lathes in a single order later. According to Hydraul, and with the benefit of hindsight, the company simply would not exist today had it not adopted the new technologies. Competition in the industry is fierce, and Hydraul without the new equipment, have found it impossible to enter the continental market to the extent it has. In the early 1980s very little of Hydraul’s output was exported, while today most of it is. Exports have been a major factor behind the firm’s success over recent years. Process of selecting and adopting FA Information on the new machine tools was widely available in books, engineering journals and trade magazines. Suppliers were only too keen to visit potential buyers, and a great deal of information was provided by them. Most suppliers could be reached by phone and visits by salesmen or engineers were usual. Visits to other users of the equipment by Hydraul personnel were common and were normally organized by suppliers. The advantages of visits are that one can see the equipment in operation and can consult with other engineers about the problems being faced, and that they provide ideas as to alternative uses and ways of organizing the factory so that the new technologies’ use is optimized. Purchasing production equipment is similar to purchasing conventional machine tools. The technologies have been standardized to a great extent and their potential and main features are now known to production engineers. It was important for Hydraul, however, to have an exact idea of the products and materials that were going to be machined so that the precise machine configuration, ancillary equipment like tools, loading-unloading methods, fixtures, and the computer hardware and software were ordered. Perhaps the main difference with conventional machine tools, so far as the purchasing

54 CONCEPTS, METHOD AND SYNTHESIS

process was concerned, lay in the wider range of additional options confronting Hydraul. In the case of the UNIX-based industrial computer bought by Hydraul there was the view among some engineers in the company that it was mispecified, as not all its power would be used due to lack of appropriate software to link all existing computing applications. The problem was that the central computer and individual machine tools have their own operating systems which are incompatible with each other. The CAD was lying nearly idle for several months until some engineers decided to give it a try; and then it was used for some time alongside with manual drawings, rather than instead of them. This is now gradually changing but hand drawings are still used. Once there was a clear idea of what the new technology would be used for, selecting the specific equipment became the next problem. Essentially, comparisons were made in terms of the price, technical performance and options and a decision taken on the best combination. The second stage was simulating how selected makes of equipment would perform, on the basis of Hydraul’s own data on major parts already being machined in the factory, and comparing unit costs, cycle times and manufacturing times. Delivery times and the back-up service of suppliers were then considered. A process of ‘consultation’ was set in motion in which engineers and accountants from Hydraul and outside were involved. Some board members were also approached at this stage. Once the final decision was reached by management, an investment proposal which included the special installations that were required before the new equipment could be operated, was submitted to the board for approval. These special installations included purpose-built foundations, which were necessary because of the weight of the equipment. A draining system also was necessary to remove metal chips or swarf and coolants. Wiring for electrical installations and computer interfacing were needed, and special fuse boxes and power controls were required to avoid potential damage by variable electrical current. 5.3 Product scale and scope Setting-up times To address the changes in setting-up times it is necessary, first, to see how a product used to be machined and how this is done today. Gearbox casings, one of the products Hydraul has been manufacturing from its inception and which therefore has gone through the most changes of technology, will be used as an example. It is also a relatively complex part and so is a good illustration of some of the different operations through which workpieces can go. In the late 1970s the machining of gearbox casings required two vertical milling operations, two boring operations, three multi-spindle drilling and tapping operations, one radial drilling operation, one deburring operation and a final wash. Each of these operations was carried out in separate machines or

METHODOLOGY 55

required resetting and replacing tools in a machine that had been used for a different operation. Not all operations were necessarily sequential, although some were. In this machining process, setting-up times consisted of machine adjusting, tool changing, loading and unloading the workpiece, fixing the workpiece to the machine tool, measuring the workpiece and test cutting. As in most other cases, these procedures were done manually, there was a set-up time on each occasion the same machining operation was carried out, but the set-up time was progressively reduced as the number of equal operations performed grew, because of repetition. The cumulative setting-up time in the machining process of gearbox casings amounted to around 3,500 minutes for a batch size of 170–80, which represents 9 per cent of machining time. This percentage was not substantially lower when batches of 1,000 gearbox casings were produced, as a significant part of the setting-up time included loading and unloading which had to be done with any product. With new technologies, the setting-up time for the same number of gearbox casings was reduced to three hours or 2 per cent of machining time. The machining centres that form part of the FMS can mill, drill, tap and deburr without the workpiece having to be moved from the machine bed. There is still need to machine on two faces of the workpiece, as one side is always clamped to the fixture so that the workpiece has to be turned round for machining. Thus, the same piece is obtained after two machining operations and washing. Setting-up times in this case arise from fixing the tools in the machining centre, instructing the machine to perform the required operations, and some electronic measuring of castings. Loading and unloading workpieces to and from fixtures and pallets can now be done off-line, but as these operations still take considerable time they need to be matched with machining times to ensure a steady flow of workpieces. If there were an automatic tool changer allowing tool loading while machining other workpieces, setting-up times could be reduced even further. Batch sizes Estimating optimal batch sizes used to be very difficult at Hydraul. Not all machines had the same set-up time and this always complicated the determination of batch sizes for parts that required cutting in several machines. Most batch sizes are today around the 150–250 range (Table 2.3). A batch size of 200 is the rule of thumb planning size criterion for products manufactured for stock or when clients want monthly deliveries below that figure. This is a figure developed through experience and is one that eases production scheduling. It is roughly equivalent to a day’s work. For instance, in the case of gearboxes Hydraul delivers 40 a month to the customer but machines 170–80. Hydraul starts production of gearbox casings whenever capacity utilization is low. There were still minimum batch sizes. These were given, for instance, by the size of the fixture. If only one small part with a low cycle time was to be machined it would not be done. CNC machine tools are better utilized if the fixture

56 CONCEPTS, METHOD AND SYNTHESIS

Table 2.3 Hydraul: batch-size composition (%), before and after

Source: Interview data.

is full of parts so that the indexing probe, which checks the position of each casting in the fixture prior to machining, does not waste time indexing empty spaces. For small parts this can mean batches of 20–30 units. It must be pointed out that while FA has reduced setting-up times, so allowing smaller batch sizes, there was, for Hydraul, another important factor which had a bearing on batch sizes: the size of orders. Given that Hydraul manufactures a diversity of components to order, changes in batch size through time have been determined also by changes in the requirements of clients. Clients are themselves trying to reduce inventory and moving to just-in-time production—while asking for shorter, and sometimes very tight, delivery times—so there has been a batch size-reducing impact due to this factor as well. Hence, at Hydraul batch sizes were the result not only of calculation on economic batch quantities but also of market demand. Product variety Product variety measured both by product family and by number of machined parts, decreased steadily until the mid-1980s following Hydraul’s policy of concentrating on high-volume products about which Hydraul could provide the detailed knowledge the customer wanted. Since the arrival of the new technologies it has been possible to reverse this earlier trend and manufacture new families of products although still within Hydraul’s main area of specialization— hydraulic valves and gear boxes (Table 2.4). Indeed all the additional output in volume between 1985 and 1992 was the result of 15 new families of products. Profits during this period increased from around £100,000 to around £800,000, or from 0.3 per cent to 1.3 per cent of total sales. FA has been crucial in allowing more product variety. Apart from leading to a substantial advance in the quality of products, by easing the machining of complex shapes and angles FA has allowed Hydraul to experiment more with product design and manufacture, so that it has been possible consistently to introduce minor product changes and new product families. For instance, the new technologies permit cutting very accurate small notches in some valve components that allow for a far more precise control of the flow rate, pressure and direction of fluid. These notches are extremely awkward to machine because they are not completely round, but need a special shape and form. Several of them are normally required. The new lathes, with their capability to position

METHODOLOGY 57

Table 2.4 Hydraul: product diversity, before and after

Source: Interview data. Notes a Total number of items, i.e. equal and different parts machined. b All types of part machined. Excludes differences arising from size. c All main types of assembled final product, like different series of directional control valves; relief, lock and check valves; and, gearboxes. d All varieties potentially available, including differences in control mechanism, size, flow and pressure rate, and material.

workpieces in different angles and to perform secondary drilling and milling operations, allow for this. Another major innovation has been the fitting of valves with electronic attachments so that they allow for remote control. These minor modifications to products can bestow an important competitive advantage because of the very specific requirements of clients in the engineering industry. While it may have been technically possible to introduce these modifications with old technologies, the amount of machining time and the testing effort required would have made them unfeasible. Customers simply would not have asked for some of the specifications they request today, nor would Hydraul have offered them. Today, Hydraul has many more options available to it, but always tries to keep within its main area of general strength. 5.4 Plant scale and unit costs Changes in plant scale Output volume, measured by the tonnage of machined castings, has increased by 167 per cent between 1985 and 1992. Measured by number of parts machined — i.e. the total number of individual items—output volume increased by 192 per cent over the same period. Plant sales, however, have only doubled since 1985. The increases in volume and the improved quality of Hydraul products notwithstanding, prices in the engineering industry have fallen dramatically. Lower prices have resulted from increases in labour productivity and machine efficiency, more intense competition as the new machines are becoming more widely available in the market, and a greater awareness by large customers of the cost structure that

58 CONCEPTS, METHOD AND SYNTHESIS

small-component firms like Hydraul face (Hydraul has an ‘open-book’ policy with major customers, all it asks from them is a ‘decent profit’). Factors affecting changes in scale Prior to the introduction of the new technologies Hydraul worked on average two shifts per week. Running a third shift was difficult to organize and meant higher labour costs due to extra allowances for ‘unsociable hours’. Depending on the cycle times of individual pieces and the number of pallets available, the new machines can be operated unmanned for at least one shift as the workpieces can be left clamped to the fixtures and programmed for automatic feeding. Given the workload at Hydraul, it pays to organize shifts on a 10-hour-day, 4-days-on, 4days-off basis rather than the traditional daily 8-hour shifts over a five-day week. This shift pattern also permits use of the machines unmanned in spells of three hours and leaving one hour per day for cleaning and daily maintenance of the machines. Machining speeds also have increased. In the case of gearbox casings, two types were machined before FA, one requiring 180 minutes accumulated machining or cycle time, the other needing 240 minutes. With the new technologies total machining time is 52 and 72 minutes, respectively, for these gearbox casings. Considering that not all pieces go through the same process and that there are far less machines involved in production, average processing time for the plant as a whole today must be around two-thirds of what it was before the new technologies arrived (Table 2.5). Another factor enabling higher scales was production management and engineering. Take the case of gearbox casings manufacturing. Before the new technologies were introduced planning and scheduling the manufacturing of gearbox casings involved selecting the individual machines, the order in which machines were to be used and routeing the parts through the machines. With a single part, this task seems relatively easy as parts may follow an orderly sequence from one machine to another. But even in this case choices arise depending on the cycle time for each individual operation and the number of machines available. If one considers that the number of machining operations could reach 15–20, that not only one part is being machined but up to several hundred, that some machines may face higher demand than others, that machines break down, that machining of individual workpieces has to be co-ordinated with others that are later joined together in assembly, that certain operations may require skills possessed by only some operators, that raw material delivery has also to be scheduled, and, that some customers have priority over others, the solution to the planning and scheduling problem involved many combinations and was extremely complex. The result was a very disorganized and inefficient factory: machine utilization ratios reached 60 per cent on average, including time lost in setting-up; operators spent considerable time away from their machines doing nothing; the factory was full of unfinished products; materials were wasted; and no one knew where to find the right tools. All of these elements were reflected in Hydraul’s lead times

METHODOLOGY 59

Table 2.5 Hydraul: sources of scale increases under old and new technology

Source: Own calculations on the basis of interview data. Notes a On the basis actual hours worked in the year. b Output per hour if machines were operating continuously. c Hours’ machines are actually operating, i.e. ‘real’ capacity utilization excluding machine breakdown, unavailability of material and/or operators, setting-up times and other similar factors. d Machine efficiency without considering time lost due to setting-up times. e Share of total time dedicated to setting-up, on the basis of estimates for gearbox casings. f Some small differences due to rounding.

(i.e. the time between receipt of an order and being ready to deliver it), for gearbox casings of thirteen weeks, and even longer for the completely assembled gearbox. With the new technologies problems of production management and engineering have eased significantly. There are far fewer machines to worry about, and the more powerful computers and software allow better co-ordination of the process. Machine utilization ratios have increased to 85 per cent. There is still the need to solve the old problems of scheduling, routeing and machine use, and new ones like tool management, as tools are becoming a significant part of costs, but the complexity of this task has been reduced. Improvements are shown in the shopfloor today as the process is much more orderly and clean, with less work-in-progress and fewer final products’ inventories, and with workers spending more of their time on the machines. In gearbox casings, for instance, lead times have fallen from 13 weeks to 2 days, excluding heating time. Changes in unit cost Unit costs have fallen substantially as can be seen in Tables 2.6 and 2.7. On average they have fallen by 32.8 per cent, mainly because of the increasing output. The largest drops were in labour unit costs, and in raw materials and energy. By contrast capital unit costs increased by around 20 per cent during the same period.

60 CONCEPTS, METHOD AND SYNTHESIS

Table 2.6 Hydraul: costs and unit costsa

Source: Interview data. Note a Exchange rate 1985: £1=US$ 1.3; 1992: £1=US$ 1.76. Table 2.7 Hydraul: unit cost structure

Source: Interview data.

Factors affecting changes in unit costs A major factor affecting production costs has been the much higher cost of the new machine tools. Since 1986, Hydraul has been investing, on average, three-tofour times a year what it was allocating to capital expenditure per annum in each of the previous five years. In part this is due to the more expensive machines Hydraul is buying. A Cincinnati Milacron vertical milling centre cost £4,675 in 1972 while today the same machine costs around £15,000. Even if one assumes that 1 new horizontal machining centre replaces about 10 vertical milling centres or a combination of 10 milling and drilling machines, the new FMT machine, costing £300,000, is still double the cost of the old machines at today’s price. Despite the much higher level of output of the new equipment, capital equipment unit costs have increased by 34 per cent (Table 2.7).

METHODOLOGY 61

Table 2.8 Hydraul: main capital investments

Source: Interview data.

Hydraul has saved on building space as the fewer new machines require less area, but as the plant had been built for the previous number of machines the result has been space available for future expansion. This space also allows Hydraul to retain older machines for use in emergencies, whereas this would not be possible had a smaller new plant been built. Hydraul has, however, engaged in replacing its company car fleet, so there have been additional costs. In unit terms, however, building and vehicle costs have fallen by 19 per cent. Labour total costs (Table 2.9) have increased due to higher costs of employment, particularly in respect of wages and training. Despite automation, manufacturing employment, including shopfloor-related overheads like production engineers and logistics, rose by 10 per cent because of the increase in orders between 1985 and 1992. Wages increased nearly in line with inflation and, although there are no separate figures for training, Hydraul has been spending more on training than ever before. Operation of the new equipment is easy to learn because of increasingly user-friendly software. Anyone could learn to operate the machines after a few hours’ training, but operating them at their best is more difficult and not even Hydraul’s best machinist has been able to make the transition, as it requires good mechanical and electrical knowledge combined with programming and data-analysis capability. Thus, a great deal of on-the-job training, as well as sending some workers to technical school, has become crucial. FA has, nonetheless, increased productivity to the extent that labour unit costs have fallen by more than half (Table 2.10). To the extent that output/equipment plus labour costs can be seen as a proxy for capital and labour productivity, there was a rise of 58 per cent on this account, too.

62 CONCEPTS, METHOD AND SYNTHESIS

Table 2.9 Hydraul: employment (total of workers)

Source: Interview data. Note na=not available Table 2.10 Hydraul: productivity changesa

Source: Own calculations on the basis of interview data. Notes a Only manufacturing employment and investment; excludes overheads. b On the basis of actual hours worked in the year.

Raw materials unit costs have fallen sharply, by 39 per cent. A major factor in this case has been the better utilization of castings. In some cases waste—mainly swarf and metal offcuts—has been reduced from 60 per cent to 1 or 2 per cent because of more precise machining and by using offcuts for other parts. On average the new machines have allowed Hydraul to save between 10–15 per cent in raw material use. Swarf now is also sold at a better price as it is much easier to discriminate between aluminium, iron and steel waste. Aluminium, iron and steel casting prices have been increasing far less than inflation, thus also contributing to lower raw-material unit costs. Inventories of raw materials, work-in-progress and finished goods were also reduced gradually over the years. At one stage in the early 1980s Hydraul had around 25 per cent of sales tied in inventory but this figure has now reduced to 2 per cent. There is little room for improvement in this respect. Consumables like tools are gaining significance as cost items, although in unit terms their costs have remained unchanged. With machines cutting metal much faster, tool life-cycles are increasingly being reduced. Despite the increasing availability of cheap tools in the market, Hydraul has a policy of buying the more expensive, quality, tools. Tool management has become important, and

METHODOLOGY 63

considerable time and effort are spent now in choosing the right type of tool for an operation so that its life can be extended as far as possible and the cost kept under control. Coolants required for the new machines are also more sophisticated and expensive. Energy and repair unit costs also have fallen significantly. Energy unit costs are lower due to the diminished energy requirement of the AC and DC motors driving the spindles of lathes and machining centres. Repair unit costs are lower because the equipment is relatively new and because the major cause of failure with the new technologies is human error, and this has been considerably reduced by the electronic diagnostics of the machines and the better training of personnel. However, FA requires constant preventive maintenance because of the many components involved, many more than in the old equipment, which Hydraul does not itself provide and for which may pay a price in the long run. Overhead unit costs have increased by 21 per cent mainly due to higher marketing and administration costs. By the mid-1980s the accounting system was inadequate to cope with the increasing information and sales being handled, and there was a need to renovate accounting procedures. Since that time, Hydraul has worked to improve its accounting department, replacing most of the accountants with information technology-based accounting experts. Marketing unit costs have also increased due to the hiring of new engineers to handle some of the new large customers who need a specialized approach. Research and development unit costs have not changed significantly, although it is difficult to distinguish some design costs from marketing costs as sales engineers design and design engineers sell. 6 Implementation of the study Implementation of the research was done in three phases. The first phase had the objective of providing background knowledge for the study. Four main tasks were performed during this phase. The first task included reviewing available academic literature and examining pertinent industry journals. Industry journals proved an invaluable source of technical and economic information. These journals, which are oriented mainly to engineers, technicians and production managers, provided technical information on characteristics and performance of individual CNC machine tools, robots, CAD, FMS and ancillary technologies, and on NPD. They provided information on use of specific technologies including main technical and economic advantages and disadvantages encountered. They also have articles based on interviews with user firms about their experiences with FA. Indeed, one of the firms interviewed in one of these journals agreed to be part of this study. Finally, they provided statistical information, annual surveys of use of FA and basic data on main equipment suppliers. A number of industry journals were consulted including: American Machinist (USA); Metalworking Production (UK); Machinery and Production Engineering (UK); European Machining (UK); Metalworking Engineering and Marketing (Japan); Machine Production (France); Machinery Magazine

64 CONCEPTS, METHOD AND SYNTHESIS

(Taiwan); and Información de Máquinas-Herramienta, Equipos y Accesorios (Machine Tools, Equipment and Attachments Information) (Spain). The second task involved contacting fellow researchers and academics, mechanical engineering and management departments, manufacturing technology laboratories and research centres specializing in FA. Such contacts provided information on the changing technological conditions and key technical problems faced by the engineering industry, on technological trends in flexible automation and how they relate to organizational techniques, on issues of management of FA, on recent research on the technical and economic impact of FA, and on the actual and potential for use of FA in developing countries. Institutions approached included: the Department of Mechanical Engineering and the Department of Management at Brighton University, at the University of Manchester and at Imperial College London, all in the UK; the University of Winsconsin at Madison, the National Institute for Standards and Technology (NIST), the National Academy of Engineering (NAE), the National Science Foundation (NSF), Office of Technology Assessment (OTA) and RAND’s Critical Technologies Institute, all in the USA; the Netherlands’ Organisatie Voor Toegepast-Natuurwetenschappelijk Onderzoek, Institute for Produktie en Logistiek (Organization for Applied Scientific Research, Institute for Production and Logistics); India’s Confederation of Indian Industry Technical Department and the Automotive Components Manufacturing Association, Centre for Technology (ACMA/ACT), Mexico’s Centro de Innovación Tecnológica at the Universidad Autónoma de Mexico (Center for Technological Innovation/ Autonomous University of Mexico) and Centro de Tecnología Electrónica e Informática (Centre for Electronic Technology and Informatics); and Venezuela’s Instituto de Ingeniería (Engineering Institute). The Hanover European Machine Tool Exhibition (EMO), was also visited at this stage. The third task was approaching suppliers of CNC equipment and industrial controls for information on key markets in developing countries, NPD and product prices. Suppliers, particularly those operating worldwide, provided invaluable information as to which companies were their most important individual customers in specific countries, arranged study visits to some of their factories and customers both in developed and developing countries, and contacted their local agents and representatives to provide information about local customers. The initial contact resulting in the case study of the previous section was made by the supplier of the equipment. Lists of potential interviewees in each of the selected countries were made on the basis of the information obtained from equipment suppliers, both at their headquarters and locally, and a large number of firms in the sample actually have CNC machine tools from the same maker with similar technical characteristics, increasing the degree of comparability between firms domestically and internationally. Two suppliers, a major Japanese machine-tool producer and a large European industrial controls’ manufacturer, were of great help in this effort, although advice was sought also from European medium-and small-sized manufacturers of tools.

METHODOLOGY 65

The fourth task, which involved a small pilot study, was performed in developed-country firms and was aimed at gaining more precise insights into the issues and the data available, and as a means of testing and improving the questionnaire. Twelve developed-country firms in the Netherlands, Switzerland and the UK were interviewed, including firms in textiles, printing and, above all, engineering (automobile and components and general equipment), as these three industries share technologies such as CAD/CAM and process-control computers. In engineering, apart from Hydraul and a UK manufacturer of book-binding equipment, several automotive component producers and two automobile plants were visited in the Belgian and Dutch Limburg region (see Alcorta 1993a). The second phase of the study consisted of identifying the consultants, identifying the firms and conducting interviews in selected countries. Because of the need to identify firms where there was some certainty that CNC machine tools were being used, it was crucial to approach researchers with a deep knowledge of and extensive contacts in the engineering industry. Thus, researchers who were invited to participate had a track record on studying the engineering industry or were consultant to firms in their respective countries. According to Wangwe (1995), the use of experienced researchers in case studybased research leads to inferences and conclusions applicable to the whole of the industry, not only to what can be gleaned from the specific case studies, hence reducing the risk of generalization inherent in the case-study approach. Moving on to the identification of the firms themselves, key roles were performed by equipment suppliers and, in particular during this phase, by local agents and representatives. Several of them provided local researchers with lists of main customers, addresses and contact persons, as well as information on type and quantity of CNC machines purchased. In addition, individual researchers requested information and statistics from local departments of engineering, manufacturing and management; from industry associations; from relevant government offices; and, in the cases of Brazil and India where there is local production of CNC machine tools, from machine-tool industry associations. Altogether, around 200 firms in the six countries were identified as targetable. Of these, around 120 were approached and, as was said, 62 agreed to be interviewed. It also turned out that, through this approach, we were able to engage firms accounting for between 1 per cent and 7 per cent of the estimated total stock of CNC machine tools in the selected countries. In order to have a sample as comparable as possible within and between countries, researchers were provided with the model case study and the questionnaire as guidelines. In addition, intensive communication took place between the research co-ordination function and the individual researchers, and between the latter and the targeted firms, to see whether specific firms wished to be part of the research. After successive approximations a ‘final list’ was decided, although there were some last minute changes as firms that had agreed to participate withdrew or found that they had different technologies from those that were being surveyed. It was during this process that it became obvious that a third industry —manufacturing of dyes and tools—should be included as there were more firms of this type than expected.

66 CONCEPTS, METHOD AND SYNTHESIS

During the interviewing stage there was a period of intense consultation and interaction between the research coordination function, individual researchers and firms. There were issues of clarification on the questions and the quantitative indicators that had to be constantly addressed. There were some preliminary results that had to be checked in the light of what was happening in other similar enterprises; when differences emerged, it was necessary to consider whether the wrong choice of firm or a real phenomenon accounted for them. Also at this stage, there was a need to replace some firms that had promised to provide certain information but in the event could not. It was likewise apparent that the degree of precision and comparability achieved through this process would not have been possible by means of a survey, further strengthening the argument for the approach selected and followed in this research. The third phase involved the presentation and discussion of the preliminary results of individual country studies at a workshop held in Maastricht in September 1994. The workshop was attended by developed- and developingcountry economists, managers, policy makers, technologists and international experts in the field. They reviewed the country studies and made suggestions to clarify and improve individual studies and the presentation of results, to check and complement results with additional information and about the overall conclusions of the project. Comments arising from the discussions were incorporated in the final versions of the respective country studies. Starting in 1995, preliminary overall results have been presented and discussed at several academic conferences in the Netherlands, the US, Chile, Turkey and India. Suggestions arising in these conferences also have been incorporated in the final draft. Notes 1 In operational terms engineering, mechanical engineering or metal working, as they will be referred to interchangeably throughout the text, are defined to include the International Standard Industrial Classification (ISIC) two-digit Fabricated Metal Products, Machinery and Equipment (38) or three-digit Fabricated Metal Products (381), Machinery Except Electric (382), Electrical Machinery (383), Transport Equipment (384) and Precision Instruments (385). 2 A survey of machine tool use in US engineering industry sectors in 1983, summarized in (AMT) Association for Manufacturing Technology (1989), showed that the motor vehicles and parts sector and the general machinery and equipment sectors accounted for 4 per cent and 5.7 per cent, respectively, of the total stock of numerically controlled, mostly CNC, machine tools in manufacturing industry. 3 Lathes are used in turning operations which essentially consist of removing material by rotating the workpiece, mainly of cylindrical form, against a single-point cutter. Drilling produces a hole in an object by forcing a rotating cutting tool or drill, normally a twist drill, against it. Boring is enlarging a hole already drilled. Milling consists of removing metal by feeding the workpiece against a rotating tool or cutter. The tool has multiple cutting edges and the axis of rotation of the cutting tool is perpendicular to the direction of the feed. Other commonly used conventional

METHODOLOGY 67

4

5 6

7

machine tools include grinding machines which are used for material removal through the application of abrasive particles contained in a bonded wheel that operates at high speeds. The US’s NMTBA—The Association for Manufacturing Technology, previously the National Machine Tool Builders’ Association—also collects similar data and for a far larger number of countries, but CECIMO’s data are more orderly and more consistently processed. For confidentiality purposes the names of firms throughout this research will be changed to some indication of their product range. Today, RCG firms are involved in contracting and construction, engineering, construction materials, property development and car dealership. The group has interests in the UK, the US and Australia. Although it was also clear that returning to the degree of the product diversity that had led to the previous two restructuring crises was not desirable.

3 THE DIFFUSION OF FLEXIBLE AUTOMATION IN DEVELOPING COUNTRIES

1 Introduction This chapter examines the extent of the diffusion of FA and identifies the main reasons for the use of FA and associated organizational techniques, particularly in the case of selected developing countries. Diffusion refers to the aggregate results of activities through which new products and processes spread across potential markets. It involves choices, simultaneous interactions and outcomes between suppliers and users which, in turn, are influenced by technological, economic, institutional and individual considerations (Hall 1994; Karshenas and Stoneman 1995; Mansell and Steinmueller 1996; Stoneman 1995). Diffusion processes may take place locally or internationally, within one firm or industry, in the case of specific technologies or across industries, when generic technologies are involved. By bringing new products and processes to users, the diffusion process provides users with embodied and disembodied knowledge and skills previously unavailable to them. The extent of the diffusion of FA and, more specifically, of CNC metal-cutting machine tools will be measured at two levels: globally, on the basis of international data on production, trade and consumption of CNC and non-CNC metal-cutting machine tools such as lathes and machining centres; and, for the six countries under study, at both the national and the firm level, on the basis of data on use of CNC and non-CNC metal-cutting machine tools and related technologies provided by international, business and government sources, by trading partners and by the firms themselves. As far as organizational techniques are concerned, data on diffusion will be presented at the firm level only as no figures are available on their international diffusion. Although the concepts underlying the new organizational techniques are widely diffused, their application is often firm-specific, difficult to observe and to establish borders and nearly impossible to cost fully, making it extremely difficult to produce measurable indicators that can be aggregated (Alänge et al. 1995). The chapter will proceed as follows. Section 2 will analyse the international diffusion of metal-cutting and CNC machine tools. Section 3 will examine the extent of diffusion of CNC machine tools in the six countries under study: Brazil, India, Mexico, Thailand, Turkey and Venezuela. Section 4 will examine

DIFFUSION OF FA IN DEVELOPING COUNTRIES 69

the diffusion of FA and, specifically, of CNC machine tools at firm level. It begins by analysing the state of mechanization and automation in the engineering industry of the countries under study prior to the adoption of FA, and continues with an investigation into the diffusion of CNC machine tools in firms classified by degree of automation, size, type of ownership and industry. Section 5 looks into the diffusion of organizational techniques and examines the question of the complementarities between them and automation. Section 6 focuses on the macroeconomic and microeconomic factors accounting for the diffusion of CNC machine tools and associated organizational techniques. Finally, section 7 examines the process of assessment and assimilation of FA by the firm. 2 International diffusion of metal-cutting machine tools 2.1 Diffusion of metal-cutting machine tools Consumption of machine tools in value terms has been increasing at an annual average of 5.7% since 1973 (see Table 3.1). Consumption of machine tools peaked at around US$42 billion in 1990 and declined thereafter following reductions in demand in Western Europe and developed Asia, especially in Germany and Japan, and the collapse of Eastern European economies. North America’s consumption growth rate, mainly the result of US purchases, was the fastest among developed countries and amounted to an annual average of 8.8% over 1973–94. The use of machine tools is concentrated in developed countries, which accounted for 64.6 per cent of total world use in 1994, around the same share they accounted for in 1973. Demand in Asian developing countries, particularly China, Korea and Thailand, has been expanding even more rapidly at an annual average of around 20.7%. By 1994, developing Asia’s consumption levels had nearly caught up with those of Western Europe and exceeded those of North America and Japan. That same year China had become the second-largest machine-tool consumer in the world, accounting for 12.7 per cent of total world consumption (see Appendix A19). Machine-tool consumption growth rates were, nonetheless, far lower in other parts of the developing world. Eastern European, Latin American and African consumption growth rates were below the world average over 1973– 94 and have been mostly negative since the early 1980s. Local production of machine tools is usually a key factor in domestic diffusion because diffusion often depends on continuous improvements in the technical and performance characteristics of the equipment, in the adaptation and modification of equipment to the demands of specific users and in the introduction of complementary devices (Rosenberg 1982). While adaptation is not normally necessary when conventional machine tools are in use, it is a must in the case of specialized equipment, transfer lines or advanced CNC metalcutting machine tools and FMS. It is also sometimes necessary with the simplest

70 CONCEPTS, METHOD AND SYNTHESIS

and more standard CNC machine tools, as the software, jigs and fixtures and tools may need to be adapted to the needs of the purchaser. Further, the environment in which new machine tools are to operate may have to be adapted to accommodate them, and a new set of management, organizational or working skills may have to be developed. Although adaptation can be undertaken internally, particularly if there are technological capabilities available within the user, it is more effective if it is done in collaboration with the producer of the equipment, as both user and producer possess unique information and knowledge on technological opportunities and user needs. User and producer knowledge, when brought closely together, can enter a process of cumulative learning and further knowledge creation, leading to improved performance in both of them (Andersen 1992; Lundvall 1992). Indeed, in countries closed to trade or where trade is severely limited, local production of machine tools and close interaction between users and manufacturers becomes the only means for their successful diffusion. Between 1973 and 1994 production of metal cutting machine tools worldwide grew in value terms at an annual average rate of 5.5 per cent (see Table 3.2). Developed and developing countries’ machine-tool production rates were higher than the average, compensating for the relatively low growth rates of East European countries. Developed countries’ high growth rates were the result in part of the emergence of a highly competitive machine-tool industry in Japan, which more than compensated for the slowdown in US production. By 1994 Japan had increased its share of world output to 27 per cent from 13.9 per cent in 1973, while the US’s share had fallen from 14.1 per cent in 1973 to 7 per cent in 1990, recovering to 13.1 per cent in 1994. Between 1973 and 1994 Germany’s share of output also fell from 18.1 per cent to 16.3 per cent (see Appendix A2). Developing countries’ production growth rate for metal-cutting machine tools is clearly impressive. Between 1973 and 1994 output growth more than doubled the world rate at an annual average of 14.5 per cent (see Table 3.2). Production is, nonetheless, heavily concentrated in Asia, in particular in China, South Korea and Taiwan, as Brazil and India, the two other large metal-cutting machine-tool producers in the developing world, account today for a small proportion of total production (see Appendix A2). Between 1973 and 1994, Taiwan’s share of world metal-cutting machine-tool production rose from 0.2 per cent to 3.8 per cent. In the case of South Korea, its share of world output increased from 0.1 per cent in 1976 to 3.5 per cent in 1994. But, more importantly, China increased its metalcutting machine-tool output from 1.2 per cent in 1973 to 5.1 per cent in 1994. China is now competing with Switzerland to become the world’s fifth-largest producer of metal-cutting machine tools. Production of metal-cutting machine tools in countries such as Argentina, Mexico and South Africa has all but disappeared. Diffusion processes often require imports that complement local production or substitute for the lack of it. While imports do not necessarily result in the cumulative joint-learning process between users and producers that arises from domestic production, as very often transactions are undertaken at arm’s length and mediated by distributors or local agencies with limited information, they may provide users with alternatives, in terms of technical features, models, sizes

Source: Our own elaboration on the basis of American Machinist data. Notes a Includes metal-cutting and metal-forming machine tools. Metal-forming machine tools shape metal by pressing, forging, hammering, extruding, shearing, bending or die casting. Metal-forming machine tools accounted for around 25% of total machine-tool output over 1973–94. See Table A1 in Appendices for the yearly share of metal-forming and metalcutting in total machine-tool production. b Includes former USSR/Russia.

Table 3.1 Value of world machine-tool consumption,a 1974–94

DIFFUSION OF FA IN DEVELOPING COUNTRIES 71

Source: Own elaboration on the basis of American Machinist data. Notes a Includes former USSR/Russia.

Table 3.2 Value of world metal-cutting machine-tool production, 1973–94

72 CONCEPTS, METHOD AND SYNTHESIS

DIFFUSION OF FA IN DEVELOPING COUNTRIES 73

Table 3.3 Metal-cutting machine-tool net imports,d 1978–94 (US$m.)

Source: Our own elaboration on the basis of United Nations (1996) and official statistics. Notes a Includes Mexico. b For 1994 includes Azerbaijan, Armenia, Georgia, Kazakhstan, Kyrgystan, Tajikistan, Turkmenistan and Uzbekistan. c For 1994 includes Belaruss, Estonia, Latvia, Lithuania, Moldova and Ukraine. d SITC 736.1, Rev. 2.

and prices, that are unavailable locally. Even more so if local users have already accumulated technological capabilities and there is a local capital goods industry that can benefit from the inflow of knowledge from other countries. Where there is no local capital goods industry, imports of CNC machine tools are the only source of supply of the new technology. Table 3.3 shows the world net imports of metal-cutting machine tools by country groupings and regions between 1978 and 1994 (see Appendices A9–A12 for full import and export figures for selected countries and products). Net imports are the difference between CIF imports and FOB exports and show the net increase (or decrease) in any grouping or region’s supply of metal-cutting machine tools. The ‘economies in transition’ grouping includes East European countries, the former USSR and more recently Russia.1 Although as a whole developed countries are, and have been for some time, net exporters of metal-cutting machine tools, there are also some significant importers among them, particularly in North America. The US, which until the mid-1970s had been a major world supplier of machine tools, became by 1978 a

74 CONCEPTS, METHOD AND SYNTHESIS

net importer of metal-cutting machine tools. Moreover, between 1978 and 1990, its net imports increased fourfold to US$1.2 billion, falling back to US$550 million in 1994 (see Table 3.3). By 1994, US imports of machine tools accounted for 18.6 per cent of the value of world metal-cutting machine-tool imports. Imports of cheaper, lower performance Japanese and Taiwanese machine tools, aimed at the small-scale firm market, underscored this trend. Other major importers of machine tools within the developed world are Germany, Italy and France, although the first two countries have a largely positive trade balance in machine tools while the latter does not (see Appendices A7 and A8). As might be expected, diffusion of metal-cutting machine tools in most developing countries has been the result mainly of imports. The most significant exception is Taiwan, which is by far a net exporter. Although there have been some regional variations, developing countries as a whole increased their net imports of metal-cutting machine tools nearly sixfold, from US$0.4 billion in 1978 to US$2.2 billion in 1994. However, China, South Korea and, increasingly, Thailand and Mexico, account for the bulk of developing countries’ machine tool net imports (see Appendix A7). Diffusion in economies in transition also has been the result of metal-cutting machine tool imports, although, as was said, figures should be taken carefully as they exclude a non-identifiable proportion of intra-socialist countries’ trade. Like African and Latin American developing countries, East European and former USSR net imports declined consistently for around ten years beginning in 1978, recovered significantly during the late 1980s only to plummet in the early 1990s. Imports of machine tools in former socialist countries accounted in 1994 for only 2 per cent of the world total. Significantly reduced imports together with the breakdown of the local machine-tool industry meant that by 1994 former socialist countries were consuming around half the amount of machine tools they had consumed in 1973. Rapid diffusion of metal-cutting machine tools in developing countries, particularly in China, South Korea and Taiwan, has been the result of supply and demand factors (Amsden 1977, 1985; Cheng 1972; Chudnovsky and Nagao 1983; Fransman 1986a; Gu 1994a, 1995; Jacobsson 1986; Jacobsson and Alam 1994; Jain et al. 1995; UNCTAD 1985a; UNIDO 1991e). On the supply side, as had been the case with Japan a few years earlier, the three countries entered manufacturing at the low end of the market, concentrating on the production of cheap, initially conventional, machine tools to be sold in their protected local markets. At varying speeds they then moved up the ladder of scale and technological sophistication. In the cases of China and South Korea, local markets were potentially very large and were eventually fully exploited, but in the case of Taiwan its local market was small, forcing local producers to move very quickly into exports. By 1980, Taiwan also had made the transition from catering for low-income South-East Asian markets to selling in the more demanding and high-income US market. The three countries increasingly used foreign licensing and foreign technical assistance as the main mechanisms to obtain product and process technology. Korean and Taiwanese firms kept abreast of technological developments in more

DIFFUSION OF FA IN DEVELOPING COUNTRIES 75

advanced countries and early on and rapidly complemented foreign technology with copying, reverse engineering and product adaptation to reduce costs and prices. They also sought ways of increasing the efficiency of their production processes through incremental changes, reaping scale economies and subcontracting, so that they or their users could successfully compete abroad. Chinese firms initially obtained technology from the Soviet Union, but as relations soured, and after an interlude of ‘de-linking’ during the Cultural Revolution, they began to turn, slowly, to Japanese, US or European technology. Chinese firms also tried to assimilate and adapt foreign technology, but excessive centralization in decision making, high degrees of vertical integration and relatively weak links with undemanding users meant that technological upgrading was far slower in China than in the other two countries. Finally, Chinese, South Korean and Taiwanese firms received support from the state, although the extent and the mechanisms used varied. Chinese firms were directly controlled by the Ministry of the Machinery Industry and were supported by a number of government research institutes. South Korean firms were helped by quantitative import quotas, preferential loans to machine-tool producers and tax and credit incentives for users. Taiwanese firms were assisted with high tariff barriers to imports and import licences, export subsidies and credits, research and technical support from partially government-funded institutes and through the provision of risk capital. On the demand side, China’s growing machine-tool consumption is the result of its own efforts, since the early 1980s, to modernize its agricultural, petrochemical, heavy equipment and electrical equipment industries (particularly, hydraulic and thermal powerstations, motors and electricity transforming and transmitting devices), and to develop the car, telecommunications equipment and electronic components industries. These industries are large users of machine tools (Jain et al. 1995; Gu 1995). South Korea’s demand arises from well-established and internationally competitive vehicle, machine tools and electronic components and accesories industries and a relatively new commercial aircraft manufacturing industry, all of which are major users of CNC machine tools. 2.2 Diffusion of numerically and computer-numericallycontrolled machine tools2 While the diffusion of locally produced metal-cutting machine tools in developing countries has been rapid it has not been accompanied by an equally intense diffusion of domestic CNC machine tools in most of them, the only exceptions being South Korea and Taiwan. Germany’s share of CNC machinetool output in total machine tool output by value increased from 18.8 per cent in 1980 to 78.6 per cent in 1994 (see Table 3.5). The US’s equivalent ratio increased from 33.9 per cent in 1980 to 65.3 per cent in 1994 while Japan’s ratio grew from 49.7 per cent in 1980 to 83.8 per cent in 1994. Between 1980 and 1994 Japan had manufactured over 660,000 units of CNC metal-cutting machine tools, three times the number produced in Germany and six times the

Source: Own elaboration on the basis of data provided by CECIMO. Note a Includes not specified.

Table 3.4 CNC and non-CNCa metal-cutting machine-tool production, 1980–94 (in units)

76 CONCEPTS, METHOD AND SYNTHESIS

Source: Own elaboration on the basis of data provided by CECIMO. Note a Includes not specified.

Table 3.5 CNC and non-CNCa metal-cutting machine-tool production, 1980–94 (US$ m.)

DIFFUSION OF FA IN DEVELOPING COUNTRIES 77

Source: Our own elaboration on the basis of United Nations (1996) and official statistics. Notes a Includes Mexico. b For 1994 includes Azerbaijan, Armenia, Georgia, Kazakhstan, Kyrgystan, Tajikistan, Turkmenistan and Uzbekistan. c For 1994 includes Belarus, Estonia, Latvia, Lithuania, Moldova and Ukraine. d SITC 731.21, 731.31, 731.35, 731.42, 731.44, 731.51, 731.53, 731.61, 731.63, 731.65, all Rev. 3.

Table 3.6 Total CNC machine-tool net imports,d 1989–94 (US$ m.)

78 CONCEPTS, METHOD AND SYNTHESIS

DIFFUSION OF FA IN DEVELOPING COUNTRIES 79

number manufactured in the US (see Table 3.4). In India, only 49.7 per cent of the total value of machine-tool production was accounted for by CNCs in 1994. In Brazil, the equivalent ratio was 35 per cent at the peak of machine-tool demand in the late 1980s (Vermulm 1994). There are no equivalent figures for China, but according to Jain et al. (1995), China was producing 172,200 metalcutting machine-tool units in 1987, of which 2,604 units were CNC. In South Korea and Taiwan, however, the production value ratio of CNC machine tools to total metal-cutting machine tools is approaching that of the US. Turning to imports and consumption of CNC machine tools, there has been a clear shift away from non-CNC machine tools both in developed and developing countries, although the trend is much more pronounced in the former. The fact that, for instance, during the early 1990s around 80 per cent of total lathe imports by developed countries were CNC lathes suggests that most of these countries have not only been producing more CNC metal-cutting machine tools but have also themselves been consuming more of them (see Table 3.6 and Appendices A14, A16 and A18).3 In the case of the US for instance, the share of CNC metalcutting machine-tool imports in total metal-cutting machine-tool imports in 1990 and 1994 amounted to 74.4 per cent and 85.1 per cent, respectively. Considering the value of locally produced CNC machine tools and the ratios of imported and exported CNC machine tools to total machine tools, total apparent consumption —i.e. excluding stocks—of CNC machine tools was around 53.5 per cent of total machine-tool consumption in the early 1990s. In Japan the equivalent figure was 57.8 per cent in 1992 (Metalworking Engineering and Marketing 1993). Unlike in developed countries, the diffusion of imported metal-cutting machine tools in developing countries is geared less towards CNC machine tools than it is to conventional ones. Although rising, only 31.1 per cent and 54.7 per cent of lathe imports by developing countries in 1990 and 1994 were CNC. In Asia and Latin America, CNC lathes accounted for around 61 per cent of total lathe imports, but in Africa CNC lathe imports were only 34.2 per cent of total lathe imports (see Appendices A12 and A14).4 Altogether, the share of developingcountry CNC metal-cutting machine-tool imports to total developing country metal-cutting machine-tool imports in 1990 and 1994 amounted to 27.8 per cent and 47.7 per cent, respectively. China is also emerging as one of the largest CNC machine-tool importers among developing countries. Such relatively low CNC machine-tool imports, together with the relatively low production share of CNC machine tools in most developing countries, suggest a level of consumption of CNC machine tools much lower than the levels in developed countries.5 In sum, the international diffusion of CNC machine tools suggests some emerging trends. Within developed countries, the US, Germany and other European manufacturers have clearly lost out, as far as production is concerned, to Japanese firms. Whether countries such as the US have fallen backward technologically is much more difficult to ascertain as the US still has an advanced —and now streamlined—machine-tool industry which seems to be recovering technological leadership and world market share, particularly in recent years. Part of this industry is today in the hands of Japanese and European manufacturers, although whether and how this will negatively affect diffusion in

80 CONCEPTS, METHOD AND SYNTHESIS

the US is an open question, especially in the context of sophisticated, costconcious and demanding users. Also, the US has very large imports of machine tools, brought in for markets that the local industry apparently does not cater for. This could only benefit all users through offering alternative technological solutions and wider choices in price, yet the extent of local CNC machine-tool consumption would seem to be lower than in similar countries. Finally, the US is still the largest producer of high value-added and knowledge-intensive electronic industrial control and instrumentation equipment, which are becoming major elements of the cost of CNC machine tools, have wider potential applications than do the machine tools themselves, and are a much faster growing industry than the machine-tool industry (Alcorta 1995a). It may have been a forwardlooking move to switch to this kind of equipment. Within developing countries there are also some new patterns. It is clear that users in South Korea and Taiwan have become nearly as advanced as are users in developed countries and have reached that position in a remarkably short period of time. China seems to be progressing into the CNC era but still has a considerable distance to go if it is to reach the levels of diffusion of South Korea and Taiwan, let alone those of developed countries. The bulk of developing countries are facing a much widened technological gap with both developed countries and their more advanced counterparts in developing countries. In some of the more advanced developing countries, further diffusion should, however, be expected, given that they are at the initial phase of their diffusion curve, i.e. the S or Sigmoid curve, and could eventually be moving to a rapid growth phase. But it must also be borne in mind, as Ehrnberg and Jacobsson (1993) point out, that developed countries also are not standing still and that, for instance, CNC machine tools are being used as part of more technologically sophisticated and efficient FMS in Europe and Japan. 3 Surveyed countries’ CNC machine-tool diffusion in an international perspective The country case studies show that the diffusion of metal-cutting CNC machine tools in Brazil, India, Mexico, Thailand, Turkey and Venezuela began in a significant way in the second half of the 1980s. Although some local production and imports of CNC machine tools had taken place in the late 1970s and early 1980s, some firms in the sample having purchased a CNC lathe or a machining centre, use was limited and sporadic and, as is pointed out in the Venezuelan study, it is not clear in all cases what these machines were being used for. It was only in the second half of the 1980s that adoption of FA began to spread throughout industry and within firms.6 Given that most countries in the study were starting from a very low base, use of CNC machine tools proceeded relatively rapidly in the early stages of diffusion, at least as measured by the flow of CNC machine tools. Between 1983 and 1994 annual growth rates in the stock of CNC machine tools ranged from 21. 2 per cent in Brazil to around 43 per cent in Thailand and Turkey (see

Notes a Stocks for 1982 were obtained from Mercado (1990a:89), to which average yearly import data from the US and Japan for 1980±91 and projected to 1994 from Watanabe (1993b:3) and 1983±94 imports from Europe reported in Eurostat have been added. b Stocks for 1982 and flows up to 1989 were obtained from UNIDO (1990b:59), to which local production minus export figures up to 1994 as stated in SOBRACON (1992), average yearly import data from the US and Japan for 1989±91 and projected to 1994 from Watanabe (1993b:3) and 1990±4 imports from Europe reported in Eurostat, have been added. c Imports in 1982 were considered as the initial stock to which average yearly import data from the US and Japan for 1980±91 and projected to 1994 from Watanabe (1993b:3), 1983±93 imports from Europe reported in Eurostat and 1994 imports from UN (1996), have been added. d Imports up to 1982 were considered as the initial stock to which local production minus export figures from Indian Machine Tool Manufacturers' Association (IMTMA) and CECIMO, average yearly import data from the US and Japan for 1980±91 and projected to 1994 from Watanabe (1993b:3) and 1983±94 imports from Europe reported in Eurostat, have been added. e Imports up to 1982 were considered as the initial stock to which yearly import data from the US and Japan for 1980±9 from Watanabe (1993b:3), import data from 1989±91 projected to 1994 by AMT (1993:F-87) and 1983±94 imports from Europe reported in Eurostat, have been added. f Imports up to 1982 were considered as the initial stock to which average yearly import data from the US and Japan for 1980±91 and projected to 1993 from Watanabe (1993b:3), 1983±93 imports from Europe reported in Eurostat and 1994 imports from UN (1996), have been added g Employment data for Brazil and Thailand are 1991, India and Mexico 1992 and Turkey and Venezuela 1993. h Figures for Brazil and Thailand are 1991, India and Mexico 1992 and Turkey and Venezuela 1993.

Table 3.7 CMC machine-tool stocks in selected countries, 1982–94 (in units)

DIFFUSION OF FA IN DEVELOPING COUNTRIES 81

82 CONCEPTS, METHOD AND SYNTHESIS

Table 3.7).7 Since the late 1980s, however, the rate of diffusion has slowed in most countries under study, particularly in Brazil and Venezuela, and, since 1994, also in Turkey, as governments began limiting aggregate demand in order to grapple with serious macroeconomic imbalances and high inflation rates. Despite relatively rapid diffusion in terms of flow, as the Brazilian study points out and as the figures in Table 3.7 would seem to confirm, the stock of CNC metal-cutting machine tools remains low as compared with those of developed countries. In 1994, the stock of CNC machine tools ranged from 10, 742 units in Brazil to 1,029 units in Venezuela. The stock of metal-cutting CNC machine tools amounted to 94,012 units in the US in 1983, rising to 204,468 units by 1989 (AMT 1989, 1993). The stock of CNC metal-cutting machine tools was 46,235 and 118,157 units respectively in Germany and Japan in 1984, 47, 000 units in the UK in 1987, and 20,000 units in Sweden in 1989 (ECE 1992c; Edquist and Jacobsson 1988). Edquist and Jacobsson (1988) point out that the sheer comparison of absolute numbers of CNC machine tools provides only a partial picture of the variance in degree of diffusion between countries. The reason for this is that such comparisons do not take into account differences in size of the engineering industries in which CNC machine tools are used, which in turn imply varying diffusion potential. To address this problem they use a ‘normalized density of diffusion’ measure which is the number of CNC machine tools installed divided by millions of employees in the engineering sector. Although comparisons may still be affected by differences in employment policies and regulations, i.e. various degrees of overmanning and difficulties in laying-off workers, the ratio provides a better basis of comparison between countries and through time.8 While Edquist and Jacobsson’s suggestion seems appropriate, a simple ratio of workers employed in the engineering industry per CNC machine tool will be used here. In practice this new ratio is the inverse of that used by Edquist and Jacobsson, without multiplying by one million, and therefore can easily be related to theirs. The figures for workers per CNC machine tool show a somewhat different picture from that arising out of the use alone of absolute numbers of CNC machine tools. Mexico and Turkey have higher degrees of diffusion than do any of the other countries in the sample, while India has the lowest (see Table 3.7). Brazil, which had the higher degree of absolute diffusion, is around the average for countries in our sample as a whole of 174 workers per CNC machine tool. Still, the comparison of the normalized ratios of CNC machine-tool use with those of some developed countries continues to suggest that surveyed countries must accelerate diffusion if they are to reduce the technological gap. The US and Germany, for instance, had reached ratios of 85 and 87 workers per CNC, respectively, already in 1980 and by 1989 the US had reduced it to 34 workers per CNC machine tool. Japan and Sweden had achieved levels of around 45 workers per CNC machine tool already in 1980, much lower than the current level of any of the countries in our sample (Edquist and Jacobsson 1988).

DIFFUSION OF FA IN DEVELOPING COUNTRIES 83

4 Diffusion of flexible automation in selected firms 4.1 The point of departure One of the key assumptions of the ‘modern technology’ literature is that the engineering industry, and more generally the manufacturing industry as a whole, are characterized by ‘mass production’ or ‘fordism’. Unlike in developed countries, where mass production is said to have grown out of historical developments since the Industrial Revolution and established itself as the main form of production organization, it is argued that in developing countries mass production emerged and became dominant as a result of the relocation of largescale production from developed countries in search of cheap labour and of attempts by governments to foster industrialization through diverse strategies since the Second World War: The industrialization strategies of many countries from the 1950s through to the late 1970s, in the form of import-substitution and export-promotion, were essentially geared toward transferring mass production industrial systems to the larger and richer Third World countries. (Storper 1991:105; see also Sabel 1986) Popular as it is, the characterization of engineering or manufacturing as ‘mass production’ or ‘fordist’ is not without its ambiguities. Sayer and Walker (1992) point out that there are several possible meanings to the use of mass production at the plant, corporation or industry levels: first, a form of production organization based on single-purpose or dedicated machines operated by semi-skilled or unskilled workers producing vast quantities of standardized goods; second, a hegemonic form of organization consisting of large integrated corporations and factories organized on the basis of volume production of standardized goods; and, third, a group of profitable volume-producing industries including automobiles, steel and chemicals which are propulsive of other industries. Sayer and Walker (1992) and Williams et al. (1987) argue, regarding the third definition, that there is little empirical evidence to support a sustained propulsive role for the industries generally regarded as mass production industries as there was no large increase in productivity as a result of their establishment, and since other sectors and industries, such as agriculture or construction, have performed an equally, if not more, propulsive role. As to the second definition, they point out that the implicit association between large firms and mass production is questionable as large-volume standardized production can be done in small firms and larger firms do small-batch production; that the relationship of dependency between large and small firms can, in principle, work both ways and that any specific relationship has to be demonstrated empirically; that the growth of firms is not linked exclusively to increases in scale but can be the result of financial

84 CONCEPTS, METHOD AND SYNTHESIS

considerations; and that many large firms are organized multidivisionally precisely to avoid the excessive centralizing tendencies of large-scale operations. Finally, concerning the first definition, they argue that no more than 10 per cent of the total workforce in the UK is engaged in mass-production activities. Indeed, the relatively minor share of productive activity being performed under ‘mass production’ conditions is confirmed also by Jacobsson (personal communication based on American Machinist data) and the Economic Commission for Europe (ECE). Jacobsson points out that in 1978 only 27.8 per cent of the total stock of lathes was automatic, i.e. could be considered as ‘fixed’ automation, and that they were in use in no more than 15 per cent of the plants. Also, in 1983 only 0.8 per cent of the total stock of metal-cutting machine tools in the US were stationtype machines or of the ‘fixed’ type, including transfer lines, and that only 5 per cent of the plants had such machines, mainly in the transport machinery industry. By 1989, this share had fallen to 0.7 per cent (AMT 1993). A review of studies made in the US and the UK by the ECE (1986) points to production in ‘jobshops’, i.e. small-batch production with general purpose equipment, accounting for 50–75 per cent of engineering output, and this does not include all general purpose equipment. Our own country studies also support the view that ‘mass production’ accounts for only a small proportion of productive activity. By and large, surveyed firms used conventional machine tools including engine lathes, drilling, milling, boring and grinding machines. Automatic lathes were used by many firms, but they were a relatively minor proportion of their total stock of machine tools.9 In the Brazil, India, Mexico and Thailand case studies some of the automotive component firms had transfer lines, but these were used only in the manufacturing of parts such as valves, pistons or piston rings. Generally, transfer lines neither accounted for a significant share of output nor could have been said to ‘pace’ or ‘dominate’ in any way the overall production process: they were just one more family of machines used for specific purposes, as are many others. Hence, the dominant form of production organization in the sampled firms, and probably to a very large extent in developing countries as a whole—as many of surveyed firms were the most automated and organized of their respective countries—cannot be said to have been ‘mass production’. This finding, in turn, suggests that a great deal of the hype over the impact of new technologies on scale and scope both in developed and developing countries arises out of a wrong assumption: that ‘mass production’ is the dominant form of production organization. Comparing ‘mass production’ and today’s forms of flexible automation, although valid in certain cases, cannot be generalized to manufacturing as such, and even less so to the economy as a whole. A more appropriate comparison would be between old forms of production organization, where conventional machine tools and varying degrees of specialization of labour are predominant, and new forms of production organization characterized by flexible automation and skilled labour. This comparison would seem to be pertinent also for developed countries.

DIFFUSION OF FA IN DEVELOPING COUNTRIES 85

4.2 Diffusion of flexible automation in studied firms Table 3.8 brings together the data on adoption of FA by type of equipment in the 62 firms studied in the six country case studies. The most widely used equipment was metal-cutting CNC machine tools and, within them, CNC lathes and CAD/ CAM were also available in all countries, although their use was heavily concentrated in firms in Brazil and Mexico, most of which had more than one ‘seat’ each. In most firms CAD/CAM was used to design and manufacture jigs and fixtures, although some firms used them also for product adaptation. Robots were in use in Mexico and Thailand, mainly as attachments to CNC machine tools for automatic feeding. In Thailand robots were used also in material handling applications, for transferring workpieces between stations and in arcwelding applications. FMS, combining machining centres, were used by four firms in Brazil and Mexico, and were probably among the few available in the developing world. Indeed, the Brazilian firm reported in that country’s case study as having an FMS claimed that it was the only one available in Brazil. Computerized production control for production planning, scheduling and routeing was in use by Brazilian, Mexican, Turkish and Venezuelan firms. As had been intended, the selected firms were among the most automated in their respective countries. The average number of workers per CNC machine tool in the sample was 20.4 (Table 3.9) as compared with an average of 131 workers per CNC for the six countries in 1993 (Table 3.7). Altogether, the studied firms accounted for between 1.7 per cent of the total estimated stock of CNC machine tools in India to 8.5 per cent of the total estimated CNC machinetool stock in Mexico. The average number of workers per CNC machine tool was the lowest in Mexico, across all firm sizes, industries or types of ownership. Selected Mexican firms were more automated than was the average US engineering firm and more than Hydraul, the UK valve manufacturer analysed in Chapter 2. Within Mexico, locally owned small capital-equipment firms had the lowest workers: CNC machine-tool ratio. Venezuelan firms were the next most automated, having nearly double the number of workers per CNC machine tool than their Mexican counterparts. The relatively high level of automation of locally owned small customized-product Venezuelan firms accounted for this result. Thailand and Turkey had very similar low ratios of workers per CNC. Unlike most other countries in our sample, diffusion of CNC machine tools in these two countries was relatively higher in the car-component industries. Finally, Brazilian and Indian firms were among the least automated of the sample, although the Indian ratio of workers per CNC machine tool more than doubled the Brazilian ratio. While in the case of Brazil the relatively low degree of automation seemed to be the result of heavily manned large autocomponent firms, in the case of Indian firms low automation seemed to be more of an acrossthe-board phenomenon. Table 3.10 shows the ‘density of automation’ by firm size, industry and type of ownership as measured by all the country studies. In this case ‘density of automation’ refers to the share of machining that is accounted for by CNC

Source: Country and firm case studies. Note a Includes other CNC metal-cutting, machine tools

Table 3.8 Diffusion of flexible automation by type of equipment in surveyed countries (in units)

86 CONCEPTS, METHOD AND SYNTHESIS

Source: Country and firm case studies.

Table 3.9 Metal-cutting CNC machine tools in mechanical engineering by firm size, industry and type of ownership in surveyed countries (ratios of workers to CNCs)

DIFFUSION OF FA IN DEVELOPING COUNTRIES 87

88 CONCEPTS, METHOD AND SYNTHESIS

machine tools. Measuring the extent of automation in this way has the advantage that it focuses exclusively on the production stage that is affected by the use of CNC machine tools. Hence, it is a precise measure of the degree of automation at the point of production and avoids the problem of varying levels of overmanning. A share of over 75 per cent can be considered as near-total automation while a share equal to or under 25 per cent can be seen as minimal automation. One first conclusion that emerges from the data and the country case studies is that large firms were not highly automated, while small firms were. Typical large firms in the sample were long-established, with significant investments and considerable experience with old technology. Scrapping of capital equipment implied major disruption in terms of production process, labour organization and skill requirements. With superior information, a more organized process of searching for and selecting new machines, and given a context of high uncertainty, as is the case in developing countries, large firms seemed to be taking investment decisions with care and adopting new technologies gradually. Some of the large firms in the sample began introducing FA as part of a significant strategic change in the direction of the firm, as in the case of Oilhydro, in Brazil, or as the result of a major organizational change in the company as in the case of Multi-liners, in Turkey (see Quadros Carvalho, in this volume, p. 187; Ansal, in this volume, p. 342). The only large firm that was highly automated was Mexico’s Glass-die, which was part of a glass-manufacturing conglomerate operating globally and with a substantial share of the world market in glass containers (Domínguez and Brown, in this volume, p. 227). Small firms were typically much younger, and many of them had started their operations with new technologies. Unlike large firms, whose production processes may include foundry, forging, chemical treatment and large-scale assembly, small firms focused mainly on machining, although some simple assembly took place occasionally. Some were spin-offs by engineers, managers and/or skilled workers from large companies or were members of large local industrial conglomerates, to which they were supplying specialized components or machining services. Thus, at least initially if not continuously, they had drawn on the information and resources available in large firms and had started with a relatively captive and predictable market, but often on condition that they adopted flexible automation. This was clearly shown in the Indian study, particularly in the cases of Telco-diffs, Telco-shafts, Telco-tools and Telco-gears (Alam, in this volume, p. 305). These companies were set up in the early 1990s, and were equipped with CNC machine tools in order to produce components for TELCO, India’s largest lorry manufacturer. They were located in an industrial estate near TELCO and two of them sold exclusively to TELCO. The four companies stated that they would not have been acceptable to TELCO as suppliers without the use of CNC machine tools. A second conclusion is that customized-product manufacturers were more automated than were their capital equipment and automotive component counterparts. Mould production is one of the most demanding activities within engineering because of the difficulty of machining operations, arising from the

Source: Country and firm case studies. Note a Share of machining accounted for by CNC machine tools.

Table 3.10 Density of automation by firm size, industry and type of ownership (number of firms)

DIFFUSION OF FA IN DEVELOPING COUNTRIES 89

90 CONCEPTS, METHOD AND SYNTHESIS

uniqueness of each mould and the intricacy of design. Machining services firms manufactured the widest product mix possible within the industry, and normally faced very unusual and precise demands. It was crucially important to the success of these firms to deliver the best possible quality, something that could be achieved only with the latest technologies. Some of these companies were operating in safety-sensitive activities or foreign markets and therefore had to meet the highest international standards to remain in business. Machines spares and Auto-machines, two of the Venezuelan firms manufacturing vital and safety parts for oil rigs and wells, were permanently assessed by Petróleos de Venezuela (PDVSA), the Venezuelan state oil company, to ensure proper standards were maintained (Alonso, Tamayo and Cartaya, in this volume, p. 274). Glass-die, the Mexican glass-mould manufacturer referred to earlier, not only had the most up-to-date manufacturing equipment but had its CAD/CAM linked via satellite to its main customers, so that their online clients could give precise instructions and introduce the exact modifications required (Domínguez and Brown, in this volume, p. 250). A third conclusion is that foreign-owned firms seemed to be no more automated than were local firms, although this conclusion must be taken carefully as there were only a few predominantly foreign-owned firms in the sample. This was a surprising finding, as one would expect that foreign companies, which in all cases were subsidiaries of German and US multinational corporations, would have access to better information and managerial and financial resources, and would therefore use more advanced technology vis-à-vis their local counterparts. However, foreign firms in the sample were mostly autocomponent manufacturers that had been in operation for 20–40 years, meaning that they were established at the height of import-substitution and of efforts to develop a local car industry. They were firms attracted by the less demanding local market and had made major investments in technology some time back and therefore were ‘locked-in’ to old technologies. In addition, although quality standards were getting stricter in the autocomponent industry, they were still not as exacting as in mould manufacturing or machining services, thus allowing component firms to pursue automation gradually and selectively, focusing on the critical parts first. Some firms had started only recently to invest in materials research and NPD and to integrate more closely with assemblers, all of which may result in additional changes in production technology at a later stage. A case in point was Mech-trans, the Brazilian-based US-owned manufacturer of mechanical transmissions for light and medium lorries. Established in Brazil in 1953, it had built in São Paulo one of the largest plants of its kind in the world but with a density of automation of only 20 per cent. The company started to invest significantly in FA around 1986–7, and soon after installed a four-machining centre FMS which was used for machining the bodies of its newest transmission. Since that time, investment in new machines had been oriented exclusively towards FA and the rate of acquisition of CNC machine tools had doubled. But there was still a long way to go to replace most of its equipment and the firm does not seem to be in a hurry to do so (Quadros Carvalho, in this volume, p. 192).

DIFFUSION OF FA IN DEVELOPING COUNTRIES 91

Locally owned and foreign-participation firms seemed to be resorting to the use of CNC machine tools as a means of achieving domestic and international competitiveness. In the case of foreign-participation firms or joint-ventures, most country studies pointed to foreign partners as having a significant influence in the level of use of CNC machine tools locally. In the case of local firms producing for original equipment manufacturers (OEM), the main client had a significant say also in the level of automation of its suppliers. In most cases this implied higher rather than lower automation, although we found one firm where the opposite was recommended.10 At Thai-brakes, following the introduction of four machining centres, the local manager decided to add a robot as a machine server because of the weight of some of the cast parts—up to 40 kilograms. The Japanese client, apparently, was not confident that Thai-brakes could handle the complexities of introducing a robot and insisted on manual loading (see Brimble, in this volume, p. 380). Pressures to automate on local firms producing for local markets were arising also from the competition of foreign firms entering the more liberal domestic markets in the countries under study, although we will return to this point later. Summing up, the evidence in our country studies suggests that CNC machine tools in the main have been replacing conventional machine tools rather than ‘fixed’ or ‘specialized’ automation. The latter did not account for a significant share of the technology in use, which suggests in turn that ‘mass production’ was not as strong a feature of production in developing countries as the literature would have us believe. Diffusion of flexible automation in the countries studied was fastest in Mexican and Venezuelan firms, owing mainly to the highly automated small capital-equipment firms. Generally, small firms seemed to be more automated than larger firms because of their age, their lack of commitment to old technologies and because of the support they received from large firms. By helping the small firms, larger firms would seem to be postponing the undertaking of major investments themselves. Customized-product firms were more automated than were autocomponent or capital-equipment firms because of the exacting quality standards their industry faced. Finally, local firms were more highly automated than were foreign firms because of the demands exercised by foreign partners or clients or by the more competitive conditions faced in local markets. 5 Flexible automation and new organizational techniques As we saw in Chapter 1, one of the most widely held views within the modern manufacturing literature concerns the complementarity of FA and new organizational techniques or concepts. By helping to reduce setting-up times through the timely delivery of inputs, by lowering the time dedicated to the preparation of equipment, the exchange of jigs and fixtures, the machine adjustments and trial runs necessary to begin production and by reducing the rate of defects, organizational techniques or concepts are meant to help ‘hard’ technologies to reduce economies of scale and increase economies of scope. In

92 CONCEPTS, METHOD AND SYNTHESIS

this section we will look at the extent of diffusion of organizational techniques or concepts in firms in the countries studied and discuss the degree of complementarity between flexible automation and organizational techniques. Tables 3.11 and 3.12 present the consolidated data on the diffusion of new organizational concepts by density of automation and by firm size, industry and ownership for the six country case studies. A number of points regarding the combination of flexible automation and organizational concepts arise. First, none of the organizational concepts had been diffused to more than half the firms in the sample, while eighteen firms had not adopted any of the organizational concepts. This pattern of diffusion was similar in all the studied countries but India, which showed a much lower level of diffusion. It suggests that the new organizational techniques are not as widespread in developing countries as one would expect, given the alleged complementarities. One possibility, recently put forward by Fleury (1995), Humphrey (1995) and Meyer-Stamer (1995), is that implementing organizational change is complex and difficult because it involves modifications in managerial structures and working practices that have been in use for a long time. Only firms that are already experimenting, for instance, with quality control and which have the ‘basics of production organization and cost control in place’ (Humphrey 1995: 156) will be capable of implementing organizational change. Change involves a learning process and firms must be willing to embrace it. Under monopolistic conditions, in long-protected markets or, one might add, in fast-expanding markets where whatever is produced is accepted by consumers, firms do not feel the need to reorganize. Indeed, this seems to be the case with many of the Thai firms in our sample which, facing increasing demand for exports, particularly to Japan, have responded by adding productive capacity rather than looking into ways of reducing any slack existing in production and improving productivity through the adoption of some of the new organizational concepts.11 There may be human resource constraints to the adoption of the new organizational techniques (Kaplinsky 1995). The successful introduction of these techniques increasingly requires an educated labour force capable of understanding the underlying technical processes. It may require sustained training also in order to develop the new skills that are necessary to implement the new technologies and organizational techniques. But an educated workforce and the resources to invest in training may not be available, an issue to which we will return. The complementarities indicated in the literature may not be as low-cost, or cost-free, as many authors would seem to assume (Hoffman 1989a; Kaplinsky 1994; Posthuma 1995). While it may be true that organizational change may be ‘divisible’, in the sense that one can introduce the amount of JIT, quality procedures or new working practices that one can afford, it nevertheless implies a considerable monetary cost. In preparing this research project a Dutch autocomponent manufacturer, building a completely new plant for manufacturing dashboard panels, was visited. The plant had been equipped with the latest automation technology and a carefully studied lay-out, and was introducing most,

Source: Country and firm case studies.

Table 3.11 Diffusion of new organizational concepts by density of automation (number of firms)

DIFFUSION OF FA IN DEVELOPING COUNTRIES 93

Source: Country and firm case studies. Note a Includes three firms for which density of automation was not available.

Table 3.12 Diffusion of new organizational concepts by firm size, industry and type of ownership (number of firms)

94 CONCEPTS, METHOD AND SYNTHESIS

DIFFUSION OF FA IN DEVELOPING COUNTRIES 95

if not all, of the new organizational concepts. To do so, workers had been hired six months earlier than they would normally be taken on when starting a new plant in order to get them acquainted with the technology, despite their fully qualified status, and involved in quality discussion groups. Quality-circle meetings were taking up to three days a week, and most workers participated. While management had no doubts about the long-term benefits of this approach, it was also fully aware that the financial cost in the short-term was equivalent to six months’ wages. Even so, the workers had to be enticed to stay in the company through competitive salaries in order for the benefits of such training to be fully reaped. Preparing the new lay-out had also required two years’ work by a specialist engineer.12 Lack of adoption of new concepts may also result from the state of the infrastructure in developing countries. JIT requires a sophisticated and efficient transport system such as is rarely available in developing countries. Some of the countries studied not only have no developed transportation system but lack basic services such as electricity or water. In a context where management does not know whether the factory is going to have electricity the next day or week, or whether delivery of raw material is going to be on time or a month late, priorities have to be organized on a different basis. Inventory holding may make a lot of sense under these conditions. Otherwise the just-in-time approach may result in just-without-anything. A second point regarding the combination of flexible automation and organizational change is that the degree of complementarity would seem to vary according to the type of organizational technique. Take the case of changes in lay-out. As pointed out before, one of the most distinctive characteristics of the new technologies is that they can integrate diverse equipment and functions into a single machine. A CNC lathe can perform turning operations at both ends of a rod, and can perform simple drilling, tapping and boring operations on the same rod. Drilling, milling, tapping and boring operations can be performed in a single machining centre. With conventional technologies, machining a gear box casing and a shaft for the gear box in Hydraul required around 20 operations in some 12–15 different machine tools. Bringing the 12–15 machines together in a cell or product type lay-out would demand considerable space and would be totally unmanageable, considering that more than one shaft and gearbox casing have to be produced as have other parts.13 Under these conditions, a functional lay-out for the factory would make considerable sense. With CNC machine tools both the gearbox casing and the shaft can be produced with 2 machining centres and 1 lathe. And so can some other parts for the gear box. In this case, bringing together the three machines and a gear-cutting CNC machine tool is enough to produce the whole gearbox and may be the most efficient, if not the only way, of organizing the factory, and thus hardware and organization are highly complementary. Indeed, nearly 70 per cent of firms modifying their lay-outs had a density of automation higher than 50 per cent. Working practices’ reorganization seem to be related also to the level of automation. As discussed in Chapters 1 and 2, working practices’ reorganization involves reduction in hierarchical levels, redefinition of job posts and functions,

96 CONCEPTS, METHOD AND SYNTHESIS

changes in pay practices, multifunctionality in tasks and enhanced responsibility for shopfloor workers. While, before, several workers were involved in machining any specific product—certainly at least one per machine was needed— today, with CNC machine tools, a few workers can take care of the whole process. Further, when large numbers of workers are involved, the need for supervision rises. Mody et al. (1992) point out that the ‘span of control’, i.e. the number of workers under one supervisor, typically varies between 7 and 10 in manufacturing firms in developing countries. With a much lower number of workers involved in any specific process there is obviously less need for supervisors. In addition, because CNC machine tools perform a wider range of operations, machinists must have some knowledge also of the larger number of functions performed by the machines and thus must become ‘multifunctional’. Finally, as the Turkish study in particular pointed out, the degree of responsibility of individual workers is linked to some extent to the value of the capital equipment they are operating. To the extent that the new capital equipment in use is more valuable than earlier machines, an issue we will return to in the next chapter, the degree of responsibility of the workers involved in using those machines should be higher. The lower the number of workers involved in any production process, the less need there is for supervision; the greater number of functions and higher responsibility involved clearly call for a redefinition of posts and functions and different pay practices. Some ‘devolution’ of responsibility can take place independently of the level of automation, but it is a necessity as automation increases. TQM and JIT would seem to be relatively independent of the level of automation. Consider the case of TQM. One of its components consists of quality checks being instituted at each stage of the production process instead of the more traditional checking of final products for defects after they have been manufactured. Quality becomes the responsibility of each individual worker and not solely of the quality department. To equip workers with the capacity to judge the quality of each part or product they are informed of the characteristics of the product and the level of quality expected at each stage, and they participate in discussions with other workers and staff to identify possible sources of problems, problems that are already emerging and ways of solving them collectively. Because errors are found at an early stage and individual or collective responsibility is identified, this approach leads to significant reductions in defect rates. Obviously, TQM concepts can be adopted irrespective of the level of automation and may be adopted even with no automation, as Kaplinsky’s (1994) review of the introduction of ‘Japanese management techniques’ in Brazil, the Dominican Republic, India, Mexico, South Africa and Zimbabwe shows. Our own data showing no significant difference for diffusion of JIT and TQM by level of automation would seem to suggest that this is indeed the case. A third point is that organizational concepts seem to be most widely diffused among large and medium-sized firms. Indeed, this seems to be the case even in India where there were only three firms adopting any organizational techniques. This finding is remarkable in the light of the finding that small firms are the most

DIFFUSION OF FA IN DEVELOPING COUNTRIES 97

automated of the sampled firms. This may be due to the fact that small firms focus mostly on machining and have a small number of machines, so that there is need neither for a sophisticated production lay-out nor for elaborate qualitycontrol procedures and practices. Being small facilitates face-to-face contact and therefore should lead to better communication. Production takes place as quickly as possible anyway, as fast response time in providing exactly what the client requests is one of the key competitive characteristics of small firms (Pratten 1991a). Large firms, by contrast, have to install procedures and develop practices that will allow them to ensure quality and rapid delivery. Organizational concepts seem to be more diffused in the autocomponent industry than in the capital-equipment and customized-product industries. One plausible explanation is that, as competition in the automobile industry gets fiercer and final assemblers increasingly move towards JIT production and stricter quality controls, autocomponent firms are being dragged into adopting the same techniques to remain in business. For JIT techniques to be fully effective requires involvement at the industry level. Note also that, insofar as car component firms are the least automated in the sample, they seem to be substituting organization for hardware, rather than complementing one with the other. The same would seem to be true for foreign-owned firms, which have low levels of automation but high levels of adoption of organizational techniques. 6 Factors underlying the diffusion of CNC machine tools in surveyed firms Two sets of factors emerged as underlying the diffusion of FA concepts in the developing countries under study. At the macro level, changes in trade regime and macroeconomic conditions and a higher level of aggregate and engineering goods’ demand were among the most significant. Growing foreign trade liberalization in many developing countries during the 1980s led to lower CNC machine-tool prices and their wider availability in producer countries such as Brazil and India. Actual and potential users increasingly found imported CNC machine tools better in quality and sophistication, and available at affordable prices, and thus were attracted to invest in them. The Brazilian study clearly points to this factor as key in firms’ decisions to acquire CNC machine tools. Also, by increasing direct competition by foreign goods to their final products, trade liberalization forced engineering firms to restructure and modernize, or else to face closure. Some of these firms included the adoption of CNC machine tools and/or new organizational techniques, foreign or local, as key strategic elements in their restructuring processes. Macroeconomic conditions such as stability and low interest rates were also factors stimulating the purchase of CNC machine tools. Macroeconomic stability and low inflation rates allowed firms to be less uncertain about the future and to have a medium-term, if not a long-term, planning horizon and, as a result, to invest in the relatively more expensive CNC machine tools. This was particularly

98 CONCEPTS, METHOD AND SYNTHESIS

true of the Latin American countries studied. Low interest rates and availability of finance were additional reasons, given especially by Venezuelan and Turkish firms, for investing in new technologies (see Alonso, Tamayo and Cartaya, in this volume, pp. 62–3; Ansal, in this volume, p. 345). Countries like India, Mexico, Thailand and Turkey also benefited from an increase in engineering goods’ demand, particularly for vehicles, resulting from both local and foreign sources. Domestic demand for vehicles increased sharply during the second half of the 1980s in India, Mexico and Turkey. Production of cars and commercial vehicles rose from 114,000 units in 1980 to 355,000 units in 1991 in India; from 490,000 units in 1980 to 989,000 units in 1991 in Mexico; and from 51,000 units in 1980 to 242,000 units in 1991 in Turkey (Jenkins 1993). A larger demand for vehicles slowly translated into a higher demand for components because of local content rules, lower local unit costs and increasing domestic technological capabilities. Mexico profited also from increasing demand due to its participation in the North America Free Trade Agreement (NAFTA), which had given a significant boost to engineering exports, particularly in the United States. By 1990, Mexico was exporting US$2.6 billion worth of vehicles and $1.2 billion worth of engines. Thailand, in turn, had found a large outlet for its engineering products in Japan, including car and motorcycle components, electrical motors, water pumps and sanitary fittings. Thailand, in addition, faced a 10 per cent real growth of aggregate domestic demand between 1987 and 1993. At the micro level, factors of quality, flexibility, machine productivity, labour costs, lack of skilled labour, lower production costs and diminishing lead times were at the basis of the diffusion of flexible automation (see Table 3.13). Quality was the most important reason for the diffusion of flexible automation in developing countries. Around 90 per cent of the firms gave this as one of the reasons for adoption and it was at the top of the list in most of them. But ‘quality’ meant different things for different types of firm. For autocomponent firms the key dimensions of quality were repeatability and durability. Except for four autocomponent firms in the sample which had transfer lines, the predominant technology in use by firms in developing countries prior to the diffusion of FA was conventional machine tools accompanied on occasions by some automatic lathes and/or drills. Often this meant that obtaining consistency across the same part depended on the capability of the machinist to perform the same operation in exactly the same way throughout the day. This was obviously not possible and, as a result, numerous defects emerged, most of which went undetected. Add up the defects of the many parts that make any autocomponent and the defect rate at the end of the line was exceedingly high. In the case of Venezuelan autocomponent firms it ranged over 5–13 per cent of all final goods (see Alonso, Tamayo and Cartaya, in this volume, p. 271). With CNC machine tools and TQM, this was no longer the case. Consistency of machining across the same part could be guaranteed without too much, if any, loss, of the manipulation capacity that the human hand gives. It was also easier to comply with pre-set tolerances and specifications: hence, developing country autocomponent firms’ interest in adopting CNC machine tools.

Source: Country and firm case studies. Note a Includes three firms for which density of automation was not available.

Table 3.13 Motives for diffusion of flexible automation by firm size, industry and ownership (number of firms)

DIFFUSION OF FA IN DEVELOPING COUNTRIES 99

100 CONCEPTS, METHOD AND SYNTHESIS

For customized-product manufacturers, ‘quality’ meant precision, as their key techno-economic problem was being able to manufacture complex parts with a high degree of accuracy. The predominant technology in use by mould producers and machining services firms prior to the adoption of FA was conventional machine tools, even more than was the case for autocomponent firms. Achieving the shapes and tolerances required depended on the skill of the machinist and the accuracy of the relatively simple measurement instruments that were used with conventional technologies. Mistakes occurred often, scrap rates were high and the production of the more difficult parts took a long time. CNC machine tools made it possible to machine to the tolerances required without human intervention in the actual cutting. The operator limited himself/herself to setting the required parameters in the control unit, some of which had user-friendly menu-driven software. If a CAD/CAM program or terminal was available, the exact design of the part and the parameters for machining could be directly transferred to the machine tool, further reducing the role of the machinist. Cutting became extremely precise, mistakes were minimized, material and time were saved, and the information could be stored for future use. ‘Flexibility’ refers to the capacity to produce a single product or range of products with speed or in quantities that correspond more closely to changes in demand, or to the ability to alter the product mix by introducing new products or modifying the properties of existing ones. Machine productivity relates to the need, expressed by several firms, to machine more output per unit of time. Reduction in lead times alludes to the desire of some enterprises to reduce the period from the placement of a manufacturing order to the moment the product was ready for delivery. Achieving flexibility was important for the surveyed autocomponent firms because of the way the car industry had been reorganized. Around 69 per cent of the autocomponent firms in the sample gave flexibility as a key reason for the adoption of FA. As final assemblers increasingly introduced JIT techniques, they tried to push inventories down to their first-tier suppliers, which in turn introduced JIT and pushed their inventories to third-tier component suppliers, and so on. In the end, it was expected that all the firms involved would minimize inventories and so reduce the costs of carrying stock and the opportunity cost of having capital tied up in production.14 The Dutch dashboard manufacturer mentioned earlier said that Ford had started a system whereby, depending on the part, Ford could request it from this manufacturer with as little advance notice as two hours. Hence, component firms needed to be able to respond very quickly to the changing demands of their customers. Before moving on to the next factor underlying adoption of FA, one further comment on the extent of changes in flexibility is required. As was discussed in earlier sections, much of the modern technology literature is based on the belief that ‘mass production’ and associated specialized and dedicated technologies, such as transfer lines, are prevalent in the engineering industry. It follows from this view that engineering production processes tend to be highly inflexible. Yet, the evidence of our country studies suggests that, by and large, production processes were based on the use of multipurpose machines such as conventional

DIFFUSION OF FA IN DEVELOPING COUNTRIES 101

lathes or drilling machines. As compared to this widely diffused technological alternative, FA can offer less, rather than more, flexibility as there still needs to be made available a technology that is as flexible as are the hands of a worker. Reasons related to machine productivity and the reduction of lead time were more important for adoption by producers of capital equipment and, to some extent, also by machining services firms, because the key constraint on production was efficiency rather than flexibility. Before FA was introduced, manufacturing of any product involved selecting the individual machines, the order in which machines were going to be used and routeing the parts through the machines. With a single part this task seems relatively easy, as parts may follow an orderly sequence from one machine to another. But if one considers that the number of machining operations could reach 15–20 and that each operation was slow, that not only one part was being machined but up to several hundred or even thousands, that these types of firm typically produced a much wider variety of goods than did, for instance, autocomponent firms, that some machines faced higher demand than did others, that machines broke down, that machining of individual workpieces had to be co-ordinated with others which were later joined together in assembly, that some operations required skills which only some operators had, that raw-material delivery was not always on schedule, and that some customers had priority over others, it was evident that the production process was extremely complex and considerable time was lost in re-routeing, reorganizing and solving one problem after the other—a ‘nightmare’. In practice, capital-equipment factories and machining services firms using old technologies were very disorganized and inefficient. Flexible automation allowed firms to produce similar products but much more efficiently than was the case prior to adoption. Labour-related reasons for the adoption of new technologies, such as the need to reduce labour costs or having to compensate for lack of skilled labour, were also mentioned in our country studies. Reduction of labour costs was important in Turkey and Thailand, where increasing real wages constituted a serious concern for firms. In the case of Turkey, the government had granted wage increases to manufacturing workers that had resulted in real wages hiking by an average of 20 per cent during the late 1980s and early 1990s (see Ansal, in this volume, p. 360). Lack of skilled labour had been particularly acute in the customized-product industry in Mexico, Thailand and Venezuela. During the 1970s and the early 1980s Venezuela had obtained skilled machinists from neighbouring countries, particularly Colombia. By the late 1980s, however, many of the Colombian workers, having saved for some years and with the prospects for work improving back home, started to return to their country, creating a serious shortfall of skilled workers in Venezuela (see Alonso, Tamayo and Cartaya, in this volume, p. 275). Finally, around one-third of the firms in the six countries pointed to the reduction in unit costs as a key objective in their adoption of FA. Throughout the engineering industry the competition was getting tougher and, with an increasing number of firms able to deliver similar quality in quick time, price competition was becoming a crucial factor.

102 CONCEPTS, METHOD AND SYNTHESIS

On the whole, the evidence in the country studies suggests that a combination of technological and economic factors underscored the diffusion of FA among selected firms. Economic factors, such as macroeconomic stability, low interest rates, growing domestic and international demand, the capacity of the new technology to reduce costs, and relative factor prices, were important determinants of diffusion of FA; but so were technological requirements for quality and the technical performance of the equipment. It is difficult to make a judgement on the relative importance of economic and technological factors, except to make the point that our case studies suggest that, where economic conditions were far from the state of equilibrium, economic factors were emphasized in the adoption decision, while in relatively stable situations firms emphasized technical attributes in their decisions to adopt FA. 7 The process of intra-firm diffusion of CNC machine tools Thus far, the diffusion of CNC machine tools in the surveyed firms has been examined as if it were a simple process. Technological diffusion is far from simple, however. Because in selecting a new technology a firm not only chooses a method for producing a good or range of goods at certain costs but acquires a potential capability for subsequent productivity-increasing improvements and adaptations and for innovation, a proper assessment and assimilation of the new technology is normally required (Dahlman et al. 1987; Enos and Park 1988). Further, the more systematic efforts that a firm makes, in obtaining information and knowledge on a specific technology, the better it can profit from it (Karshenas and Stoneman 1995). Hence, the process of assessing and assimilating a new technology normally requires several phases or steps; and many of the firms in the study engaged in such steps, albeit with varying intensity. Once the surveyed firms had identified their needs and motivations for acquiring CNC machine tools, four additional steps were generally undertaken: obtaining information; assessment and selection of equipment; approval of selection and investment; and installation and start-up. The first step necessary in the adoption of a new technology is the obtaining of information on its availability and its potential (Edquist and Jacobsson 1988; Karshenas and Stoneman 1995).15 The process of dissemination of information about new technologies normally involves efforts by both suppliers and users, and it requires also: mechanisms such as the use of technical or trade publications; personal and institutional exchanges on technical characteristics and performance; user experiences and prices of the new equipment; use of information provided by foreign partners or licensors; and diverse marketing efforts, including publicity, trade fairs, visits to other users and the establishment of sales representatives or distribution agencies. The most important source of information among surveyed firms in the country case studies were visits to trade fairs. Most of the country studies point to information gathered from local and international fairs, of which the EMO

DIFFUSION OF FA IN DEVELOPING COUNTRIES 103

International Fair in Hanover figures prominently. In the cases of Brazil and India, given the existence of domestic CNC machine-tool production, local machine-tool fairs were an important source of information. The Brazilian study, however, points to some differences in accessing information through fairs experienced by large autocomponent firms, on the one hand, and by small customized-product firms on the other, mainly because of the larger financial and technological resources of the former (see Quadros Carvalho, in this volume, p. 197). Information on CNC machine tools was obtained also from local sales representatives and distributors, although the Venezuelan study points out that sales representatives and distributors took some time to establish themselves in the country, meaning that many potential users had no option but to travel abroad. Jacobsson (1986) points to a similar limitation to obtaining information on CNC machine tools in Argentina in the mid-1970s. Lack of availability of local representatives normally results in delays in installation of equipment and in repairs and maintenance, as the technicians have to be flown in from abroad. When established, sales representatives provide technical information on their product range, technical advice on the choice of machinery and ways of introducing the equipment into the plant, assistance in equipment installation, programming, operation and maintenance, and training of personnel. Clearly, representatives of Japanese suppliers had been far more successful than representatives of suppliers from other countries in establishing local sales offices and distributors, something that also accounted for their large share of the CNC machine-tool market in the sampled countries. Some of the studies point out that, although useful, the information provided by representatives of foreign manufacturers, or by distributors of local makes for that matter, had to be carefully assessed, as it was usually biased in favour of the products they sold or were knowledgeable about. The Brazilian study mentions the case of a small customized-product manufacturer which was led into installing far more capacity than was necessary as a result of poor information on machine efficiency provided by the sales representative. Subsidiaries of multinational corporations, licensees of foreign companies or firms linked to large domestic conglomerates with contacts abroad normally obtained information from their headquarters, licensor or ‘parent’ company. Local affiliates or subsidiaries often used the information obtained from headquarters’ suppliers of technology, something that was particularly true of Japanese companies which worked closely together with Japanese CNC manufacturers, as became apparent in the Thai study. Foreign customers, particularly from developed countries, were another source of information on availability and new developments in CNC machine tools. Unlike developed-countries’ firms and Hydraul (the UK firm discussed in Chapter 2), sampled firms did not generally resort to visiting other firms or to experimenting with equipment prior to its full incorporation into production. A couple of firms in Turkey had sent engineers to other firms to familiarize themselves with the new equipment, and one firm in Mexico had bought one or two CNC machine tools to experiment with prior to adoption; but, by and large,

104 CONCEPTS, METHOD AND SYNTHESIS

the studied firms did not obtain information or knowledge through these means (see Ansal, in this volume, p. 345; Domínguez and Brown, in this volume, p. 227). Also, with the exception of a few Brazilian firms, enterprises generally did not approach specialized research institutes or universities to obtain information and knowledge on the potential of CNC machine tools. This would seem to be the result in part of the fact that many of the sampled firms were among the first in their localities to use the new technologies. But it would seem to be the result also of an approach to information searching that relies to a significant degree on the uncritical acceptance of whatever salespersons of machine tools stated. To the extent that ‘learning by viewing’, ‘learning by using’ and ‘learning by experimenting’ (Bell 1984; Rosenberg 1982) are key sources of information and knowledge and, therefore, of improved performance, it is quite possible that the surveyed firms had not been using their CNC machine tools to their full capabilities. Turning to the stage of assessing and selecting specific equipment, with the exception of a couple of large and medium-sized Turkish firms and a large multinational company in Brazil, few firms reported an elaborate, if any, costbenefit analysis or technical evaluation of the different equipment available prior to the decision to purchase a CNC machine tool. The only major criterion that emerged in the research as crucial in the decision to purchase was local availability of service. Indian firms pointed to ‘little difficulty’ in choosing equipment, but several admitted that wrong technological and economic choices had been made. An important source of this mistake was an exaggerated emphasis on low-cost machines where the most appropriate course of action was the obtaining of additional information for comparing technical features and costs and benefits (see Alam, in this volume, p. 313). The Thailand study also states that, for CNC machines to meet companies’ expectations fully, it would have been necessary to adapt them to their specific conditions (see Brimble, in this volume, p. 381). One plausible reason for the lack of elaborate assessment and adaptation of the new equipment, put forward by Dahlman et al. (1987) and by Karshenas and Stoneman (1995), is that proper evaluation and adaptation of any new technology, as was the case also with the adoption of new organizational techniques, are not cost free processes. At Hydraul, the process of selection of technology involved several days of work by accountants, engineers and managers on the financial returns to the investment and on the advantages and disadvantages of the different technological options available. Only very large domestic or multinational firms in the developing countries studied could have afforded the time and human resources that a proper evaluation normally requires. Even seeking cheaper outside advice was beyond the financial possibilities of many of the smaller firms in our sample. In the few firms with elaborate systems for technological assessment in place, the approval of the selection was subject also to a number of procedures, including approval by headquarters in the case of Brazil-based multinationals (see Quadros Carvalho, in this volume, p. 198). In many cases, notably in Mexico and Venezuela, the evaluation and selection was concentrated in the

DIFFUSION OF FA IN DEVELOPING COUNTRIES 105

owner-manager resulting, particularly in the early phases of diffusion in Venezuela, in decisions to purchase specific makes of CNC machine tools based solely on the country of ancestry of the owner (see Domínguez and Brown, in this volume, p. 243; Alonso, Tamayo and Cartaya p. 231). Installing and starting to operate CNC machine tools, as the case studies show, is neither an easy nor a cost-free process. Training for up to two months’ was normally required for operators, technicians and engineers who would be involved with the new machine tools. A couple of firms in Venezuela stated that they had sent their personnel abroad for training prior to and during installation of the equipment (see Alonso, Tamayo and Cartaya, in this volume, p. 266). Training was normally provided by the equipment supplier. Often new personnel had to be hired to operate the new machine tools, and they were not readily available in the market. Indeed, lack of skilled workers knowledgeable about CNC machine tools was clearly perceived as a major limitation to their wider diffusion in studied countries, except for Turkey. Older firms faced some resistance from workers while installing FA, presumably because of their fear of losing employment or of being unable to acquire the new skills necessary to operate the CNC machine tools. Worker resistance often delayed the installation and start-up period and required additional training and reassurance efforts if major disruptions were to be avoided. A problem that emerged everywhere during installation, and was prolonged well into the phase of operation, particularly when maintenance was required, had to do with the repair of the printed circuit boards and other electronic parts of the machine tools. Suppliers were extremely zealous about users repairing or replacing boards with cards made locally or even by other foreign manufacturers. According to a machine-tool supplier, the main reason for this approach, which was common to all machine-tool producers, was the high cost of the electronic components in relation to total machine costs, because if anything went wrong other expensive electronic parts may have also been damaged. In addition, manufacturers were concerned that users or other manufacturers could develop the ability to copy their control cards’ designs, for these, to a large extent, determined the technical capabilities of the machine tools. Users believed that electronic card reparation and replacement were simple and should have been allowed, as obtaining an original card required time and normally delayed unnecessarily the repairing and maintenance period. None of the firms in the sample admitted to having attempted to repair or replace a printed circuit board, but according to machine tool suppliers this was a common occurrence, leading to friction between users and producers and, on occasions, the loss of guarantees from producers. To sum up, the process of adopting CNC machine tools by the firms in the sampled countries was not devoid of complexity. It involved obtaining information which was not always readily available, making judgements on technical and economic performance that were subject to great uncertainty and which firms did not or could not make systematically, and, occasionally,

106 CONCEPTS, METHOD AND SYNTHESIS

installing equipment in the context of worker resistance, lack of training and skills, and friction with suppliers. 8 Conclusions The aim of this chapter was to examine the degree of diffusion of CNC metalcutting machine tools and related organizational techniques in developing countries especially. Since the emergence of CNC machine tools in the mid-1970s there has been a clear shift towards them worldwide. The examination of internationally available data on production, trade and consumption of machine tools since the early 1970s has been seen to suggest several emerging diffusion patterns. As far as developed countries are concerned, the US, Germany and Japan emerge as the main users of CNC, although the first two countries have lost their predominance as producers to the third. As far as developing countries are concerned there has been rapid diffusion in countries like Taiwan and Korea which are large users of CNC machine tools and also significant exporters. China, too, has a very large and growing consumption of metal-cutting machine tools and has began switching to the use of CNC machine tools but not yet at a rate as fast as those of developed countries, or of Taiwan and Korea. Diffusion in low-income Latin American and African countries has been very slow, particularly during the early 1990s. Diffusion in the six studied developing countries has been growing rapidly since the mid-1980s, although these six seem still to lag behind China. Brazil was seen to have the largest number of CNC machine-tool units in stock but once the size of its engineering industry is accounted for the extent of FA diffusion is less than in each of the other countries but India. Mexico and Turkey emerge as the countries with the widest extents of diffusion once the size of each country’s industry is controlled for, while diffusion in Thailand and Venezuela is somewhere in the middle. CNC machine tools have mostly replaced conventional machine tools in the six countries, though they cannot be said to have replaced transfer lines or other types of specialized equipment which were little used by the sampled firms. ‘Mass production’ was not, therefore, very much in evidence in developing countries. Small customized-product local firms were found to be among the most automated because of their young age and the exacting technical requirements being imposed on them. Large autocomponent and general capitalequipment firms seemed to adopt FA more gradually and thoughtfully, given the amounts of investment, learning and possible disruption involved. The diffusion of CNC machine tools was complemented by only some of the available organizational techniques, such as CM and WPR. CM and WPR are complementary to FA because, by arranging the order of machines and providing workers with the necessary skills, functions and responsibility, machine productivity can be enhanced. Other organizational techniques such as TQM, JIT and the aplication to international quality standards would seem to be used to substitute for rather than to complement the use of CNC machine tools, as their

DIFFUSION OF FA IN DEVELOPING COUNTRIES 107

application requires no automation, flexible or otherwise. Indeed, their use was normally linked to low levels of diffusion of hardware. The main motives for the diffusion of CNC machine tools and associated organizational techniques were macroeconomic conditions and microeconomic factors such as quality and flexibility. Low and stable price and interest rates and high aggregate demand, whether foreign or domestic, were crucial to induce investments in, and obtain steady returns from, CNC machine tools. Increasing quality requirements throughout the engineering industry has meant that firms have had to adopt the new technologies or else risk losing clients to competition. For autocomponent firms ‘quality’ meant repeatability, while for customizedproduct firms it meant precision. Achieving flexibility was important because of changes in the demand schedules of clients, i.e. smaller and more frequent deliveries, particularly in the vehicle industry. The chapter concluded by observing that during the diffusion process firms face a number of—sometimes severe—information, training and labour- and supplier-relations’ limitations. Hence, the question we must turn to now is whether the use of CNC machine tools by the firms in the six developing countries have resulted in higher product, plant and firm optimal scales and economies of scope. Notes 1 Since the 1960s East European countries, the former USSR and more recently Russia have made haphazard reports of trade data to the UN. To build a consistent time series of trade figures from these countries, reported data have been used for those years and countries which made them available and are added to figures for imports and exports obtained from trading partners, the ‘mirror statistics principle’ (for an early application of this principle to machine-tool trade data, see ECE 1992a). This approach has two implications. First, trade between East European, the former USSR and Russia is not included except when reported data were used. Second, some of the imports are FOB rather than CIF. Despite these limitations, the data herewith presented constitute one of the few, if not the only, comprehensive metal-cutting machine-tool import and export time series for these countries thus far estimated. 2 Numerically controlled machine tools include three types of machine tool. First, there were what was known initially as numerically controlled machines, where information to produce a particular part was put into a punched or magnetic tape which was then fed into the control unit. These machines were developed in the 1950s. Second, there were machine tools in which the control unit was based on minicomputers. This generation of machine tools started to diffuse widely in the early 1970s. Third, micro-computer-based machine tools, in which the control unit became a microcomputer, started to appear in the market in the mid-1970s (Edquist and Jacobsson 1988). All of today’s production of numerically-controlled machine tools is CNC machine tools. Because available statistics do not distinguish between the two types of machine tool, and for simplicity, we will refer to both NC and CNC machine tools as CNC machine tools.

108 CONCEPTS, METHOD AND SYNTHESIS

3 Table 3.6 is based on the UN’s SITC Revision 3 codes. For the years 1990 and 1994 this table is comparable with the SITC Revision 2-based Table 3.4 in the sense that each subgroup is part of the larger group, i.e. CNC machine tools are a subset of metal-cutting machine tools. 4 In 1994 the proportion of CNC lathes in total lathes imports was 76.1 per cent in Brazil, 44.9 per cent in India, 64.1 per cent in Mexico, 33.7 per cent in Thailand, 73.2 per cent in Turkey and 41.9 per cent in Venezuela. 5 In 1994 the share of imported metal cutting machine tools accounted for by CNC machine tools was 58.9 per cent in Brazil, 38 per cent in India, 60.9 per cent in Mexico, 30.9 per cent in Thailand, 69 per cent in Turkey and 24.4 per cent in Venezuela. 6 To the extent that meaningful diffusion of CNC machine tools in developed countries began in the mid-1970s, there would seem to be an approximate ten-year lag in CNC machine-tool use by the countries in our sample. 7 The annual growth rate of the physical stock of CNC machine tools in the US was 16 per cent between 1978 and 1985. In the UK the equivalent figure was 18 per cent for 1976–83 (ECE 1992c; Edquist and Jacobsson 1988). 8 Alternatively, value added could be used in the denominator, but this would be subject to the limitation of comparing situations and countries with different trade regimes (Edquist and Jacobsson 1988). 9 According to Amstead et al. (1986), most automatic lathes should not be considered, strictly speaking, as ‘fixed’ automation because, though tools are fed automatically, an operator is always required to feed the part to be machined into the lathe and to remove it afterwards. Also, although they take some time to set up, they can be used in the turning of a variety of workpieces and thus are not ‘specialized’ in the way a transfer line is. 10 It is rather paradoxical that while local firms were being asked to automate by multinational corporation clients, foreign-owned firms were not. 11 Meyer-Stamer (1995) also makes the interesting point that there may be different paths to competitiveness, some of which may not require organizational change at all. 12 Relative prices may also play a role here. Where interest rates are low and land inexpensive, then pressures to introduce JIT and to reduce inventories may be lower than where interest rates and land are dearer. Also, under situations of high inflation and negative interest rates, stocking physical assets is much more economically rational than making savings in financial assets. 13 When the number of conventional machines involved is small, an arrangement along cellular/product lines may be feasible. 14 For problems with the implementation of JIT in the car industry see Rhys (1993). 15 Often the two phases of identifying a firm’s needs and obtaining information on a new technology overlap significantly as these needs are, at least partially, defined with reference to technological changes elsewhere.

4 IMPACTS ON SCALE AND SCOPE

1 Introduction This chapter is concerned with the impacts of flexible automation and associated organizational techniques on scale and scope, and the technical and economic reasons for such an impact. According to the discussion in Chapter 1, the predictions of the existing literature for product scale and scope are of a reduction in product scales due to lower setting-up times and a wider variety of goods and economies of scope arising from those lower setting-up times and from the flexibility of the new technologies. With regard to plant scale, there are widely divergent views, some pointing to lower optimal plant scales and economies of scale, or ‘de-scaling’, because of the divisibility and lower fixed costs of flexible automation, while others pointing to higher optimal scales and economies of scale, or ‘scaling-up’, due to higher efficiency and capital fixed costs. Finally, regarding firm scale, the main expectation is for higher scales arising from larger R & D and marketing fixed costs. The chapter proceeds as follows. Section 2 addresses the issue of changes in product scale and examines the relationship between batch sizes and setting-up times. Section 3 addresses the question of product variety or scope, also in relation to setting-up times, and discusses the extent of changes in scope with regards both to what is meant by ‘new products’ and to vertical integration. In section 4 the issues of plant and firm scale as well as unit costs are considered in relation to the conceptual framework developed earlier. Finally, section 5 analyses the factors underlying changes in scale. The first part of section 5 analyses technical factors including: the conditions of operation in job-shops and batch manufacturing; FA’s capacity to integrate different types of equipment, to produce at higher rates and to save on raw materials; and the role of TQM, JIT and cellular manufacturing in improving on efficiency. It examines also the degree of divisibility of CNC machine tools. The second part discusses changes in the relative prices of CNC machine tools vis-à-vis the technologies they replace and their impact on capital, labour, raw materials and overhead costs.

110 CONCEPTS, METHOD AND SYNTHESIS

2 Changes in product scale As discussed in Chapter 1, one contention on which there is agreement in the literature is that FA will lead to a reduction in batch sizes or ‘product de-scaling’ due mainly to lower setting-up times and costs. So, are setting-up times and batch sizes falling in developing countries? What do the country and firm case studies tell us regarding changes in setting-up times and batch sizes, as old technologies have been replaced by CNC machine tools and associated organizational techniques? Of the 48 firms that gave figures for changes in setting-up time, 44 reported reductions in setting-up time, more than half reporting reductions of at least 75 per cent. This result was consistent across countries, density of automation, industries, firm sizes and types of ownership. Reductions in setting-up times, however, tended to be lower in some Indian firms because their CNC machines did not have automatic workpiece feeding. Setting-up time increases were recorded by two Turkish manufacturers of moulds, Tin-moulds and Glass-moulds, and by one Thai manufacturer, Gear-shafts, all of which had undertaken the production of far more complex designs with very small tolerance levels, requiring longer setups and several trial runs until engineers and machinists were satisfied that the correct measurements had been achieved The evidence on batch sizes presented in the country studies, however, is less clear-cut than it is for setting-up times. Of the 48 firms that furnished figures for setting-up times, 48 per cent showed reductions in batch size while 16 per cent showed no change and 38 per cent showed batch sizes increasing (see Tables 4.1 and 4.2). There were some indications that increases in batch size were associated with the level of automation, but the number of firms involved was too small to reach a clear conclusion. ‘No change’ or ‘increases in batch size’ were prominent results among small firms, among customized-product firms and among firms stating that they had adopted new technologies to increase machine productivity. A number of conclusions emerged from the data on setting-up times and batch sizes. First, despite the economic and technical advantages of CNC machine tools vis-à-vis specialized equipment and conventional machine tools, there are still some limits to the reduction of batch size due to setting-up times and costs. In machining a single part on a conventional machine tool the operator usually performed the following operations: plan the sequence of operations; select the cutting tools and workholding devices; set and change the tools; select speed, feeds and depth of cut; load and unload the workpieces and position the work relative to the tool (de Garmo et al. 1984).1 Although with CNC machine tools many of these operations can be done ‘off-line’, i.e. while the machine is operating, and are incorporated into the software program that instructs the machine, it is still necessary to do some testing and gauging of the tools, workpiece and software program before each new batch. The Thai firm Plumbing-parts, one of the few firms facing an increase in batch size, stated that for very small batches it was not worth setting-up a CNC machine tool as there

IMPACTS ON SCALE AND SCOPE 111

were still some significant setting-up costs (see Brimble, in this volume, p. 384). Air-brakes, a Brazilian autocomponent manufacturer, always produced a minimum of three months of sales to amortize setting-up costs. Mediumservices-A, a customized-product Brazilian firm, similarly always produced a minimum batch of 8 days for each 10-hour set-up (see Quadros Carvalho, in this volume, p. 201). In Turkey, many of the surveyed firms operated with batch sizes equivalent to one day’s production because they preferred to adjust and program machines in the morning and to use them continuously through the second and third shift if necessary (see Ansal, in this volume, p. 348). Second, it does not necessarily follow from a reduction of setting-up time that batch size will fall. CNC metal-cutting machine tools do provide firms with the potential to ‘waste’ less time in preparing machines and so to reduce costs, but whether or not this capacity is used by a firm depends on other factors. Nor is it the case that reductions in batch size are determined exclusively by setting-up times and cost factors, as the optimal batch-size model suggests. While, as was pointed out above, the economic rationale underlying the optimal batch-size model has a bearing on the batch size decision of some firms, it is not the only factor, nor even the most significant one, at least for a good number of firms. Indeed, the evidence in our country studies clearly suggests that a number of other factors besides cost considerations determine batch sizes. The size of demand, delivery schedules and availability of castings and forgings were commonly mentioned in the case studies as additional factors influencing batch size. Size of demand was evident in the Indian and Venezuelan case studies. Induvalves calculated batch size on the basis of the monthly demand for each type of valve. Small standard valves in high demand were produced in large batches while large and/or specialized valves, some of which were custom-made, were produced in small batches (see Alam, in this volume, p. 316). Gas-parts, the Venezuelan monopolistic manufacturer of standardized household gas valves, increased its batch size fourfold following a sharp rise in the demand for its products. This manufacturer was dedicating all its CNC machine tools to the manufacture of a single product, albeit with variations in the size of the gas valve (see Alonso, Tamayo and Cartaya, in this volume, p. 281). On the whole, if batch sizes were too small or too large in relation to demand, they were adjusted accordingly.2 In the case of some autocomponent firms the batch size was determined largely by the delivery schedules insisted upon by the final assembler. Automobile manufacturers used to place bigger orders which resulted in larger batches being manufactured. Since FA arrived, final assemblers have reduced the time-span between deliveries as part of their JIT strategies, causing suppliers’ batch sizes to fall, as in the case of some Mexican firms where batch size decreased from several thousand units to around 100 units (see Domínguez and Brown, in this volume, p. 234). Although the reduction in setting-up times and costs facilitated the trend towards smaller, more regular, deliveries of components, it was not clear that this trend would not have emerged anyway in

Source: Country and firm case studies.

Table 4.2 Changes in product scale by firm size, industry and type of ownership (number of firms)

Source: Country and firm case studies.

Table 4.1 Changes in batch size by density of automation (number of firms)

112 CONCEPTS, METHOD AND SYNTHESIS

IMPACTS ON SCALE AND SCOPE 113

the absence of CNC machine tools, as even component manufacturers without new technologies are being press into delivering JIT. Indian firms also pointed out that the availability of castings affected their batch sizes. Good quality castings and forgings were available often in small numbers only, so firms had to keep this in mind when choosing batch sizes. Also, their supply was erratic, often forcing firms to run smaller batches than desirable (see Alam, in this volume, p. 317). Summing up, CNC machine tools and associated organizational techniques seem to have the potential to significantly reduce setting-up times and, as a result to reduce batch sizes (‘product de-scaling’), although there are still limits to further reductions because of the need for testing and gauging before each new batch. Whether the potential was actually realized by the firms studied depended on a number of other factors, of which the demand and the demand schedule for each individual product seem to have been the most prominent. 3 Changes in product variety or scope Another key finding of the research is that the adoption of CNC machine tools and associated organizational techniques resulted in an increase in the scope of products among most of the firms in the sample that provided data. Tables 4.3 and 4.4 show the number of different final products and product families, i.e. functionally different final products as defined by the companies, and the number of different parts machined, i.e. different parts and components that enter final products. Product variety increased across all levels of automation, firm size, industry and ownership-type, although it was slightly more pronounced among autocomponent firms. In the case of autocomponent firms the adoption of FA was concomitant to the reduction in the life cycles of many vehicles, which resulted in an even larger demand for new products by final assemblers, and to an expansion into foreign markets. Increases in product variety were as high as those in HT-shafts, one of the autocomponent manufacturers in Mexico, which duplicated the number of brake models to 180, raised the number of shafts from 5 to 13 models and the number of axles from 42 to 203 models (see Domínguez and Brown, in this volume, p. 237). The Brazilian enterprises Transmissions and Piston-rings stated that they had launched more new products since the adoption of FA than they had since their establishment in the 1950s. In capital-equipment manufacturing and customized-product industries, product variety had always been a feature, but firms were moving now from a rigid, catalogue-based, product mix to a more flexible customer-oriented supply of goods and services. Giving customers exactly what they wanted became a distinct competitive advantage in these industries (see Quadros Carvalho, in this volume, p. 202). Many country studies revealed also that some of their new products simply could not have been economically produced without the use of CNC machine tools. This was either because the new products required excessive long

114 CONCEPTS, METHOD AND SYNTHESIS

machining times, and thus were costly in tool and labour use, or because the resulting quality was below the standard acceptable to the market. It must be emphasized, however, that while there were clear indications of increasing product variety, at least two qualifications are necessary. First, CNC machine tools have brought about new fixed costs arising out of the need for specialized jigs and fixtures. Some Mexican autocomponent manufacturers interviewed had arranged for customers to pay for jigs and fixtures because of their high cost and their ‘uniqueness’ to specific customers. The importance of these fixed costs had resulted in the widespread use of ‘setting-up rules of thumb’ in determining whether or not to manufacture a new product. Mexicanvalves operated with a ‘rule of thumb’ which stated that investment in jigs and fixtures should be no greater than 7 per cent of production costs, otherwise the part was not produced. In general, producing a new product in a CNC machine tool requires investing in fixtures, jigs, tools and system planning, some of which could be used for certain products only and which were generally unprofitable if the expected output was too small or the selling price too low. Second, greater variety did not always mean new products. The Mexican study clearly pointed out that, particularly for autocomponent producers, figures on product variety included a combination of some new products with variations of existing products or variations in the dimensions of such products (see Domínguez and Brown, in this volume, p. 238). Some products that were measured, for instance, in both inches and centimetres, according to the system of measurement of the destination market, were often considered to be different products. In terms of product technology, a pressure valve produced with dimensions in both inches and centimetres may make little difference; and producing both dimensions may have required only a small adjustment to the machinery, particularly since FA arrived, but many companies still considered them as different models. Gearshafts, a Thai autocomponent firm, made a similar point. FA greatly facilitated the introduction of new variants of the same product, it was particularly easy shifting in terms of different sizes, and dimensions and into higher levels of precision, but, with few exceptions, the introduction of new products had not been usual (see Brimble, in this volume, p. 387). Fuel-sys and Transmex, two Mexican autocomponent firms which had developed new products, pointed out that developing completely new products required significant efforts in product development (see Domínguez and Brown, in this volume, p. 238). Previously, they had copied designs from other manufacturers, but as soon as they began developing their own products they had to start their own research and development and engineering departments and link closely to car assemblers. Transmex, previously a domestic leader in manual and automatic transmissions, had been unable to improve its product and process technology in automatic transmissions and was rapidly losing market share, forcing the company to specialize in the production of components only, not the fully assembled automatic transmission kit. These findings suggest the need for caution as to the extent of product diversity or scope increases resulting from FA. Indeed, a close look at the

Source: Country and firm case studies. Note a Includes two firms for which the density of automation was not available.

Table 4.3 Changes in product diversity or scope by density of automation (number of firms)

IMPACTS ON SCALE AND SCOPE 115

Source: Country and firm case studies.

Table 4.4 Changes in product diversity or scope by firm size, industry and ownership (number of firms)

116 CONCEPTS, METHOD AND SYNTHESIS

IMPACTS ON SCALE AND SCOPE 117

historical evolution of Ford automobiles, the archetypes of mass-produced goods, reveals long-standing reasons for caution. Examining model changes since the Ford T, as described by Abernathy (1978), one finds that between 1914 and 1973 car models increased from 2 to 12. But if the number of basic body frames only is considered the increase was from 2 to 5, with some bodies having parts added in different positions to obtain different configurations. In fact, the average number of models or body frames being manufactured in any one factory fell from 2 to 1.5 between 1900 and 1975 and most new Ford factories produce only a single model. What has changed is that nowadays a modern plant can introduce up to 5 variations on the same body frame, up to 4 engine sizes, 2 types of power transmission mechanism and a range of colours, altogether around 20 different ‘models’, without stopping a production line. What Ford has been offering over the years in terms of product variety would seem to be, essentially, changes in style, specifications and colours, not so much ‘new’ products, at least not in terms of product and process technology. Designing automobiles that look very different but actually do not require costly factory retooling is one of the major challenges of any car company. Japanese car makers have been particularly successful in ‘changing models’ in this way, but even they are increasingly reducing, rather than expanding, the number of available models because of the complexities and the automation and operational difficulties, and therefore expenses, involved (Clark and Fujimoto 1991; Rhys 1992).3 Another important finding regarding product diversity was the increase in vertical integration. Together with the figures on product diversity, the research obtained data on changes in the number of parts or components manufactured in the sampled firms. Most of the firms that showed increases in product diversity also had rises in the number of manufactured parts and components that were part of those products. But, more importantly, the rate of increase in manufactured parts and components was much higher than that of new products. Considering that not all the new products were actually ‘new’ but were variations of existing ones and on the assumption that the number of parts and components had not changed significantly, one can conclude that firms were integrating vertically into the production of their own parts and components. In the case of Hydraul, for instance, the number of new products as measured by an increase in product families manufactured, was 100 per cent; yet the increase in the number of manufactured parts that went into those new products was 400 per cent. Indeed many sampled firms in India, Mexico, Brazil, Thailand and Turkey stated the same. They were themselves taking on the production of all but a few of the main parts and components, and even some of the less important ones. Some Thai firms declared that they were progressively integrating backwards because they could now produce some of the more complex components which previously they imported or sourced from other domestic firms (see Brimble, in this volume, p. 387). Machine-sub2, one of the two Mexican machining-services firms, pointed to its own large customers, also equipped with CNC machine

118 CONCEPTS, METHOD AND SYNTHESIS

tools, as their fiercest competitors as they now had the capacity to produce many components themselves (see Domínguez and Brown, in this volume, p. 240). What is the impact on subcontracting of higher vertical integration? Best (1990), Piore and Sabel (1984) and van Dijk et al. (1992) argue that recent technological developments, including CNC machine tools, are resulting in patterns of industrial organization characterized by increasing subcontracting activity. To answer this question, however, it is important to distinguish effects at the plant and industry levels. Industry-level issues, including issues of entry and exit of firms, will be dealt with in the next chapter. At the plant level, one obvious effect of larger internal component production is the reduction of subcontracting. There are solid interrelated technical and economic reasons for such an effect. On the technical side, as Thai firms pointed out, the sophistication of the new CNC machine tools allowed companies to cut pieces that previously were not technically feasible. The enhanced technical capabilities of the CNC machine tools brought added advantages in terms of input use, as Book-tech, a UK manufacturer of book-binding equipment visited in the preparatory stages of this research, pointed out. Because of CNC machine tools’ capability to handle a wide range of workpiece sizes, the firm could manufacture 10 components, out of a total of 110 components, from the offcuts of metal plates that had to be scrapped when the old machines were in use. Manufacture of these components used to be subcontracted. Hence, some components now could be produced at the cost of machining alone. On the economic side, increasing machine productivity combined with low capacity utilization and high capital costs—issues to which we will return later—meant that firms had to engage in the production of more components, particularly those with the highest value added, in order to minimize costs. Although there were clear indications of subcontracting activity being reduced in all countries, a number of qualifications are necessary. First, subcontracting activity was still significant in the case of large firms with lower levels of automation. These firms were subcontracting work from small new technology manufacturers of specialized components or from machining services. Second, subcontracting was sought also by firms reaching a maximum in capacity or to deal with peaks in orders. However, this was only an occasional phenomenon, as most firms were facing severe capacity underutilization problems. Finally, some firms engaged in subcontracting as a way of avoiding ‘social’ costs. One Turkish firm started to subcontract some of the simpler production stages to its exemployees in order to reduce production cost, since these subcontracting small firms usually did not carry the burden of social security and other social taxes (see Ansal, in this volume, p. 304). In sum, although FA has resulted in increased scope it has led also to variations in the dimensions of products and vertical integration. Producing completely new products would seem to require product innovation capabilities which were beyond the majority of the firms in our sample.

IMPACTS ON SCALE AND SCOPE 119

4 Changes in plant and firm scale Turning to the impact of new technologies on scale and minimum optimal scale, our evidence shows a very clear pattern, at least as far as scale is concerned To begin with, 29 out of the 32 firms that provided figures on changes in scale of output or production capacity reported increases in scale, with 16 firms increasing their output or capacity by more than 50 per cent (see Table 4.5). In the case of 11 Indian firms and some Thai and Brazilian firms, although not furnishing precise figures, there was also evidence that, with FA, scale was increasing. Therefore, out of a total of 53 firms where there was evidence on changes in scale, 50 firms would be facing scale increases. Two firms that had more than one plant stated that scale had increased in all their plants. Body-parts, a firm in Venezuela, which claimed scale reductions, pointed out that it was having labour problems so severe that it was considering divesting or closing down the factory (see Alonso, Tamayo and Cartaya, in this volume, p. 282). The two Mexican manufacturers showing reduction or no change in overall output or capacity were operating at 30 per cent of capacity, by far the lowest in the whole sample (see Domínguez and Brown, in this volume, p. 238). An increase in scale was not, however, enough to conclude that optimal plant scale had also increased. It was thus necessary to find out whether scale had increased at lower or at equal unit costs. Or, getting back to the conceptual framework outlined in Chapter 1, whether firms were producing at any point within the shaded area to the right of line AY*1 in Figure 4. Although unit-cost figures were difficult to get, some firms gave us the data on or indications of the trends in unit costs after the adoption of flexible automation. Table 4.6 shows that of the 32 firms that provided unit-cost figures, around 60 per cent had unchanged or decreased unit costs. Actually, in the great majority of these firms unit costs had fallen. All of them were firms where scale had increased. Reductions in unit cost and increases in scale were recorded in all the industries, types of ownership and firm size studied. In addition, Indian firms provided some quite revealing calculations. Rather than providing actual figures on output and unit cost they estimated the changes in capital and labour unit costs due to the adoption of FA, given a fixed level of output. In 9 of 11 cases unit costs would have increased unless higher levels of output were reached (see Alam, in this volume, pp. 321–3). Firms where unit costs were higher stated a number of reasons for such a result. Oil-hydro, the Brazilian manufacturer of oil valves which had increased its output at higher unit cost, said that the increase was in part the result of the better quality of its products (see Quadros Carvalho, in this volume, p. 208). There was now more capital, as measured by the number of machining operations performed in the manufacture of each product, and more skilled labour per ton of output. Had the number of operations and the input of skilled labour remained at the level it had been with old technologies, unit costs would have fallen. Oil-hydro also faced a large degree of unused capacity, pressing unit costs upward.

120 CONCEPTS, METHOD AND SYNTHESIS

Table 4.5 Changes in scale of production by density of automation (number of firms)

Source: Country and firm case studies . Note a Includes 1 firm for which density of automation was not available and 20 other firms for which no specific amount was given. Table 4.6 Changes in unit costs (number of firms)

Source: Country and firm case studies.

Other firms that had seen their unit costs increase mentioned the problem of low utilization rates. Mexico’s machining subcontractor Machine-sub2, whose unit costs had increased by 15 per cent, was operating at 70 per cent of capacity while Family-brakes, a Turkish manufacturer of car components, which saw a similar increase in unit costs, was operating at 43 per cent of capacity (see Ansal, in this volume, p. 358; Domínguez and Brown, in this volume, p. 239). Some firms stated that at a more ‘normal’ capital-equipment utilization rate unit costs would have actually fallen. Mexican firms in the sample were operating at an average of around two-thirds of capacity. Turkish firms had a utilization rate of around 80 per cent, but this was about to change drastically as the Turkish economy dipped into recession towards the end of 1994. Identifying a ‘reasonable’ level of capacity utilization is no easy task. Theoretically, a CNC machine tool cutting the same product could operate at around 4.4 shifts, each of 37.5 hours per week, at just under the 100 per cent utilization rate—to take account of the initial setting-up time—if it is continuously used over a year.4 According to CNC equipment manufacturers, however, their machines perform better and last longer if they are used on average by three shifts during weekdays and at a utilization rate of at least 85 per

IMPACTS ON SCALE AND SCOPE 121

cent. Hydraul, the UK valve company, was operating at the manufacturers’ recommended utilization rate but was working at an equivalent of 3.3 shifts of 37.5 hours per week, as they were manufacturing also on Saturdays. In fact, Hydraul operated with two 10-hour shifts per day over a four-day working week i.e. 40-hour shifts. The factory stopped only on Sundays, for summer holidays and for maintenance and repair. Most of the sampled firms with an equal or higher scale and lower unit cost than before the adoption of flexible automation were operating between 2.5 and 3 shifts and had an effective utilization rate of over 70 per cent. Although an increase in capacity utilization can be seen as an indication of higher efficiency and success in the market, by the standards of Hydraul and the manufacturers of CNCs mentioned above, sampled firms seem to have some way still to go in terms of efficiency. Further, the fact that only two of the firms providing relevant data reached the manufacturers’ recommended utilization levels, and seeing that the sampled firms were among the most advanced in their respective countries, suggest that the ‘average’ or ‘going’ utilization rates and working hours are much lower than in developed countries. Nevertheless, given that any improvement in marketing efforts or a recovery in domestic or foreign demand could result only in higher utilization rates and, as a result, in further reductions in unit cost, as many firms pointed out themselves, one can safely conclude that for the 19 firms shown in Table 4.6 with an equal or lower unit cost than before the adoption of flexible automation, optimal scale had increased. The ‘equilibrium’ point, inasmuch as there is one, has shifted to the shaded area to the right of line AY*1, in Figure 4. Considering the potential for increasing the working hours and utilization rates, the fact that some of the CNC machine tools could be run unmanned for several hours, and the savings that could be made in inputs, it is conceivable that at least some of the firms that had not reduced unit costs should be capable of doing so significantly as distribution improves and demand picks up. There was considerable room for improvement because many of these firms were working less than 1.5 shifts of 37.5 hours per week and had a utilization rate of 45–70 per cent. Also, as to the Brazilian and Thai firms which reported scale increases but which had not provided figures on changes on unit costs, there is no reason to believe that their performance would have been any different from that of similar firms in the other countries studied. Brazilian firms were operating with the same types of technology and under similar economic conditions and, where figures were available, were obtaining results comparable to those of their counterparts elsewhere. Thai firms were facing an expanding and very competitive foreign market which suggested that they should have been operating more efficiently and at higher utilization rates than the others. Hence, even in the case of firms reporting a larger scale but failing to furnish figures on unit costs, one would expect that some should have been able also to reduce unit costs and, as a result, increase optimal scale. Furthermore, for those few firms where scale and unit costs actually increased, no conclusions can be reached either way regarding changes in optimal scale. On the whole, therefore, the evidence suggests that

122 CONCEPTS, METHOD AND SYNTHESIS

flexible automation and associated organizational techniques seem to have resulted in higher plant optimal scales rather than in plant ‘de-scaling’. At firm level, however, the main conclusion is somewhat different. By and large, firm scale does not seem to have changed much as a result of the adoption of flexible automation and associated organizational techniques. The main reason for this is that, in general, firms did not engage in significant product development and innovation but either copied designs or obtained them through OEM relations. There was hardly any investment in marketing and advertising. Studied firms were, therefore, mostly ‘passive’ in terms of product development and marketing strategy, having focused mainly on production. Not many attempts were made to establish distribution networks and other marketing channels. R&D, marketing and management ‘fixed’ costs actually changed very little, or declined when they did, in most firms. This conclusion seems to be applicable both to firms owning a single plant and to firm owning several plants. By way of an overall conclusion to the last three sections, a comment can be made on the relationship between technical change, economies of scale, economies of scope and multiproduct or plant economies of scale by answering the question: have economies of scope replaced economies of scale? The evidence from the case studies suggests that firms were pursuing a variety of options with regard to the source of their economies. Some firms were using flexible automation in the context of a relentless pursuit of pure product-scale economies of the traditional type. Indeed, Venezuela’s Gas-parts had its equipment almost entirely dedicated to the production of a single kind of valve. Although this may have been slightly more expensive than producing the same valve, let us say, with a transfer machine, it reduced the uncertainty that a potential change in the technical characteristics of the product or in demand would engender. Other firms were resorting to the exploitation of economies of scope. Sometimes this was done by producing new products which were not very demanding in terms of changes in product and process technology, but more often it was done through manufacturing variations of the same product or by producing a larger number of the components internally. In between these options there were other firms that seemed to be taking advantage of both product economies of scale and economies of scope, albeit to different extents. Whatever the extent of the product scale or scope economies involved, it is clear that in most cases they were accompanied by an increase of plant economies of scale or, to use Baumol et al.’s (1988) term, multiproduct scale economies. 5 Factors underlying increases in plant scale 5.1 Technical factors Production processes in the mechanical engineering industry, prior to the diffusion of FA, were classified into three types: job-shop; batch; and volume or

IMPACTS ON SCALE AND SCOPE 123

mass production (Amstead et al. 1986; Groover 1987 1996; Woodward 1965). Although precise borders between types of process are always difficult to define because of the many combinations possible between individual technologies, process lay-out, skills and levels of output, job-shop production was characterized by low volumes of output, multipurpose or conventional equipment, a functional plant lay-out where machines of the same type were grouped together, and the employment of highly skilled workers able to perform a range of different work assignments. There was a ratio of one person per machine (De Garmo et al. 1984). Jobbing included a range of activities such as the production of units to customers’ requirements, production of prototypes and the fabrication of large equipment in stages (Woodward 1965). Typical examples of jobbing include the manufacture of large items such as planes and ships, and moulds and specialized spare parts as illustrated by many of the customizedproduct firms in this research. Batch production differed from job-shop production only slightly. Multipurpose or conventional equipment was widely in use, although some of it had been modified or adapted to increase productivity (De Garmo et al. 1984; Groover 1987, 1996). Manually fed ‘automatic’ lathes would fall into this category. Process layout was similar to jobbing and so were the skill requirements of workers, although where so-called automatic equipment was in use some of the skills of the operator, including the selection of speed and direction of movement, were embodied in the machines. However, there was still one worker per machine. Perhaps, the most important difference was that production took place in lots or batches and decisions had to be made regarding the batch size. Typical batch products included consumer products, such as household appliances, and industrial equipment, such as pumps, electric motors, some autocomponents and most types of valve.5 Volume or mass production differed widely in all dimensions from the previous two types of production process. According to Groover (1987, 1996) equipment, specialized tooling and, occasionally, the whole plant were dedicated to the manufacture of a single product, which was normally in very high demand. Skills tended to be low as most of them would be embodied in the equipment. Plant lay-out was organized on the basis of a flow lay-out, where the processing and assembly facilities were situated continuously along the line of flow of the product. Typical products included screws, nuts, bolts, nails, or slightly more complex products such as automotive engine blocks. Because, as we have seen, mass production constituted a minority of manufacturing operations and was not, by and large, in use by the firms studied, it is important to recall that the firms in our sample were predominantly of the job-shop and batchmanufacturing types. Also, all sixty-two firms sampled for this research were by definition involved in machining. Some of the largest firms, particularly in Brazil and Mexico, had their own foundries, but this was not the case in most sampled firms. Heattreatment facilities were much more common, but a significant number of firms outsourced this operation. Finishing was done in-house by all firms as at this

124 CONCEPTS, METHOD AND SYNTHESIS

stage the final quality and marketability of products were determined. Autocomponent and capital-equipment firms also had assembly operations. Prior to the use of FA the conditions of operation in the sampled firms were similar to those in many other developed and developing countries which used the same conventional equipment. According to De Garmo et al. (1984), quoting research on job-shops by C.F.Carter, parts were worked at for only 5 per cent of the time, while for 95 per cent of the time they were ‘in-process’, i.e. moving or waiting until machines became available. In a ‘typical’ batch-production factory conventional machine tools were utilized for only 20 per cent of the time for cutting metal while for the remaining 80 per cent of the time they were idle. Wasted time was accounted for by positioning, tool changing or transferring, which took 10 per cent; gauging, which took 15 per cent; loading-unloading, which took 15 per cent; setting-up, which took 20 per cent, waiting or idle, which took 13 per cent; and repairing and addressing technical problems, which took the remainding 7 per cent. Time utilization could have been improved with the aid of some special-purpose automatic equipment—indeed in the case of Hydraul it was 35 per cent—but there was a maximum utilization rate of 50 per cent achievable with totally dedicated automatic equipment. De Garmo et al. (1984), again quoting research by C.F.Carter, pointed out that over the course of a year in a ‘typical’ batch-production factory, 34 per cent of the time was lost in weekends and holidays, 44 per cent was lost due to incomplete use of the second and third shifts; 2 per cent of the time the equipment was just idle for no apparent reason, 12 per cent was spent in setting-up and loading and 2 per cent in testing cutting conditions. For only 6 per cent of the time was the equipment actually engaged in productive activities. The authors added that, at factory level, productive time was even lower because of the elaborate schedules and complex planning requirements of production, which were further complicated by delays and disturbances caused by engineering design changes, vendor failures, material shortages, emergency orders and machine failures: ‘As this kind of system grows to any size, the levels of cost and chaos increase…’ (1984:930). Generally, with old technologies, the result of all this lost time was that lead times, i.e. the time between an order being received, sent to production, fulfilled and made ready for delivery, were very long. Or, put in different terms, overall factory efficiency was low. In the case of Hydraul, operators used to spend considerable time away from their machines doing nothing, the factory would be full of unfinished products, materials were wasted and no one knew where to find the right tools—all of which was reflected in a lead time for gearbox casings of 13 weeks, and longer still for the completely assembled gearbox. In the Mexican case study, lead times as high as one month for brakes and two weeks for transmissions are reported (see Domínguez and Brown, in this volume, p. 242). Lead times in Venezuelan firms were higher again, and amounted to 90 days in the case of Oil-valves and 75 days in Auto-shock (see Alonso, Tamayo and Cartaya, in this volume, p. 271).

IMPACTS ON SCALE AND SCOPE 125

The advent of FA and associated organizational techniques resulted in significant operational efficiency gains for sampled jobbing and batching firms. In processing, efficiency gains were made particularly in machining operations. CNC machine tools’ capability to integrate different operations meant that the overall number of machines was reduced. Generally, by integrating drilling, milling, boring, facing and other operations, CNC machining centres combined in a single machine 3–4 independent machines. There were significant improvements also in CNC lathe technology, meaning that lathes could undertake not only turning operations but also secondary drilling, milling and boring operations. Furthermore, some advanced CNC lathes could operate simultaneously at either end of a rotating cylindrical workpiece. Three to four independent machines were reduced to one in this case also. CNC lathes and machining centres were equipped with automatic tool changers which exchanged the tool in the spindle of the machine. Workpieces and tools, all except for the tool cutting at any one moment, could be fed in off-line. Machining centres, for instance, had two or more separate pallets that could be presented to the cutting tool. While machining was performed on one pallet, the other pallet, located away from the cutting spindle, could be safely loaded by the operator. The tool magazine or drum was also located away from the spindle, so the operator could replace other tools without stopping the machine. As already pointed out, the main advantages of integration, automatic tool changing and off-line feeding was that workpieces no longer wasted time waiting in between machines as all cutting took place in the one machine. Also there were far fewer set-ups because there were fewer machines. CNC machine tools’ production rates per unit of time were higher than those of the conventional machines they replaced.6 This was the result of CNC machine tools’ faster machining speed and their ability to change tools automatically, to move and rotate the workpiece bed continuously along several axes (three cartesian and three rotational), and to adapt to the changing conditions in the material of the workpiece.7 With conventional machines, interruptions in order to re-adjust were constant. In the case of Hydraul, two types of gearbox casing used to be machined, one requiring 180 minutes’ accumulated machining or cycle time, the other needing 240 minutes. With CNC machine tools total machining times were 52 and 72 minutes for each type of gearbox casing, respectively. Some thirty-six firms in our sample said that machining speed had increased with FA. One of the Turkish firms also furnished specific figures on machining times for twenty-two different standard parts (see Ansal, in this volume, p. 352). Savings in machining time averaged 68 per cent for these parts, ranging from 46 to 86 per cent. Savings in machining time for all those developing-country firms that provided figures ranged from 15 to 900 per cent. CNC machine tools’ accuracy improved the input utilization rate by allowing the recycling of what previously had been treated as waste. Processing efficiency was therefore increased and higher output obtained. The accuracy of a CNC machine tool is the result of the capability of its control unit to divide the range of an axis, or several of them simultaneously, into closely spaced points and position the machine bed and spindle at the exact location desired. Positioning at

126 CONCEPTS, METHOD AND SYNTHESIS

the same point can be repeated as many times as necessary as the information remains stored in the control unit (Groover 1987). The case of Book-tech, where 10 per cent of components were manufactured from offcuts of other components, has been mentioned. Most of the firms in our sample pointed to the better utilization of castings: specifically, waste—mainly swarf and metal offcuts—was reduced by firms in Venezuela by 60–90 per cent and by up to 40 per cent by firms in Mexico (see Alonso, Tamayo and Cartaya, in this volume, p. 287; Domínguez and Brown, in this volume, p. 244). Several Mexican firms had achieved scrap rates of less than 1 per cent per year (Domínguez and Brown, in this volume, p. 243). Cellular manufacturing (CM) arrangements further increased machine and factory processing efficiency. In essence, CM is an attempt to combine the flexibility of the functional lay-out with the efficiency of the flow lay-out (Groover 1987; Steudel and Desruelle 1992). CM involves two stages. The first, commonly known as group technology (GT), is based on the identification, classification and coding of families of parts with similar design and manufacturing characteristics.8 The second furthers the GT concept by devising a ‘composite’ part with all the design and manufacturing attributes possessed by the various individuals in the family. A machine cell is then designed to provide all the necessary attributes. There are different types of cell, but the most common among the sampled firms was the group machine cell, consisting of more than one machine used collectively to produce one or more families of parts, with human operators performing the material-handling functions. Although firms were not precise as to the extent of adoption of CM, and none of them actually provided figures on the impact of this organizational technique, it was widely accepted that by taking the machines to the part rather than the part to the machine, machine efficiency is raised, and travel and waiting times are lowered, thus reducing lead times for the parts and increasing overall factory efficiency. Savings in production time were reaped following the introduction of JIT delivery. By organizing the movement of parts on a ‘pull’ rather than a ‘push’ basis through the production process, less time was wasted in each workstation (Cheng and Podolsky 1993). With the ‘push’ methods used in the past, whatever part was manufactured in one station was ‘pushed’ on to the next station irrespective of whether the receiving station was prepared for it, resulting in confusion, mistakes and high levels of work-in-progress (WIP). WIP refers to the amount of workpieces, parts or components that are being processed or waiting to be processed. In JIT ‘pull’ systems, the part is moved to the next station only when it is needed: it moves on when the receiving station requests it. This is then repeated down the whole production line. Efficiency is further enhanced by organizing the delivery of raw materials and the dispatch of final products so that their on-site stay is minimized. This involves increasing the frequency of deliveries of raw materials but a reduction in their size, the introduction of very precise schedules of delivery and a reduced number of suppliers which offer quality at reasonable prices. Altogether, 11 of the 14 firms that provided figures on changes in raw materials’ inventory showed reductions, with Venezuelan and Indian companies’ inventories falling by an average of 65 per cent and 50 per

IMPACTS ON SCALE AND SCOPE 127

cent, respectively (see Alam, in this volume, p. 324; Alonso, Tamayo and Cartaya, in this volume, p. 287).9 In inspection and testing operations additional efficiency gains arose out of the introduction of TQM. Apart from reduction in the number of defects that emerge from the technical capabilities of CNC machines, and the more precise electronic measuring equipment available, introducing systems and procedures to ensure quality throughout the activities of the firm can further reduce the need for reworking or scrapping. According to Groover (1996) quality has two dimensions: product features and freedom from deficiencies. Product features result from a number of attributes such as product design, aesthetic appeal, function and performance, reliability and dependability, durability and reputation of product or producer. Freedom from deficiencies means that the product performs the functions for which it is made and has no technical defects. Traditionally, quality control in engineering was the responsibility of the quality department. The problem with this approach was that, because quality control employees were not involved in the actual production operations, it was difficult for them to identify the causes of errors and to suggest solutions to problems (Steudel and Desruelle 1992). In TQM each operator or group of operators is made responsible for the quality of their output, so that it becomes easier to identify sources of errors and problems and to get informed suggestions on how to solve them. Other TQM practices introduced by some of the sampled firms included: obtaining regular information about customer satisfaction; instituting mechanisms of control over incoming material; and establishing production process control and evaluation procedures such as material movement control, preventive maintenance and statistical process control. Not all aspects of TQM were, however, applied by the surveyed firms. Since little product design took place among surveyed firms, there was not much improvement in quality arising out of this aspect. The intensity of application varied sometimes. Worker involvement in quality control was more intense among some of the Mexican and Brazilian firms, although statistical process control (which involved regular sampling and charting of key characteristics or attributes of a part or product and analysing its variations around a band considered acceptable) was widely used in all countries. Maintenance of CNC machine tools was generally considered ‘easy’ by the sampled firms, although it is unclear from the country studies if this means that preventive maintenance was carried out. Despite some diversity in the application of TQM approaches, there were some tangible indications of improvements in quality and efficiency, as final product rejection rates fell in India to around one-third of what they were with conventional technologies, while in Venezuela they were down by 63–98 per cent (see Alam, in this volume, p. 320; Alonso, Tamayo and Cartaya, in this volume, p. 287). Finally, efficiency gains arose in overall process control operations and in management of plant-level activities. The complexities and difficulties for planning, scheduling and managing manufacturing in the context of disorganized and ‘chaotic’ factories have already been pointed out. With the arrival of FA,

128 CONCEPTS, METHOD AND SYNTHESIS

however, production control and management eased significantly. There were fewer machines to worry about and, in some firms, computers and production management software allowed better planning and scheduling of the process. JIT approaches to inventory reduction meant that there was less inventory to look after, further simplifying planning and scheduling. Labour management also became simpler because there were less production workers, one worker overseeing more than one machine. A few new problems were encountered, for example in tool management—as tools have become a significant part of cost, but in general the complexity of production management and control has been reduced. All in all, efficiency gains in processing operations, in inspection and testing, and in process control and management arising out of flexible automation and associated organizational techniques have had a major impact in plant efficiency, and therefore in output or production capacity, of the sampled firms. Improvements could be seen in shopfloors as the production process was much more orderly and clean, with less work in progress and final product inventories, and with workers spending more of their time on the machine. Machine utilization ratios had increased in 14 firms of our sample by between 7 and 98 per cent. In four Indian firms average machine utilization had increased from 79 to 89 per cent (see Alam, in this volume, p. 321). Labour productivity increased by between 4 per cent and 325 per cent in Turkey, by between 16 per cent and 137 per cent in Brazil, by between 16 per cent and 650 per cent in Mexico and by between 75 per cent and 810 per cent in Venezuela (see Alonso, Tamayo and Cartaya, in this volume, p. 280; Ansal, in this volume, p. 354; Domínguez and Brown, in this volume, pp. 243–4; Quadros Carvalho, in this volume, p. 212). As a result, reported lead times fell significantly. All Mexican and Venezuelan firms reported reductions in lead times to an average of 25 per cent of what they were before (see Alonso, Tamayo and Cartaya, in this volume, p. 293; Domínguez and Brown, in this volume, pp. 242–3).10 Before moving on to the next section one last remark is called for on the ‘smallness’ and divisibility of CNC machine tools. Regarding the ‘smallness’ or physical size of the new CNC machine tools, a feature extolled in the ‘modern technology’ literature, most firms in the sample acknowledged having saved on building space, as the lesser number of new machines required less area compared with older conventional technologies. In this sense, CNC machine tools are ‘smaller’. Firms pointed out that as plants were built to house the previous number of machines, the result had been more empty space available for future expansion. Old machines could be retained for use in emergencies for a much longer period than if new smaller plants had been built. The reduction in physical size of the new machines does not of itself entail reduced plant size. The issue of the ‘divisibility’ of CNC machine tools is slightly more complex. If one considers that the trend among CNC machine tools is to integrate different functions into a single machine, then there is a loss of ‘divisibility’, at least as far as functions are concerned. But, divisibility needs to be looked at in terms of the productive capacity of the equipment, and this is quite difficult to determine with metal-cutting machine tools.

IMPACTS ON SCALE AND SCOPE 129

For instance, even in the case of narrowly defined CNC horizontal lathes, makes and models differ considerably in characteristics such as design, maximum length of workingpiece bed, capacity to move the workingpiece bed, depth of cut, types of computer control, measurement instruments and motor power, among others. Hence, it is not obvious which parameter should be used for measuring productive capacity. Trade journals define the capacity of a CNC lathe in terms of the dimensions and weight of the workpiece that the machine is capable of turning (Machine Production 1993). The problem with this indicator is that it gives no clue as to capacity in relation to output volume. Others use the weight per unit or the motor power (presumably both are positively correlated) as the indicator of productive capacity because it is the motor that determines the cutting speed and power of a lathe (Jacobsson 1986). The main limitations with this indicator are that motor power is positively correlated with the dimensions and weight of the workpiece and that the power required to machine a specific workpiece varies in relation to the hardness of the material being cut or to rigidity of the machine tool used, the shape of the workpiece and the depth of cut.11 Because most smaller workpieces can still be turned in a powerful lathe, although obviously not the other way round, and hence the full power of the motor applied to comparable workpieces, the second indicator is normally accepted as a proxy for productive capacity in terms of output. Examining the motor power capacity of around 700 different CNC horizontal lathe models manufactured by 112 companies from France, Germany, Italy, Japan, Korea, Switzerland, Taiwan, the UK and the US (Machine Production 1993), it was found that there was a wide range of productive capacities. Motor power as measured in kilowatts (kW) ranged from 1kW to 1,000kW, with nearly all the values in the range up to 100kW represented.12 Does this mean that CNC lathes are divisible? It has already been pointed out that the less-powerful CNC horizontal lathes cannot be used for large workpieces, which introduces a comparability problem across levels of power. But, more importantly, when the equivalent indicator for 248 conventional horizontal lathes from manufacturers from the same countries as the CNC lathes, also available in Machine Production (1993), was estimated, it was found that motor power ranged from 0.4kW to 110kW, with around 50 per cent of the values being less than 10kW. Similarly, conventional drilling and milling machines were equipped with motor power ranging from 0.9kW to 12kW (Machine Production 1993). Motor power in machining centres used to replace drilling and milling machines ranged from 2kW to 135kW, with most of the values concentrated below 65kW (Machine Production 1993).13 This suggests that machine tools have always been ‘divisible’, in so far as they can be. Conventional machines operated across a range of productive capacity and so do CNC machine tools. It is true that CNC machine tools have a wider range of power possibilities than conventional equipment, but they also have a wider range of functions. If anything, the minimum capacity seemed also to be increasing with CNC machine tools.

130 CONCEPTS, METHOD AND SYNTHESIS

5.2 Economic factors The most important economic factor underlying increasing optimal plant scales was the price of FA. Table 4.6 shows the changes in unit costs for a number of firms in our sample. Three-quarters of the firms that reported data on changes in unit costs stated that capital unit costs had increased. Increases were reported across all countries, types of industry and ownership and sizes. This was the result in part of increases in capital costs, although the extent varied widely. Nine Indian firms reported that capital expenditure increased with FA by 10–250 per cent for the same level of output as with old technologies (Alam, in this volume, p. 322). For one Mexican firm the adoption of new technologies implied a threefold increase in investment (Domínguez and Brown, in this volume, p. 247). Oilhydro, a Brazilian manufacturer of oil valves, declared an average annual investment in CNC machine tools of US$1.1 million since acquisitions started, which was higher than any previous record of capital investment (see Quadros Carvalho, in this volume, p. 211). Two of the firms that had reduced their capital unit costs had capacity utilization rates of over 90 per cent while the other seven firms showed capital unit-cost reductions smaller than the reductions in total unit costs, suggesting that capital costs were becoming a larger proportion of unit costs. In nearly all firms that provided data the share of capital unit costs in total unit costs rose. At the base of the increase in capital unit costs would seem to lie the relatively more ‘expensive’ CNC machine tools vis-à-vis conventional equipment. To address the issue of the relative prices of new and old machine tools it is necessary to look into the trends in price over the years. The approximate unit price of a locally produced German, US and Japanese CNC lathe in 1975 was US$150,700, $129,700 and $48,600, respectively (see Table 4.7). Differences in price reflected differences in the technical characteristics of CNC lathes, with German and US manufacturers also supplying the largest capacity and the more sophisticated lathes, while Japanese producers concentrating on smaller capacity lathes. German and Japanese CNC lathe prices have nearly doubled since 1975, mainly as a result of currency appreciation and of German manufacturers’ continued strategy, increasingly also being pursued by Japanese companies, of manufacturing ever more technically complex and high performance CNC lathes.14 Japanese CNC lathe prices were converging with US lathe prices by the mid-1990s. CNC lathe prices in Italy, Korea and Taiwan also show an upward trend. US prices show a more cyclical trend, increasing up to the early 1980s, falling back by the late 1980s, then increasing again the early 1990s only to fall again after 1992. The only case of a clear-cut downward trend in prices is India, whose CNC lathe prices are approaching those of Taiwan and Korea, the cheapest in 1994.15 At first sight, the data in Table 4.7 do not suggest a significant reduction in CNC lathe prices over the years. Nonetheless, one has to be careful with this conclusion as the choice of equipment depends on two additional considerations:

IMPACTS ON SCALE AND SCOPE 131

the price actually being faced in the market, including imports, and the price of the alternative technology. In economies open to machine-tool imports, users of equipment face not only the market price of their locally made equipment but also the prices of CNC lathes made elsewhere. In 1987, a US lathe purchaser faced the choice of a US $240,700 German lathe, a US$138,700 locally made lathe, a US$70,800 Japanese lathe or the much cheaper US$36,500 Taiwanese lathe. There will have been functional, technical, performance, after-sales service and reputational differences between each CNC lathe, and transport costs to the US had to be added to the final price; but there was also a 6.5 times price difference between the most expensive and inexpensive lathes. The existence of substitution possibilities between lathes from different sources means that when examining price trends, all countries need to be included. When this is done, it becomes evident that there was indeed a downward price trend in CNC lathes, with prices dropping by two-thirds between 1975 and 1994. Turning to relative prices of CNC and conventional machine tools, a clear downward trend can be observed. According to Ayres (1991), when numerically controlled lathes began to enter the US market around 1973 their price relative to non-numerically controlled lathes was around 8–10 times more: machining centres were around 18 times the cost of the conventional machine tools they replaced. Since that time, both the unit values of CNC lathes and machining centres and their unit value relative to non-CNC machines have been coming down, but not exactly in the same proportions. Taking the ratio of CNC lathe prices to conventional lathe prices first, there has been a significant drop over the years (see Table 4.8). In the US for instance, the relative price of CNC lathes fell from an approximate eight times in 1975, to around four times in the mid-1980s and to around 2.5 times in 1994. This trend is present in all producing countries except Taiwan. Yet, everywhere, CNC lathe prices are still considerably higher than conventional lathes, and they will not continue to fall for the simple reason that fewer and fewer conventional lathes are being produced and, therefore, there will be an insufficient number of the latter with which significant comparisons can be made. In 1980 Germany, Italy, Japan, France, the UK and the US produced 72,801 conventional lathes (see Appendix A3). In 1994 the same countries produced 8,623 units. India, which produced 5,899 conventional lathes in 1980, had reduced its output to 2,049 units in 1994. Even Taiwan, the largest producer of conventional lathes, manufactured in 1994 only 70 per cent of its peak production in 1981. As the French trade journal Machine Production very aptly put it: ‘Plus les jours passeront, plus il sera dificile de trouver autrement qu’en material d’ocassion des tours destinés a une conduite entièrement manuelle’ (‘As time goes by, it will be more difficult to find a completely manual lathe, except perhaps at the second-hand market’) (1993: 189).16 Trends in the price of machining centres were slightly different. Since 1986, machining centre prices have fallen in France, Italy, India, Korea, the UK and the US but have remained similar or have increased in Germany, Japan and Taiwan (see Table 4.7). Currency appreciation and better quality may be factors behind Germany and Japan’s machining centre price rises. Taiwanese machining

Source: Our own elaboration on the basis of data provided CECIMO.

Table 4.7 Unit values of CNC metal-cutting machine tools in selected countries, 1975–94 (in US$1,000s)

132 CONCEPTS, METHOD AND SYNTHESIS

IMPACTS ON SCALE AND SCOPE 133

Table 4.8 Relative unit values of CNC and non-CNC lathes

Source: Our own elaboration on the basis of data provided by CECIMO. Table 4.9 Relative unit values of machining centres and non-CNC machine tools (excluding lathes)

Source: Our own elaboration on the basis of data provided by CECIMO.

centres, much in the way that Japanese machining centres had advanced ten years earlier, also began moving up the sophistication ladder, resulting in higher prices. Considering the substitution possibilities between machining centres from different sources, it is conceivable that, as a whole, machining centre prices have fallen over the years. However, prices of conventional drilling or milling machines seem to have been falling more rapidly than have prices of conventional lathes or machining centres, despite manufacturers’ attempts to tap into the parts’ repair and the more serious ‘do-it-yourself markets. Thus, machining centre prices, although falling, remain around ten times the cost of the conventional machines they replace (see Table 4.9). In sum, what the preceding discussion suggests is that while prices of CNC lathes have been falling, reductions are partially because of entry of new,

134 CONCEPTS, METHOD AND SYNTHESIS

cheaper, lower performance machine tools from abroad, initially from Japan and then from Taiwan. Hence, price reductions were not primarily the result of any clear-cut downward trend in costs due to standardization. It suggests that whatever the reduction in costs or degree of technological advancement, CNC machine tools have remained more expensive to produce relative to the technologies they replace. It suggests that there are some differences between relative price reductions of CNC lathes and machining centres, with the former’s price relative to that of conventional lathes falling much more since their introduction into the market around 1975. This is the result of the fact that machining centre technology was much more ‘revolutionary’, in terms of its technical and economic potential, than that of the conventional technologies it replaced. Thus, it was able to keep relative prices higher than those of lathes, which, despite the advances arising out of microelectronics, still remained a comparatively well-established technology. Finally, it suggests that whatever their relative prices, conventional machines have been, and will continue to be, displaced from the producer market and that in order to manufacture engineering parts, components and products CNC machine tools will have to be increasingly in use. 6 Other price and efficiency effects Thus far the chapter has been concerned mainly with the impact of FA on scale and scope. Yet, there are a number of additional effects on overall efficiency, variable costs and prices that go beyond scale and scope but may still have important consequences for location of production in developing countries and on which the study was able to throw some light. One effect relates to whether CNC machine tools are capital saving, i.e. whether they use less capital per unit of output than do conventional technologies. Our data were not conclusive but merely suggestive in this respect. The survey showed both decreases and increases in capital unit cost. The Turkish firm Transaxle had a reduction in capital unit costs of 33.3 per cent in rear axles and 28.6 per cent in front spindles. In the case of Hydraul, capital unit costs increased by around 20 per cent, but if inflation is adjusted for there were savings in capital per unit of output. In Turkish firms like Trans-axle, Localvalves and Glass-moulds, and in Venezuelan firms such as Auto-machining, Oilvalves, Water-valves, Machine-spares and Gas-parts, there were savings even if inflation is not adjusted for (see Ansal, in this volume, p. 360; Alonso, Tamayo and Cartaya, in this volume, pp. 284–7). Generally, CNC machine tools proved to be capital saving in firms where the amount of installed capacity closely matched demand, the equipment was utilized as efficiently as technically possible and for as many hours as necessary, and if there were no constraints on skill availability. For the sample as a whole, adjusting for inflation and increasing capacity utilization would have meant that some additional firms would have saved capital unit costs, but there was a significant number of others that would not.

IMPACTS ON SCALE AND SCOPE 135

These results are similar to those reported by Soete and Dosi (1983), Edquist and Jacobsson (1988) and Jacobsson (1986). For instance, on the basis of their own research on Swedish engineering firms and a study by Boon on Dutch and Swedish firms, Edquist and Jacobsson found both increases and decreases in capital unit costs. However, it was found out that a reduction in capital unit costs depended also on the extent of reductions in work-in-progress and inventory and shift working practices prevalent in the countries in question. In Sweden there were insufficient skilled workers willing to work shifts with conventional machine tools, meaning that only one shift was possible with conventional technologies, while at least two were possible with CNC machine tools because CNC machine tools were capable of operating unmanned, while a single worker could oversee more than one CNC machine tool. The second effect concerns labour and training costs. If there are reservations about the potential of CNC machine tools to save capital, there are certainly none with respect to their labour-saving potential. Labour unit costs fell significantly in most sampled firms, with the exception of certain firms in Turkey that were facing major government-determined real-wage increases. Changes in employment were a key factor explaining reductions in labour unit costs. There was a net loss of 168 jobs or 1 per cent of total employment in the 31 firms that gave figures for changes in employment. Of these 31 firms, 16 reduced their personnel by a total of 3,183 positions. Most of the firms that gained in employment were Thai, and these were facing large increases in demand due to exports. Indian firms pointed out that use of CNC machine tools should have led also to a reduction in employment, but because of government restrictions to laying-off personnel there had been no reduction (see Alam, in this volume, p. 323).17 Many firms pointed out that training costs were increasing as a result of the need to adapt the skills of workers to the new technologies and organizational concepts. The question of whether the new technologies required more or less skills figured prominently during the interviews. In Brazil, Mexico, Thailand and Venezuela CNC machine-tool operators were considered more skilled than those operating conventional machine tools, and were paid and trained accordingly. Workers had more responsibility over the machine, could perform simple programming operations, were involved in quality control and were generally aware of what was happening in other sub-processes. In countries such as India and Turkey, where workers limited themselves to adjusting the workpiece and starting the CNC machine tool, the skills required by the operator were seen to be fewer, as the programming and overseeing of the operation were carried out by an engineer. In most countries, however, the proportion of engineers in the total workforce had increased since the adoption of FA. All together, the interviews suggested that the combined number of skills required to operate a CNC machine tool is higher, but that these skills do not have to be ‘embodied’ in the individual machinist, but rather have to be available in the firm. Many firms pointed out, however, that not all the combinations of skills were generally available in the market and that that was a factor limiting further and efficient use of flexible automation.

136 CONCEPTS, METHOD AND SYNTHESIS

Thai firms, in addition, pointed out that, because there were less machines to operate and because a single worker could oversee several machines at the same time, the total number of skilled personnel required was lower than with old technologies. This was seen as a great advantage of CNC machine tools, particularly since Thai firms faced an acute shortage of skilled machinists or engineers. Thus, in this sense, the new technologies could be seen as ‘deskilling’. Furthermore, if one considers that productivity per operator increases significantly, the skill content per unit of output decreases even more, as Edquist and Jacobsson (1988) have already pointed out. One final consequence of FA, repeatedly raised by firms, was the drop in the unit prices of engineering goods. Seven Venezuelan, three Mexican, one Brazilian and two Turkish firms claimed and provided figures indicating either no change or else a reduction in the prices of their goods. Often the price reductions were higher than were the falls in unit cost, thus squeezing profits. In Mexico reductions ranged from 15 to 40 per cent, while in Venezuela they ranged from 30 to 75 per cent (see Alonso, Tamayo and Cartaya, in this volume, p. 293; Domínguez and Brown, in this volume, p. 251). All but one of the Mexican firms claimed reductions in profitability of up to 50 per cent. In Brazil the cost of a water pump rose from US$2,431 in 1985 to $2,500 in 1993, a reduction in real terms (see Quadros Carvalho, in this volume, p. 177). In Turkey, the price for the bottom mould for a 5kg tin box fell from US$4,243 in 1989 to $2,317 in 1994, while the cost of a rear axle fell during the same period from $36.6 to $22.7 (see Ansal, in this volume, p. 357). There were a few firms that reported price increases but in these cases product technology also had changed significantly. One of the Mexican firms could raise prices by 30 per cent but only after moving from a carburettor to a fuel-injection-based system (see Domínguez and Brown, in this volume, p. 252). This last point suggests that the use of CNC machine tools does not necessarily lead to increases in unit price. Competition throughout the engineering industry seemed to be getting stiffer and increasingly more firms were acquiring CNC machine tools, thus making it possible to offer the same or better quality at even lower prices. Most firms were resorting to price competition as their main competitive weapon, which further indicated that, unless firms expanded their design and marketing capabilities, it would not be possible for them to reap the full advantages of the new technologies. 7 Conclusions This chapter aimed to examine the impact of FA and associated organizational techniques on scale and scope, at the product, plant and firm levels. At product level, by reducing setting-up times and costs FA had the potential to reduce batch sizes and therefore to reduce the ‘optimal product scale’, although whether such potential was actualized depended on other factors, particularly the size of the demand being faced. It was found that FA does allow for wider product scope and resulting economies of scope, although such diversity was often the result of

IMPACTS ON SCALE AND SCOPE 137

variants of the same product and the production of components previously outsourced. In addition, although setting-up times and costs have fallen significantly, some setting-up operations are still required and there are also some new fixed costs linked to each new product, which continues to put some limits on the extent of product variety. At plant level, FA has led to higher ‘optimal plant scale’ or ‘scaling-up’ as most firms which furnished relevant indicators showed higher levels of output at lower unit cost than were achieved with old technologies. Higher levels of output were achieved in nearly all the sampled firms, accounted for in part by an increase in the production of components. It was also found that higher scales had been accompanied by higher unit costs in some cases, meaning that no firm conclusion could be reached in terms of optimal scale. Nevertheless, it was revealing that many of the firms facing higher scales and unit costs claimed that with improved capacity utilization, unit costs would have fallen beyond those achieved with conventional technologies, thus also leading to ‘scaling-up’. Indeed, to varying degrees, the problem of relatively low capacity utilization rates seem to have been present even in the most efficient firms in the sample At firm level, FA did not lead to further increases in ‘optimal scale’ as the majority of firms did not significantly increase R&D, marketing and administrative costs after the introduction of the new technologies. Enterprises were able to concentrate exclusively on production, leaving activities such as design, development, advertising and distribution to business partners. Technical and economic reasons were identified as underlying increases in optimal plant scale. Prior to the adoption of FA surveyed firms were organized along job-shop and batch-manufacturing lines. These forms of organization involved the use of several slow conventional machine tools which resulted in long setting-up, production and workpiece waiting times and complex production planning, scheduling and routeing problems which often were not fully resolved, leading to confusion and further delay. The arrival of CNC machine tools meant the availability of equipment which has higher production rates per unit of time, which integrates different operations into a single machine saving on setting-up time in each machine and on waiting time between machines, and which is much more accurate, so improving input utilization rates and allowing for the use of material that previously was wasted. The larger output that can be obtained from advances in the technical characteristics of machines can be expanded further by the adoption of CM techniques, which reduce workpiece travelling time; JIT techniques, which increase productive efficiency and reduce confusion by providing the necessary raw material and work-in-progress exactly when needed; and TQM techniques, which can provide additional reductions in defect rates. The new CNC machine tools are, however, much more ‘expensive’ than the technologies they replace and therefore have to produce large volumes of output if they are to be fully amortized. Other key findings of the chapter are that the change from old technologies to CNC machine tools has meant little in terms of the ‘divisibility’ of machine tools, and that it has led, if anything, to larger minimum capacities. Also, while it was normally the case that CNC machine tools were labour and skill saving, it was often

138 CONCEPTS, METHOD AND SYNTHESIS

the case that, when used as efficiently as possible, the new technologies resulted in capital savings. One final conclusion is that, as the use of CNC machine tools extended through the economy, the ensuing competition stiffened, forcing companies to drop prices significantly and so reduce their profitability rate. The implications of this and other findings for the location of production in developing countries will be the topic discussed in the next chapter. Notes 1 To complete one machining operation and begin another, the operator must, in addition, monitor or control the cutter path during cutting, reposition work or tools after each operation including starting and stopping the machine and inspecting the parts. 2 Scherer et al. (1975) point also to what they call ‘real world’ complications to the optimal lot model. Some paint factories operate on the basis of the ‘80–20' rule; that is, the best selling 20 per cent of products account for 80 per cent of total sales’ volume. ‘Low-volume’ paints have to be produced anyway, despite the fact that setting-up costs in one case they studied trebled the unit cost of production, because not doing so would imply losing customers for ‘volume’ products. In many ‘volume’ shoe factories, the best-selling model accounted for 50 per cent of total output with the remainder being divided among 10–20 models. A further complication emerges in beer packaging because of the variety of containers: ‘For low volume packages, brewers face a Hobson’s choice among small machines, poorly utilised special machines, high set up costs, less frequent processing and hence higher inventories (with the attendant physical deterioration), or forgoing the sales that the special package permits’ (Scherer et al. 1975:305). 3 A very relevant case in point is the joint-venture between Volvo and Mitsubishi, located in Born, in The Netherlands. The plant actually produces two different brands of car, Mitsubishi and Volvo, in several ‘models’, all coming out of the same production line. However, although the cars look very different to the consumer, more than 95 per cent of components in all ‘models’ are exactly the same. Achieving such commonality in components involved several years of jointdesign and product development and very large investments by both companies. 4 Utilization rate was defined for the purposes of this research as the time for which machines are actually operating, i.e. excluding setting-up times and stop-overs for breakdowns, maintenance and absence of operators. 5 According to Groover (1987), around 75 per cent of all engineering manufacturing takes place in lots of fifty pieces or less. 6 Production rates per unit of time are sometimes referred to as ‘cycle times’, in which case they would be lower than the cycle times of conventional machine tools. 7 Some manufacturers of machine tools point out that certain transfer lines are faster than CNC machine tools when used to machine a workpiece. However, bearing in mind the retooling costs of a transfer line, many manufacturers were using CNC machine tools for the production of specialized goods: in the long run, it was said, CNC tools would prove more cost effective.

IMPACTS ON SCALE AND SCOPE 139

8 Primrose (1992) points out that ‘family planning’ techniques, where a group of machines were dedicated to the production of one family of crucial components, has been in use in the UK since the 1930s. He indicates that GT techniques were used in Russia at around the same time and introduced in Europe in the late 1950s. 9 Other unit costs, which include the major cost of energy, rose in a significant number of enterprises, notably in Venezuela (see Alonso, Tamayo and Cartaya, in this volume, p. 290) and Turkey (see Ansal, in this volume, pp. 361–2). Despite the much lower energy requirement of the AC and DC motors driving the spindles of CNC lathes and machining centres, the share of this factor rose in 11 firms. This was the result of price liberalization and privatization of public services, particularly in Venezuela, which had resulted in huge jumps in the prices of utilities. 10 A Mexican firm stated that lead-time reductions depended also on reductions in administrative lead times (see Domínguez and Brown, in this volume, p. 244). It could reduce its production lead time from a month to a day, but getting the paperwork ready for the sale still took a week. As a result the firm was considering undertaking a major administrative reform prior to undertaking additional investment in CNC machine tools. 11 The second limitation could be controlled for by using a ‘standard’ workpiece and tool and by stipulating ‘standard’ operation conditions. 12 Motor power is the only technical characteristic common to all lathes and other machine tools. 13 It is argued that flexible automation is ‘divisible’ not only with regard to old technology in terms of productive capacity but also in terms of investment, in the sense that it is not necessary to disburse in one single payment the full cost of the new equipment. This seems to be correct in the very restricted case of an FMS replacing a transfer line because the potential user can ‘build-up’ over a period of time an FMS equivalent to a transfer line. But until the FMS is up and running it may not be possible to manufacture the same good and the lathes or machining centres would have to be dedicated to the production of other goods and components, making the comparison between old and new technologies, strictly speaking, inapplicable. 14 In Germany, prices of CNC lathes fell during the early 1980s following the recession of German industry after the second oil price shock and the competition from Japanese producers. In general, demand conditions are also an important determinant of machine tool prices. Manufacturers face regular sharp changes in demand resulting in considerable stock variations and/or in large price fluctuations. To protect themselves, manufacturers generally operate with very large unit profits which they can cut drastically during the downswing of the economic cycle. It also gives them considerable leeway for discounting when they want to clinch a sale. 15 In 1990 the cost of a Brazilian made CNC lathe manufactured by ROMI was US $90,000 while a universal manual lathe cost around US$40,000 (see Quadros Carvalho, in this volume, p. 210). 16 In order to protect themselves from the loss of technological dominance conventional lathe manufacturers, at least the few that remain producing only conventional lathes (most have switched to manufacturing CNC lathes), have reacted by switching emphasis to the parts’ repair and service market, departments or firms that specialize in prototypes and manufacturers of luxurious metal

140 CONCEPTS, METHOD AND SYNTHESIS

products. There is a growing market also for ‘classical’ machine tools among collectors and households. Taiwanese manufacturers have turned also to poorer developing country markets, where the repair and service market is large. 17 Labour unit costs were affected also by changes in real wages. In Latin America, Venezuelan firms experienced large reductions in real wages but in Brazil and Mexico real wages had been kept constant or increased slightly. In Turkey real wages were significantly increased by the government.

5 FLEXIBLE AUTOMATION AND LOCATION OF PRODUCTION IN DEVELOPING COUNTRIES

1 Introduction This chapter focuses on the impact of FA on the location of industrial production in developing countries. It analyses the direct effect of changes in scale and scope, which have been our main concerns. However, in order to make a proper appraisal of the impact of FA on the location of production, other effects arising out of the diffusion of FA, but not necessarily through scale and scope, and which may have an important bearing on location of production, are addressed. ‘Location of production’ refers to the spatial locus of industrial operations. It is normally discussed in the context of regional development; but in this case a broader approach is taken so as to focus on countries particularly developing countries, rather than regions. Perhaps the most important distinction between regions and countries is not so much their different sizes, both in population and space terms, but the broader range of options and policies available at the country level. True, trade liberalization and globalization have the effect of restricting the freedom with which nation states operate, but national borders still remain very much in evidence, providing some room for manoeuvre both to national governments and domestic firms. Furthermore, the fact that location decisions involve the sequential consideration of factors at successively more limited geographical scales, beginning with countries and only then moving into regions, suggests that a country focus is quite valid (Chapman and Walker 1990). Focusing at the country level, therefore, implies that the analysis should concentrate not only on factors traditionally affecting location such as costs, including transport, and demand but also on other important determinants of location such as choice of technique and industry entry considerations. Section 2 of this chapter examines the impact of FA on production and cost conditions in the engineering industry, and seeks to determine whether the new conditions favour or hinder the location of manufacturing firms in a developing country. Section 3 focuses on the impact of changes in scale and scope on barriers to entry to the engineering industry. In particular, the minimum efficient scale, concentration, minimum investment requirement and vertical integration barriers to entry will be discussed. Section 4 addresses the issues of whether and how transport costs, just-in-time approaches and the availability of basic

142 CONCEPTS, METHOD AND SYNTHESIS

infrastructure affect the establishment of industrial operations in developing countries. Section 5 looks into the impact of changes in factor prices and biases, while Section 6 looks into the importance of domestic demand in attracting firms. The chapter ends with an overall assessment of the impact of FA and other factors on location of production in developing countries. 2 Flexible automation and production and cost conditions in engineering As we have seen in previous chapters, it has been commonplace to characterize the mechanical engineering industry as ‘mass production’. While mass production conditions were available to the engineering industry, we have argued that they were by no means extensive or dominant, certainly not in developing countries; rather, engineering has been based on job-shop and batch-production conditions including the use of various conventional or multipurpose equipment and skilled labour producing in relatively small lots. The predominance of jobshop and batch manufacturing meant that production conditions were characterized by disorganization, considerable time wastage and low efficiency. Even under mass production conditions, engineering factories could be, at best, operative for half of the time available during a year. According to Freeman (1982) and UNIDO (1991b) the movement from batchbased towards flow-based production processes resulted in generally significant technical and economic improvements. Freeman (1982) gives the example of the introduction of catalytic cracking in the petroleum industry, beginning in the 1920s, which changed the production process from one based on cooking heavy fractions of oil in a vessel at atmospheric temperature to one where the fluid feedstock travelled under controlled pressure and temperature along a number of tanks and tubes aided by a chemical catalyst. Over the years, the introduction of the new process resulted in the erection of much more standardized large-scale capital-intensive plants built by specialized contractors. It led to reductions in labour costs of around 98 per cent, which were associated with increases in labour productivity of nearly 800 per cent, savings in raw materials and energy costs of more than 50 per cent and higher quality and more uniform products. There was also a reduction in the variance of these cost and efficiency factors across the industry. But it has led also to intense competition and overcapacity in the industry, driving profitability downwards, particularly in the area of bulk petrochemicals. Although engineering products are still far from being made entirely by fluid processes and also far from being as income-inelastic as are petrochemical ‘commodities’, there are some changes in the production and cost conditions arising from the use of FA in engineering which resemble the gains ensuing from the transition from batch-based to flow-based production processes.1 To begin with, production processes are becoming much more homogeneous across the industry. Unlike in the past, when there was much more diversity in the technologies in use between and within large and small batch producers,

FA AND LOCATION OF PRODUCTION 143

today both are acquiring the same two basic technologies: CNC lathes and machining centres. Production organization also is converging across firms as a result of cellular manufacturing. Admittedly, CNC lathes and machining centres have diverse capacities, but the basic technologies remain the same. Larger producers may increase the degree of integration between the equipment and move into FMS in order to have higher production rates or be able to operate unmanned for several hours, but this involves changes only in the material handling and transport equipment, and not in the basic equipment. The fact that even producers of standardized goods are replacing their previous equipment with CNC machine tools and FMS (because of the costs of switching and the uncertainty attached to transfer machines and moving assembly lines) is further increasing the degree of homogeneity across industry. Engineering processes, particularly machining operations, are, therefore, increasingly technologically similar and ‘look alike’, whatever they produce or wherever they are located. Second, like petrochemicals beginning in the 1920s, engineering cost and efficiency parameters have radically changed and also would seem to be converging across the industry. Labour and raw-material costs are losing out relative to capital costs. This is due to significant increases in labour productivity as detected not only in our case studies but in most research on developed countries. Research by UNIDO (1993a, 1993b) found that, particularly among small and medium-sized batch producers in developed countries, productivity had increased between 50–100 per cent when moving from conventional machines to CNC machine tools and between 250–650 per cent when moving from conventional machines to FMS. It is also the result of an increase in the capital intensity of engineering firms, as evidenced by our own case studies and studies by the Bureau of Industry Economics (1988a), Pratten (1991a) and Rhys (1992) for developed countries. Third, average capacity utilization ratios also are finding a higher ‘threshold’ level. Although the sampled firms in our study still needed to improve their capacity utilization ratios, it was clear that capital utilization ratios had increased significantly in most firms. This is consistent also with machine-tool manufacturers’ data on machine efficiency increases of more than 100 per cent and findings for developed countries by Ayres (1991) and UNIDO (1993a, 1993b) which point to average plant utilization ratios of at least 80 per cent in firms that had adopted FMS. Fourth, product quality is improving and becoming more homogeneous across the industry. According to Ayres (1991) both product reliability, as measured by the defect rate, and product performance, as measured by the technical features, precision and durability of products, have been significantly enhanced in both developed and developing countries. Finally, in very much the same way as happened with petrochemicals some years ago, profit margins are narrowing down. This was evident in the case of Hydraul and in the Venezuelan, Mexican, Brazilian and Turkish country studies, and is consistent with what is happening throughout engineering, particularly in the automobile industry, where OEMs are demanding that component

144 CONCEPTS, METHOD AND SYNTHESIS

manufacturers cut prices and costs by fixed amounts (Boston Consulting Group 1991; Pratten 1991a). These changes in production and cost conditions have implications for location of production in developing countries. A first implication has to do with competitive strategies available to firms. Porter (1980) argued that there are three generic strategies available to firms with which to cope with the forces of competition: overall cost leadership; differentiation; and focus. The overall cost leadership is based on vigorous pursuit of cost reductions by building scaleefficient facilities, making rapid progress along the learning curve and keeping tight control on labour and overhead costs. The differentiation strategy involves creating some intangible advantage that makes the firm unique industrywide. These advantages may arise from design or brand image, quality, product technology, customer service, dealer network or other similar factors; and firms that have these advantages are able to charge significant premium prices. The focus strategy is based on concentrating on a narrowly defined product or market which enables a firm to compete either on the basis of lower costs or differentiation, so we will not address it any further. Engineering firms in developing countries and newly developed countries have competed domestically and internationally mainly on the basis of overall cost leadership strategies. It was the relentless pursuit of cost reductions by Japanese and, later, Taiwanese machine-tool manufacturers that accounted for their very successful worldwide performances described in Chapter 3. Brazilian, Mexican and Thai exporting firms in our sample were able to compete because of their much lower costs abroad. Metales Leve, a successful Brazilian manufacturer of piston rings and major supplier of automobile transnationals worldwide, also based its strategy on low costs (Stal 1993). Even in the heyday of import substitution most engineering firms tried to compete locally on the basis of lower costs, imports of equivalent products being much more expensive because of tariffs or other means of protection. The convergence in production and cost conditions, together with increasing liberalization of international trade, mean that firms can no longer base their competitiveness exclusively on cost-cutting exercises. While in the past firms could compete on the basis of incremental process innovations, improved forms of organizing the production process and better skills of the workforce, or by depressing wages, the increasing homogeneity, standardization and transparency of production and cost conditions across industry and countries means that there is little room for competing on cost-cutting alone. There may be more space for cost-cutting competition where production processes involve significant assembly operations and thus a larger labour content, but even assembly operations are now being automated. On this account, engineering products are becoming like bulk chemicals. Neither will it be possible to compete exclusively on the basis of differentiation strategies. At least as far as product quality is concerned, the capacity of FA for ‘equalizing’ quality across industry and countries may mean that customers will be increasingly reluctant to pay premium prices for allegedly better quality products. Some other differentiation strategies may retain

FA AND LOCATION OF PRODUCTION 145

competitive power but they will have to be well-entrenched if they are to compensate for the large price differentials generally associated with these kinds of strategy. Plainly, firms will have to compete increasingly on the basis of a combination of cost-reduction and differentiation strategies, which is indeed what Japanese firms are trying to do nowadays in order to maintain their leadership in the machine-tool industry. For developing-country firms, the need to compete on the basis of both cost reduction and differentiation does not bode well. As we saw in the Chapter 5 there seems to be considerable X-inefficiency among firms in these countries judging by the lower ‘threshold’ levels of capacity utilization, higher levels of overmanning and longer lead times. Scientific management skills, in a broadly non-Taylorist sense, also seem to be lacking as was evident in the Venezuelan study. Further, efforts will have to be made in other areas underlying differentiation strategies, such as brand imaging, marketing, product research and development, establishment of dealers, among others, at which developingcountry engineering firms, with some exceptions, are demonstrably inefficient. A second implication of the increasingly homogeneous production and stricter cost conditions for developing countries has to do with the extent or duration of the learning period. ‘Economic learning’ refers to the process whereby knowledge and skills are acquired by individuals or organizations and which result in higher efficiency (Arrow 1962; Bell 1984; Maxwell 1981). Learning has two main forms (Bell 1984). The first form arises from experience or ‘learningby-doing’ and is characterized by passivity, i.e. it happens automatically and flows directly from ‘doing’, and lack of cost. Thus, for our purposes, it can be seen as an externality following on from undertaking manufacturing production. The second form of learning arises from purposeful effort and expenditure in acquiring skill and knowledge. Such learning is an investment. Under the old technologies, even during the early phases of diffusion of new technologies, engineering industry learning processes tended to be long. They involved acquiring skills and knowledge relating to a larger range of equipment and about organization of production in highly disorganized and inherently timeconsuming processes, and providing the necessary training to the workforce, all of which took considerable time. Jacobsson (1993) examined learning-period duration in the Korean engineering industry since the 1970s, including the production of semiconductors, machining centres and CNC lathes, and hydraulic excavators, and concluded that it was excessively long, taking around twenty years, despite many of the firms having been established for some time. According to Enos and Park (1988; see also Jacobsson 1993) it took Daewoo, a large Korean manufacturer of vehicles, some fifteen years, starting in 1970, to master diesel-engine production technology. Learning to design the engine itself, it is argued, would have taken take much longer. With FA, the duration of the learning process should be reduced. There are many less machines to worry about; machines tend to be similar and ‘embody’ many of the skills required of workers in the past; and the production process flows in a much more orderly manner. There are also fewer workers to control and the overall complexity of the production process has been reduced. To the

146 CONCEPTS, METHOD AND SYNTHESIS

extent that, like catalytic cracking in the petroleum industry, FA is bringing the engineering industry to a stage of ‘maturity’, in the sense that equipment is more standardized, the production process is more homogeneous, and knowledge of the technologies, organization and ‘best-practice’ standards more widely diffused, the duration of the learning process is shortened accordingly.2 In one sense, this should be good news for developing-country firms. A shorter learning period means that they can, in principle, be more rapidly brought up to best-practice standards. Obviously, this capacity is available to firms everywhere, but at least developing-country firms would be on an equal footing with firms elsewhere—something much more difficult to achieve with longer and more complex learning processes. Also, the fact that production technologies are more mature, and thus easier to learn in the context of costcutting competition worldwide, may favour foreign technology transfer and relocation of international production to developing countries, particularly where local demand is large or where transnationals already have established bases. Nonetheless, whatever the duration of the learning process, there are some requirements for acquiring knowledge and skills which developing-country firms may not be able to satisfy. Early research, on learning in the Colombian metalworking industry, by Dudley (1972) suggested that a major factor in accelerating the learning process in firms was the systematization and development of control procedures arising from the learning process itself. Jacobsson’s research (1993) on the duration of the learning period in Korea also makes the point that ‘it is not merely production skills which are required but a range of skills such as research and development (R&D), design, production engineering, marketing skills etc.’ (p. 409). Lieberman’s work (1984) on learning curves in the chemical industry found that learning and R&D spending were closely associated. Investment in the acquisition of technological information, R&D, and other areas of the firm is a ‘conscious exertion to use technological information and to accumulate technological knowledge to choose, assimilate, adapt or create technology’ (Bell et al. 1984:107–8). Unfortunately, as was seen in Chapters 3 and 4, the process of researching new technologies and the amounts of investment in R&D and marketing by firms in developing countries, as illustrated by our case studies, would seem to be minimal. This may not, however, be the case for internationally established firms that can establish a manufacturing base in a developing country and undertake R&D expenditure elsewhere. In addition, a reduction in the duration of the learning period may not be such good news for late-entrant firms, from the viewpoint of the industry as a whole. Longer learning periods mean that, at any one point in time in the evolution of an industry, there will be a larger number of firms engaging in learning processes than there would be if learning periods are shorter. There are fewer ‘established’ firms relative to ‘learning’ firms and competition proceeds on the basis of less information and transparency and a diverse range of strategies, much in the way Porter (1980) suggested. By contrast, shorter learning periods mean that late entrants have to compete with the more established firms, some of which may already be benefiting from lowest costs possible and from first-mover advantages such as patent protection or brand name. As Mueller, surveying some

FA AND LOCATION OF PRODUCTION 147

first-mover advantage models indicates: ‘buyer uncertainty over the quality of a new product will lead some buyers to continue to purchase a pioneering brand whose quality is known to them, even if a second brand of unknown but equal quality is available at a lower price’ (1991:10). Developing-country firms, which tend to be late entrants, therefore have less room for manoeuvre, and so for correcting mistakes, which is a common occurrence during the learning process. They have to establish themselves quickly and concomitantly in a range of areas, including production, marketing and R&D, which makes the learning process much more vulnerable to errors. Obviously, the degree of competitive pressure entrants encounter will depend also on the rates of entry and exit of firms, with high rates of exit of established firms vis-à-vis entry of new firms reducing the competitive pressure, and on the ‘age’ of the industry, with younger industries allowing more ‘learning’ firms at any one point in time than would happen with ‘mature’ industries. But, as was said earlier, engineering is a relatively mature industry. In sum, FA has led engineering to take the first step in the movement from a ‘batch-based’ to a ‘flow-based’ industry. Increasingly, production and cost conditions are converging and becoming more homogeneous across the industry. As a result, firms today have to compete on a combination of cost reduction and innovation, and duration of learning processes has shortened. This has opened up some possibilities for location of production in developing countries, as information and knowledge are also available to local firms and because intense cost competition may result in the international relocation of production by transnationals. But, it means also that, at least so far as local firms are concerned, much greater efforts than so far demonstrated, particularly in R&D and marketing, will be required if their countries are to benefit also from the location process. 3 Scale and scope and firm entry One of the key conclusions of Chapter 5 was that plant ‘optimal scales’ were increasing due to higher capital costs. What does this imply in terms of firm entry? Are barriers to entry increased? To address this question it is necessary to examine first the impact of higher optimal scale and capital costs on minimum efficient scales, minimum investment requirements and vertical integration. By ‘barriers to entry’ is meant the degree of disadvantage potential entrants may encounter in relation to already established firms in an industry (Clarke 1985; Lyons 1988). Perhaps one of the most important entry barriers for small developing-country firms in the engineering industry emerges from rising minimum efficient scales.3 Minimum efficient scale (MES) refers to the minimum level of output or the capacity at which a firm may achieve minimumcost operation, expressed either in units of output or, more usually, as a share of industry output or capacity (Hay and Morris 1991).4 For a stagnant industry, or in slow-growth industries such as mechanical engineering, higher optimal scales imply that the MES increases at the industry

148 CONCEPTS, METHOD AND SYNTHESIS

level. A higher MES, in turn, implies that less firms can operate efficiently in the industry. While small deviations from the MES may not seriously affect entry as cost differentials may not be significant, large deviations would.5 Other things being equal, potential small-scale entrants producing at suboptimal levels, and therefore with higher unit costs, will find it difficult to compete in an industry increasingly characterized by low unit-cost large firms. In addition, industrial concentration also may increase. The degree of concentration depends on the definition of the industry and the output or sales distribution of firms within that industry. Generally, however, the engineering industry for a long time has been considered concentrated. In the US, using fourand five-digit industry classifications, the largest four firms accounted for more than 75 per cent of output in industries such as passenger-car manufacturing and turbines and turbine generators in 1982. In the same year the largest eight firms accounted for more than 50 per cent of output in manufacturing industries such as aircraft, farm machinery and equipment, and motors and generators (Scherer and Ross 1990). In developing countries, available research as surveyed by Kirkpatrick et al. (1984) shows that, while the extent of concentration is generally higher than in developed countries, there is considerable similarity with them in the concentration ranking of industries. Thus, higher levels of optimal plant scale may lead to even fewer enterprises having larger shares of the market in developing countries. Larger market shares, in turn, may result in higher profits, which could then be ploughed back into additional investments in flexible automation, into advertising and into other forms of protection of market power, further limiting entry. ‘Minimum investment requirements’ (MIR) refer to the minimum outlays that are necessary to purchase a new piece of equipment or a complete plant if fullscale entry is envisaged. Rising minimum investment levels imply that ‘corporations with an ample cash flow and/or good links to capital markets are more likely entrants than some totally new enterprise’ (Scherer and Ross 1990: 396). In perfectly working capital markets ‘good links’ are available to everyone as each investment project is evaluated on its merits and the best, in theory, should be selected; where they are not, other factors such as size, reputation or collateral determine access to finance. It was shown in the previous chapter that the investment requirements of the CNC machine tools are much larger than those of previous technologies. It was also the case that the levels of investment by many firms in our sample were far higher than at any time before in their history. Most small developing-country firms, and even some large ones, may simply lack the capital, or local capital markets may not be able to provide the finance to purchase the ‘more expensive’ new equipment. The limitations of developing-country financial institutions and markets which render them unable to provide long-term financing—any financing in the case of small firms—are well known. Indeed, the Venezuelan case study clearly points to the importance of subsidized financing in facilitating the acquisition of FA (see Alonso, Tamayo and Cartaya, in this volume, pp. 262– 3). On the whole, therefore, the higher capital costs of FA may act as a deterrent

FA AND LOCATION OF PRODUCTION 149

to entry to small developing-country engineering firms, particularly in the presence of pervasive capital market imperfections. Barriers to entry may arise also from larger vertical integration into component production. As was demonstrated in the Chapter 5, many sampled firms were using the capability of FA to switch tasks towards the production of a large number of components previously farmed out to subcontractors. In general, vertical integration limits entry because it forces potential entrants to produce and consume inputs and output or to purchase inputs at often discriminatory prices, in rationed quantities or under restrictive contractual agreements (Edwards 1955, 1956; Hay and Morris 1991; Schmalensee 1988). But in the case of the engineering industry there are other complications that need to be looked into prior to arriving at any conclusion. Of all industries, engineering is perhaps the one that involves the largest number, and possible combinations, of manufacturing operations. Considering only processing and assembly, there are at least twelve basic operations, each one of which involves several subprocesses or suboperations; these subprocesses may require one or a number of transforming steps or stations, each of which, in turn, may require one or a combination of machines and workers.6 Different combinations of machines, subprocesses and operations are possible for the manufacture of the same final product. And, for the same final product, different combinations of backward integration and external sourcing also are possible. Whether, and to what extent, a plant undertakes some or all of these operations depends on: strategic decisions taken during the history of the firm; the technological capabilities of the firm; the technical complexity of the product to be manufactured; and the expected returns for each individual machine and for the process as a whole. Let us take the case of the automobile sector to illustrate the processes of vertical integration and ‘de-integration’ taking place throughout the engineering industry. Generally speaking manufacturing a car involves four major stages: material extraction and manufacture, including production of steel, non-ferrous metals, plastics, textiles and paints, accounting for around 11 per cent of auto manufacturing’s value added; materials processing, including universal parts such as nuts and bolts, castings and some component parts, accounting for around 28 per cent of the value added; component manufacturing, including production of engines, transmissions, piston rings, carburettors, steering, brakes, suspension, glass, tyres and accessory options, accounting for around 36 per cent of the value added; and, vehicle assembly, involving body and final assembly, accounting for around 24 per cent of the value added (Boston Consulting Group 1991). Final assemblers, or OEMs, manufacture more than 40 per cent of components, most of which are produced in the assembly plants themselves. Rising costs and Japanese competition have put final assemblers under significant pressure to increase efficiency. Typically, a Europe-based volume car company has a much higher degree of vertical integration than does a Japanese car manufacturer. On average, value added as a percentage of sales in Europebased OEMs amounts to 47 per cent, while in Japanese OEMs the equivalent figure is 36 per cent (Boston Consulting Group 1991). In addition, Europe-based

150 CONCEPTS, METHOD AND SYNTHESIS

car producers have between 800 and 2,000 direct suppliers as against only 160– 300 in the case of Japanese manufacturers. In order to reduce costs, particularly of labour as workers in final assembly are the highest paid of the whole industry, to reduce complexity in manufacturing and to simplify co-ordination of supplier delivery, most car manufacturers are reducing the number of ‘first-tier’ component suppliers and gradually shifting in their direction the tasks of producing more components and of assembling component systems which can then be easily and more rapidly fitted into the car, rather than delivering a large number of individual components.7 In this way, the number of direct suppliers is reduced but the number of parts supplied by each one increases (Boston Consulting Group 1991; Financial Times 1995). Despite this trend towards the ‘de-verticalization’ from final assemblers to ‘first-tier’ component manufacturers, final assembly plants typically still involve a large number of complex steps (Abernathy 1978, Mair 1991; Rhys 1992). The body-assembly process involves a press shop with many lines of shearing and punching machines, followed by lines of automated presses for stamping the larger body parts and additional side lines with smaller, sometimes handoperated, presses for the smaller parts; a body-shop where main body parts are brought together and transported through several spot-welding stations by a moving line and progressively joined to other body parts transported by overhead conveyor belts; and a paint-shop where the car is coated with anti-corrosion chemicals, protective primers, and sealers and fillers, and painted in various colours. The final assembly-shop is the section of the plant where an additional 2–3,000 parts are fitted into the car along a moving line of several hundred JITsourced work stations. Although there are differences in the degree of automation, routeing, use of labour and the organization of the process among plants, by and large the car-assembly process worldwide has a uniform pattern of vertical integration. This pattern is the result of a long history of knowledge accumulation in the automobile industry and is the most efficient form of car assembly thus far developed. It has resulted over the years in the need to produce no less than 250,000 cars per year if all the necessary steps are to yield an adequate return (Altshuler et al. 1984; Rhys 1992). Entry to the final assembly stage is possible only at volumes lower than 250, 000 cars per year in the luxury or specialist market segments of the industry, as what the consumer seems to be paying for here is status rather than a means of transport. However, even some of the luxury or specialist market manufacturers, such as Jaguar in the UK or Saab in Sweden, have been taken over by volume producers. As far as developing-country firms are concerned, internationally efficient entry into the final assembly stage has been successful only in the case of the large Korean conglomerates Hyundai, Daewoo and Kia, and possibly also the heavily state-supported Malaysian Proton car. There are internationally efficient assembly plants also in Mexico and Brazil but these are owned by international large-volume producers. Attempts by many developing countries to install car assembly plants, mainly by focusing only on final assembly by importing semi-knocked-down (SKD) or completely knocked-down (CKD) vehicles, have not been auspiciously successful and have resulted in very high

FA AND LOCATION OF PRODUCTION 151

production costs and sales prices, despite hefty protection and the co-operation of international volume producers. Turning to first-tier component suppliers, it is clear that, as with some of the developing-country firms in our study, technological changes and the need for industrial restructuring have already resulted in such suppliers having a larger share of the industry’s machining and assembly and, possibly, of product design and development too. It is perhaps in the first-tier component producers’ segment of the automobile industry—which is in any case characterized by the dominance of older and larger firms, many of which are still benefiting from first-mover advantages—where the effect of increasing MES, MIR and concentration barriers to entry will be felt more significantly as diffusion of FA progresses. To the extent that first-tier component manufacturers need to be more involved in product development, which is a distinct possibility given their increasing role as ‘component system integrators’ and the closer interactions between design and manufacturing arising from FA, there will be demands for additional R&D expenditure resulting in even higher MES and technological capability barriers to entry. In the words of the Boston Consulting Group (1991: 12): ‘Design delegation will mean a heavier R&D burden for component suppliers which requires critical mass/economies of scale at the component producer level.’ In addition, first-tier component suppliers are also taking upon themselves the task of purchasing and co-ordinating a much wider variety of parts and suppliers, which will require the mastery of a broader range of technologies. Hence, given higher MES, MIR and concentration barriers, and the breadth of experience in product design and development, potential developingcountry entrants to this segment of industry will face heightened barriers to entry. The second-tier component suppliers’ segment of the industry, which includes some of the smaller autocomponent manufacturers and customized-product producers in our sample, are facing a rather mixed prospect, at least as far as vertical integration and entry are concerned. Like first-tier component suppliers, they are under increasing pressure from their customers, and from their own larger manufacturing capacity, to undertake additional machining and even some simple assembly operations, which should increase the MES, MIR and the concentration barriers, and the technological capability barrier, to entry. But at this ‘third’ level of the engineering industry there are more entry possibilities than at the other levels, particularly for small firms, arising out of the timing of adoption by first- and second-tier component suppliers. As was discussed in Chapter 3, larger autocomponent firms are slower to adopt FA than are the smaller autocomponent firms. Larger firms are economically and technologically committed to old technologies and thus find it more difficult to replace them, while smaller firms, particularly new entrants, have been quick to adopt FA. Indeed, it was suggested that large autocomponent firms were encouraging and supporting—even financially or by offering a guaranteed market —the entry of highly automated small firms as a way of postponing their own investments in FA. The relevance of this finding for our present discussion is that a number of small firms are actually entering the industry despite the

152 CONCEPTS, METHOD AND SYNTHESIS

growing limitations to entry. However, it is not old technology firms that are entering engineering but only those that have adopted FA. This suggests that small-firm entry at the ‘third level’ of the engineering industry is related to the timing of FA adoption. Small firms adopting FA early on in the diffusion cycle seem not only to be capable of successfully entering engineering but also to be necessary presence to compensate for the relatively slow adoption of FA by larger firms. Small-firm entry possibilities may arise also from the production of products for which the market is limited and where there are few competitors, and by differentiating themselves through providing additional services, such as adapting products to specific requirements and being prompt in delivery, giving advice, repair and maintenance. Products manufactured and services provided by small firms included, for instance, moulds and specialized moulds for prototypes or luxury cars, stampings and parts used only in estate cars or automobiles made for the disabled, jigs and fixtures for use by robots, air-conditioning components, after-market component manufacturers and service and repair shops. According to Pratten (1991a), these smaller firms can coexist with the more efficient large firms in the same industry because they are not competing with them but serving them or producing complementary products which the larger firms are uninterested in manufacturing because markets are too small. However, as Pratten (1991a) himself adds, this is a role which small firms have always had and one that is independent of the state of technology at any particular moment. It is worth emphasizing that while small-firm entry is possible because of precise timing or complementary production, to the extent that MES and vertical integration are increasing throughout the industry anyway, and as medium-sized suppliers of component become more automated, the exit rate of other small, or even some medium-sized, old technology engineering firms should exceed the entry rate, resulting in even fewer small or large enterprises remaining in the industry. A case in point, although not from the automobile industry, is Booktech, the British manufacturer of book-binding equipment referred to in earlier chapters. The company stated that, before it acquired FMS all its suppliers employed around thirty staff. All those jobs have now gone because the company is producing all components in-house, except for very basic ones like screws, nuts and bolts, which were being outsourced already anyway. Small-firm entry may be possible also in the area of electronic autocomponents. Together with the application of new materials, such as ceramics, new metal alloys and glass, one of the fastest-growing automobilerelated areas is the development of electronic control systems and instrumentation for different car functions. According to the Boston Consulting Group (1991), major innovations and developments are taking place in areas such as electronic fuel systems, anti-lock braking, air conditioning, steering, suspension and computerised driving and monitoring, and will continue to do so. Growing demand from automakers for the new electronic products and specialized software is creating entry opportunities for myriad small design and manufacturing firms producing the new components and their parts.8

FA AND LOCATION OF PRODUCTION 153

It must be noted, however, that many of these new components, strictly speaking, belong to the electronics industry, which is an expanding industry rife with innovation, rather than to the automobile industry itself, which is an industry facing the effects of electronics and FA. Perhaps more importantly, as was discussed in the Chapter 5, very few firms in the sampled developing countries are actually engaged in innovation, let alone in electronics-related innovation, despite some of them actually having a quite advanced automobile industry. The combined mass of skills in mechanical engineering and electronics, as well as the financial backing for innovation, to engage in these types of activity were simply not available in these countries.9 To sum up, developing-country firms, large and small, face increasing barriers to entry to the engineering industry. Minimum efficient scales, concentration barriers, minimum investment requirements and vertical integration are all rising as FA diffuses throughout engineering, making it ever-more difficult for developing-country firms to establish themselves in the industry. Heightened barriers, however, do not mean foreclosure on entry. Large domestic firms with ample funding and significant technological capabilities, achieved perhaps in other areas of industry, may be able to enter even the most complex stages of manufacturing, as the cases of the large Korean car manufacturers illustrate, but this will increasingly be possible only on the same conditions as and at efficiency levels comparable to those of industry competitors. A few small firms, which adopt FA early on, complement rather than compete with larger efficient firms; those which specialize in car electronics, also should be able to enter the engineering industry, but this will probably be at the expense of many more old technology firms. 4 Transport costs, just-in-time and infrastructure Traditionally, transport costs have been viewed as an important determinant of the country of location for production. Essentially the view on how transport costs relate to location is based on the comparison between the weight of the final product plus the distance it was carried and the weight of the raw material plus the distance it was carried. Generally, the greater the weight lost in the manufacturing process, the lower the transport costs for final products, although the shape and fragility of the final product may somewhat alter this relationship. Four possible combinations arise: high transport costs of the raw material relative to the value of the final product, with low transport costs of the final product relative to the value of the final product; low transport costs of the raw material relative to the value of the final product, with high transport costs of the final product relative to the value of the final product; and the two limit cases of high/low raw materials and high/low final product transport costs (Sutcliffe 1971). Engineering products would be somewhere between low raw material transport costs and medium-to-high final product transport costs, meaning that location of production should be at some point between ‘not-too-far’ and ‘close’ to the final market.

154 CONCEPTS, METHOD AND SYNTHESIS

Chapman and Walker (1990) indicate that, over the years, freight costs for finished goods have risen more rapidly than have those of primary products. The reasons for this are the application of economies of scale principles to transport, particularly in the case of primary commodities, as a result of larger ship sizes and the use of unit trains which transport a single commodity from point to point. Final products, on their side, have required increasingly more care and special handling and storage spaces; and a larger proportion of higher valueadded manufactured products are being transported, affecting not only transport costs but insurance rates. These factors have meant that location of production has to be nearer to final markets. Thus, to the extent that developing countries provide a large enough local market, production will be located in them. Road transport also has been growing over the last few years, providing an alternative to shipping or rail transport, particularly for short hauls. Indeed, over certain distances, manufacturers have a choice of modes of transport, including water, air, road and rail, and therefore of speed and cost, which are becoming increasingly important factors in determining location of production. Such choice is an important criterion for location and normally implies an elaborate, diversified and expensive transport infrastructure, not always available in developed countries and even more difficult to find in developing countries, in part because of their adverse physical features (Chapman and Walker 1990). Another factor affecting location, and one directly related to our discussion of flexible automation and related organizational techniques, is the expansion of JIT techniques. There are two points in this connection. The first point has to do with whether the growing use of JIT supports proximity of the location of production to final markets. As was discussed in Chapter 5, JIT techniques involve the introduction of ‘internal’ and ‘external’ methods of organizing delivery. ‘External’ JIT, which refers to delivery from outside the firm, involves complex co-ordination and scheduling, and communications problems between the demands of each individual factory and the capacity of suppliers to satisfy those demands precisely, regularly and often at short notice. To address these problems automobile final assemblers, for instance, are not only reducing the number of suppliers but, more importantly, are putting pressure on suppliers to locate closer to them (Jones and Womack 1984; Hoffman and Kaplinsky 1988, Mair 1991; Womack et al. 1990). The best example, perhaps, is Toyota, which has all its main component suppliers located within 20 kilometres of the main factory in Japan. Although such proximity is not always possible, mainly because suppliers often are located elsewhere and relocation may prove extremely expensive, Mair (1991) points out that most new component plants are being built near to the final assemblers. Meanwhile, suppliers are having to leave their components in warehouses nearby final assemblers, and at their own expense.10 The implications for location of component production in developing countries are, therefore, not very promising. Japanese auto-assemblers, for instance, not only do not source significantly from foreign component manufacturers, either from developing or developed countries, because they do not see any economic or other rationale for purchasing components from

FA AND LOCATION OF PRODUCTION 155

developing countries, not even from the Asian region (Hoffman and Kaplinsky 1988). Japanese final assemblers only source abroad in countries where they have production facilities, and even in these cases, they normally do so from fully or partially owned Japanese component suppliers (Hoffman and Kaplinsky 1988). Nevertheless, as is illustrated by the Thai case study, this approach is beginning to change, particularly outside of the automobile industry, presumably because of the revaluation of the Japanese Yen, among other factors. The second point has to do with the demands and costs that JIT can put on society. According to Rhys (1993), JIT approaches have some ‘inherent problems’ of application. In Japan, JIT delivery to a typical truck assembly factory is based on the use of light commercial vehicles (LCVs) with a payload of no more than 1.5 tonnes, so that small batches can be transported. This approach implies frequent journeys, which adds to the serious urban and suburban congestion and results often in JIT suppliers being stuck in traffic. It is also in part the result of the fact that 94 per cent of vehicle registration in Japan is of vehicles under 3 tonnes. On several occasions assembly lines have had to be stopped due to shortages of parts, and assemblers have to hold buffer stocks to compensate for delays due to traffic congestion. Suppliers are dispatching LCVs earlier so that now they have to wait in the factories before unloading, thus increasing their costs and defeating the whole purpose of the exercise.11 In Europe, by contrast, JIT is based on lorries with payloads of around 24 tonnes. Applying a ‘Japanese’ approach to Europe would entail a 500 per cent increase in the roadspace requirement in Europe; and, more practically, it could imply much higher costs, as operating costs tend to decrease with the size of the vehicle. Although JIT approaches require adaptation to specific circumstances which may ease their application, it is still the case that they involve a number of additional costs that are normally overlooked. A ‘Japanese’ approach normally involves traffic congestion and air-pollution problems, and increases Japan’s reliance on imported oil. The ‘European’ approach requires the availability of more spacious road infrastructure. Both require heavy investment in either smaller LCVs or large articulated lorries. Thus, whether it is a Japanese, a European or a local version of JIT, there will always be a number of economic and social costs to be borne by the society, government or firms implementing JIT, and which many developing countries are not prepared or may not be able to afford. In addition to transport infrastructure, the use of FA requires an efficient telecommunications infrastructure. Research on the impact of telecommunications availability on location of production in developed countries generally finds a positive correlation (Saunders et al. 1994). With few exceptions, telecommunications facilities are not widely available in developing countries; or, when they are, they are congested and unreliable. The use of FA and associated organizational techniques does, however, require a growing use of telecommunications facilities and services. As the Mexican study shows, Glass-dies, a glass mould manufacturer, needed to be ‘online’ with many of its customers in order to modify designs and technical specifications. To do so,

156 CONCEPTS, METHOD AND SYNTHESIS

however, Glass-dies had to purchase its own satellite dish. The growing diffusion of CAD/CAM indicates that more production and design information is available in electronic form and can be transmitted over the phone, easing understanding and communication between suppliers and customers. But the phone lines have to be available. Equally, the use of FA requires a continuous reliable source of energy and water. A constant flow of electricity is fundamental to the smooth running of CNC machine tools. The equipment is very sensitive to variance in current intensity, something that is common in developing countries, and sudden power break-downs can have a very damaging impact on the long-run functioning of the machine tools. The use of uninterrupted power supply equipment and oilbased sources of power can compensate only partially in the event of a break in the continuous flow of electricity, and in any case adds to the cost of the machine tools. Water and water installations, in turn, are needed to clear the swarf and offcuttings from the CNC machine tools. Lack of water or of a constant flow of water can result in machine tools overheating and getting stuck, sometimes beyond repair. 5 Factor prices and biases One important criterion for determining industrial location is cost of labour (Chapman and Walker 1990). Two aspects are normally considered: the share of labour costs in total costs; and the spatial variations in labour costs. Generally, a high share of labour costs in total costs, together with large spatial variations in the cost of labour, would favour location in a low-wage country, while a lowlabour share in total costs and little spatial variation in labour costs mean that location could take place in a high-wage country. Note that labour costs refer not only to wages but to all other labour-related costs, including social security, pensions, health, training and other ancillary costs. Also, where the costs of raw materials and energy are significant, the location decision would involve the evaluation of labour, raw-material and energy costs together. The increasing diffusion of FA throughout engineering is changing the relative importance of labour costs in total costs. As Chapter 5 indicated, labour unit costs and labour’s share in total costs were falling significantly in the sampled firms, although the precise magnitude is difficult to ascertain. The reduction in labour unit costs and labour’s share in total costs are phenomena confirmed for most of the research done on this topic both in developed and developing countries (see, among others, Ayres 1991; Boston Consulting Group 1991; Soete and Dosi 1983; Edquist and Jacobsson 1988; Kaplinsky 1991; Womack et al. 1990). The main reason for this reduction is the inherent labour-saving bias of FA, which translates into lower employment in engineering firms. It must be noted, however, that the employment impact on the engineering industry and at the aggregate level will depend also on the growth opportunities opening in other subsectors and in other industries and sectors, and on the extent of the expansion in capital productivity and output.

FA AND LOCATION OF PRODUCTION 157

Location of production in developing countries should be adversely affected by the reduction in labour costs. However, the impact on location should not be expected to be uniform across all operations and functions within engineering. While in machining operations labour costs over total cost reductions are large, this does not seem necessarily to be the case in assembly operations, where there are significant capital and labour substitution possibilities. A robot is still not necessarily more cost efficient than a worker in all circumstances. For instance, Mital (1992) compared the manual and automated costs of assembling a compressor cylinder-head valve specially designed for assembly and found that manual assembly costs were sometimes lower, even at high volumes of output. Nonetheless, whether relocation of assembly operations takes place will depend also on the complexity of the process and the degree of vertical integration within assembly, with assembly processes such as those for automobiles unlikely to be relocated on labour-cost considerations alone, while relatively easy processes such as those of simple consumer durables, pumps, valves and some autocomponents are more likely to be relocated on the basis of labour-cost differentials. Developing countries can, however, take comfort in the fact that FA saves on skills. One of the key advantages, discussed in Chapter 5, of the new technologies vis-à-vis conventional equipment is that there is both an absolute and a relative decrease in skill requirements. The absolute level or ‘mass’ of skills required to use CNC machine tools is reduced, because although there is an increase in the individual skills required by skilled workers, supervisors or engineers, their absolute number is reduced as the ratio of one worker per machine is broken and the overall number of machines falls drastically. Furthermore, according to Jacobsson (1986) and Edquist and Jacobsson (1988), the skill-saving potential of CAD is immense, particularly in the area of design, as the engineer is provided with a range of options on which to build. The relative level of skills also is reduced due to large increases in output. These reductions in skill requirements do not mean, however, that the skills’ problem is solved for developing countries, or that it will lead to rapid relocation of production in developing countries. As many sampled firms stated, even the relatively few knowledgeable workers and engineers required are hard to find in the market, and this is still preventing firms from acquiring some of the FA technologies. What it does mean is that the magnitude of the problem is reduced. Another factor possibly favouring relocation of production in developing countries is the potential saving in capital per unit of output arising from the use of FA. Obviously it was not possible in this research, as also in many other studies, to give a general answer to the question of whether or not CNC machine tools are capital saving. But it was also clear in this and others’ research that where operating conditions for CNC machine tools were optimal, they did prove to be capital saving. To the extent that FA is not only labour saving but also capital saving, the capital shortage constraint to growth becomes less rigid, making it more attractive for all firms, local firms included, to replace old technologies by new ones.12 Further, Soete and Dosi (1983) and Freeman (1992) argue that the capital-saving features of new technologies may result in further

158 CONCEPTS, METHOD AND SYNTHESIS

substitution of old by new technologies. Improving labour and capital productivity may result in higher profitability and/or international competitiveness which, other things being equal, can lead to further growth and employment: The full capital-saving and employment-generating potential of the new technologies can best be realised through a high level of participation in the design and implementation of new systems, whether they are flexible manufacturing systems (FMS) in industry, new office systems or new social services. (Freeman 1992:173) 6 Is large local demand still necessary? It is well established that, as real incomes and standards of living rise, demand falls for ‘mass’ or standardized goods and consumers seek customized and fashionable goods and services. There is a proliferation of niche markets which can be exploited by innovative and automated small firms (Acs et al. 1990; Acs and Audretsch 1990a; Carlsson 1989b; Gilder 1988; Piore and Sabel 1984). The ‘niche proliferation’ argument, however, does not seem to be applicable to the engineering industry in developing countries. As Boon (1985) had already pointed out, the demand structure in developing countries is far less diverse than in developed countries, being geared towards more standardized and mature products. This is in part the result of the far lower average incomes and standards of living compared to those of developed countries. The relatively small increases in diversity, understood in terms of new products, found in the case studies, the fact that many of the FA technologies were being dedicated to the production of ‘standardized’ products and that product scale was increasing in many sampled firms suggests that this was indeed the case in developing countries. Further, there continues to be a large, untapped, growing demand for cheaper standardized goods, as the Venezuelan study clearly points out. As far as the entry of small firms is concerned, as discussed in previous sections of this chapter, it would seem to be fairly limited. Moreover, as diffusion of FA progresses, larger ‘optimal scale’ developing-country firms will increasingly have the flexibility that makes it possible for them to compete on differentiated markets, too. This is already taking place in some developed countries. In a recent review of this issue, Harrison (1994), referring to US-based research, pointed out that large metal-working plants were as likely to produce a range of customized products as were small and medium-sized firms. Thus, small firm entry will be, if anything, even further restricted. So, are large local markets still necessary to induce localization of production? The short answer would seem to be, yes. To begin with, larger optimal plant scales still require larger ‘aggregate’ markets. Although the capacity to produce variety does provide some room for manoeuvre, there are clear limits to scope.

FA AND LOCATION OF PRODUCTION 159

Butter and canons, to use Samuelson’s well-known but extreme dichotomy, cannot be produced by the same production line. Neither can fridges and automobiles, televisions and metal furniture, car engines and boat engines or sametype but differently sized valves or pumps. Thus, for as many different models, varieties or sizes of automobile, car engine, television or video as can be produced, plants will still need to produce 250,000 cars a year, 600,000 engines a year, or around 500,000 TVs or video recorders to be efficient. Or, if design capabilities are to be included, around 2 million cars should be produced by firms (Rhys 1992, 1993). There must be the demand for those cars, engines, televisions and video recorders. Therefore, if minimum cost is what is desired, then the right market size has to be there. Second, large local markets, other things being equal, attract foreign engineering firms. As Shapiro (1994) argues, Brazil was in the initial stages of developing a motor-vehicle industry, able to offer manufacturing transnationals an initial market large enough for assemblers to begin producing vehicles at ‘reasonable’ sales prices. In addition, large vehicle markets allowed for component markets also to develop so that the costs of domestically produced parts were not significantly higher than international prices, something that clearly attracted transnationals. A more recent example is China, where transnationals seem to be ‘queuing’ to enter the very large automobile, powergeneration and machine-tool markets. A very large market would permit several firms to be operating at any one time, meaning that most of the key players operating in the oligopolized engineering industries should be able to enter, particularly since engineering firms tend to follow each other internationally. Finally, large markets are sought also by transnationals because they make worthwhile modifying and adapting production processes and products to local tastes and conditions. Third, large developing-country markets normally have a significant growth potential in volume terms. In a market of 1 million vehicles, like the Brazilian or the Mexican markers, a growth rate of 5 per cent per year means that every fourand-a-half years a new optimal plant can be installed. Today the Brazilian and Mexican vehicle and components markets allow several optimal scale plants to operate in each. Indeed, transnationals often have just a foothold in large markets to monitor their growth and to enter them at the right time. Large and growing developing-country markets could provide a recipient developing country with a better leverage in negotiations for the localization of new production. 7 Location of production in developing countries: an assessment It is difficult to come up with a clear-cut assessment of the impact of FA on location of production in developing countries. On the side, there are a number of factors at play, few of them operating in the same direction, and many offsetting one another. On the other, the notion of ‘developing countries’ encompasses diverse economic realities, something that entails differentiated

160 CONCEPTS, METHOD AND SYNTHESIS

impact. Nonetheless, an attempt will be made to provide an assessment both in overall terms and in relationship to three areas: firm size; firm ownership; and country size. Taking the overall assessment first, the preceding discussion suggests that FA would not seem to favour location of production in developing countries’ engineering industry. First, although production and cost conditions are converging across the industry, easing access to information and knowledge, developing countries still face tight competitive conditions and do not seem to be undertaking the tasks necessary to address them. Second, MES, MIR, concentration and vertical-integration barriers to entry are growing throughout engineering. There are some entry possibilities as second-tier component manufacturers, provided that the timing of adoption is right and that the products are complementary. Third, the high costs of transporting engineering goods, together with the growing use of JIT approaches, means that production has to be located closer to final markets, most of them in developed countries. Fourth, the fact that FA is labour saving adversely affects location of production in developing countries, although this may be partially or totally compensated by the skill- and capital-saving characteristics of the new technologies. Finally, large markets are still important. Against this rather negative overall background there would seem to be some opportunities for location of production in developing countries that are opening to large domestic firms, to foreign companies and to large developing countries. If there are any firms capable of surmounting the growing barriers to entry into engineering and of locating production either at home or abroad, they are the large domestic firms. Many of them will already be established in other industries, or even in engineering itself; will have some domestic or even international recognition; and will be members of large local conglomerates which should give them access to finance. Although perhaps not always internationally competitive, they often are the most efficient firms within the local industry. Large domestic firms will be well-informed of what is happening in their industry, both locally and abroad, and are increasingly automating and becoming more flexible. They will have ‘strategic thinking’ capabilities and be involved in co-ordinating suppliers and small firms related to them.13 Some have experience with product and process innovation, although this does not seem generally to be a strength of firms in developing countries, large or small, and would probably require additional incentives for investment in technology or the procurement of foreign technology. Indeed, potential foreign partners may prefer to link with large and well-established local firms. Some local firms are exporting, and have developed marketing and distribution channels home and abroad.14 Another potential source of firms interested in the location of production in developing countries is typified by the large engineering transnational corporation that has a significant proportion of its assembly operations in developed countries. Many such transnationals are now facing large labour costs in their base countries and have a need to restructure. While the more complex, vertically integrated, processes may still require a significant local market, there

FA AND LOCATION OF PRODUCTION 161

are many assembly processes that are relatively simple and do not involve many steps. There is a diversity of mechanical engineering equipment and products, such as fluid power, refrigerating, ventilating and heating equipment, weighing machinery and portable tools, the assembly, and even the manufacture of the simpler components, of which could be undertaken in developing countries. There is immense potential also in the assembly of electronic products such as televisions, video recorders, hi-fi equipment, telecommunications and even in the manufacture of some of their components. As Edquist and Jacobsson (1988) pointed out, the competitiveness of the manufacturers of many of these products is improving in developed countries, but the fact that they still involve a significant labour use suggests that some serious calculations need to be done before taking a location decision. Indeed, many transnational companies have relocated, and could continue to relocate, production of mechanical engineering equipment and electronic goods for exporting back to their home bases.15 Developing countries with large populations and per capita incomes, such as China, India, Brazil, Mexico, Malaysia, Indonesia and Turkey, can ‘trade’ their big and expanding domestic markets with local and foreign firms and increasingly become recipients of new investment in engineering. Some of these countries already have a fairly well-developed engineering industry on which they can build further growth and demand. In particular, these countries have, or are in the process of developing, a vehicle industry which is an important demand ‘pull’ factor for engineering products and equipment. Equally, middle-income developing countries such as Thailand, Argentina, Colombia, Venezuela, Egypt, Nigeria and South Africa could ‘trade’ their smaller but still significant and often growing local markets.16 In the case of these countries, a focus on specific engineering products and industries around the exploitation of natural resources, such as oil in Venezuela and Nigeria or minerals in Colombia and South Africa, can result in significant location of production decisions. Notes 1 It should be noted also that industrial automation, flexible or otherwise, by replacing human effort with machines and by introducing feedback control, i.e. the procedure for measuring and inspecting or ‘sensing’ a process, evaluating and processing the information received in relation to a theory or algorithm of that process, and providing an output of instructions as a response if required (Kaplinsky 1984; Ramtin 1991) —in effect attempts to convert discontinuous production processes into continuous or flow-type processes. 2 The notion of maturity is being used here in the ‘product life-cycle’ sense, but applied to production processes. A product is considered ‘mature’ if standardization opens possibilities for achieving economies of scale, there are few uncertainties as to the supply of components and inputs, there are substitutes for the specific product, product information is widely available and cost consideraions are paramount (Koutsoyiannis 1982). 3 The concept of MES was originally coined by Bain (see Carlton and Perloff 1994; Hay and Morris 1991; Scherer and Ross 1990).

162 CONCEPTS, METHOD AND SYNTHESIS

4 Minimum efficient scale (MES) should be distinguished from minimum optimal scale (MOS). Although both refer to the cost-minimizing level of output, the former is an industrial organization concept and thus is normally expressed in terms of industry share, while the latter is a microeconomic concept. 5 Rhys (1992, 1993) argues that new technologies may reduce the steepness of the long-run average cost curve and thus reduce cost penalty for entry at lower MES, but that there is no evidence for this. 6 The twelve basic operations are: (in processing) casting or moulding, particulate processing, deformation processing, material removal, heat treatment, surface treatment and coating and deposition; and (in assembly) welding, brazering and soldering, adhesive bonding, threaded fastening and permanent fastening (Groover 1996). 7 A case in point is the dashboard manufacturer based in the Netherlands to which we referred earlier. The company was asked by a final assembler to manufacture and supply dashboards in which airbags had already been fitted. What used to happen was that an OEM acquired the airbags and dashboards from separate manufacturers and assembled them itself. 8 According to research by Audretsch (1995) on US start-up firms, it is in new electronics-related industries where the largest number of start-ups, particularly of small firms, are taking place and where entry is relatively less affected by the business cycle and macroeconomic conditions. 9 Autocomponent manufacturers’ average R&D expenditure on sales amounts to 4 per cent in Europe, 3.5 per cent in Japan and 2.1 per cent in the US (Boston Consulting Group 1991). Only three Brazilian firms, Mech-trans, Piston-rings and Diesel-engines, of all the firms in our sample, possessed that R&D spending capacity (see Chapter 7). 10 An illustrative example of the complexities involved in JIT delivery is provided by Mair (1991): ‘The physical size of a part and its batch size also play an important role. Larger parts can be delivered in smaller batches cost-effectively (e.g. one truck-load). But for smaller parts a truck-load may contain sufficient parts for several days or weeks of production. Thus small parts cannot be delivered at short intervals, unless, as one manager joked, they can be delivered by taxi’ (p. 61). 11 Some specialists and component manufacturers cynical about JIT point out that the JIT approach is a way to push inventories down the supplier chain: ‘Some observers believe that the burden of maintaining inventory has simply been transferred from the final assembly plant to the suppliers’ (Groover 1987:778). 12 There will continue to be a minimum capital-requirement barrier to entry. 13 Small firms may have a lot to gain in terms of access to information, local and international markets and finance from close collaboration with large firms, very much in the vein of the Japanese keiretsu, although there are some dangers, too, as often large firms ‘download’ adjustment costs (Alcorta 1989). Of course, direct cooperation between small firms is an alternative route for gaining access to information, market and finances, but the relative costs and benefits of each route would have to be carefully evalued. 14 Apart from the Korean car manufacturers Malaysia’s Proton car, for instance, is being exported to Europe, and the Indian private engineering TELCO Group is also starting to export four-wheel-drive vehicles to Europe.

FA AND LOCATION OF PRODUCTION 163

15 In 1995, China exported US$17 billion in electronic goods including products such as computer peripherals, electronic calculators, telephone sets, colour and monochrome TVs, radio and radio recorders, tape recorders, electronic watches and clocks (Gu and Steinmueller 1997). 16 This classification coincides roughly with UNIDO’s type A and type B categories for countries with, respectively, large domestic markets for capital goods and high bargaining power, and medium-sized markets and moderate bargaining power (see also Edquist and Jacobsson (1988)). It is increasingly difficult to consider countries such as Korea or Singapore as developing countries.

6 CONCLUSIONS AND POLICY RECOMMENDATIONS

The purpose of this book is to examine the impact of FA and associated organizational techniques on economies of scale and scope, and to draw out the implications for the establishment of industry in developing countries. By doing so we hope to contribute to the debate on the impact of microelectronics-based automation on the product, plant and firm dimensions of scale and on economies of scope by establishing a solid conceptual framework and gathering and examining empirical evidence. The conceptual framework was based on standard microeconomic theory and was developed in Chapter 1. It specifically addressed the issue of the impact of a change in technology on optimal scale. Essentially, the framework argued that the final outcome of such impact involves interactions between individual product, plant and firm scale economies and economies of scope. It pointed out that changes in optimal scale imply the modification of the level of total output, product variety and unit costs. ‘De-scaling’, or reductions in optimal scale, takes place only when, following a change in technology, lower levels of output are produced at lower unit costs. By contrast, ‘scaling-up’, or increases in optimal scale, takes place only when, following a change in technology, higher levels of output are produced at lower unit costs. The key underlying determinants of changes in optimal scale are the level of fixed costs and the possibilities for unit cost reductions due to joint-production. The ideas discussed in the conceptual framework were then operationalized in a methodological discussion in Chapter 2. Given the absence of industrial survey data, the chapter pointed to enterprise-based case studies arising out of in-depth semi-structured interviews together with internationally available statistics as the most appropriate means of carrying out research on scale and scope in developing countries. The chapter indicated that a comparative statics approach, where a ‘before’ or old technology situation and an ‘after’ or new technology situation are compared, provides the kinds of indicator necessary for significant conclusions to be reached. As to the industry, the chapter concluded that the focus should be on the engineering industry because of the widespread impact of microelectronics-based technologies on engineering, its key role in technological progress and its historical impact on conceptualizations of manufacturing. It concluded that because FA encompasses a diversity of technology, the research should concentrate on firms using CNC machine tools for machining, one of the core processes in manufacturing. Main associated organizational techniques

CONCLUSIONS AND POLICY RECOM MEND ATIONS 165

included in the research were total quality management (TQM), just-in-time (JIT), cellular manufacturing (CM) and working practices’ reorganization (WPR). Finally, the chapter indicated that developing countries selected for this research would include those with medium-to-high levels of diffusion of CNC machine tools and with some variance in macroeconomic conditions. Hence, Brazil, India, Mexico, Thailand, Turkey and Venezuela were chosen. Three questions have guided Part I, the empirical part, of this book, the more general results of which have now been presented; the individual country studies comprise Part II. The first question examined the extent and motives for diffusion of FA, particularly in developing countries, so as to provide contextual information on the possible effects of FA (Chapter 3). The second looked into the product, plant and firm dimensions of the impact of FA on scale and scope, and the reasons for such impact (Chapter 4). The third analysed the effects FA will have, working mainly although not exclusively through scale and scope, on the location of industrial production in developing countries (Chapter 5). Extent and motives for the diffusion of flexible automation The examination of the data on production, trade and consumption of machine tools suggested that there has been rapid diffusion in countries like Taiwan and Korea which are not only large users of CNC machine tools but significant exporters. China has a very large and growing consumption of metal-cutting machine tools and has began switching to the use of CNC machine tools but not yet as fast as have the developed countries, Taiwan or Korea. Mexico and Turkey also showed growing use mainly of imported CNC machine tools and has the highest level of diffusion of all the studied countries, though it still seems to be behind China. The extent of diffusion in Brazil and India although growing rapidly, particularly in the case of the latter, was still lagging behind all the other studied countries. Diffusion in other Latin American and African countries was extremely slow. ‘Old technologies’ being replaced by ‘new technology’ CNC machine tools in the engineering firms studied were mostly conventional machine tools and thus there was little evidence of mass production. Highest levels of CNC machine tool use among the surveyed firms were found in small customized-product local firms which faced exacting technical demands. In contrast, large autocomponent and general capital-equipment firms were automating at a much lower pace because of the significant associated investment and learning requirements. Diffusion of CNC machine tools was complemented by organizational techniques such as CM. Other organizational techniques such as TQM and JIT would seem to be substituting for rather than complementing the use of CNC machine tools and hence were not as ‘associated’ as initially envisaged on the basis of the literature. The main motives for the use of CNC machine tools were macroeconomic conditions, including low and stable price and interest rates, and high aggregate demand and microeconomic factors, such as quality and flexibility.

166 CONCEPTS, METHOD AND SYNTHESIS

Flexible automation impact on scale and scope As expected, the reduction of setting-up times and costs arising from the use of FA and ‘associated’ organizational techniques meant that product optimal scale was reduced in many firms. Nonetheless, it was also found that, in practice, sampled firms decided on batch sizes not only on the basis of cost considerations, but perhaps more importantly, on grounds of the demand being faced for specific products. Also, as expected, the flexibility of the new CNC machine tools allowed for wider product scope and resulting economies of scope, although the wider diversity was often the result of variations of the same product and the production of components previously outsourced. Turning to the more contentious impact of FA on ‘optimal plant scale’, the main conclusion of the research is that use of CNC machine tools had resulted in ‘scaling-up’ because, where relevant indicators had been made available, firms showed by and large higher levels of output at lower unit costs. Furthermore, an improvement in capacity utilization rates, which were low even among the most efficient firms, would have meant that ‘optimal scales’ would have increased in sampled firms where the evidence was inconclusive. At the firm level, FA did not lead to further increases in ‘optimal scale’ as the majority of firms did not significantly increase R&D, marketing and administrative fixed costs by leaving these activities to business partners. Underlying the changes in optimal plant scale were a number of of technical and economic factors. On the technical side, the key factor was the higher efficiency of the new CNC machine tools and organizational techniques. The higher efficiency was the result of the new technologies’ superior production rates per unit of time; it arose in part from the capacity of the new equipment to integrate different operations within a single machine, saving on setting-up time in each machine and on waiting time between machines; and it was in part a consequence of the greater accuracy of the CNC machine tools, which improved input utilization rates and allowed for the use of material previously wasted. Organizational techniques added to the efficiency of the equipment by further reducing workpiece travelling time (through CM), by reducing confusion and by providing the necessary raw material and work-in-progress exactly when needed (JIT) and by making possible additional reductions in defect rates (TQM). On the economic side, the new CNC machine tools involved larger capital ‘fixed’ costs than had the technologies they replaced, thus requiring even higher volumes of output to amortize these costs. Implications for location of production in developing countries On the whole, higher plant optimal scales and, more generally, FA’s diffusion, do not seem to bode well for the establishment of industry in developing countries. Competition in the engineering industry, which is increasingly global, involves nowadays a combination of price competition and the creation of intangible advantages, mainly through innovation. Yet, as was seen, developing-

CONCLUSIONS AND POLICY RECOM MEND ATIONS 167

country firms are barely managing to be efficient let alone create new products. The engineering industry is moving into a mature stage, where such barriers to entry as minimum efficient scales, minimum investment requirements, concentration and vertical integration are all growing, posing further difficulty of entry for new firms. There is also a trend to locate factories near to final markets, though this is mainly in developed countries, spurred by the high costs of transporting engineering goods and the growing use of JIT approaches. Large markets continue to be as attractive as ever. Finally, FA’s labour-saving characteristics make location of production in developing countries relatively less attractive. But there are some ‘windows of opportunity’ for certain developing-country firms and developing countries themselves. FA is not only labour saving, but skill and, possibly, capital saving. From a developing-country perspective this could be quite attractive since, despite large initial investments in skills and equipment, FA would save on their relatively scant resources. Also, there are still some labour-intensive assembly operations within engineering that may prove attractive to multinational companies considering whether to relocate in developing countries. There are entry possibilities for large established developing-country firms that have some experience in engineering or related industries and can use their local position to expand domestically or abroad, either alone or with foreign partners. Large and small engineering firms may still find business opportunities in large and expanding developing-country markets such as China, India or Brazil, or in those, like Mexico, that are involved in trade agreements with advanced countries. Large firms may find it easier to enter engineering as second-tier component manufacturers. Small firms may enter as third-tier component producers, where there is always some demand for new firms, provided the timing of adoption is right or if they produce complementary products. Areas for further research In the course of doing this research and writing the book additional work areas requiring further research were identified and whose investigation is considered necessary to understand fully the diffusion and impact of new technologies in developing countries. A first area of work would be the production of annual or, at least, quinquennial national surveys on the use of technology, and of the new technologies, by the manufacturing sector. Some of the studied developing countries, like Mexico and Turkey, have begun work in this area; indeed, in the case of Mexico preliminary results have reached some researchers. It became clear, however, that further, and more refined work, would be necessary before the data could be fully used. Clearly, organizing and financing national surveys are beyond most individual researchers’ means, and would have to be undertaken by national governments —at least by those with larger resources. Benefits will go, however, well beyond the possibility of making international comparisons or

168 CONCEPTS, METHOD AND SYNTHESIS

better empirical and ‘generalizable’ research, and will surely include improved private and public sector policy formulation and evaluation. Another area for further work is developing understanding of the underlying processes and wider implications of the diffusion of new technologies in developing countries. Although our main concern in this research lies with the impact of FA on scale and scope, an attempt has been made to provide some idea of the complexity involved in the decision by a firm to adopt a new technology, particularly in Chapter 3. Also, some discussion has taken place about some of the wider implications of the use of FA, especially in Chapters 4 and 5. But a lot more would seem to be needed. A number of topics arise. One is the question of whether firms or the owners of firms deciding to adopt FA have some specific characteristics. Some differences due to size, type of industry and ownership have already been explored, albeit in a limited way. But it was also pointed out in Chapter 3 that in the case of very early adopters, the choice of technology was related to country of ancestry of the owner, which in turn prompts the question of whether there are significant differences in the choice of technology by entrepreneurs of dissimilar origins or from divergent professional backgrounds. Bearing in mind that developing countries have a large proportion of owner-managed firms, the question seems relevant. Other enterprise factors that may lead to differences in choice of technology are age, technological capabilities, organizational features, skills and information availability, and profitability. A second topic is the relationship between large and small firms in the diffusion process. As was seen in Chapter 3, an important factor behind the high levels of automation achieved by small firms was the support granted by large firms. While in the diffusion and innovation literature large and small firms are seen as substitutes, this research suggested a much more complex, interdependent and, perhaps, complementary relationship. Whether or not this is the case in general and in developing countries in particular, and what would be the implications for diffusion theories and practice if it were, would seem to require further investigation. A third topic that needs to be thoroughly investigated relates to the reasons for the underlying dynamic and causal interactions that emerge between investment, technological choice, cost and scale changes, innovation and demand.1 Obviously, increases in optimal output do not take place timelessly, automatically or linearly, as implied in our ‘before’ and ‘after’ comparisons. They are the result of an accumulation of complex decisions, actions and effects, combined in various ways. It may well be the case that firms exhibiting higher scales have reached those heights as the result of past success with old technologies which permitted building firm-specific knowledge in metal-cutting technology and obtaining high levels of profitability; thus generating the technical and financial capacities necessary to invest in new technologies and increase optimal scale. Hence a key factor pertinent to both the diffusion of the new technologies and increases in optimal scale would be prior achievement of technological and financial capabilities. Higher levels of optimal output may have been the result also of a continuously rising domestic or external demand

CONCLUSIONS AND POLICY RECOM MEND ATIONS 169

which created simultaneously the conditions for the use of new technologies and an increase in output. Reaching the desired levels of optimal output is not easy. As was seen in Chapter 3, the installation of new technologies involved making judgements with poor information and great uncertainty, and occasionally in a context of friction with workers and/or suppliers. But this was only the beginning. Surely, whatever the duration of the learning process, getting the new technologies to function with the requisite minimum of efficiency must have demanded intense learning efforts, often resulting in disappointments and failures, which would have affected output and costs. It is hard to believe that most interviewed firms were at all times operating efficiently; one supposes rather that they often faced discontinuities and technical and economic difficulties. This may be particularly true for firms in developing countries as the environment in which they operate is less uniform and more uncertain than in developed countries. Indeed, the ability to live through such circumstances may have an important bearing on the subsequent success of most developing-country firms. Hence, understanding fully the links and processes involved in reaching new heights in efficiency would seem to be a key task for future research. Finally there is the issue of economic and social implications of the use of new technologies. On the international trade side, the use of new technologies may have an important effect on the comparative and competitive advantages of firms and nations. Most exporter firms in our sample pointed out that it would not have been possible to access foreign markets were it not for the new technologies. A change of technological specialization to higher value-added and new products seemed to be increasingly contingent on the use of new technologies, but the precise links between adoption and foreign trade have yet to be explored. For instance, is it necessary for firms to begin exporting cheap-labour goods prior to the adoption of new technologies? Or, conversely, can firms ‘leapfrog’ to the use of new technologies quite early on in their exporting trajectory? If a transition is required, how does it take place and is it any different from what happens to nonexporting firms that adopt FA? Are wage differentials necessary as between ‘high-tech’ developing-country exporting firms and domestic-market firms? On the social and employment side a key issue is the impact on employment of the diffusion of new technologies, an issue particularly relevant in the context of the labour-saving nature of the new technologies and the large supply of labour existing in developing countries. Also, there is much worry as to the potential impact of new technologies on employment in developed countries, though their impact may be as intense, or even more intense, in developing countries. On the whole developing countries do not manufacture the new technologies and thus cannot compensate in this way for reductions in employment due to their local diffusion and to their impact through foreign trade. Policy areas There is a clear need for a policy on providing information about the advances, advantages and limitations of new technologies. Because of their closed or small

170 CONCEPTS, METHOD AND SYNTHESIS

domestic markets, developing-country firms do not have access to the array of suppliers and sources of information available elsewhere. Unless a potential user has access to foreign sources of information, which seems to be restricted to very large firms, it faces limited options and a ‘sellers’ market’. There were several cases in this study where firms were actually misled into purchasing the wrong number of CNC machine tools; and it was shown also that early adopters bought new technologies not so much for their technical characteristics but because the hardware came from the owners’ country of ancestry. While it may be the case that as markets grow and as diffusion ensues information will increasingly be made available, there is, nevertheless, a clear need for providing potential users with information on the economic and technical potential of the new technologies, as well as available alternatives, especially in early the phases of diffusion. This can be done by government agencies, universities or research institutes and by private business associations, or by a combination of them. It must, however, be provided. Another area in need of policy intervention is finance. There are three possible directions in this regard. The first involves equalizing access to finance so that large and small firms have equality of opportunity. As was seen in Chapters 4 and 5 the new technologies require significant capital investments, amounts that are readily available only, perhaps, to a few large enterprises. In addition, given the role of small firms in early stages of diffusion, lack of finance becomes a crucial barrier towards widespread diffusion. The second direction confronts the cost of finance. Clearly, largely subsidized interest rates lead to overinvestment and the purchase of equipment that often does not match needs, as the Venezuelan case study illustrated; but, equally, real interest rates that are too high result in underinvestment or no investment at all, as a proportion of new technologies is normally bought with borrowed finance. Particularly in the Latin American region, where exceedingly high real interest rates predominate as a result of economic stabilization attempts, there would seem to be need to reduce real interest rates at least to the levels of advanced countries, if not lower. There may also be the case for differentiated treatment of certain kinds of technology, on the proviso that the extremes of import substitution approaches, where everything is subsidised, are not reached. The third direction takes issue with the terms of lending. While there is no generally agreed time limit by which CNC machine tools, for instance, fully depreciate, it is also true that the depreciation period tends to be long; and it can be argued that the terms of lending should match this length of depreciation time. Yet, again, availability of finance under the appropriate conditions does not seem to be widepread among developing countries. There is need for policy intervention also in education and training. While the skill content per unit of output may be falling, there is need still to provide training to the workers who will be operating the new technologies. Most case studies pointed to an insufficient number of skilled workers as a major limitation to the successful use of CNC machine tools. Creating a sufficient ‘mass of skills’ in the economy requires investment in education and training, and the adaptation of training resource requirements and methods to the specifc demands of the new

CONCLUSIONS AND POLICY RECOM MEND ATIONS 171

technologies. The fact that CNC machine tools are in early stages of diffusion, especially in developing countries, suggests that on-the-job training schemes designed to make best use of workers’ tacit knowledge—which perforce emerges with the diffusion of new technologies, particularly if it is speedy—should be encouraged. A final area for policy intervention, particularly relevant to developing countries, is infrastructure. Again, while the use of energy inputs such as electricity may be lower than with older technologies, they must be available, which is not everywhere the case, and with the necessary regularity. The same is true for water supplies, road infrastructure and haulage, and port facilities. No doubt, efficient use of new technologies in developing countries may be helped by other policies, including incentives for investment, involvement in economic integration agreements that enlarge the market or provide access to other large markets, as in the case of NAFTA or any other similar approach. Export promotion schemes and policies have been particularly successful in the past, although they will now have to be reconsidered in the light of new international trade agreements. More generally, developing countries may compensate for most of the obstacles to efficient use of new technologies by introducing imaginative and innovative policies, of whatever kind. Note 1 I would like to thank an anonymous reviewer for pointing this out.

Part II THE COUNTRY STUDIES

7 THE IMPACT OF FLEXIBLE AUTOMATION ON SCALE AND SCOPE IN THE BRAZILIAN ENGINEERING INDUSTRY Ruy de Quadros Carvalho1

1 Economic and technical change: 1980–93 1.1 Manufacturing Industrialization was the driving force of stable economic growth in Brazil until the 1980s. From the end of the Second World War, Brazilian manufacturing grew at an average rate of more than 8 per cent per year, supported by an import substitution policy aimed at building up production capacity to serve the highly protected and cartelized internal market. This trajectory was interrupted in 1981. The structural weakness of Brazilian industrialization, particularly its technological fragility and the slowdown in productivity growth, combined with a deteriorating macroeconomic situation, showed that the import substitution model had exhausted its potential.2 These limitations became more evident in the face of the worldwide acceleration of technical change and the increasing importance of international competitiveness as a strategic policy objective in industrialized countries. Brazilian industry suffered three successive recessions—in 1981–3, 1987–8 and 1990–2—with intervening periods of industrial stagnation. Given the downturn in the domestic market, industry started to put greater emphasis on exports, initially as an outlet to utilize idle capacity. However, as the government extended export promotion policies, and as some industrial segments succeeded in international markets, Brazilian manufacturing exports grew from an average of 28 per cent of total exports in the 1970s to 53 per cent in the 1980s. Although the rate of investment in Brazil declined substantially in the 1980s, private industrial firms, and particularly the leading exporting firms, undertook systematic technological and organizational modernization.3 Firm restructuring was based on the selective adoption of a mix of electronic industrial equipment, updating products, and organizational innovations such as just-in-time and total quality control which supersede the more rigid principles of ‘scientific management’ regarding specialization and division of labour. The major aim, at

174 THE COUNTRY STUDIES

the level of the firm, was the enhancement of quality standards and the achievement of greater production flexibility. Since this technological upgrading was concentrated in a few firms, the result was an increase in the technological heterogeneity of Brazilian manufacturing. In the 1990s, Brazil has moved rapidly towards a more liberal economic policy. Privatization, market de-regulation, the pursuit of fiscal discipline, the reduction in the role of the state in providing economic infra-structure and the opening of the economy have affected manufacturing. Trade liberalization and the gradual reduction in import tariffs have been adopted to increase substantially the competitive pressure on the internal market and to promote Brazil’s integration with the world economy. International competitiveness became the priority of trade and industrial policies. As the case studies show, actual or potential competition from imports has led industrial firms to substantially revise their marketing, investment and production policies in order to reduce costs and prices, and to further enhance quality. Firm restructuring has become both deeper and more extensive, encompassing not only exporting firms but companies which primarily serve the domestic market. The processes of policy and economic change and firm restructuring have been promoted by the policy makers and the media under the somewhat loose description ‘economic modernization’. These initiatives have generated a lot of excitement in business circles, and with some justification. After seven years of stagnation, labour productivity in Brazilian manufacturing increased by approximately 30 per cent over 1992–94 (Bernardes and Lemos 1995). But so far the new industrial policy has emphasized trade liberalization, deregulation and the promotion of competition almost exclusively. Although it was envisaged to include programmes promoting industrial technological capabilities, little has been done in these respects. One consequence of leaving industrial investment entirely under the direction of market forces is the deepening of the technological and managerial divides between large and small Brazilian firms. 1.2 The engineering industry Brazil’s economic growth in the 1970s was driven primarily by expansion in the engineering, chemical and metallurgy sectors. Within the engineering sector, autoparts and mechanical capital goods were the most dynamic industries. They were also the most affected by economic crisis in the 1980s, as shown in Table 7.1, in which the ‘mechanical industry’ comprises capital goods producers and the ‘transport material industry’ comprises the motor-vehicle industry (including the autocomponent industry).4 The steep fall in output in the ‘mechanical industry’ (as from 1981) reflects the drop in investment. A recovery started in 1993, but with different impacts in the various sectors.

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 175

Table 7.1 Brazil: industrial manufacturing output by sector, 1980–93 (1980=100)

Source: IBGE. Table 7.2 Brazil: the mechanical engineering capital goods industry, 1980–93 (US$ billions)

Source: ABIMAQ/SINDIMAQ

Capital goods Within the mechanical industry, the capital goods segment was affected most by the recession in the Brazilian domestic market in the 1980s. Table 7.2 reveals a steady decrease in capital goods sales and employment since 1980, with a short

176 THE COUNTRY STUDIES

period of partial recovery (1985–8). By the end of 1993, employment in the capital goods segment was just above half the level it had been in 1980. Although the diffusion of FA in Brazilian industry was (and still is) limited, the capital goods sector is one of the major users. The leading capital goods producers also started to introduce new management practices from the mid-1980s, and the process has intensified following trade liberalization. Technical changes, and the growth in capital goods imports, suggest that employment in this sector may fall further, even if there is a sustainable economic recovery in Brazil. Machine tools are a particularly important form of capital goods. Because of trade barriers, government requirements for product nationalization, and the paucity of local suppliers of cost-effective quality parts, the machine-tool industry in Brazil became very much vertically integrated, and produced a large diversity of products in small numbers. Machine-tool companies can be classified into three groups which cater for distinct segments of the domestic market. The first group comprises multinational producers (such as Traub and Index) and the largest national machine-tool maker (Romi). This group supplies machine tools (standard or customized) with a high technological content. These firms were the first to introduce computerized numerically-controlled machines (CNCs) in Brazil, and have now adopted FA quite intensively, in their production as well as in their products. A second group of large and mediumsized local firms supply conventional machine tools and simple CNC machine tools. A third group of medium-sized firms produces cheap and simple conventional machines for less-demanding customers (Vermulm 1994). The economic crisis, technical change and the opening of the Brazilian economy have brought substantial changes in the machine-tool industry. Successive recessions have eroded the domestic market, and the export market was not open to most local producers, given their technological limitations. In fact, Brazilian exports of machine tools fell from some US$70–80 million in the early 1980s to just US$24 million in 1989. With tougher competition from imports, some firms left the industry, while others merged with larger firms. Output in 1993 was less than 60 per cent of the 1980 figure. Import liberalization is inducing firms to specialize, to reduce their vertical integration, and to reduce also local content. The incorporation of microelectronics is demanding higher technological capabilities, larger investment in R&D and engineering activities, and larger scales of production. The tendency is towards polarization between large producers of CNC machine tools, on the one hand, and producers of conventional machines for the lower end of the market, on the other, with the intermediate-level firms being squeezed out. As the need for local linkages grows, some of the smaller machine-tool makers are opting to become specialized suppliers of parts to the larger producers. Autoparts In contrast to the capital goods segment, the Brazilian autoparts sector has grown significantly in the past twenty years. Its share in industrial GDP rose from 2.9

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 177

Table 7.3 Brazil: sales and exports of the autoparts industry, 1980–93 (US$m.)

Source: SINDIPECAS Note a Sales (in US$m.) per 1,000 employees.

per cent in 1975 to 7.2 per cent in 1991, despite the crisis which affected the whole automobile industry during the 1980s. Since the 1950s, autopart producers have developed in parallel to car assemblers in Brazil. In 1958, Brazilian vehicles had an average of 50 per cent local content, rising to 96 per cent in the late 1960s but then falling, from the early 1990s, due to trade liberalization. Government policies to promote exports began in the 1970s, and were an important support for Brazilian autopart producers at a time when the globalization of the world automobile industry was beginning. Involvement in international markets was very important in upgrading local producers’ technology and products, and helped the industry to deal with the recession in car manufacture and the steep reduction in the domestic market for autoparts in the early 1980s. After a period of declining sales, growth resumed due to increased exports. In 1993, autopart exports had increased to almost 400 per cent of the 1980 level (Table 7.3). The share of production going to domestic car assemblers fell to less than 60 per cent in 1992, from 71 per cent in 1980. In 1994 there were approximately 750 firms in the autoparts industry, producing more than 10,000 items. The vast majority of these firms are small or medium-sized, and controlled by local entrepreneurs. Only 5 per cent employ more than 1,000 workers, and half employ less than 200 employees. However the large producers of autoparts are technological leaders in the industry and account for most of the exports. Fifteen autopart producers account for 75 per cent of Brazilian autopart exports, and ten of these are European or American multinational corporations (MNCs). Brazil is among the top exporters among the newly industrializing economies, but exports are strongly concentrated in a few product lines (engines, engine and gearbox parts, and automobile radios). The

178 THE COUNTRY STUDIES

largest external market has long been the USA, which takes 40–50 per cent of Brazilian autopart exports), but exports to the Argentinean, Mexican and German markets have grown more rapidly since 1992. The autoparts industry employed 176,000 direct workers in 1993, as compared to 214,000 in 1980 (Table 7.3). Until 1986, employment levels followed output fluctuations. But firm restructuring, based on rationalization, FA and the new organizational practices, has led to substantial employment cuts. Labour productivity, as measured by the sales: employment ratio, almost doubled between 1980 and 1993. In the 1980s, the restructuring process affected exporting firms in the main, since firms catering only for the domestic market were still protected. At that time, firms upgraded product design, product quality and production processes to keep their external markets. After trade liberalization, the restructuring process deepened (Posthuma 1994) and has included a process of firm concentration. Many mergers and acquisitions, and the departure of the more fragile firms, reduced the number of firms from 2,000 in 1989 to 750 in early 1995. Brazilian autopart producers have adopted a restructuring strategy with a dual focus: the limited and selective adoption of electronically controlled machines, such as computer-aided design equipment (CADs) and CNC machine tools, in the production areas with the highest quality and precision requirements; and the more widespread adoption of new organizational practices. The latter have brought significant improvements in both productivity and quality. Techniques such as statistical process control, Kanban control and cellular manufacturing have been incorporated in most leading firms, although the extent and quality of utilization vary considerably from one firm to another. Polyvalent (multi-task) working has also been spreading in the autoparts industry, which is in line with the employment decrease in the past three years (Posthuma 1994). Another emerging tendency is a reduction in vertical integration and an increased tendency to obtain parts and machining services from other firms. This restructuring has been unevenly spread. It has helped the most dynamic areas and firms while less-competitive products and firms have become more fragile because of trade liberalization. 1.3 The diffusion of FA in Brazilian manufacturing New production technologies have diffused steadily in Brazilian manufacturing, although the rate of diffusion is slow compared to OECD countries and some of the Asian NICs. CNC machine tools, robots, CADs and other specialized electronically controlled machinery in the engineering industry have been adopted primarily to improve product quality for export markets. Various factors contribute to the current low level of diffusion, including macroeconomic instability, which has had a negative effect on investment, and especially the contraction of the internal market for capital goods. Unfortunately there is little reliable current aggregate data on the diffusion of FA in Brazil. According to SOBRACON (1992) data there were 7,141 CNC machine tools and 1,664 CAD

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 179

and CAE (computer-assisted engineering) units in use in 1992, but these figures underestimate the current stock because they are based on a survey which does not cover the entire population of user firms and does not include imports since 1988. Asea-Brown-Boveri do Brasil estimated that there were 345 robots in use in 1994. The figures for CNC machine tools and CADs are quite respectable. Brazilian manufacturing firms are relying increasingly on programmable automation. In the engineering industry, as attested in our case studies, mediumsized and large CNCs are becoming an increasingly important component of investments in new machine tools. This seems to be an established trend. According to the SOBRACON (1992) survey, 49 per cent of the users of industrial automation (including equipment other than CNC, CAD, CAE and robots) are in the engineering industry. The same survey suggests that 45 per cent of industrial users are large firms, and 20 per cent are medium-sized firms. However, programmable automation techniques have been adopted less widely in Brazil than in advanced industrialized countries and some of the EastAsian NICs. Quadros Carvalho (1993) showed that this gap is systematic, in terms of both stock of equipment and density of diffusion. In 1985, the density of diffusion in the OECD countries ranged from 10,505 to 28,619 for CNC machine tools, 1,188 to 22,011 for robots and 1,385 to 9,047 for CADs, whereas in Brazil in 1987 the density figures were 2,100 (CNC), 60 (robots) and 431 (CAD).5 The corresponding Korean figures, for 1985, were 5,176, 2,100 and 1,437. This gap appears to have widened. Since 1989, the worsening economic crisis in Brazil has led to a steep fall in the rate of diffusion of programmable automation. Annual sales of industrial equipment for the automation of manufacturing (CNCs, CAD, robots and programmable logic controllers) dropped from a peak of US$400 million in 1989 to approximately US$220 million in 1992 (SOBRACON 1992:15). The local production of CNC machine tools, which accounted for the largest part of the supply in the 1980s, also fell, from a peak of 1,052 units in 1989 to an annual average of 500 units in the 1990s. However, it is important to consider also the quality of diffusion. The literature on the Brazilian experience suggests that the engineering industry has been incorporating FA selectively, combining old and new techniques within a single plant. More rigorous competition forced exporting companies to adopt programmable automation techniques to meet international quality, precision and homogeneity requirements (Fleury 1988; Quadros Carvalho and Schmitz 1989), but they have not yet found it necessary to proceed to full plant automation. Prado reports that in the Brazilian autoparts industry ‘the adoption of microelectronics-based equipment has been mainly determined by quality requirements and is occurring at critical points in the productive process, where the existing machines do not meet the required standards’ (Prado 1989:34). Similar conclusions were drawn in studies by Fleury (1988) and Posthuma (1991), regarding the autoparts industry, and by Quadros Carvalho and Schmitz (1989), with regard to the car-assembly industry. The pattern of adoption in Brazil is related in part to relatively low labour unit costs, and relatively high capital unit costs (including high interest rates),

180 THE COUNTRY STUDIES

although Brazilian authors have pointed to economic instability as the most significant cause for low rates of diffusion in Brazil (Coutinho and Ferraz 1994; Prado 1989; Quadros Carvalho 1993). The low labour costs are due not only to low (direct and indirect) wages but to great labour flexibility (Quadros Carvalho 1993). Labour cost is particularly significant as regards the adoption of robots (Edquist and Jacobsson 1988:156). However factor price seems to have been less important in persuading firms to adopt FA than have quality and performance requirements. The case studies in this chapter suggest that it is only since trade liberalization that the adoption of FA and new organizational practices has been driven primarily by the search for cost reductions. 2 Technical change in the Brazilian engineering industry: the case-study evidence 2.1 The sampled firms The sampled firms belong to two distinct segments of the Brazilian engineering industry: producers of autoparts for the car assembly industry; and the pump and valve manufacturers, which are part of the wider capital goods segment. A subgroup in the autoparts industry supply machining services for autopart producers. Except for the small suppliers of machining services, the sampled firms can be described as major competitors in their respective market segments in Brazil. Furthermore, some of the autopart firms export much of their output. The sample includes local firms and multinational subsidiaries, and large, medium-sized and small firms. Some characteristics of the firms are outlined in Table 7.4. Production processes in sampled firms Firms in the autoparts and the capital goods segments themselves carry out most phases of the manufacture of their products. The casting and forging of steel, iron and, to some extent, aluminium are done internally by Water-hydro and Pistonrings (casting) and by Mech-trans (forging), while the other firms buy their metallic raw material (including steel bars) from steel mills, forges and casting firms. In Water-hydro, Piston-rings and Mech-trans, the casting and forging are carried out, at separate plants, by independent business units which supply external customers also. The machining of the parts is the core of the production process. Except for Piston-rings, which deals only with relatively small parts and does not need machining centres, all autopart and capital goods producers carry out a full range of machining operations, from turning, milling, drilling and boring through to final deburring and grinding. The components of a particular product

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 181

Table 7.4 Brazil: basic characteristics of the sampled firms

182 THE COUNTRY STUDIES

Source: Interviews and company reports.

generallyare processed in parallel, each shape going through a particular machine route.The average number of operations per part varies widely, from an average of5 machining operations per part (Oil-hydro, Water-hydro and Air-brakes) to 15operations (Piston-rings). In the case of producers of major autoparts (Mechtransand Diesel-engines) the average is 10 operations per part. All firms then use high-temperature chemical treatments to improve cohesion and durability. Piston-rings, Mech-trans and Diesel-engines carry out these processes in-house. Some parts also have a chemical treatment to improve their anti-oxidation properties. Piston-rings and Mech-trans do this in-house, the other firms by subcontracting. The complexity and organization of assembly varies from firm to firm and from product to product. Generally assembly is organized around fixed assembly lines, with products moving on tables and being passed on by workers, with mechanical aids to help move heavy pieces. The assembly of standard valves and pumps (whether hydro or oil) is relatively simple, with 2–8 stages of assembly. The assembly of customized hydraulic pumps is more complicated and requires some knowledge of mechanics. Thus it is organized in assembly teams and not in lines. The assembly of major autoparts (Mech-trans and Diesel-engines) involves a much larger number of parts and up to 30 assembly stages, organized in something resembling a car-assembly line. Although most of the sampled firms have adopted TQC procedures, involving production workers in quality production and control, a final check on quality is necessary before the products are packaged. In the case of more complex products (gear pumps, hydraulic pumps, transmissions and diesel engines) each product goes through performance testing involving electronic monitoring.

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 183

Simpler products still pass through visual inspection (Piston-rings, for instance) and sample testing. The sixth stage is packaging. The production processes in the firms providing machining services to order are different, as would be expected, involving only machining, quality control and packaging. Often, in the case of small service suppliers, the client provides the raw material to be processed. Generally speaking, machining operations are simpler in these firms. The average number of operations per part is below five and these operations do not require integrated machining (machining centres or transfer lines). Medium-sized service firms carry out their own quality control inspection and are involved also with just-in-time deliveries. 2.2 The adoption of flexible automation Our case studies show that firms’ responses to economic instability and to a shrinking and more competitive internal market were clearly influenced by the particular constraints faced by their industrial segment. Capital goods firms Following the general falls in investment and in local demand for capital goods, the market for pumps and valves declined. At the same time, competition became more rigorous because of import liberalization. Table 7.5 shows that sales for Oil-hydro and Water-hydro in 1993 were still below their 1985 levels, in spite of the general industrial recovery in that year. This point was emphasized by these firms in explaining their strategies of product diversification, greater customization and quality improvement. For Oil-hydro, this change consisted of an increase in the diversity of products manufactured, from 150 in 1982 to more than 350 types in the early 1990s. This diversification was accompanied by substantial ‘complexification’: many of the new product lines relate to more complex higher-value-added hydraulic devices. Moreover, the new strategy involved a shift from a rigid, catalogue-based, product mix to a more flexible, customer-oriented, supply of products and services. Oil-hydro spends an average of 2 per cent of sales on product development, and has accumulated some capability in the design of quick couplings, control valves and gear pumps. Thus the firm had the technical basis to add manufacturing according to client specifications to its traditional supply of standard parts. Moreover, Oil-hydro’s aim is to integrate its sales and engineering departments with clients’ processes of detail engineering. In the management’s view, this would represent the transition to service-like industrial production. Water-hydro has gone further in the direction of customization of serial hydraulic equipment. Its German parent group is a world leader in water hydraulics and centrifugal pumps. The Brazilian affiliate has acquired significant experience inproduct adaptation and development, and could become

Source: Interviews and firm reports. Note a US$1,000 per employee.

Table 7.5 Brazil: sales, employment, productivity and exports in selected engineering firms, 1985 and 1993

184 THE COUNTRY STUDIES

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 185

the central engineeringunit for Water-hydro Americas. Its outlay on product development in 1993 was 5per cent of sales. At least 40 per cent of the firm’s sales are based on specialpumps manufactured to customer specification, but a large proportion of thestandard pumps are supplied in dimensions also determined by clients. Crisis andhardening competition has led this firm to look more carefully to the service sideof its business, intensifying the integration of its technical areas (including sales)with clients’ engineering activities. In addition to product diversification and customization, these firms stressed high quality standards as the key to competitiveness in this market. Water-hydro is seen as a pioneer in quality because it was one of the first suppliers of Petrobrás, which is known for tough quality requirements. The firm has been a Petrobrás quality-assured supplier since 1984. It was undergoing an audit to qualify for an ISO 9000 certificate at the time of our visits, and plans to introduce a total quality maintenance programme. Oil-hydro also has a high quality standard, and has developed a quality tradition in the market. The firm developed its own formal quality-audit system in 1978, which it is now being adapted to match the ISO 9000 standard. However, the emphasis on product development and quality should not be taken to imply that price competition is not an issue in this market. Particularly since the reduction in import tariffs, both firms have felt the pressure of cheaper imports on profit margins, and have taken action to reduce their costs. One response, which was still being implemented at the time of the visits, was the introduction of computerized production control systems to enhance cost control. The firms began to incorporate FA (particularly CNC machine tools) in the mid-1980s, and have continued to do so since then. The management in both Water-hydro and Oil-hydro considered that the adoption of FA, particularly CAD systems and CNC-based machining, was crucial to their new marketing strategies. In addition to integrating product development with manufacturing, only the new automation technologies could provide the flexibility required for frequent product changes. Of the sampled Brazilian firms, Oil-hydro and Water-hydro have by far the highest densities of diffusion of CNC machine tools, with 75 per cent of the parts produced at Oil-hydro machined on CNC machines, and 60 per cent of parts at Water-hydro.7 Oil-hydro has the second largest stock of CNC machines (59 machine tools and 7 machining centres) in the sample. This firm started to experiment with FA in 1983–4, with the purchase of two Index CNC lathes. The results were considered so promising that Oil-hydro has maintained a permanent programme for the adoption of CNC machines. While the largest purchases of new machines were made between 1985 and 1988, the latest acquisition occurred in 1991. At that time, CNC machine tools and machining centres had replaced 100 conventional machines. At Water-hydro, adoption was intensified only after 1990. Management considered import liberalization as a major incentive to adoption, since CNC machine prices were halved after tariff reduction. In 1993, this firm imported two sophisticated Mazak CNC lathes, equipped with one Mitutoyo electronically controlled laser micrometer. Given the versatility and precision of these

186 THE COUNTRY STUDIES

machines, this move was considered by management as a qualitative jump in the machining capability of the firm. The utilization of information technology in project and management areas also took off in the 1990s. The information system which Water-hydro was implementing at the time of the interviews is one of the most integrated of any of the comparable systems among the sampled firms. The subsystem for design and engineering was upgraded with the introduction of an Intergraph CAD system, with three seats. This system is integrated with a Digicon-Smartcam system for CNC programming, which was introduced with the first CNC machines. The most comprehensive move towards informatization, however, is the adoption of SIMEX, a network system, based on a Hewlett Packard (HP) supermicro computer supporting 100 terminals, which will be spread around the factory (one for each plant supervisor) and offices. The SIMEX software is designed to integrate production planning and programming, production control in real time (utilizing bar codes on products), cost control, sales and inventories. This system will allow Water-hydro to obtain a finer control of production costs and times, a weak area in most Brazilian firms. In this respect, Oil-hydro is behind Water-hydro, but moving in the same direction. In addition to the utilization of integrated CAD/CAM systems, the firm has started to introduce direct numerical controls (DNC) in most CNC machines, to integrate them with the project-programming systems. This is seen as an important step to improve cost control. In each firm, the adoption of new production and design technologies was accompanied by and integrated with organizational innovation. Before the adoption of FA, production lay-outs were organized along functional lines, with conventional machines clustered by type of operation and size. The typical layout included one area for conventional automatic lathes (single and multi-spindle) and another for conventional transfer machines, followed by an area for drilling, boring and milling machines, and then a fourth area for grinding machines. Batches of products of different shape went through this route before assembly. Water-hydro changed to cellular plant lay-out when it incorporated the first CNC machines, in 1985–6. At that time, the major aim of both innovations was to reduce lead time, to gain flexibility and quality, and to expand capacity within the plant. Later the firm widened the scope of production workers’ tasks, including the delegation of greater responsibility for quality, while keeping a conventional atomized job classification. Inventories also were reduced, although management does not consider that the firm meets the standard of internal JIT. Oil-hydro started a process of comprehensive organizational change in 1989– 90, with the technical support of a renowned management consultancy and training institute. This included a shift to cellular organization, changes in production job titles and an increase in average wages (the lowest machineoperator wage was levelled with the top wage for operators). A participation campaign was started, aimed at obtaining workers’ goodwill towards change and innovation. After two years, the experiment was considered problematic and was reversed. Instead of physical production cells, Oil-hydro adopted ‘featuring cells’, which meant returning to a functional lay-out and adopting a concept of

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 187

production programming that defines the optimal route to be followed by a batch, in order to reduce lead times. In spite of the organizational problems at Oil-hydro, and the lack of change in some crucial areas such as human resource management, these firms have gone beyond the narrow localized use of a few techniques. The systematic nature of the changes described above is evident.8 Autopart firms Autopart manufacturers have been facing much tougher price competition than have the producers of hydraulic devices. Sales diversification towards export markets during the 1980s had required the autopart firms of our sample to modernize and diversify their products and to improve quality to international standards. While the domestic market was protected, lower international prices were partially counterbalanced by higher prices and margins on their domestic sales. However, import liberalization and the intensification of global sourcing by local car assemblers have confronted autopart manufacturers, in the local market, with the hard price competition which marks the global market (Posthuma 1994). Since the late 1980s, as car assemblers intensified global sourcing, autoparts manufacturers have been required to increase their integration with assemblers’ product development activities and to take more responsibility for the basic design of new parts. Mech-trans, Piston-rings and Diesel-engines have been involved with product innovation. Their spending on product development varied from 3 to 7 per cent of their annual sales in 1993. Since the mid-1980s, these efforts have related mainly to materials research and product design, in connection with product performance and product diversification. Although the manufacture of autoparts in Brazil is typical of the mass production of standard parts, it has followed the general tendency in the car industry towards diversification within large volumes. Thus Mech-trans and Piston-rings have launched more new products in the last eight years than they had in all the previous years since their establishment in the 1950s. Air-brakes, the fourth autoparts firm in our sample, has also developed greater product variety, but without any product development activities, which are entirely located in the company’s European affiliates. Product upgrading and diversification have required changes in production and design methods. Before the introduction of FA, plant lay-outs were largely functional, grouping together types of machines and operations. However, at the huge plants of Mech-trans and Piston-rings, division by major product lines preceded functional division. Moreover the heavier parts of transmissions and engines were machined in very large transfer lines which constituted separate production areas. In all cases, the vast majority of machining operations were carried out in rigid, conventional, automated machine tools. The four autopart firms began using FA, particularly CNC machine tools, in the early 1980s and increased the rate of adoption in the period 1986–8. Initially, FA was selectively incorporated, mainly for quality- or product-related reasons. The use of CNC-

188 THE COUNTRY STUDIES

based machining for certain operations was tied to product design innovation. As autopart firms realized that the market tendency (like assemblers’ demands) was towards greater diversity and smaller batches, CNC machine tools and machining centres became the major technological option in the companies’ investment programmes. The adoption of FA helped firms to attain the flexibility required by more frequent product changes, reduced economic batches and lead times, and increased product homogeneity. Quality and flexibility were the motives also for Mech-trans and Piston-rings’ decisions to start JIT and TQC programmes in 1985–6. Since then, these firms have largely shifted from functional production lay-outs to production cells for groups of parts of similar shape, in which the responsibility for quality is partially delegated to production workers. At Diesel-engines and Air-brakes, JIT and TQC were systematically adopted only with the 1991 recession and import liberalization, although these firms had already improved their quality-control procedures. The new economic policy climate in the early 1990s induced the autopart firms to increase and to deepen the process of organizational change. This time, their primary goal was the attainment of cost efficiency. The four firms embarked on what they called ‘downsizing’, which embraced many different actions aimed at cost reduction, such as the reduction of management layers, the outsourcing of production, maintenance and administrative jobs, and the intensification of work in the production cells. In this respect, Air-brakes stands out. This firm underwent an organizational revolution in a short period (1991–4), combining downsizing and the adoption of systemic JIT and TQC methods. Table 7.6 shows that autopart firms have a lower density of diffusion of CNC machine tools than do firms in the pump and valve business. The share of output machined with FA varied between 10 and 30 per cent at the time of our inquiry. Mech-trans has the largest stock of CNC machines in the sample, although its density of diffusion is just 20 per cent. This is because the firm has the highest stock of conventional machines as well, with approximately 1,200 machines in operation. This company is the only one in the sample (and one of very few in Brazil) which has adopted a Flexible Manufacturing System (FMS), comprising four Grob machining centres, one head changer and a system for automatic transfer and feeding. The FMS machines the bodies of the firm’s latest product. Mech-trans’ adoption of CNC machines started in 1980 and intensified in 1986– 7, when the firm introduced a cellular lay-out. Since then, investment in new machines has been oriented exclusively to FA. Management stressed that import tariff reduction, and the resulting fall in CNC machine prices, contributed to accelerate diffusion. Mech-trans invests an average of 3–4 per cent of annual sales in new machines (US$5 million). Air-brakes started to adopt FA in 1982, and Diesel-engines did so only in 1987. Diesel-engines is now building an entirely new production line for the machining of a more compact and efficient engine (to be launched in 1995), in which 95 per cent of production will be by CNC machines. Piston-rings started to adopt CNC machines only in 1986, one year after beginning the change from a functional production lay-out to production cells.

Source: Interviews and firm documents. Notes na=not available. a Number of CNC machine tools adopted in the plant. b One of these machines is served by a laser micrometer. c Four CNC machining centres are integrated into a flexible manufacturing system. d The density of diffusion is the percentage of output due to CNC machining. e In these columns (x) refers to adoption, whereas (−) means no adoption.

Table 7.6 Brazil: diffusion of microelectronics-based automation in selected engineering firms, 1994

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 189

190 THE COUNTRY STUDIES

This firm has the lowest density of diffusion and one of the smaller stocks of CNC machines in the sample. Yet it is the only firm to incorporate automated handling and feeding devices, which are dedicated to serving the CNC machine tools. Management argued that the production of piston rings demands special machines, given the need to machine very large batches of small parts of a peculiar shape, and that the application of FA in these processes has not developed much worldwide. The firm manufactures most of its production machines in-house. Nevertheless, the manager of the Ring Processes and Methods Division stressed that ‘after the introduction of cellular organization and of the JIT philosophy, and with the fall in CNC prices, the choice and adoption of CNC equipment became imperative’. The introduction of CNCs in the machines built at Piston-rings improved the machining precision and allowed the control of product homogeneity to be automated. With conventional technology, this control demanded considerable attention and time from workers. The new technology freed workers to assume other tasks in the cell, contributing to increased productivity. Given the high share of labour in the unit cost of piston rings, labour productivity is a central concern for management. This explains why 80 per cent of all planned investment in new machines in 1994 was to be in CNC machines. This connection between new organizational practices and a greater urgency in the adoption of new production technologies is one of our most interesting findings. The autopart producers in our sample are among the most advanced Brazilian firms in terms of the breadth and depth of the organizational innovations they have developed and adopted in the past ten years. Firms such as Piston-rings and Mech-trans are showpieces of organizational change, and have inspired other companies and served as parameters within and outside the industry. It is not possible to give a thorough account of their achievements here, but we do want to emphasize the simultaneity and the links between organizational innovations and the diffusion of information technologies. The far-reaching organizational ‘revolution’ that has taken place at Air-brakes is another case in point. The changes included the full adoption of production cells and mini-factories in connection with new-cost control procedures; a new flatter and more integrated job classification aimed at motivating workers to participate in Kaizen groups; and just-in-time delivery to major clients. In the interviews, Air-brakes management emphasized that the new organizational format and practices made it more urgent to adopt FA. Some of the old equipment generated bottlenecks for the balancing of cells (due to the high set-up and machining times of conventional equipment). Two Okuma CNC lathes were bought in 1993 and 1994 to overcome such problems. A similar tendency could be found in the adoption and integration of information systems. Most autopart firms had adopted CAD/CAE systems in the 1980s to provide the flexibility required by constant product changes, and at the time of our visits they were investing substantially in extending and upgrading their information networks to increase computerized control of production on the shopfloor (in real time), and to integrate this control with the design and inventory areas. As the industrial director of Mech-trans put it:

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 191

A good information system is high on our agenda. Information is a basic tool for competitiveness today: it is critical to reduce lead time in new product development. It is also essential for agile cost control and to keep cell teams informed. We want to display as much information as we can at the shopfloor. The pressure for cost reduction has increased the importance of computer networks in production control. Plant management at Mech-trans keeps the cells informed by displaying graphs of the cells’ performance in terms of productivity, quality, absenteeism and safety. The firm is investing US$2.4 million this year in an ORACLE network system to integrate engineering, production and general management areas. The network comprises 3 HP Risc computers which provide support to 11 CAD seats (using Computervision software) and 150 PCs in offices and on the shopfloor. Piston-rings and Air-brakes had similar strategies. Machining service firms For the machining service firms, tough price competition is the central feature of the market. This is due mainly to autopart firms’ struggle for cost efficiency. Mech-trans and Air-brakes are among the major clients supplied by the sampled machining services firms. In the search for cost-effective production, the autopart producers substantially increased the outsourcing of machining jobs, while pressuring the suppliers of such services for price reductions, for example by threatening to import parts, or by actually doing so. The medium-sized service firms fear competition from Chinese and Taiwanese competitors. Machining service firms are in turmoil, and face a major challenge. As one interviewee put it, ‘we have a lot more work, but our margin is nil’. They also have less of the financial and technical resources needed to overcome their problems than do the other sampled firms. Having no product lines of their own, the machining service firms do not carry out product development activities or participate in their clients’ product designs. They work from detailed product drawings supplied by the customers. However Medium-services-B was upgrading its product engineering area because the technical relationship with clients was becoming more sophisticated. Market evolution and performance in the service firms have closely followed changes in the autoparts industry. After the sudden drop in sales provoked by the 1981–2 recession, the growth in Brazilian autopart exports was responsible for a corresponding increase is machining services, which peaked in 1986–7. The service firms began to incorporate CNC machine tools and introduce more reliable quality-control procedures at that time. The adoption of microelectronicsbased machining was suggested, if not required, by autopart manufacturers. The attainment of precision in certain operations, as well as product homogeneity, demanded a gradual diffusion of FA. In the case of Medium-services-A and B, clients’ orders were also becoming smaller, more frequent, and the required delivery time was becoming shorter,

192 THE COUNTRY STUDIES

requiring the firms to reduce lead times and increase plant flexibility. Smallservices-A and B had always delivered individual piece-work. Thus, when some clients started to require just-in-time deliveries in 1990, the machining service firms were already flexible enough to adapt. However the pace of diffusion in the service firms has been less regular than in the other groups, because firms’ investment programmes are more affected by the economic climate. As in the autopart firms, FA is incorporated selectively, with a relatively low density of diffusion (15–30 per cent of output; see Table 7.6). However, the area in which service firms seemed to be lagging behind most is the adoption of organizational innovation. Even in the medium-sized service firms, which produce quite a diversity of parts, the conventional (functional) lay-out, based on the grouping of similar machines and operations, had been retained in most areas, although the need for cost reductions in the 1990s had accelerated restructuring. Autopart producers began to require new management practices from their suppliers, such as cellular organization and statistical quality control methods. Our interviewees complained about their lack of managerial and financial resources to meet clients’ demands and about the lack of effective support received from them. The manager of Medium-services-A criticized the short-run price-oriented approach of his larger clients (autopart producers): ‘They are not developing true partnership; at the slightest price variation, they shift to another supplier.’ Medium-services-B is changing employment practices to reduce the cost of labour and attain cost effectiveness and price competitiveness in an industry which is labour intensive. It has replaced many male workers with female workers, who are paid 20–30 per cent less for the same job (50 per cent of production workers were women at the time of the interview), and it has hired exemployees as outworkers to carry out simpler machining jobs. Informal labour is less costly, since the firm avoids social security and other social taxes. This may be a sign of the development of a fourth tier in the car industry—the outworkers. It is clear that the introduction of new production concepts is just beginning in these firms, and is far from demands. Although quality procedures have been upgraded, improvization is still the dominant managerial model. Of the two medium-sized service firms, Medium-services-A has the larger stock of CNC machine tools and the highest density of diffusion (Table 7.6). This firm introduced its first CNC machine tool in 1986, and after a short trial period bought 13 machines with supplier finance. However, although Mediumservices-A’s sales in 1993–4 were at their highest, the CNC machine tools were substantially idle, which suggests that the firm had hastily purchased more CNC machines than it really needed. A similar problem occurred at Small-services-B, with two aggravating factors: here, the idle capacity of CNC machines was higher and the financial burden of the investment was driving the firm towards disaster. There would appear to be a general problem of insufficient knowledge and information for the adequate planning for the adoption of FA in small and medium-sized local firms. On the other hand, Medium-services-B and Smallservices-A revealed a healthier adoption process.

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 193

Given their limited project capabilities and lack of technical resources, the utilization of information systems in these firms also is limited. The mediumsized service firms rely on Digicon equipment for the generation of CNC software, and no plans for the more extensive utilization of information systems or networks were found. Flexible automation and organizational change: an overview The findings described above underline the organic links between the diffusion of FA and that of the new management practices. At least in the leading firms of the Brazilian engineering industry, microelectronics-based automation, organizational restructuring in line with JIT-TQC methods and the more recent accelerated adoption of information systems can best be understood as interrelated parts of a continuing process of change, rather than as separate experiences. The particular balance of techniques and organizational forms adopted varies according to each firm’s possibilities and capabilities, and is influenced by macroeconomic and sectoral factors. The fact that new production technologies have been widely adopted in the capital goods firms, and more selectively so by the autopart firms, is related to their market situation and product strategies. However the sampled firms’ planned and realized investments in new machines suggest that new production technologies are likely to be adopted more rapidly in both sectors if the Brazilian economy resumes sustained growth. Our evidence shows also that the continuing increase in the intensity of competition is contributing to more rapid diffusion of information technologies, at least in the leading firms. 2.3 Selecting and adopting flexible automation: procedures and problems As regards sources of information and procedures for the adoption of new industrial automation technologies, there is a clear gap between the machining service firms and the rest of the sample, a gap which is in line with the disparity in technical capability between these firms. In the autopart firms and the producers of hydraulic devices, machines are selected on the basis of information which is continuously collected by the firms’ industrial engineers from the literature, from regular visits to international and national fairs and from contact with machine suppliers. At the MNC subsidiaries, the technical process of searching for and selecting machines is entirely carried out by the local engineers. The interviews suggested that MNC subsidiaries (Water-hydro, Air-brakes and Mech-trans) include a more accurate and formalized cost-benefit analysis in their decision making than do local companies (Oil-hydro). In at least two of the MNC subsidiaries, the final decision requires the agreement of the parent company. Local engineering teams are the major actors also in the gathering and processing of information about organizational change, even in MNC affiliates.

194 THE COUNTRY STUDIES

For instance, Mech-trans is considered a pioneer in the introduction of management innovations in Brazil. This firm has a long-established relationship with the engineering departments of the University of Campinas and the University of São Carlos. These departments help the firm to train technical personnel in new management practices. Other firms have also sought outside support from local and foreign engineering and management institutes, such as the North American Just-in-Time Institute. Machining service firms are less able to monitor available technical information, and are thus more dependent on machine suppliers. Selection and decision making are empirical and not at all formalized. This has led two of the four service firms to overestimate their need of FA, and probably explains why service firms cited more problems related to the utilization of FA than did the other, better organized, firms. The major problems they faced included the death of workers skilled in dealing with CNC machines and new concepts of production, the higher maintenance costs of FA, and delays in obtaining imported parts, particularly electronic parts. This last was the only problem cited by autopart and hydraulic equipment firms. Problems with human resources have been less acute for these firms, mainly because they have invested in training. 3 The impact of flexible automation on scale and scope 3.1 Changes in product scale and scope The literature has emphasized the contribution of FA to achieving economies of scope. FA’s programmability and flexibility are associated with reduced set-up and machining times, greater accuracy and reduced scrap and re-working, so that a firm can economically manufacture a diversity of products or parts in smaller batches. Our case studies provide further evidence in support of these points. Changes in setting up times and machining cycles The reductions in setting-up time and machining cycle time were major goals of the sampled firms in adopting microelectronics-based automation. Improvements in both respects enabled batch sizes and lead times to be reduced, and product diversity to be increased: factors which were crucial to product and process flexibility. Moreover, setting-up times and machining cycle times, together with improved inventory turnover, were the most important factors in cost-benefit analyses of FA. To gain a better understanding of the implications of FA for scale and scope at the product level, our survey collected some sample comparable data on settingup and machining times for the manufacture of single parts as they were before and as they have been since the introduction of CNC machine tools or machining

Source: Interviews. Note a Percentage reduction in setting-up time in the manufacture of the part, after the introduction of FA. b Percentage reduction in machine cycle time, after the introduction of FA. c Ratio of setting-up time to machine cycle time, before and after FA.

Table 7.7 Brazil: gains in setting-up and machining cycle times for single parts in selected engineering firms, 1985 and 1994

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 195

196 THE COUNTRY STUDIES

centres. These comparisons confirm the enormous potential gains associated with FA, as indicated in Table 7.7. For pieces produced in smaller batches (up to 100 pieces), machining with the old technology involved several different operations, carried out in separate machines. At Water-hydro, for instance, the machining of hydraulic pump axles used to involve 7 distinct machines and 5 workers, who carried out cutting, premachining, turning, milling and finishing operations. Each operation required a separate setting-up, comprising adjustment of the machine, changing tools, loading and unloading and fixing the piece in the machine tool. Some aspects of the set-up of conventional machines were particularly time consuming, such as tool changing and fixing workpieces in drilling machines. The cumulative setting-up time for smaller batches was as high as 340 minutes (in the Water-hydro case). For the manufacture of large batches, we collected cases showing up to 1, 000 minutes of cumulative setting-up time. Tool changing and machine adjustment, in these cases, were the most time-consuming set-up activities. With the adoption of CNC machining, setting-up and machining times were substantially reduced, as shown in Table 7.7. Previously separate operations are now grouped together in one CNC machine, so that setting-up is required for fewer operations. At Water-hydro, machining pump axles now involves only 3 machines and 2 workers. The core machining operations are carried out on the Mazak CNC lathe, with a laser micrometer, which produces such high-precision machining that the finishing operation has been dispensed with. Moreover, setting-up is simpler and easier with FA. There is little time involved in machine adjustment and tool change is considerably simplified. The organization of production in cells also has made an important contribution. As a result, the setting-up times for smaller batches at Water-hydro fell to 60 minutes following the introduction of FA. For parts produced in large batches, the maximum cumulative setting-up time among our sampled firms, after CNC machining had been adopted, was 600 minutes—in the Piston-rings case. Another important aspect of FA which has contributed to plant efficiency and reduced lead times is the increase in machining speeds. For the sampled parts considered above, machining cycle times fell by up to 80 per cent. The implications of these improvements in setting-up and machining times for plant efficiency will be considered further in the section 3.2. Reductions in batch size Most of the sampled firms reported significant reductions in batch size, following the adoption of FA and cellular lay-outs. Table 7.8 compares batchsize distributions before and after technological change in those firms which provided detailed information on this topic. In addition to the cases presented in Table 7.8 above, Mech-trans reported a 75 per cent reduction in the average batch size, Piston-rings mentioned a reduction of 35 per cent, and Small-services-A and Medium-services-B also reported reductions in clients’ order size, with consequent reductions in batch size.

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 197

Table 7.8 Brazil: distribution of machining batch sizes in sampled firms, 1985 and 1994

Source: Interviews

Until the mid-1980s, most of the sampled firms worked with large inventories of raw materials, parts and finished products. Most firms reported having had no criteria for optimal and minimum batch sizes, although they used to machine enough components for several months of sales. Air-brakes and Mediumservices-A did have guidelines: Air-brakes produced at least 3 months’ sales, and Medium-services-A had a minimum batch size of 8 days’ production for a 10hour set-up. With the adoption of FA and a substantial reduction in the size of orders, this situation changed completely. Most firms reported that the central rule is to keep inventories as low as possible (15–30 days’ sales). Oil-hydro mentioned the need to improve cash flow as a crucial argument in favour of this rule. Other firms reported substantial increases in inventory turnovers. At Mech-trans, the operating inventory turnover rose from 5.9 in 1989 to 11 in 1994, which was made possible with reduced batch sizes and lead times. However, even with FA, firms have not introduced economic criteria for optimal and minimum batch sizes. The most common rule-of-thumb was to produce whatever was necessary to meet the order plus the amount required to complete the minimum inventory. Thus batch sizes are determined primarily by demand. In fact Oil-hydro sometimes

198 THE COUNTRY STUDIES

Table 7.9 Brazil: increase in product diversity (distinct types) in sampled firms, 1985 and 1994

Source: Interviews. Note a The figure refers to number of basic shapes.

produced very small batches as a service to clients, irrespective of the costs involved. The policy was to attempt to pass on these costs in prices. Other firms confirmed the difficulty of establishing economic parameters to define optimal batches, given the lack of information on set-up costs and the large fluctuation in interest rates. These problems seem indicative of a more general fragility regarding cost control in the sampled firms. This point is discussed further in sections 3.2 and 4.1. Increased product variety We have seen that the introduction of FA was crucial for the sampled firms if they were both to increase product diversity within their main area of specialization and to improve their capacity to meet clients’ demands for more customized products or the development of new components for the clients’ new products. The increases in product diversity in the sampled firms are presented in Table 7.9. The adoption of CNC machines combined with CAD/CAM systems allowed the producers of hydraulic devices to manufacture pumps, valves and connections tailored to customers’ technical requirements. Oil-hydro’s industrial director said that FA had permitted a radical change in the firm’s business philosophy. The firm had introduced three new product families of more complex valves and gear pumps, and had begun to adapt its catalogue products to clients’ drawings. At the time of the interviews, this firm was able to manufacture ten times the number of different product sizes that it used to manufacture. Water-hydro had introduced four new products and was now able to meet clients’ dimension specifications. The firm’s customer-oriented marketing policy and the flexibility and quality resulting from the new production techniques had also facilitated an increase in sales of special pumps.

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 199

In the autopart firms, greater product variety has been important to develop new clients and to satisfy the increasing product diversification by current clients. The 50 per cent increase in the number of products at Piston-rings is directly related to the diversification of Piston-rings’ international clients. The reduction in the number of product families at Air-brakes is due to a recent redefinition of the concept of product families. As to the service firms, Table 7.9 shows only Medium-services-A, with a 66 per cent increase in the diversity of parts machined, but other firms also reported increases. Medium-services-B reported a 40 per cent increase in the number of component sizes machined, whereas Small-services-A stressed the greater diversity of discrete items subcontracted for autopart clients. The interviews confirmed that the use of FA may be a necessary condition for the introduction of new product designs. Management at Air-brakes and at Water-hydro emphasized that some of the parts designed for new products could not have been economically manufactured with old technology. To sum up, the section on changes in product scale and scope has shown that the incorporation of FA, in association with organizational change, is allowing the sampled firms to benefit from economies of scope. They are producing a greater diversity of products, in smaller batches, and in a more efficient manner. The question to deal with now is whether or not these economies of scope are displacing economies of scale. To answer this question, we need to examine what has happened to plant scales in the innovating firms. 3.2 Increases in plant scales It has not been easy to gauge plant scales in the sampled firms, since most firms were operating with idle capacity at the time of the study, and because they had output records but no records of their capacity. Except for Piston-rings, which publishes data on plant capacity in the company reports, plant capacities had to be estimated. In the sampled firms for which some information was available, we estimate that overall plant scales increased with the adoption of FA. At Piston-rings, capacity has been systematically expanded, from 100 million rings per year in 1986 to 118 million in 1993. The 1993 report says that a 30 per cent increase in the production of cylinder liners at the São Bernardo plant was obtained in that year alone, due, first, to the shift to a production cell lay-out, and, second, to the adoption of FA. Except for 1990 and 1991, this firm has operated close to full capacity for the past ten years. The output of the other sampled firms fluctuated rather more. In 1993, a year of recovering sales for the industry, the output at Oil-hydro, Water-hydro, Mechtrans, Diesel-engines and Medium-services-A was still below their 1985 levels. These firms had substantial idle capacity in 1993. Thus, the estimates of changes in plant capacity in Table 7.10 had to take account of output levels, firms’ estimates of idle capacity and the varying number of shifts operating in 1985 and in 1993. Table 7.10 shows that the increases in production capacity at Water-hydro and Medium-services-A were substantial (50 and 39 per cent, respectively). The

200 THE COUNTRY STUDIES

Table 7.10 Brazil: estimates of plant capacity in sampled firms, 1985 and 1993

Source: Interviews. Notes a Capacity in tonnes per year per shift. b Capacity in number of pumps per month per shift. c Capacity in transmission units per shift per day; Mech-trans data refer to 1986 instead of 1985. d Capacity in number of engines per shift per year. e Idle capacity was calculated considering the plant running in 3 shifts,9 except for Dieselengines (2 shifts). The night shift capacity averages 80 per cent of day shifts’.

engineer in charge of the implementation of new production methods at Waterhydro considered that 70 per cent of the increase in plant capacity was due to the introduction of CNC machines, while the shift to cell organization accounted for the rest. Capacity at Mech-trans and Diesel-engines increased by 18 and 25 per cent, respectively. However, a further increase in plant capacity was expected at Dieselengines with the implementation of a new engine production line based almost entirely on CNC machines. In 1996, total plant capacity will grow to 50,000 engines per year per shift, a 33 per cent expansion on the current level. Mechtrans also is considering investing in a new line of transmissions for small cars, which will entail further capacity growth. At Oil-hydro, capacity in tonnes of raw material machined did not expand much, but since this firm switched to manufacturing more complex and diverse parts, the number of machining operations per tonne increased. This is reflected in the increase in the value of sales per tonne of product, from US$20,500 in 1985 to $30,400 in 1993. In short, this firm is adding approximately 50 per cent more value per tonne. Since Oil-hydro has benefited from scope economies and productivity gains due to the new CNC machines, the increase in the number of machining operations per part did not imply greater capacity utilization. The output per working hour, which was close to 1 kilogram per hour in 1985, has

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 201

changed little. The conclusion is that, if the machining per tonne of output is considered, capacity at Oil-hydro increased by more than 9 per cent. The idle capacity in these firms is striking. At Water-hydro this was an intentional part of its modernization. The firm believes that export sales will increase in the near future and that it must be ready to hold its market share when the Brazilian economy returns to sustained growth. Although aiming at capacity expansion, the firm was unwilling to increase the area of the plant. A similar situation was found at Mech-trans and Diesel-engines. At Oil-hydro the capacity increase was not intended, but was welcomed since the firm believes that sales have been low due to the crisis in the capital goods sector and that its potential demand is considerably higher than the actual demand. In contrast, Mediumservices-A apparently was not expecting such a large capacity increase and its CNC machining capacity is now too far above the projected demand. We found signs in the other firms of increased plant capacity, although the information was not sufficiently full for the increases to be estimated. At Airbrakes, for instance, sales per employee rose from US$37,000 in 1985, when FA adoption began to accelerate, to $62,000 in 1989, and then to $88,000 (see Table 7.5) just after the implementation of organizational innovations. Considering that its product range and complexity did not change very substantially (as compared to Oil-hydro’s, for instance), that management said the market had not allowed any increase in the prices of finished products, and that management also reported greater idle capacity for the 1990s than for the 1980s, it seems that output and plant capacity also must have grown. But since Airbrakes has substantially increased the subcontracting of component manufacturing, we cannot estimate the capacity growth from the sales: employment ratio alone. The sales per worker at Small-services-A, on the other hand, increased from US $18,700 to $24,000 over the same period, a 28 per cent growth. Management revealed that this figure was in line with the expansion in output (tonnes of parts machined) and that, although the firm is machining a greater diversity of parts, there was no increase in the average complexity. Small-services-A is the final tier in the subcontracting network of the autoparts industry, and has not contracted-out any machining operation. Thus the capacity expansion must be close to the change in sales per employee. Medium-services-B provided no information on changes in output or idle capacity, but management said that one reason for incorporating CNC machines was to add new production capacity in order to supply increasing export sales. Overall, we can conclude that, in most of the sampled firms, plant scales have increased since the diffusion of FA started. In some firms this expansion was planned, and in others it was unanticipated. 3.3 Factors determining changes in plant scales In the sampled firms, the introduction of FA and of new forms of production organization are connected to the increase in plant scales in four ways. First, the

202 THE COUNTRY STUDIES

Table 7.11 Brazil: indicators of efficiency gains (%) in total plant setting-up time for sampled firms, 1985 and 1994

Source: Interviews.

increase in machine speeds and the decrease in the number of machines involved per part contributed to a reduction in average processing time, at a given level of output. At Small-services-A, management estimated that the total processing time of the plant had been reduced by 30 per cent with the introduction of CNCs. Although we were not given access to indicators of total plant processing times in the other sampled firms, some examples showing substantial gains in machining speeds for single parts were obtained (see Table 7.7). Shorter machining cycles resulted in the greatest increases in plant scale in firms such as Oil-hydro, Water-hydro, Air-brakes and Medium-services-A, where the density of diffusion is high (see Table 7.7). A second cause of increased plant capacities was the improvement in total plant efficiency stemming from decreased setting-up times. Table 7.11 presents indicators of these gains for some firms, in terms of the estimated ratio of total setting-up times to total machining times. Further evidence was found at Piston-rings, which reported a 20 per cent reduction in total setting up time; at Medium-services-A, which estimated a 30 per cent reduction; and at Small-services-A (also 30 per cent). Medium-servicesB pointed in the same direction, but did not provide precise information. This suggests that workers are less occupied with setting-up, thus contributing to increased capacity. Another interesting example was found at Mech-trans. This firm has kept records of the average setting-up time since the introduction of technological and organizational innovations. In 1985, the average setting-up time was 1.8 hours. Following the introduction of SMED (single minute exchange die) in 1987 this was reduced to 1.5 hours. However, according to the industrial director the reduction became very marked after the introduction of more CNC machines and a reduction in batch sizes following 1987. By 1990, the average setting-up time was down to 0.92 hours, and it was reduced to 0.4 hours in 1992. The third factor behind increased plant scales is the reduction in reject rates and reworking time, due in part to the use of CNC machines but more especially to the adoption of new quality-control techniques. At Piston-rings, the rejection rate fell from 8 per 100 rings to just 1 per 100 following the introduction of the total quality programme. Management stressed that this change led to a 10 per cent increase in output. At Air-brakes, the reject rate fell from 7,300 to 1,900 parts per million.

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 203

Fourth, workers’ greater motivation and the intensification of work have made an important contribution to reducing downtime, and therefore to increased output and capacity. The larger and more organized sample firms have been implementing new forms of work organization and new labour control practices. At Mech-trans, Piston-rings, Water-hydro, Air-brakes and Diesel-engines, the managements had introduced multi-tasking in job design, enlarging workers’ capacities to take on additional tasks related to the same or to different machines. Greater labour flexibility has been directly reflected in work intensification and downtime reduction. Another way of increasing workers’ contributions was the successful adoption of suggestion schemes at Mech-trans, Air-brakes and Pistonrings. Some workers’ suggestions had produced significant gains in plant efficiency. For instance, Air-brakes reported a number of examples of reductions in setting-up time (by up to 90 per cent) due to changes in set-up procedures suggested by workers. Thus the incorporation of FA and new organizational forms has been associated with increasing plant capacities, due in most of the sampled firms to gains in plant efficiency. However, this is not in itself evidence against the descaling argument. The question is whether greater plant capacities and output are still important for achieving cost reductions. This is the crucial point in the argument that FA should reduce the minimum economic scale. To address this question, we will examine the evolution of cost structures in the sampled firms. 4 Flexible automation and costs 4.1 Changes in total unit costs Cost control seemed to be the weakest management area in most of the sampled firms. For a number of important items, firms simply had no accurate knowledge of real costs. For example, in the pump and valve manufacturing firms, labour costs are estimated from standard times established by engineers rather than from a knowledge of real manufacturing times. The costs of raw materials, consumables, and so on are estimated in the same way. At Small-services-B, management reported that the firm subcontracts to an accountant who estimates costs and prices, but this information cannot be used for management purposes, since the management does not understand the cost-estimating procedures or the meaning of cost figures. In short, the information on costs in this survey must be treated with caution. While high inflation was mentioned as a major obstacle to cost control, some of our interviewees admitted that the environment of low competition which prevailed until recently in Brazil was responsible for the lack of interest in cost control. ‘The market just accepted that we passed our inefficiencies on to prices; so controlling costs with precision was not a priority.’ This may in part explain

204 THE COUNTRY STUDIES

why unit cost was for us the most difficult topic about which to obtain information. Three firms did provide detailed data on unit cost structures (see Table 7.12). The data reveal that unit costs have fallen since technological and organizational change started at Medium-services-A and Air-brakes, by 10 and 15 per cent, respectively. These reductions were smaller than the reductions reported from advanced industrialized countries (Alcorta 1993b). At Medium-services-A the major reductions in unit cost were in overhead costs, raw materials and labour, in that order. These were offset by increases in capital, maintenance, consumables and energy unit costs. At Air-brakes, most cost components were reduced, with the exception of consumables (and presumably capital costs, although no information was provided). Thus the changes in cost structure vary from firm to firm, for reasons which will be discussed below. There is also potential for further gains, because these firms were operating with significant levels of idle capacity (see Table 7.10). If output increased, capital and overhead unit costs would be further reduced. In contrast to the other two cases, Oil-hydro’s production and total costs increased substantially (by about 35 per cent). However, as we have seen, Oilhydro by 1994 had become virtually a new firm. The increase in the number of machining operations has required more capital, and more skilled labour per tonne of output, so capital and labour unit costs have increased. However, the increase in total unit cost is connected also to the high level of idle capacity, and increased utilization might well reduce fixed unit costs sufficiently to offset other rising costs. Cost reductions were also reported at Piston-rings, Mech-trans and Smallservices-A, although these firms refused to provide detailed cost data. At Pistonrings, operational costs were cut from 60 per cent of sales in 1986 to 43 per cent in 1993, which represents a reduction of almost 30 per cent. The most important factor was increased labour productivity, which reduced the labour unit cost. This firm has maintained a high level of investment in R&D (3 per cent of sales) throughout the past 15 years, but it was successful in reducing other indirect costs as percentages of both costs and sales. According to management, the substantial growth in output and sales since 1985 has been important in reducing the indirect costs per product. This change has allowed Piston-rings to face the toughening of international competition in recent years. Mech-trans also reported a decrease in total unit cost, by 15 per cent. Once again, the most important factor was increased labour productivity, partially offset by a substantial increase in all items of indirect cost except capital. As in the cases of Oil-hydro and Medium-services-A, Mech-trans would benefit from reduced indirect unit costs if the idle capacity which was expected in 1993 was in fact utilized. The management of Small-services-A estimated that the total unit cost, markups and prices had decreased by 15 per cent. Again, the major factor was higher labour productivity and reduced labour unit cost, which more than compensated for increases in capital, energy and training cost. If small-services-A was not producing and selling on a larger scale than previously, it would not have been

Source: Interviews Note a Capital costs not included.

Table 7.12 Brazil: unit cost structures in sampled firms, 1985 and 1994

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 205

able to compensate for increases in fixed unit costs. In fact increasing output was even more important for cost efficiency, following the introduction of FA, in the smaller sampled firms. The reasons for this will be examined below.

206 THE COUNTRY STUDIES

The general conclusion from this analysis of unit cost structures in the sampled firms is that increased fixed production costs and overhead costs, in the presence of substantial idle capacity, reinforce the importance of expanding output scales to reduce costs. 4.2 The impact of flexible automation on the structure of unit costs and firm scales Impact on capital unit costs The evidence suggests that the firms which adopted FA more intensively (Oilhydro, Medium-services-A and Small-services-A) suffered increasing capital costs, mainly due to the higher cost of CNC machine tools and machining centres. According to data collected from the sampled firms, CNC machines, whether imported or locally produced, are more expensive than conventional machine tools. Prices of CNC machine tools have fallen dramatically in the past two years, but during the 1980s, when most of the investment in FA was made, prices in hard currency terms were increasing. Oil-hydro, for instance, reported paying US$138,000 for an Index GE42-NC automatic lathe, made in Brazil, in 1981. In 1983, the same machine cost $140,000 and, in 1986, $220,000. The prices of CNC lathes made by the largest Brazilian manufacturer, Romi, went up in the 1980s by about 50 per cent. In the late 1980s Romi launched a simpler model (Romi Cosmos) costing approximately US$90,000. Prices of the simpler conventional machine tools varied from $40,000 (universal manual lathes) to $80,000 (single-spindle automatic lathes). Direct comparisons between different machines are problematic, since they are not able to perform the same tasks. Water-hydro paid US$500,000 to import a Mazak SQT30 CNC with laser micrometer, but this machine is superior to any locally produced machine tool. In broad terms the simpler CNC machine tools produced in Brazil were until recently at least 50 per cent more expensive than the simpler conventional lathes. This helps to explain why capital costs increased in the sampled firms when the adoption of FA accelerated. Oil-hydro, to take the most striking example, began increasing its outlays for FA in 1983. Over the past ten years the firm has replaced most of its conventional automatic lathes with CNC equipment. Its expenditures on CNC equipment amounted to US$5.69 million over 1983–6, and $4.18 million over 1987–91. An average of $1.1 million per year for machines alone is well above previous capital investments. The increased capital unit cost at Oil-hydro reflects the impact of this investment effort on depreciation costs. Although the shift to CNC machines yielded savings on building space, this had no impact on Oil-hydro’s unit costs. Capital unit costs could be reduced only if Oilhydro’s output increased, so as to utilize the current idle capacity. Similar factors accounted for the rise in capital costs at Medium-services-A and Small-services-A, although the situation of the small sampled firms requires some additional remarks. For firms like Small-services-A, the shift from

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 207

conventional to CNC technology entails an increase in capital costs and outlays, because the trade-off is not between automated machines and programmable automation but between manual lathes and CNC machine tools. For a larger firm choosing between fixed automation and programmable automation, the fixed option may require a more concentrated capital outlay at the outset, even where it is cheaper. For instance, the management of Diesel-engines said that an additional reason for adopting CNC machines in its new engine line was that the investment could be spread over time, allowing a more interesting cash-flow than was possible from rigid automated transfer-lines, which would require greater expenditure in the short run. For small firms with conventional manual lathes, the move to flexible automation entails an increase in capital investment outlay. The situation in the three firms discussed above is not, however, representative of Water-hydro or most of the autopart firms in the sample (Mech-trans, Pistonrings and Air-brakes). These latter had a more conservative investment policy regarding new equipment, often based on annual investments, largely for CNCs, equivalent to the historic annual depreciation cost. This explains why investment in FA has had little impact on the capital unit costs of these firms. Nevertheless, there is room for a reduction in capital unit costs, since most of these firms operate with substantial idle capacity levels. Impact on labour unit costs With the exception of Oil-hydro, labour unit costs decreased in all of the sampled firms following the adoption of FA and the new organizational practices, which led to higher labour productivity. Employment decreased between 1985 and 1993 in all firms which provided information for both years, except for Small-servicesA (see Table 7.5). Since sales remained constant (for the producers of hydraulic devices) or even increased (autopart and service firms), there was a general growth in productivity in terms of sales per employee. Table 7.5 shows that sales per employee increased by between 18 per cent (Oil-hydro) and 154 per cent (Mechtrans). The available data about physical productivity confirms this picture. Physical productivity rose by 12 per cent at Water-hydro and by 28 per cent at Medium-services-A between 1985 and 1993. Higher labour productivity was the main reason behind the decline in labour unit costs, although increases in direct and indirect wages may have partially offset the productivity gains. Medium-services-A reported that total hourly wages (direct wages plus social security taxes) increased from US$3.50 to $5.00 between 1985 and 1994. Piston-rings gave similar figures, and Mech-trans, Airbrakes and Small-services-A confirmed that hourly wages had risen. Several reasons for the higher wages can be given. First, the operators of CNC machines and the workers involved in cellular manufacturing undergo longer training programmes and are considered more skilled than their colleagues working with the earlier technologies. They receive a wage differential for this. Second, indirect wages have generally increased in Brazil in recent years because the government has created new taxes affecting firms’ payrolls to compensate for the decline in the total value of payrolls. Third, metal workers in Brazil’s

208 THE COUNTRY STUDIES

engineering industry are among the most organized in the country, and their unions were able to improve their wage situation in the late 1980s. Fourth, there was a substantial increase in training and education expenses in most of the sampled firms. An interviewee at Medium-services-B said that the firm had invested more in training in the 5 years prior to this study than it had over the previous 20 years. This might be an exaggeration, but it does reflect the emphasis that firms have recently been putting on training Oil-hydro differs from the other sampled firms in that it had taken the alternative route of substantially increasing the value added to output, which required not only more physical capital but the use of more skilled labour. Produc-tion workers at Oil-hydro, in 1994, were more skilled and earned higher wages than the workers who were employed there in 1985. Physical productivity (as measured in tonnes of output per worker), however, changed little, leading to a significant increase in labour unit costs in this firm. Impact on input unit costs Most firms in the sample agreed that FA and the new organizational methods contributed enormously to efficiency in the use of raw materials. Scrap was reduced and better quality in production processes reduced the need to rework pieces. This led to a fall in raw material unit costs at Medium-services-A and Air-brakes (Table 7.12), although the former considered that this was due in part to decreasing international prices for raw materials. Piston-rings reported major savings in raw materials of US$2.5 million in 1992, and attributed these to the TQM programme. Medium-services-B also stressed the contribution FA had made to its 20 per cent savings on raw materials. However, most of the sampled firms said that the cost per unit for consumables had increased. Table 7.12 shows that the cost of consumables rose following the introduction of FA at Oil-hydro, Air-brakes and Medium-servicesA. This is due mainly to more expensive tooling. Water-hydro, for instance, estimated that the cost of tools went up from US$0.48 to $1.20 per machining hour with the shift from conventional automatic lathes to CNC machines. For Medium-services-B, unit tool costs increased by approximately 50 per cent. Most firms also reported increases in energy costs, as can be seen from the figures for Oil-hydro and Air-brakes in Table 7.12. The management at Waterhydro said that the energy cost for CNC machines is US$1.85 per machining hour, compared with $0.29 per hour for conventional automatic lathes. Mediumservices-B and Small-services-A also said that energy unit costs had risen by 25 and 40 per cent, respectively. In contrast with the experience in developed countries, most firms reported that the maintenance cost per unit of output rose after the introduction of FA. The exceptions in this respect were Air-brakes and Water-hydro (see Table 7.12).

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 209

Impact on overheads and firm scales It was possible to gather information on the evolution of overhead costs (marketing, R&D and administration) in a number of firms. This showed that increasing competition and technical change had led to increasing overhead costs for R&D, engineering, marketing and client services. This tendency is clearer among the larger firms in the capital goods and autopart industries, but the service firms which have not been able to increase their marketing and engineering efforts are lagging behind. In the case of the hydraulic equipment firms, the need to provide more customized products and client-oriented services led Oil-hydro (see Table 7.12) and Water-hydro to expand project (R&D), marketing and client service activities. At Water-hydro, two-thirds of its employees in 1993 were engaged in indirect activities. Management reckoned that at least one-third of these were involved with marketing and project activities. One interviewee said that the company was becoming more and more like a project firm. The most significant change in Oil-hydro’s overhead costs was in marketing, where expenditures had been increased by 50 per cent to obtain new clients and take advantage of the firm’s recently developed production flexibility. The autopart producers have been equally affected by the new marketing and R&D requirements. These firms are under pressure from car makers to diversify products, to shorten product life cycle and to assume a greater share in design expenditures. Moreover, the increase in global sourcing requires the intensification of marketing efforts abroad. Thus it is not surprising that indirect expenditures at a firm such as Mech-trans have increased from 4.2 per cent of sales in the mid-1980s to 8.3 per cent of sales in the early 1990s. This rise in indirect costs can be broken down as follows: (process) engineering, from 2 to 4 per cent; marketing and sales expenses, from 1.3 to 2.3 per cent; technical assistance to clients, from 0.14 to 0.36 per cent; and R&D, from 0.7 to 1.6 per cent. At Piston-rings, general indirect costs had fallen but R&D expenses remained constant at 3 per cent of sales. Management mentioned that R&D outlays were increasing because of clients’ demands for high performance and the use of new materials. Piston-rings can face these new competitive conditions only by increasing sales (and scales). The situation is similar at Diesel-engines, where management declared that it had cost about US$20 million, over four years, to develop its new truck engine. This firm’s R&D expenses have been around 7 per cent of sales in recent years. Thus the reduction in the overhead cost component of unit costs at Air-brakes seems exceptional. This is in part because all Air-brakes’ project activities are developed abroad. Of the service firms, only Medium-services-B reported increased overhead costs. Management mentioned that clients now require some additional ‘technical content’ in machining services, meaning that it is better for business if the subcontractor can provide an improvement also (even if it is just a small improvement) in the client’s design. This is why Medium-services-B was implementing a product project department at the time of our visit. In contrast,

210 THE COUNTRY STUDIES

Medium-services-A was slashing overhead costs (Table 7.12) and had no plans for project upgrading or entering export markets. The smaller service firms reported no significant change in overhead costs. Yet it was easy to observe, during our visits, that the small entrepreneurs were very busy (overwhelmed, apparently) trying to perform all the new activities that clients now demand, that is, quality-control measuring and reporting, training workers, attending clients, and so on. The implication of the trends in overhead costs for firm scale and scope is straightforward. In the current market conditions, firms are required to increase substantially their expenditures for R&D, engineering and marketing, a requirement which represents a further burden on overhead costs. To compensate for this (and to avoid increasing overhead unit costs), firms need to expand their output scales. In short, at the level of the firm, the minimum economic scale is larger now than before FA. 5 Conclusion: industrial restructuring, flexible automation, scale of production, and the prospects for the location of production in Brazil Clearly, leading firms and their suppliers in the Brazilian engineering industry are making a substantial effort to restructure their product and production policies, to adapt to changes in the economy and in economic policy, and to become more competitive. However, not all of them have been equally successful. There is a clear gap between the larger and better-structured firms in the autopart and hydraulic equipment segments, on the one hand, and the smaller and more fragile firms in machining services, on the other. The latter had less capability, and lacked the technical and financial resources required to undertake a balanced process of technological and organizational change. In addition to changes in their product strategies, the leading firms had implemented (and continue to implement) a broad transformation in their production concepts and practices, combining the adoption of microelectronicsbased automation and new management practices. The case studies reveal that the diffusion of FA has been growing. The fact that most of the sampled firms’ investment in new equipment has been for FA suggests that diffusion will accelerate if investment increases. The study found evidence that for the leading firms in which technical change was comprehensive and systemic, technological and organizational innovations were integrated. Progress in one respect required simultaneous progress in the other. The particular pattern of adoption of FA and the means by which was combined with organizational change seemed to depend on firm’s market situation and competitive strategy. The leading firms adopted FA and new forms of organization initially to improve quality and increase product and process flexibility. This was necessary to adapt to the new conditions of competition in their markets. Moreover, they required their suppliers also to initiate the process of adoption. Quality and flexibility were achieved. Our case studies show

FA AND THE BRAZILIAN ENGINEERING INDUSTRY 211

substantial gains in terms of shorter setting-up times and machining cycles, which favoured substantial reductions in batch size and lead time, and increases in product diversity. The result was falling product scales. However, smaller product scales did not lead to reductions in plant scale. On the contrary, plant capacities in most sampled firms increased as a result of the incorporation of FA and the new organizational methods, and of the faster machining cycles and shorter setting-up times that flowed from adoption. Since demand in Brazil had stagnated since the early 1980s, and most firms in the sample exported little, the increase in plant capacity entailed an increase in idle capacity. Although in a few cases the growth in plant capacity was one of the objectives of the technical change, in the other firms it was not planned and, indeed, affected their ability to benefit from potential reductions in unit cost. Among the machining service firms were examples of ill-planned processes of adoption in which firms overestimated their need for CNC machines. Overall, firms could benefit to some extent from decreases in variable costs (mainly labour unit costs and raw material unit costs), but they will be able to compensate for the increasing capital costs and overhead costs associated with the adoption of FA only by increasing output and operating close to full capacity. Therefore, the evidence from this study suggests that optimal scales in the Brazilian engineering industry are increasing. The finding that the diffusion of FA is not associated with de-scaling has several implications for the location of production in developing countries. First, although the access of developing country firms to FA is facilitated by the greater divisibility of investment (because FA can be gradually or selectively incorporated), it does not follow that scale has ceased to be a barrier to entry. The case studies suggest that for developing countries to reap the benefits of the economies of scope associated with falling product scales, demand for their total output must be large enough to allow full utilization of their increased plant capacities. Second, the problems encountered by machining service firms in adopting FA reinforces the scale argument, since this contrasts with the better results obtained by the larger firms. Apparently it is easy to learn how to operate FA, but it seems much more difficult to become a good user. Requirements such as the technical capability to plan for the adoption of FA, the simultaneous adoption of organizational innovations, and the resources necessary to maintain a well-trained workforce may be beyond the reach of small and medium-sized local firms. It would seem advisable to shift the emphasis placed by the Brazilian Service for the Support of Small and Medium-sized Enterprises on organizational issues to give also support to these firms in adopting microelectronics-based automation. Large firms might also have an important role to play in strengthening the capabilities of small and medium-sized firms, by establishing links with them based on technical co-operation. Policy initiatives, in import liberalization and the promotion of competition, have accelerated the adoption of FA by firms in Brazil, but the crucial role of the expansion of demand in enabling firms to reduce idle capacity and so reap the cost benefits associated with this diffusion must not be overlooked.

212 THE COUNTRY STUDIES

Notes 1 I would like to thank Ludovico Alcorta, project co-ordinator, for his valuable suggestions throughout the realization of the survey and the preparation of this paper. I am grateful also to Jorge Katz, for helpful comments on the earlier draft, and to Eliane Rosandisky and Noela Invernizzi, from the Department of Scientific and Technological Policy, Campinas University (DPCT/UNICAMP), for their assistance in data collection and a review of the literature on the Brazilian engineering industry. 2 On technological fragility in Brazilian manufacturing, see Quadros Carvalho (1992). 3 From an average of 23 per cent of GNP in the 1970s, the investment rate fell to 16 per cent in the late 1980s. 4 The classification in Table 7.1 is that used by the IBGE (Instituto Brasileiro de Geografia e Estatística), the Brazilian federal agency in charge of the economic census and industrial surveys. 5 For the purpose of this comparison, density of diffusion is defined as the stock (in number of units) of equipment per 1 million employees in the engineering industry. This definition is derived from Edquist and Jacobsson (1988). A different definition is adopted in section 3 of this paper to measure the diffusion of FA in sampled firms. 6 Brazil, Argentina, Paraguay and Uruguay. 7 These are estimates made by the firms’ industrial directors and engineers. Our observation during plant visits confirmed that the density of utilization of FA in machining operations was substantially higher at the hydraulic device plants than at the autopart plants. 8 For a description of systemic organizational and technological change as opposed to the adoption of isolated techniques, see Fleury (1988). 9 Although night shifts entail higher labour costs and lower productivity, they are not an unusual practice in the Brazilian engineering industry in periods of growth.

8 THE IMPACT OF FLEXIBLE AUTOMATION ON SCALE AND SCOPE IN THE MEXICAN ENGINEERING INDUSTRY Lilia Domínguez and Flor Brown

1 Introduction: macroeconomic and industrial performance 1.1 Macroeconomic trends and poticy issues During the last twelve years the Mexican economy has undergone significant transformations due to a shift in government policy goals from import substitution to export promotion. The Mexican economic crisis of 1982 made it clear that the import substitution model applied during the previous four decades could not continue, even though it had allowed very rapid economic growth and the creation of an important industrial base. The transition to export promotion industrialization which began after the crisis of 1982 has been a slow and difficult process. In 1983 a stabilization package of drastic short-term economic policies was applied. A five-year strategy, complementing this package, aimed to restructure the economy to give it an outward orientation and make it more capable of generating the necessary foreign exchange. Subsidies were virtually eliminated and a greater emphasis on market forces was adopted. The long-standing regulatory and defensive policy towards foreign investment and technology imports changed to active promotion, and markets were opened to external trade. Although these changes were expected to proceed in a predictably selective and gradual manner, in practice import liberalization accelerated rapidly in 1985, so that by 1988 it extended to almost all sectors of the economy. After 1988, although these macro-economic policies continued, a moderate rate of growth was pursued. Between 1983 and 1988 the GDP growth rate was zero, and between 1988 and 1992 it was 3.4 per cent. In other respects it was a period of ‘deepening’ the 1983 structural change policies. These policies were intended to create the economic and institutional conditions required for the integration of the Mexican economy in international markets, emphasizing the importance of private investment leadership, foreign investment promotion, and North America Free Trade Agreement (NAFTA) negotiations. The NAFTA agreement was finally signed in 1993.

214 THE COUNTRY STUDIES

During these years, there were several achievements. The key macroeconomic indicators were transformed. The public deficit declined steadily, becoming a surplus by 1992. Inflationary pressures started receding in 1988, with the annual rise in the price index falling to 13 per cent in 1993. The renegotiation of foreign debts lowered the foreign debt. Finally, the structure of export revenues was reversed: while revenue from oil exports declined due to lower world prices, manufacturing exports tripled from 20 per cent of exports in 1983 to 60 per cent in 1992. On the negative side, moderate growth was not sustainable over the entire period. Economic activity declined again in 1993, with a 0.8 per cent growth rate. Employment growth over the whole period was extremely low, indicating that the rate of growth in the domestic market was insufficient. Moreover, the opening-up of trade had deleterious effects. Despite the growth in manufacturing exports, the external deficit quickly increased, reaching $23 billion in 1993. This has made the economy extremely vulnerable, since the external deficit is sustained by incoming external capital. It is clear, however, that the industrialization trajectory has changed. Unexpectedly, from the standpoint of textbook economics, most Mexican exports are not from labour-intensive sectors. Automobiles and autoparts account for almost half of manufacturing exports, with computers, electrical and electronic equipment, secondary petrochemicals and steel next in importance. A significant proportion of these exports are intra-firm transactions by multinational corporations. These exports tend to be tailored to the specific needs and standards of the parent company by an affiliate which is closely integrated with its parent company’s network. Thus, Mexican industrial transformation cannot be explained solely by internal economic conditions. It is part of the global pattern of competition. 1.2 The engineering sector Mexico’s engineering sector is of particular interest, since the most dynamic firms, and those which were severely affected during these transitional years, are concentrated in this sector. From 1983 to 1988, the engineering sector (sector 38) registered an annual growth rate in real terms of 4.9 per cent, while the rate for manufacturing as a whole was 2.3 per cent. Economic recession severely affected some subsectors. In particular, the situation was critical in metal furniture, structural metals, electrical household appliances and equipment, and transportation materials, all of which had negative growth rates. During this period many firms closed down and others merged. In contrast, growth rates for automobiles and motors, and for automobile accessories, were far above average, at 15.5 and 7.5 per cent, respectively. From 1989 to 1992 the engineering sector grew by 9.6 per cent per year, while manufacturing grew by 4.1 per cent. Heterogeneous growth arising from the new conditions became more acute. Automobile manufacturing continued to outperform the other industries, with an average annual growth rate of 22.8 per cent.

FA AND THE MEXICAN ENGINEERING INDUSTRY 215

Its share of total engineering output increased from 23.2 to 32.7 per cent. Electrical household appliances and structural metals also grew rapidly, at average rates of 10.2 and 9.0 per cent per year in the same period. Equipment and transportation materials registered a negative growth rate of −5.2 per cent, and other metal products, excluding machinery, had a growth rate of only 2.8 per cent. In terms of employment, the engineering sector registered a 1.1 per cent annual growth rate from 1983 to 1988, slightly better than total manufacturing employment, which grew by 0.8 per cent. In 1989–92, employment in the engineering sector grew by 0.6 per cent per year, while in manufacturing it declined by –0.6 per cent. Automobiles and electrical apparatus had positive employment growth, while employment declined in equipment and transportation materials, non-electrical machinery and equipment, and other metal products excluding machinery. The engineering sector has played an important role in the modernization of Mexican industry, although there are great differences between the technological levels of different economic activities. Productivity growth may be taken as an indicator of the extent of transformation in this sector. According to a recent study, average total factor productivity and labour productivity growth rates from 1984 to 1990 for manufacturing were 4.7 and 3.3 per cent, respectively (Brown and Domínguez 1994); the rates for the engineering sector were 7.7 and 7.6 per cent. Automobiles had the highest growth in total factor productivity and labour productivity, with 18.3 and 18.4 per cent, followed by motor vehicle bodies and autoparts, with 6.5 and 6.2 per cent. Other metallic products, transport equipment and materials, and electrical machinery and equipment all suffered falls in productivity. 1.3 The diffusion of FA in the manufacturing industry The use of FA in manufacturing is no doubt one ingredient in the updating process that is under way in Mexican industry. Although the diffusion of FA must be considered as still relatively limited, evidence shows that there is a nucleus of firms for which new technologies are increasingly important. The introduction of FA started during the late 1970s, and has been increasing in recent years. Several studies have accounted for their appearance and outlined their most important features (Domínguez 1993; Mercado 1990a; Ramírez 1993). Flexible automation is closely related to firms’ export strategies, and is having profound effects on their production processes, product mixes and labour organization. A recent survey among 5,000 enterprises found that the engineering sector accounts for 40 per cent of the CNCs and 45 per cent of the robots (INEGI 1994). Most of the literature available deals with technical change within the automobile industry (Micheli 1994; Sandoval 1986; Shaiken and Herzenberg 1987). Given the significance of this industry in manufacturing exports, its leadership in the modernization process is an established fact in Mexican

216 THE COUNTRY STUDIES

industry. Most studies have referred to the role of FA in fulfilling various requirements of the new competitiveness, with a clear emphasis on the impact of FA processes on organizational changes, modifications in the labour process and the greater flexibility of industrial relations.1 Patterns of diffusion The patterns of FA diffusion have changed during the last few years. The first CNC machinery was purchased to service the domestic market. Mercado (1990a) examined the acquisition of CNC machinery by 18 Mexican plants which owned 179 units in 1983. He found that local firms accounted for the majority of CNCs bought prior to 1981, while foreign-owned firms had the larger share for 1982–3. Domínguez (1993) explored the acquisitions made by a sample of user firms, observing factors of nationality, sizes and sectoral origin, and relating them to changes in the Mexican economy. She found that in 1980, when there was active public sector investment (which tended to favour tenders from local contractors) the great majority of the users were local firms. With the switch to export promotion, the diffusion of CNC machinery has become in recent years increasingly linked to export activity. Since multinational subsidiaries are taking the lead in export activity, it is understandable that the proportion of CNCs owned by local firms fell rapidly, from 81 per cent in 1980 to 37 per cent in mid-1987, when the study was carried out. The study found that in the early 1980s over 80 per cent of the CNC units were used by small and medium-sized firms, but their share fell to 45 per cent in 1983 and to 28 per cent in 1987. In the meantime, the share of the very large firms increased from 11 per cent in 1980 to 17 per cent in 1983, reaching 41 per cent by mid-1987. The automobile industry accounted for 22 per cent of the units in 1980, but by mid-1987 its share had risen to 51 per cent. In contrast, the shares of the electrical appliances industry (44 per cent) and the machine-tool industry (33 per cent) in 1980 fell to 17 per cent each in 1987. Evidence provided in case-study analyses by Shaiken and Herzenberg (1987), Sandoval (1986) and Carrillo (1993) confirms that the automobile industry was the most important user of FA. Export-oriented plants were found to have a higher index of automation than did domestically-oriented plants. Some 65 per cent of the export-oriented plants were fully automated, whereas only 30 per cent of the domestically-oriented plants were even semi-automated, the rest had having low degrees of automation (Ramírez 1993). Suppliers located in Northern Mexico were more highly automated than those located in the centre. Additionally, in the maquiladora enterprises (bonded export firms) along the northern Mexican border, there are electronics and autopart firms, mostly large and foreign-owned, which have introduced FA in their production processes. Brown and Domínguez (1991) found a group of twelve users in Ciudad Juárez. Although the use of CNC machinery was found to be selective and combined with labour-intensive processes, the number of CNC machines and robots in use by these firms is considerable.2 Similarly, a study in Tijuana found a group of

FA AND THE MEXICAN ENGINEERING INDUSTRY 217

plants with high labour productivity coefficients and a higher proportion of technicians per worker using intensive CNC machinery (Ramírez et al. 1988). Although these indicators show that multinationals are important users, this does not mean that local firms have not had access to FA. There are groups of mostly large, local firms which have gradually become more automated, particularly with the introduction of programmable logic controllers (PLCs) in the food and chemical industry. Most of these firms export either through subcontracting agreements, or through open market, arm’s length, exports. One possible explanation of the relative lag in the response of small and mediumsized firms to the market signals of the open economy is the differential effect of the recurrent recessions, since 1982, by firm size. The overwhelming importance of large foreign subsidiaries in the transition to an open economy, and the reduced linkages between large and smaller firms in this process, may be even more important. It is important also to stress that flexible automation is related not only to export markets but to the needs of the domestic market. As shown by the consecutive waves of acquisition since 1977, when imports started, macroeconomic trends play an important role. Thus, Mercado (1990a) found that three-quarters of the units in his sample were acquired between 1980 and 1982, the years of the oil boom. Domínguez (1993) found that 68.3 per cent of the units in her sample had been bought in 1984 and 1985, years of economic recovery after the debt crisis. Most acquisitions in our own sample were made after 1989, when economic growth resumed. 2 Technical change in the manufacturing industry: casestudy evidence 2.1 Sampled firms Sampled firms were selected in order to achieve the highest possible intensity of use of FA. An initial list of 30 firms was defined, following the advice of machine distributors. Sampled firms were expected to have a high ratio of CNC to conventional machinery.3 We selected 10 firms which had indicated that FA contributed more than 50 per cent of the value of production. The actual ratio of CNC machines to conventional machines may be much lower, since firms retain their old equipment. Nevertheless, as Table 8.1 shows, the density of FA usage in the sampled firms averaged 80 per cent. We attempted to include firms with different characteristics in terms of capital ownership and firm size. Our respondents in the firms were production and human resource managers. We made an effort to select interviewees who were knowledgeable about FA and techniques and their organizational implications. Of the 10 firms selected, 4 produced components for the automobile industry and 1 for heavy transportation; 3 produced capital goods and 2 were machining

Source: Interviews. Notes a L = local ownership; J = joint-venture. b As defined in Chapter 3.

Table 8.1 Mexico: characteristics of sampled firms, 1994

218 THE COUNTRY STUDIES

FA AND THE MEXICAN ENGINEERING INDUSTRY 219

subcontractors producing for both engineering and automobile industries (see Table 8.1). The sample included 4 large firms with more than 500 employees and workers, 4 medium-sized firms with 100–500 workers, and 2 small firms with less than 100 workers. In terms of capital ownership, there were 3 joint-ventures and 7 firms with local ownership, 2 of which had been joint-ventures at some time in the past. Three firms had been established during the last 12 years, 6 initiated operations during the 1970s, and one had existed since the late 1960s. Most of these firms were the major or the sole local producer of their particular products and were important in the local industry. But with the opening up of the economy, and now NAFTA, firms also consider exports an important element of their competitive strategies. Two firms had export ratios of less than 10 per cent, but growing. Another 4 firms exported 10–20 per cent of production, and the remaining 4 exported more than 20 per cent of production. Some export activity occurred indirectly, through the exports of automobile producers. This is not accounted for in the export ratios. Each of these firms can be characterized as having a high technological level within its own industry. Most firms manufacture their own tools, fixtures and jigs. Although it cannot be assumed that these firms are at the technological forefront, since there may be other sectors or industries with higher automation rates, they do demonstrate a common dynamic response to the new economic conditions, in the form of the introduction of FA. 2.2 Flexible automation and the production process The production process was broadly similar in each of the sampled firms. The first stage is die casting and forging. Glass-die, Machine-sub1 and Transmex are supplied by other members of their industrial groups; Fuel-sys executes this operation internally; and the rest of the firms contract the work out. The next stage comprises machining operations, which range from very complex processes to auxiliary operations; machining centres are used for complex operations requiring several tools. This is followed by thermal and chemical treatments, carried out in most sampled firms. The next stage is assembly, which is more complex in the production of glass machinery than in the production of dies. Lastly, testing and quality control are carried out by electronic monitoring. As will be seen later, firms were seeking to co-ordinate all these stages of production to reduce stocks and downtime. The firms had introduced FA mainly in machining operations, although electronic controls also were important in thermal processes and testing procedures. Table 8.2 shows the types of machine installed in the sampled firms: 668 units in all. Four firms had added PLCs to machines in addition to the acquisition of new CNC machinery. In some cases, PLCs have moved from being simple digital controllers, acting in isolation, to powerful signal processing devices capable of communicating with other machines. Diffusion is increasing in quantitative terms and in complexity.

220 THE COUNTRY STUDIES

Table 8.2 Mexico: types of machine in sampled firms, 1994

Source: Interviews

The sampled firms owned 162 machining centres.4 Three firms owned flexible machining centres (FMSs). At Joint-shafts, the FMS consisted of a number of CNC machines and transport facilities including automatic rail-guided vehicles and robots, pallets combined with a computer control system and computerized production scheduling. In Glass-die and Machine-sub1 the FMS was limited to the operation of small cells of CNC machines or machining centres with transport facilities. Each firm possessed at least one CAD unit. In seven of our firms it was used in the engineering department for product design. The other firms owned a CAD unit, but did not indicate whether it was working on product design. Additionally, CAD was reported as very important in the design of tools, jigs and fixtures. Some firms have considerable experience in the use of FA, which they have been operating for the last eight years; four firms had installed its first machine in 1982 or before. However, in the past five years adoption of CNC machine tools has intensified: the number of units has increased fivefold. Most firms were awaiting the arrival of additional CNC machine tools or PLCs, indicating that FA diffusion is continuing. The patterns of introduction of FA varied between firms. One important factor is the phase of development which a firm had reached before beginning to automate. The newer firms—the machining subcontractors (Machine-sub1 and Machine-sub2) and the producer of dies for motor vehicle bodies (Body-dies)— were established with flexible automation and incorporated the latest possible technology as a competitive strategy. The other seven firms acquired CNC machine tools gradually. The characteristics of a firm’s original production processes also will have influenced the way in which that firm adopted FA; and these characteristics should be viewed against an outline of the engineering sector in Mexico. This industry was never technologically comparable to those of developed countries. The fragmentation of the capital goods industry inhibited producers from taking advantage of economies of scale or more cost-efficient methods (NAFIN 1985). There were too many firms for the size of the Mexican market. The autoparts industry in Mexico has traditionally had scale problems due to the great number

FA AND THE MEXICAN ENGINEERING INDUSTRY 221

of models being produced by the assemblers, given the size of the domestic market. The result was that most subdivisions of the engineering sector used mostly conventional technologies, except for the automobile assembly industry which used rigid technologies and transfer lines. Most units consisted of lathes of different types (parallel, vertical or revolver), grinders, drillers and stampers. Most of the sampled firms had some production units which combined specific automatic machinery with conventional machine tools (lathes and grinders). Of these firms, the autopart producers have the greater proportion of automatic lathes. Old technologies tend to be retained. Even among the more automatized firms, the rate of scrapping is very low. Most firms have large unused facilities and have retained a great part of their conventional machine tools for use in combination with new equipment or in the event of breakdown of new equipment.5 Firms reported that it was not always profitable to use CNC machines. They assign specific machining operations or products to conventional machinery. For developing countries, these grey areas, identified by Carlsson (1984), in which transfer lines, conventional and CNC machine tools compete, may be more important than in developed countries. This, as Bhalla and James (1988) observe in their analysis of technological blending, is in part because of the difference in labour cost between developing and developed countries. An additional reason is the higher cost of flexible automation in relation to the existing stock of conventional machinery within the firm. CNC machinery entails higher investments. The firms seek the optimal combination of old and new machinery to minimize average unit cost. For example, Glass-die used conventional machine tools for simple machining operations such as simple repairs or making parts for the replacement parts market. HT-shafts used their semi-automatic machines whenever machining operations did not require frequent changes. Most sampled firms have a high level of engineering skills, and have devoted considerable resources to machine retrofitting. Retrofitting seems to have played an important role in four firms, with appreciable savings in capital costs. It has eased the introduction of flexible automation in a period of macroeconomic turbulence with great financial restrictions. 2.3 Reasons for flexible automation adoption Flexible automation was purchased to improve product quality, flexibility and productivity, and to reduce response times. Originally, firms may have purchased new machines to keep abreast with the most ‘up-to-date’ technology, but subsequent purchases seem to have been made according to firms’ specific needs. Attainment of flexibility and quality was the most important reason for purchasing FA, though speed also was frequently mentioned. However, firms invariably mentioned that these requirements had to be met at the lowest variable cost. This contrasts somewhat with the findings of Domínguez (1993), that cost factors were considered a secondary or tertiary requirement after quality, and that reductions in material waste, product defect rate and inventory were

222 THE COUNTRY STUDIES

considered more important than labour-cost savings. International competitiveness was behind these motivations. We did not find significant differences between the motivations according to firm size or capital structure, but there were important differences between autopart producers and capital goods producers. In our opinion these differences spring from the original technical conditions of the plants in these industries. For autopart firms, the most important objectives were flexibility and quality. Before introducing FA, these firms, which originally had a greater proportion of automatic and semi-automatic machinery, dealt with standardized designs. Automobile producers began working with stringent JIT systems, so that autopart producers needed to attune their production schedules to smaller orders and quicker response times. They needed to comply with certain tolerances and specifications, and low rejection rates. The old technology could not meet these requirements, because automatic machinery and production lines involve high change-over costs and cannot be used for small batches, and because conventional machine-tools depend on the dexterity of workers and do not produce the consistent quality required by new market conditions. Autopart producers also reported being pressured to lower costs, particularly from automobile producers. In other words, while flexibility, reliability and quality are important considerations, cost is also an important variable in deciding whether to invest in the technologies, and which machines to buy. The capital goods producers, making moulds, glass, machinery, valves, pumps, and the subcontractors working for them and the autoparts industry, all faced irregular work flows and suffered from uncertain delivery times because of the reliance of conventional machine tools on the dexterity and responsibility of workers. One motivation for adopting FA was to reduce the trial and error costs associated with the manufacture of complex parts with a higher degree of precision. In other words, quality means repeatability and consistency, which lead to lower waste. An additional reason was employers’ concern with labour reliability. Lower dependence on labour was mentioned by Mexico-valves and Machine-sub2 as an advantage of FA. Each firm had started with a workforce which lacked the skills required for machining tasks with conventional machinery. This would suggest that skill saving is one important advantage of the new technology, since these skills are scarce in developing countries. These firms considered that FA provided flexibility, a faster response time and efficiency. Machining centres or FMS, in particular, have increased flexibility, permitting shopfloor stocks to be drastically reduced. Glass-die and Machine-sub1 gave this as a reason for purchasing the FMS units, although these firms conceded that hardware changes had to be complemented by organizational change towards group technology, or cellular lay-out. Glass-die’s main motivations for adopting FMS were downtime and labour-cost reductions, and higher quality. FMS was intended also to reduce stocks on the assembly line, thanks to its production programming capabilities. Although well-organized production with conventional machine tools could assure a certain degree of flexibility, the response time would have been much greater and the possibilities for programming production were limited.

FA AND THE MEXICAN ENGINEERING INDUSTRY 223

2.4 Process of adoption Regarding the process of adoption, we found no differences between capital goods and autopart producers. There were slight differences between firms according to when they were established and type of capital ownership. In the newer firms, Machine-sub1 and Machine-sub2, the founders had assembled information on various types of machinery from trade fairs and suppliers before purchasing FA. In the already established firms, purchasing procedures evolved from discussions within the production and engineering departments. In particular, Glass-die, Fuel-sys, Transmex and Joint-shafts mentioned a long process of search, deliberation and evaluation. Fuel-sys, for example, bought its first units on an experimental basis. At that time, the firm was producing water pumps for the domestic market and the owner, foreseeing the need to modernize before the opening up of its market, acquired two units after an extended period of examining the opportunities offered by FA in terms of new products. Those firms with some foreign capital (HT-shafts, Mex-mufflers, Joint-shafts, and Body-dies) also received input from the overseas’ partners or parent company. In particular, the parent company provided information on different machines and suppliers, and contributed to financing. Other studies have stressed the role of foreign capital in the diffusion of FA.6 However, local partners and the entrepreneurship of the local owner were also important. For example, in Mex-mufflers, the introduction of FA and the modernization process began when the company switched to a more dynamic Mexican partner, a powerful industrial group. The cases of Glass-die, Transmex, and Joint-shafts would suggest that modernization can be aided by affiliation to a local industrial group, which provides financial and technological support. Finally, it is important to mention that foreign firms can play an indirect role as customers. For example, Fuel-sys mentioned the encouragement and financial support of its main customer, an automobile producer wishing to develop local suppliers. The initial learning process, from the time when the first order was placed until the production of the first batch of products using the new machines, was similar in all the firms, needing 6–12 months on average. Four to six weeks of training was needed for engineers and operators. Machine start-up, and the production of the first batch in satisfactory condition, took 3–8 weeks, depending on the characteristics of the firm. Established firms faced greater resistance and had to make specific efforts to help their workers adapt to FA, so they required more time. Old practices are not easy to abandon. Firms said that the problem was not the complexity of the machines, which may in fact be easy to use. However, the cost of human mistakes can be high. As we shall see in section 2.5, the adoption of flexible automation was accompanied by organizational change.

224 THE COUNTRY STUDIES

2.5 Organizational change and flexible automation Lay-out and JIT systems One advantage of FA is the ability to produce small batches. This contrasts with the old practices, which had large containers of raw materials, semi-finished inputs or work-in-progress waiting in line (a ‘just-in-case’ organizational form). This is obviously a waste of capital. Organizing the flow of materials with JIT programmes decreases the amount of capital tied up in work-in-progress. Another advantage of FA is speed. But if machinery is not properly laid out, there will be an inadequate flow of materials and delays will follow, so that the advantages of the speed of new machines will not be achieved. Thus, changes in lay-out are a necessary part of the introduction of FA and of JIT itself, although it is true that some lay-out changes may be unrelated to the FA. Therefore JIT and lay-out changes are necessarily linked to FA: as Coriat (1992) maintained, they are one of the pillars of lean production systems. It is important to add that lay-out changes and JIT, particularly the latter, contribute to quality awareness on the part of the workers. They are often considered part of a general quality strategy. As several analysts have noted, FA is never independent of organizational form (Edquist 1992; Hoffman 1989a). Nevertheless, the relation may go in either direction: an organizational change may lead to or facilitate technological change, or vice versa, or the two may accompany and condition each other. In Mexico, we find two extreme cases. Ford Mazda started with a complete automation programme, while General Motors has preferred to emphasize organizational change. But General Motors has invested in the CNC machinery necessary to complement its organizational changes, and Ford Mazda has introduced organizational changes as well (Micheli 1994; Carrillo 1993). We need to identify which organizational changes are inextricably linked to CNC machinery and which are not. Table 8.3 shows the changes in machine lay-out. Production lines of consecutive operations have evolved into production cells that group machines by similar products and processes. The objective is to reduce downtime and unnecessary product flows. For example, in Glass-die, production cells are technological groups of 17–20 CNC lathes and grinders, surrounding a single machining centre that produces a range of similar products. Manual labour is used to handle the materials and finished products. In Mex-mufflers there are two production cells grouped around assembly machines that turn out a family of products, and an additional line for specific operations. The lay-out aims to link machining operations to assembly lines. Before FA was introduced, operations were physically separated, implying unnecessary transport costs and higher rents. With the changes in lay-out, the transfer of materials is so simple that there is no need for conveyors. Joint-shafts offers a particularly interesting example. The firm has 6 production lines, each manufacturing 10–12 models of half-shafts, with a mix of operations and CNC lathes and machine centres. The movement of materials and

FA AND THE MEXICAN ENGINEERING INDUSTRY 225

Table 8.3 Mexico: organization of production processes in sampled firms before and after flexible automation adoption

Source: Interviews. Note a The year of transition varies: 1987 for HT-shafts; 1988 for Glass-die and Transmex; 1989 for Machine-sub2, Mexico-valves and Fuel-sys; 1990 for Body-dies; and 1991 for Mex-mufflers.

semi-finished goods is fully automatic and flexible (FAF). Materials are usually transported within the line. If a machine is down for some reason, the materials are redirected out of the line to an available machine. Afterwards, materials may return to the line, if necessary. In the sampled firms, lay-out changes have been made gradually and by trial and error. As the number of machines has increased, more radical lay-outs have become possible and downtime is further reduced. We conclude that lay-out changes increase the efficiency of the machinery. Besides modifying the machine lay-out, six of the sampled firms, in particular autopart producers, are developing JIT (kanban-style production), aimed at reducing stocks and increasing workers’ quality consciousness. Workers handle a much-reduced stock of materials (as little as 10 per cent of previous volumes) at any one time. The results are startling: Glass-die reduced programmed stocks in assembly by 50 per cent; Mex-mufflers reduced stock on hand from 6 months to 45 days; Joint-shafts is working to achieve a goal of zero stocks; Transmex has reduced the cost of keeping inventories by 42 per cent; and Fuel-sys reduced

226 THE COUNTRY STUDIES

inventories from 120 to 55 days. Although JIT is undoubtedly important in reducing capital costs, some firms working with FA have not introduced it. For example, those whose principal product is customized, as it is with Machinesub1, Body-dies and Mexico-valves, are not under pressure to have JIT programmes. But our findings show that the scope of organizational change is increasing among firms from different industries: whereas in 1987 changes were concentrated within the automobile industry.7 Labour and organizational changes We found firms working on four important aspects: (1) work systems’ organization in teams; (2) changes in work incentives; (3) emphasis on training and changes in training philosophy; and (4) qualitative modifications in command and control structures. (1) Work systems’ organization in teams. Even if rigid machinery was marginal before the introduction of FA, firms tended to organize labour in a rigid manner. Workers used to perform specific operations, since machines were laid out functionally, as was the assembly line. The organization of workers in teams is intended to create a new category of worker. In place of specialized labour, the goal is a multi-skilled and more participatory labour force, with workers able to take on several tasks (see Table 8.4). For example, in Mex-mufflers, assembly is moving towards a ‘hand-to-hand process’, in which each worker adds several components to the assembly and, instead of using containers, the piece passes from hand to hand.8 Organizational changes are intended also to stimulate workers’ participation. Two firms had instituted problem-solving teams: informal groups of workers, technicians and engineers who gather to discuss specific problems. In HT-shafts, before the firm buys new machinery, a group usually is formed to analyse the existing problems and the role of the new acquisition in solving them. They examine the production process to distinguish between rigidities deriving from existing conventional automatic technology and rigidities arising from defective organization. Thanks to this analysis, they locate the specific problems found in labour routines and those that can be solved through the substitution of machinery. The group ceases to exist as such when its objective is attained. Thus the reorganization of workers has given firms a degree of flexibility that goes beyond that bestowed by CNC machines. Teams tend to form more in autopart producers than in capital goods producers, though Glass-die is an exception in this respect. (2) Changes in work incentives. Some firms have found that, as time passes, wage structures become biased towards the higher job categories, due to training or seniority. This has hampered moves towards higher productivity and quality. Five firms (HT-shafts, Transmex, Mex-mufflers, Joint-shafts

Source: Interviews.

Table 8.4 Mexico: labour, organization and administration in sampled firms

FA AND THE MEXICAN ENGINEERING INDUSTRY 227

228 THE COUNTRY STUDIES

and Glass-die) have therefore modified their incentive systems, hoping to link wages to productivity or problem-solving capabilities, as opposed to seniority. In particular, Mex-mufflers and Glass-die reduced the number of wage categories by bargaining with unions to modify previous practices. The issue itself, however, has not been completely resolved, especially in firms that have been in business longer. Unions, in these firms, are more resistant to such changes and, due to the government anti-inflationary programme which has been in effect since 1988, there is an element of wage restraint. Some firms (Mexico-valves is an example) have been able to avoid problems with unions by moving the firm to another location and hiring new workers, while retaining the essential workers.9 (3) Emphasis on training and changes in training philosophy. Training has increased at all levels, whether measured in hours or in terms of the quality of workers’ training. In some firms there was also a change in focus, because of the emphasis on direct teamwork training in new production systems. The most interesting example was at Glass-die, a firm with a long tradition of training of all kinds. Previously, workers received universal instruction (general education) which was not linked to the specific problems encountered in production. This meant that learning was not connected to worker’s productivity or attitudes, a problem that the new training programmes have corrected. (4) Qualitative modifications in command and control structures. We observed some streamlining of production management, particularly in three of the larger firms: two autopart makers, Transmex and Mex-mufflers, and Glassdie had eliminated 2 of their 7 levels of top management. The firms said that they had reduced supervisors also. In one very small firm, the trend has been towards self-supervised workers, who are responsible for meeting quality standards for tasks within a relatively flexible schedule. Nonetheless, large firms may find that there is a limit to the number of supervisors that can be eliminated. Joint-shafts concluded that, even if workers are more responsible, the function of supervisors is not just to observe mistakes, but to offer encouragement and leadership. Thus supervisors have not been eliminated at Joint-shafts. In conclusion, the technologies in our sampled firms have become more sophisticated. Use of conventional machinery initially was not accompanied by flexible organization of labour. In the first phase of the introduction of the FA, firms felt no need to change their labour organization. However, most firms now realize that the machines by themselves cannot achieve the objectives which justified their original purchase. The benefits of the introduction of CNC machinery can be fully reaped only if its introduction is combined with a range of organizational techniques. Increasingly, production and labour have been radically reorganized to improve productivity, customer satisfaction and labour relations.

FA AND THE MEXICAN ENGINEERING INDUSTRY 229

3 Economies of scope 3.1 Changes affecting the scope of production Changes in setting-up time Setting-up times have been drastically reduced because of the programmability of flexible automation. This leads to smaller batch sizes and a greater diversity of products. FA machines can be easily and rapidly switched to different configurations or standards, so reducing downtime, lead times and setting-up times. The effects differ between firms depending their previous production methods and on the type of product. For example, setting-up times for autopart producers have dropped from hours to minutes. When HT-shafts (a producer of truck shafts and brakes), switched from non-flexible automatic transfer lines to CNC lathes and machine centres, the setting-up time dropped from 16 hours to 10 minutes. These reductions are due to the simplification of machine adjustments and tooling changes. Mex-mufflers reduced setting-up time from 15 hours to 10 minutes, and have set a goal of 2 minutes. In Joint-shafts, setting-up time went from 8 hours to 3 minutes with their FA machines. This firm not only introduced CNCs, but also retrofitted electronic controls to other lathes, constructed more versatile auxiliary equipment, and built an electronically controlled transport system which could ensure optimal utilization of the production machinery. However Joint-shafts emphasized that flexibility was not just a matter of machinery. Proper programming of machinery, timely availability of raw materials and multi-task workers were also necessary. The reduction in setting-up times allows different models to be produced, as we will see. Capital goods producers, whose production is very heterogeneous, had average figures. For example, in Machine-sub1, the average setting-up time was 4 hours when the firm used conventional lathes as opposed to 40 minutes for CNC machinery. There has been a marked improvement also with the introduction of a machining centre with a multi-pallet system. This system programs the sequential production of several orders, which may have different batch sizes and involve different tasks. The advantages are: the reduction of downtime due to the programming of the different machining tasks for each piece, and the reduction of the role of the worker, since tasks such as fixing pieces, changing tools and manipulation of the pieces are automated. The cost of producing such small batches in isolated CNC machines, let alone conventional ones, would be much greater. Once a group of orders has been completed, the machine needs to be reset. The setting-up time is lower than would be required for isolated orders. The machine stops working at night. The interviewee said that programming was done every 2–3 weeks.

230 THE COUNTRY STUDIES

In Mexico-valves, the old machinery required as much as 15 hours of settingup time for products of quite different types, whereas with CNC machinery only 4 hours are needed. The setting-up time between two similar products has been reduced to less than 2 hours. Batch sizes Firms in the capital goods industry and those engaged in customized production were used to receiving smaller orders than were other industries. Now, with increasing competition, the typical order has been further reduced (Table 8.5). For this reason, Machine-sub1 and Machine-sub2 have been forced to increase the number of customers they serve by increasing the heterogeneity of their output. The challenge for them is to lower setting-up times and to program each order so as to avoid downtime. In Machine-sub1, the reduced and variable size of batches encouraged the firm to introduce a machining centre with a multi-pallet system, whose advantages were mentioned above. In Glass-die, given the large number of components and the small number of orders, batches in their FMS unit may be as small as a single piece. This unit works as an independent technological group, storing up to 30 pieces, and can be programmed to work continuously. It has 120 different tools, machining 2 pieces simultaneously. The FMS unit stops at night, because the machine needs to be monitored when working. The firm is considering adapting the machine for an unmanned shift. Glass-die used to subcontract some of the pieces now machined in the FMS. The advantage of introducing the FMS instead of subcontracting these small batches is the possibility of accurate programming, and reduced stocks of work-in-progress and downtimes. In autopart firms, the change towards smaller batches is more pronounced. Automobile manufacturers used to place bigger orders, with batches of as many as 1–2,000 units, and were apparently not conscious of the cost of having large stocks. However, during the last 10 years the nature of competition in the automobile industry has changed. Automobile producers are working with muchreduced stocks and autopart producers have been required to adapt their technology.10 In the autoparts industry we found that the minimum batch size is somewhere around 100 pieces. Given the large proportion of all Mexican CNC units in use in the autoparts industry, this would confirm Mercado’s (1990a) findings regarding the correlation between the demand for CNC units and medium-sized batches. We detected some limits to batch sizes. Even if capital goods firms may need at times to produce a single piece, it does not follow that doing so is always profitable for these firms. Changing from one model to the next involves some related investments in fixtures, jigs, tools and system planning and maintenance. These investments may be unprofitable if the batch size is small and the frequency of orders is low, given the price of the product. In other words, some of our firms had encountered limits in reducing their batch sizes, with the minimum size depending on investments in jigs and fixtures, setting-up times and product price. Fuel-sys had bought some of its CNC units to satisfy specific

FA AND THE MEXICAN ENGINEERING INDUSTRY 231

Table 8.5 Mexico: batch sizes in sampled firms

Source: Interviews.

orders for automobile assemblers. Given the price of the product, it was not profitable to reduce batch sizes further, unless the customers were willing to pay for the investment in jigs and fixtures. Machine-sub2 hopes to specialize in medium-sized batches, since their stock of jigs and fixtures has been reduced. The difference in change-over costs for the firm is important, and depends on the availability of tooling. When tooling is available, setting-up times are a few minutes only. But when the firm does not have the proper tooling, setting-up may take three or four hours. Given the

232 THE COUNTRY STUDIES

situation of the capital goods market in Mexico (recession and fragmentation of demand), it is unprofitable for the firm to continue investing in tooling. Therefore, the firm’s strategy has been to increase market share as an autopart subcontractor —since such firms have larger batches. The firm did occasionally produce batches of very small size, if the price of the product was significant or the potential customer was important. In Mexico-valves, where CAD is used for the design of the company’s own jigs, fixtures and tooling and there are good stocks of these components, there is a rule-of-thumb to estimate the minimum economic batch size: the investment in jigs and fixtures should not be greater than 7 per cent of production costs (see Table 8.5). The exact level of this percentage may be unimportant, but the fact that the firm produced such a figure shows its concern to establish when such investments are profitable. In short, the sampled firms in the capital goods industry are using CNC units to reduce batch sizes, but these firms appear to be in this respect exceptional, with most CNC units being devoted to medium-sized batch production. This would confirm Mercado’s findings (1990a). 3.2 Changes in product diversity Flexible automation has allowed producers to reduce setting-up times and to work with smaller batches and a wider product mix. All of the autopart producers had increased their product diversity (see Table 8.6). Joint-shafts reported twenty changes in design per year. HT-shafts had doubled the number of models of brakes which it produces, and the number of models of shafts and axles has more than doubled over the last five years. Mex-mufflers had increased the number of different mufflers it makes from 180 to 240 over the past three years. Transmex produced five models of transmission in 1989, and at the time of the interview had added three component kits entailing 100 different parts. Fuel-sys reported that it had not introduced a single new product between 1978 and 1986, but since 1990 there had been more than 400 variations. After 1989, when the market for carburettors declined suddenly, the firm began production of a range of fuel injection systems for new automobile models. Fulsys has engineers permanently assigned to working with assemblers and R&D teams to design components for new car models. Transmex, the leader in manual transmissions until the beginning of the 1980s, has retained its leadership for heavy vehicle transmissions but has lost most of its share of the automatic transmissions market. The firm produces replacements with its own trademark for the aftersales market, and has been able to develop export models of manual transmissions for racing cars and military equipment. The latter are highly profitable, but volumes have been reduced. So far they have not been able to begin production of automatic transmission systems. Kits of transmission components account for most of Transmex’s exports. Capital goods firms reported that customer requirements had forced them to reduce batch sizes and offer a greater diversity of products. Glass-die had always

FA AND THE MEXICAN ENGINEERING INDUSTRY 233

Table 8.6 Mexico: changes in product diversity in sampled firms

Source: Interviews.

worked on a customized basis, but the firm reported that thanks to FA they were now able to comply with most customer requirements. Die production in Bodydies is a similar case. In fact, both firms have CAD facilities connected via satellite with their customers, to comply with customer requirements. Mexicovalves produced ball valves in two families: flanged and threaded. Each was produced in different sizes and for various working pressures. At the time of the interview, the firm had doubled the number of sizes and working pressures, and had added a new family of valves—butterfly valves—in six sizes and four working pressures. In short, our evidence shows a clear increase in the number of products manufactured in all firms. However, increased diversity does not always mean new products. The greater diversity in most autopart firms is accounted for largely by variations on existing products, though Fuel-sys and Transmex have both increased variety and added new products. The creation of new products is also the result of efforts in product development. In the past, these firms tended to copy designs, but both are now assigning a greater percentage of their sales to their engineering departments. CAD has contributed to increased product diversity. In conclusion, the evidence shows that the introduction of flexible automation has created the conditions for the emergence of economies of scope among

234 THE COUNTRY STUDIES

firms. Setting-up times have fallen dramatically within all firms, batch sizes have been cut and the diversity of products has increased. Firms are able now to switch to the production of different goods within the same production facilities. However, some firms have found that there are limits to how far batch sizes can be reduced. Increases in product diversity among sample firms, with some important exceptions, have involved more variants of existing products than it has the creation of original new products. Thus, while FA has no doubt allowed firms to increase product diversity, there is some risk of overstating the advantages to the Mexican firms of the flexibility permitted by FA. It remains to ask whether these scope economies imply that scale is no longer an important factor in cost reduction. This is analysed next. 4 Plant and firm scale 4.1 Production volumes and capacity utilization We found that our sampled firms aimed to use their CNC machinery more intensively. Table 8.7 shows production volumes and capacity utilization. No firm reported precise numbers for the increase in productive capacity, but they did claim to have greater productive capacities due to FA. Six firms (Fuel-sys, Transmex, Mex-mufflers, Joint-shafts, Mexico-valves and Machine-sub2) were producing the same or higher volumes of output in spite of having idle capacity and a smaller labour force. Capacity increases ranged, in rough figures, between 100 and 300 per cent. Thus, production scales have increased with the introduction of FA. Due to the slowdown of the economy, increased capacity has not necessarily led to greater production volumes. The most extreme example was found at one of Glass-die’s plants, where capacity utilization had fallen to 30 per cent of 1988 levels. Body-dies’s production is increasing rapidly but its volumes are still unstable. HT-shafts was restructuring its activities and did not produce volume figures; however, it reported having increased its capacity. Except for one firm (Joint-shafts), capacity utilization was low at the time of the interviews. Firms in the autoparts industry reported capacity utilization rates of 50–70 per cent. The capital goods industry was facing greater problems: Mexico-valves and one of Glass-die’s plants reported utilization rates of 30 per cent, and Machine-sub2 reported 70 per cent. These low rates may be due to the firms’ increased capacity, with the introduction of FA, in combination with the decline in economic activity in Mexico since 1993, when GDP growth was just 0. 8 per cent.

FA AND THE MEXICAN ENGINEERING INDUSTRY 235

Table 8.7 Mexico: volume of production and capacity utilization in sampled firms

Source: Interviews.

4.2 Factors affecting scale Cost factors The firms reported that they were utilizing the new technology machines more intensively than the rest of their equipment, partly because of higher depreciation costs, and partly because they had had to accept higher expenses and greater sales-related efforts to meet customer demands. For example, Glass-die and Body-dies have a service that links their CAD facilities by satellite with their customers. These factors may force them to increase the aggregate plant and firm scale. In times of low demand (as in present-day Mexico) some firms undertake outsourced machining work to cover these depreciation costs. Part of this outsourced work is for external markets (Glass-die, Fuel-sys, HT-shafts and Mexmufflers). Firms considered this a last-resort strategy, since profit margins in this activity are low.

236 THE COUNTRY STUDIES

Institutional factors There are other factors which do not directly lead to scaling-up, but do restrain the decentralizing of stages of production, which would eventually result in descaling. Firms do not always find it efficient to do outsourced work, although there are some small firms with FA, including some in the sample, that are doing subcontracted work. However, these may still be considered isolated examples. In general, small firms in the engineering sector lack both the flexible automation and the modern management systems, thus limiting their ability to comply with quality requirements and guarantee delivery times. The sampled firms also contract-out only a minor part of their production, and Glass-die is now producing components that it previously contracted-out. There are not enough efficient small subcontracting firms to support distribution networks serving these firms. Machine-sub2 reported that competition was tight, and that often its worst potential competition came from its own customers, some of which are larger firms that also have FA. Thus, even if these small firms have a comparative advantage, given their greater flexibility, in taking on outsourced work on new products they always face the possibility that their customers will decide to produce the parts in house. Machine-sub2 was therefore planning to combine subcontracted work with the manufacture and marketing of agricultural machinery transmissions. Technical advantages of flexible automation FA means higher machining speeds, which is one important factor in reducing lead times. With conventional machinery organized in functional lines, changing to another model required long setting-up times. FA has considerably reduced setting-up times. FA adoption is normally accompanied by changes in production lay-out to reduce unnecessary and time-consuming flows. There is greater intensity and efficiency in the use of machinery and labour. Labour productivity increases because of itself, better organization and a smaller workforce. We found that lead times in the sampled firms have been considerably reduced. Among the autopart producers, Mex-mufflers reported that the lead time for a muffler had dropped from 3 weeks to 4 days, and Transmex’s lead time for a transmission system had dropped from 2 weeks to 4 days. Fuel-sys reported that the lead times for several of its products had fallen to an average of one-sixth of the previous level. HT-shafts did not produce figures for lead time reduction, but reported that the delivery time in its brake department had fallen from 1 month to 1 week. Joint-shafts illustrates how lead times can be reduced by integrating machines to form a fully automatic flexible system which organizes the work flow and the workload of each machine to reduce downtime to a minimum. With the old technologies, planning and scheduling the manufacturing involved selecting the individual machines for the 3 stages of production required for the production of half-shafts: soft material machining (which involves 4 operations); thermal treatment; and reboring (involving 3 different machining operations). The old

FA AND THE MEXICAN ENGINEERING INDUSTRY 237

system involved inefficiency and considerable loss of time. Now, priorities are assigned to the different machines and the sequential transport of pieces is programmed, as also is the flow of materials. Lead times have dropped from 6 hours to 6 minutes. Of the firms in the capital goods industry, Machine-sub2 reported that the average lead time per piece had fallen by 50 per cent, and Glass-die reported a 20 per cent drop in lead times after the introduction of FMS in 1988. Machinesub1’s delivery time dropped from 6 weeks to 2 weeks. Only two of the firms interviewed could not produce productivity figures, citing the lack of precise statistics given the customized character of capital goods production. Data from other firms are shown in Table 8.8. Some aspects make comparisons difficult. Products are radically different, and while some capital goods producers, such as Glass-die, have been undergoing change for years, others have only recently changed from conventional to FA. In addition, the proportion of its production which is produced using CNC machinery and the complementary organizational techniques, has evolved over time, and it is impossible to disentangle their impact on productivity. Nevertheless, the trends show a 40 per cent increase in productivity per worker since 1988. In Mexicovalves, output per worker has doubled. Machine-sub2 did not produce figures for the productivity of labour, but machine productivity had risen from 2 to 7.5 pieces per machine-hour over the last five years. Of the autopart firms, Joint-shafts reported that output per employee had doubled during the past three years, and Transmex’s physical productivity went from 133 to 299 transmissions per worker per year between 1985 and 1994. For Fuel-sys, the number of water pumps produced increased from 400 to 3,000 per day, while the number of workers fell by 30 per cent over 1989–93. There was a change also in product technology. The firm used to produce 20 carburettors per man-day with old technologies. With FA it produced 70 units per man-day.11 HTshafts increased the production of brakes per employee by 90 per cent. The increase in output per employee since 1988, for the firm as a whole, was 16 per cent. Mex-mufflers’s productivity per worker went from 30 mufflers to 49 per day in the last three years. 4.3 Other competitive advantages Most interviewees emphasized that the benefits of FA and organizational change went far beyond reductions in lead time, productivity increases and faster delivery rates.12 As the time required to respond to orders falls, there is a drop in inventories, and thus in financial costs. Delivery time also has become a competitive weapon in its own right, along with quality and reliability. With the liberalization of imports after 1988, the reduction of delivery times has become part of the export-oriented strategy. These advantages result in a combination of cost reduction and better quality. Some firms have had statistical production control for a number of years.

Table 8.8 Mexico: gains in lead-time reduction, labour productivity and quality in sampled firms

238 THE COUNTRY STUDIES

Source: Interviews

FA AND THE MEXICAN ENGINEERING INDUSTRY 239

240 THE COUNTRY STUDIES

Transmex, for example, has had TQM programmes for the past twelve years. Some outstanding results in quality improvement have been achieved. Sampled firms not only reduced the external rejection rate, but implemented programmes to reduce the time devoted to corrective machining operations and to removing scrap. For example, HT-shafts has almost eliminated external rejections, which are down from 2 per cent in 1988 to 0.01 per cent. Mex-mufflers eliminated the reworking department and reduced scrap by 40 per cent in one year. Machinesub2’s rejection rate has been less than 1 per cent during the past five years (see Table 8.8). It is noteworthy that CNC machinery accounts for only a part of these reductions. Other programmes are needed. For example, five years after having introduced CNC machinery, Joint-shafts reduced scrap from 109 parts per million to 20 parts in one year with a specific programme in the assembly line to reduce small mistakes by workers. Four firms had reduced the physical space required for production and storage. This occurred both because each CNC machine displaced several traditional machines and workers, and because they implemented JIT. Delivery times depend also on administrative changes; the reduction in lead time may not flow through to a significant drop in delivery time unless the firm updates its administrative systems. For example, HT-shafts was able to reduce lead times from 30 days to 1 day, but found that, because of administrative procedures, the minimum delivery time was 1 week. This explains the interest of firms in undertaking administrative changes to complement the introduction of CNC machinery. To sum up, firms have expanded productive capacity thanks to FA and new organizational techniques. Where volumes had not increased, the firms blamed adverse demand conditions. So when economic activity improves, firms will be producing greater volumes with reduced space and a labour force which, as we shall see below, is around 60 per cent of the 1988 workforce. 5 Costs and prices We have seen that FA has created the conditions for reaping economies of scope within the sampled firms: the reduction of batch sizes and greater diversity. It is important to apply cost analysis to see whether in fact this can be carried out efficiently. On the other hand, our interviews did not suggest that scale has ceased to be an important factor. The argument that FA brings scope economies rather than scale economies implies that smaller batches and greater variety result in unit-cost savings. Scale economies are realized when unit costs fall as production volume increases. Four firms had increased their volumes and one maintained the same level, but was working at only 30 per cent of its capacity. This would indicate that optimal production scale had increased in these firms. It remains to ask whether there is any relation between unit cost reductions and economies of scale or scope which result from the new production capabilities.

FA AND THE MEXICAN ENGINEERING INDUSTRY 241

Table 8.9 Mexico: changes (%) in price and total unit cost, in local currency real terms, in sampled firms (1980=100)

Source: Interviews. Notes a The firm reported this figure as meaningless given the customized character of its product. b The product is different and more complex.

5.1 Unit cost and prices Mexican firms continued to enjoy some measure of protection until 1987, when tariffs were cut drastically as part of an anti-inflationary effort. Most of those interviewed conceded that profit margins during the protected years were in excess of 10 per cent for most firms. Since the coming of trade liberalization, competition has been stiff, and international prices now prevail. Sampled firms provided estimates of the drop in their average prices, as shown in Table 8.9. All the firms which provided figures, except for Fuel-sys, registered price decreases of 15–40 per cent in real terms during the period for which changes were analysed. Two firms did not provide price figures, and while Fuel-sys reported a higher price for its main product, this was different and more complex than the previous product. Fuel-sys is an example among autopart producers of competitiveness based on a new product. In the other firms, the increased diversity has not resulted in higher prices for differentiated products. This is not a failure in the utilization of FA. The conditions for product differentiation depend not only on the manufacturer alone, but on market conditions. As Soete (1985) noted, international competition and the diffusion of FA are making domestic markets more competitive, forcing innovating firms to cut prices. Given these market conditions, the autopart producers in our sample can be considered part of the success story of surviving firms in Mexico. The possibility of charging higher prices for new and differentiated products depends not only on the use of FA, but on design capabilities and firms’

242 THE COUNTRY STUDIES

marketing strategies. The lack of these capabilities may limit the potentialities offered by FA. Not all firms were willing to give unit-cost figures. Some, particularly in the capital goods industry, said that their production was too heterogeneous. Others considered unit-cost figures to be irrelevant. For example, Body-dies and Machine-sub1, two suppliers of specialized capital goods which are not competing on price said that fast response time was more important than unit costs as a measure of performance. We did obtain cost figures for six firms, but not the total unit cost for each product. Thus Table 8.9’s figures are approximations, only showing the trends in unit costs. Unit cost declined in three firms, remained constant in one, and rose in the other two for which we have data. These variations are not linked with differences between capital goods and autopart producers. Of the capital goods firms, one was unchanged and there was a 15 per cent increase in Machine-sub2, explained by the fact that products are not really comparable and are perhaps more complex. On the other hand, Mexico-valves reported an 8 per cent reduction, which is explained by the shift to a new plant away from Mexico City and the introduction of FA (density of 85 per cent), with a new and reduced labour force. Among the autopart producers, Fuel-sys reported an increase in total unit cost but warned that this was not comparable with the previous figure since a different product is now being made. Mex-mufflers and Joint-shafts reported large reductions in unit cost, of 36 and 40 per cent in real terms. The firms with the higher figures are those which started their modernization programme more recently. The 80 per cent reduction in Mex-mufflers corresponds to a restructuring process starting in 1989. Joint-shafts’s 40 per cent occurred over a period of eight years. This firm combines several favourable factors: an automated production process; important changes in labour organization; and full capacity utilization. From the technical and economic evidence, we conclude that smaller batches and a greater diversity of products have been accompanied by constant or declining costs in four firms, suggesting the presence of economies of scope. There was a clear relation between higher output and reductions in unit cost in three firms, suggesting economies of scale. We found two firms with increases in unit cost; but, especially in the case of Fuel-sys, prices also had increased indicating greater value added as the result of the greater complexity of the product. This was achieved by greater use of physical capital (flexible automation), the production of a new product, and greater diversity. Although the relation between higher output, diversity and reduction in unit cost is not as clear as in the first examples, the data from these two firms do not contradict the possibility of economies of scale and scope. That there is still unused capacity and pressure from costs such as capital and overheads suggests potential for higher output and variety, and for unit-cost reductions. Finally, in one firm unit cost remained constant in spite of reductions in output, thanks to the ability to produce small batches with FA, and

FA AND THE MEXICAN ENGINEERING INDUSTRY 243

Table 8.10 Mexico: structure of costs (%) in sampled firms, 1989 and 1993

Source: Interviews.

organizational changes. The fact that this firm produces customized products suggests that this may be considered a clear case of economy of scope. 5.2 Cost structure An analysis of the different items in the cost structure highlights elements which are important to understanding the dynamics of unit costs. Table 8.10 shows that there was a relative increase in direct costs in four of the six firms which gave cost figures. Firms reported using their raw materials more efficiently and that this implied cost savings. The cost of raw materials increased as a proportion of total costs in three firms (Fuel-sys, HT-shafts, Mexico-valves), and fell in three others (Glass-die, Transmex and Mex-mufflers). We do not know if the increases are explained by increases in the prices of raw materials or by increases in volume. In any event, the share of raw material costs in total costs depends also on the changes in the other items. Capital and labour costs Depreciation cost increased significantly in absolute terms and as a percentage of total cost in all firms, as seen in the fourth column of Table 8.10. This is the result in part of increased investments. For example, Glass-die’s investment in machinery increased by almost 300 per cent between 1986 and 1993. The increases in depreciation costs were higher in Fuel-sys, which reported 50 million dollars in investment, and Mexico-valves, which reported investing 20

244 THE COUNTRY STUDIES

million since 1989. Machine-sub2 did not provide information on cost structure, but reported that depreciation costs had increased threefold. Four firms reported a reduction in wage cost as a share of total cost. Transmex reported no change and Glass-die reported a relative increase in wages, from 14 to 18 per cent of total cost, between 1988 and 1993. These changes in the share of wages in total costs can be explained by changes in employment levels and wage rates. Concern has been expressed about the impact of FA on employment. Domínguez (1993) observed that the number of workers was reduced among sampled user firms. However, most of this reduction occurred during the recession which began in 1983. Only a small part was due to the introduction of CNC machinery as such. Some firms had made organizational changes which enabled them to limit the increase in labour volume accompanying the revival of demand following the recession. The study suggested that future employment losses might be minor. Among our sampled firms, the number of workers has been reduced (see Table 8.11), partly because of the effects of lower demand, and partly because of the introduction of FA and resulting organizational changes.13 The introduction of CNC machinery and organizational changes has not been a once-and-for-all process, but rather a dynamic response in which firms undertake small changes on a trial-and-error basis and gain additional cost-reduction opportunities through learning by doing. All but two firms registered a decline in total employment. Mexico-valves is a special case because the firm laid-off its labour force before moving the plant. This explains the higher percentage reduction in total employment. Although firms were eager to say that they had made general reduction in all employment categories, the fourth column of Table 8.11 shows that the decline was generally greater among blue-collar workers. Firms have retained their specialized technicians and engineers. Transmex is an exception. This firm has been cutting employment levels for the last ten years: at the beginning of the 1980s it had nearly 5,000 workers and employees. Initially there was a decline in the proportion of blue collar workers, due to the introduction of FA and the retention of technicians and engineers. In recent years, administrative changes have meant that more of the redundancies have affected white-collar workers, although there have still been significant reductions among blue-collar workers. With the drop in employment, we would expect a reduction in wage costs, and perhaps also a drop in the share of wage costs in unit costs. Some firms reported paying higher salaries to retain their workers, but on the whole the government’s wage-freeze policy has kept wage rates on the low side. Wage bills had decreased in real terms in 5 of the 6 firms that provided figures, by amounts ranging from 8 to 58 per cent. Glass-die reported a bias towards higher wage categories because of a training policy in past years, which it was trying to solve. Nevertheless its wage bill fell by 16 per cent. The wage bill at Fuel-sys increased by 8 per cent in real terms from 1989 to 1993, despite lay-offs, because of wage rises. In sum, as employment dropped considerably in most firms, the wage bill also fell. The share of wages was reduced in most, but not all, firms.

FA AND THE MEXICAN ENGINEERING INDUSTRY 245

Indirect costs Indirect costs dropped significantly in relation to total costs in 2 of the 6 firms, and in the other 2 the changes were minor. From an analysis of the items for which we had data, it appears that employment in supporting functions decreased in absolute terms, along with the support infrastructure. One major effect of the introduction of FA (and of organizational changes) was a shift in the command and control structures. We have already mentioned the streamlining of production and the trend towards flatter hierarchical structures in several of the firms (Transmex, Mex-mufflers and Glass-die). Glass-die also centralized several departments (sourcing, sales, accounting, engineering) for its four plants, laying-off administrative workers. Another reason for the reduction of salaried employees is the introduction of computers in office work and management systems. Hence there were considerable cost savings. The implication is that the firm performs the same tasks for a greater diversity of goods, and at a lower cost thanks to a leaner centralized staff. Not all the items contained in indirect costs in Table 8.10 were reduced. Some of them rose, particularly financial, marketing and distribution costs, and R&D expenses. Firms were not willing to give detailed information on these costs. However Fuel-sys, HT-shafts, Glass-die and Mexico-valves reported that these items increased during the period being considered. In relation to financial costs, there was a general consensus that conditions in the Mexican financial market are not competitive in international terms and that credit is very expensive. Even loans backed by the Mexican development bank (NAFIN) have very high interest rates (around 30 per cent). Firms reported that marketing and distribution expenses were now more important. This was because of the greater diversity and differentiation of products following the opening up of the market. We have already pointed out how crucial today are rapid response, the capacity to introduce new products and the ability to anticipate the market. We observed considerable efforts by the firms to acquire these capabilities. Firms have had to establish closer communications with their customers. Fuel-sys, Transmex, Mex-mufflers and Joint-shafts have been making continuous visits to their customers’ automobile plants, while Glass-die and Body-dies have begun communicating with their customers by satellite, which probably lowers distribution and marketing costs. The increased R&D and engineering costs were explained by the need to take a more active stance in the competitive struggle in relation to product diversity and changes in processes. Five of these firms spend 5–6 per cent of turnover on research and development. Most mentioned that these activities had been much less important before, when they copied products. Now, in spite of having technology transfer agreements, there are areas in which they have produced their own technology.

Notes a ‘Before’ is 1988 for Glass-die and Transmex; 1989 for Machine-sub2, Mexico-valves and Fuel-sys; 1990 for Body-dies; and 1991 for Mex-mufflers. b The reduction in the share of blue-collar workers in Glass-die may be overestimated, since the firm has recently rehired retired engineers and qualified operators, under temporary contracts, as part of its organizational restructuring towards cells of production. The experience of the firm with the production process was thought to be especially valuable during the organizational change. All these workers are shown in the salaried workers’ payroll.

Table 8.11 Mexico: changes (%) in employment levels in sampled firms

246 THE COUNTRY STUDIES

FA AND THE MEXICAN ENGINEERING INDUSTRY 247

Table 8.12 Mexico: changes in sales, price, total cost and profit margin (all %), in local currency real terms, in sampled firms (1980=100%)

Source: Interviews. Notes a The firm gave no precise figure on its profit margin, but stated there was no change b The product is different and more complex.

5.3 Profit margins The empirical evidence given in 5.1 and 5.2 shows that firms are taking advantage of both scale and scope economies, achieving corresponding increases in productivity. Thus, costs for various items have dropped. These cost reductions, however, were not necessarily reflected in higher profit margins. We found no specific pattern among industries. Table 8.12 shows that four firms (Fuel-sys, Transmex, Mexico-valves and Machine-sub2) had suffered declining profit margins during the previous five years. Joint-shafts observed no changes, and Glass-die and Mex-mufflers reported an increase. The latter two firms had made important internal changes. Due to Glass-die’s strategy of heavy investment, restructuring production in cells, centralizing support functions and reducing employment, the firm increased its profit margin from −0.33 to 7.83 per cent. Mex-mufflers’s situation reflects change at several levels resulting from its incorporation in an industrial group in 1989. The firm increased its profit margin from 6.2 to 9.1 per cent. Joint-shafts, which was the only firm which had introduced changes in both equipment and labour organization while working at full capacity, reported maintaining its profit margin by cost reductions, in spite of the fall in prices after 1986. The situation of the firms which faced falling profit margins reflects the effects of greater competition, but there are differences relating to the particular circumstances of each firm. Machine-sub2 had FA from the time of its establishment. Thanks to its vigilant cost control, this firm succeeded in defending its profit margin, which fell by only 2 percentage points between 1989 and 1993, in spite of declining prices for its product. However, the firm reported

248 THE COUNTRY STUDIES

that its scope for further cost reductions was limited, and it complained about financial costs in Mexico. Mexico-valves may be considered as exemplifying the opposite: the changes in the firm are recent, and costs have dropped by an average of 45 per cent as compared with the old plant, yet its profit margin was reduced from 12 to 6 per cent. The firm said this was due to the opening up of the market. Its sales fell by 47.9 per cent between 1989 and 1993. The firm was heavily indebted, since the Mexican partners had recently acquired equity from the foreign partner. Transmex has been through a long period of change in which FA has played an important part; the firm has also reduced its size considerably, from 5,000 workers in 1980 to 767 in 1993. Nevertheless, it has suffered badly from strongly fluctuating domestic sales, despite healthy exports. Its loss of market share in automatic transmissions was mentioned earlier: sales fell by 17 per cent from 1988 to 1993, and profit margins dropped from 14 to 8 per cent. Lastly, Fuel-sys is a good example of a firm that has made changes with considerable success. The interviewee said that the opening of the market had increased competition and pressure on product prices, to the extent that ‘we are in a different world’. The firm’s costs are now higher, but the products are not strictly comparable with its previous products. Its profit margin may be considered high (13 per cent), in spite of the decline of 12 per cent since 1989. Our results show that firms are facing international competition, expressed in demands for higher quality, shorter delivery times and lower prices. In the context of depressed domestic demand, this competition is leading to reduced profit margins in most of the firms. This decline in profit margins could be halted, either by reducing costs and increasing markets or by developing new products with higher profit margins. The analysis of unit costs produced evidence of a relationship between greater diversity of products, higher output and reduced unit costs in three firms. In two firms, although unit costs increased, sales levels and the value added to output both increased because their products had become more complex. This was achieved by making more use of physical capital (flexible automation) and by adding new products and greater variety. Although the relation between higher output, variety and reduction in unit costs is not as clear as in the first examples, the evidence from these two firms does not contradict the possibility of economy of scale and scope. Only one firm provided a clear case of economies of scope, with constant costs at a reduced output. Cost-structure analyses suggest that the cost strategy of most sampled firms was based on reducing the wage bill and indirect costs. Within the category of indirect costs, the wage costs for support functions, such as sourcing, sales and accounting, were reduced, but there were increases in depreciation, financial and distribution costs. At the firm level, we found contradictory evidence regarding the impact of FA on scale. The reduction in wage costs for the support functions may be considered as evidence of a reduction in scale at the firm level, since it implies that leaner firms are able to perform the same tasks for a greater diversity of goods. This would be evidence of economies of scope and de-scaling at the firm

FA AND THE MEXICAN ENGINEERING INDUSTRY 249

level. However, our evidence is not sufficient to draw definite conclusions on future trends. The increases in other indirect costs, such as marketing, engineering and R&D costs, would imply a scaling-up effect. Thus far, the cost savings coming from the administrative restructuring of firms have outweighed cost increases in marketing, engineering and R&D, but this may not continue. Lastly, the lower profit margins in most firms are due to several factors. First, the depressed domestic demand is an obstacle to reaping the potential advantages of FA. Second, our sampled firms, with few exceptions, do not have the market power to establish their product prices. Third, the lack of product development technologies cannot be overlooked. 6 Main findings and conclusions Our research shows that flexible automation has been decisive in the response of the sampled firms to the effects of competition, particularly the demands for more frequent model changes, higher accuracy, more stringent quality control and more rapid delivery at competitive prices. The density of FA among sampled firms was relatively high, but most firms also reported that the machines by themselves could not achieve the objectives which justified their purchase. Thus we found organizational changes in the majority of sampled firms. The production process and the workforce have been radically reorganized in most of the sampled firms to increase productivity and customer satisfaction and improve labour relations. Several technical advantages of FA were reported by our sampled firms. The machines can be easily and rapidly switched to different configurations or standards, reducing setting-up times; and their higher machining speeds are important in reducing lead times. Another factor is the machines’ programming capability, which permits greater accuracy and predictability of work flows. Lastly, some firms have been able to integrate machines, allowing the work flow to be planned and scheduled to reduce down time. Average lead times have been reduced dramatically. The reduction of setting-up time allows the production of smaller batches. In the capital goods industry, batches now range from one or two pieces to 300, and in autoparts, where batches could be as high as 2,000 with old technologies, they have dropped to 100. The diversity of products has increased as a result by at least 30 per cent, as compared to the product range before the introduction of FA. One finding that deserves attention is that most firms have increased capacity of production, by amounts ranging from 100 to 300 per cent. Increased capacity does not necessarily lead to greater production volume, but six firms were producing the same or higher volumes of output, despite the slowdown in the economy. All but one had idle capacity. The technical advantages of FA are meaningless if they cannot be translated into economic benefits. But empirical evidence on economic advantages was difficult to obtain. Not all firms were willing to give detailed information. Of the six that

250 THE COUNTRY STUDIES

did give information on costs, 3 reported reductions in unit cost; 1 reported that unit costs had remained constant, and 2 firms’ unit costs were higher, but were not strictly comparable because the product had become more complex. Thanks to these unit cost reductions, firms have been able to cope with the opening up of the economy and the fall in product prices which it brought. Cost structures have changed, with absolute and relative reductions in direct costs due to the reduced labour costs, largely because of smaller workforces. Direct costs also were reduced by using raw materials more efficiently. Indirect costs dropped in relation to total costs in four firms, mainly because of cuts in support staff, in both absolute and relative terms. This was the result of the streamlining of production, the trend towards flatter hierarchical structures, the centralization of support functions and the introduction of computers. The empirical evidence on technical advantages and economic benefits suggests the presence of economies of different types. The reduction in batch sizes, and greater product diversity with decreasing or constant unit costs imply that production has become more efficient. Flexible automation has created the potential for economies of scope at the plant level within the sampled firms. The reduction in indirect costs is further evidence of the growing importance of economies of scope at the firm level. However reductions in unit costs linked to volume increases suggest that economies of scale are still an important factor. We found a clear relation between higher output and reductions in unit costs in three of our firms. In two firms, in which units costs were higher, both sales and value added had also increased, so these firms cannot be adduced as evidence against economies of scale. The fact that there is still unused capacity, together with the high level of overheads, suggest that there is a potential for higher output and further unit-cost reductions. In one firm, unit costs were constant, in spite of reductions in output. The fact that this firm produces customized products suggests that this may be considered a clear case of economy of scope. Although it has been suggested that the introduction of FA reduces the impact of scale factors, we found some factors exerting pressure for larger production volumes. These factors are cost related, institutional and technical. The technical advantages have already been mentioned. On the cost side, the introduction of FA has been accompanied by an increase in depreciation costs in all the sampled firms, and in marketing engineering and R&D costs. The reductions in total indirect costs would suggest a de-scaling effect, but since these particular indirect costs have risen, it is not possible to draw definite conclusions on future trends. Institutional factors restrain the decentralization of the stages of production, which would eventually result in de-scaling. Firms do not always find it efficient to subcontract outside the firm, and there are not yet enough small subcontracting firms to generate the distribution networks which are an important support for these firms in other countries. Thus, the existing industrial organization inhibits de-scaling through decentralization. In relation to economies of scope, it is true that firms could reduce their batch sizes with FA, but some reported having encountered limits to batch-size

FA AND THE MEXICAN ENGINEERING INDUSTRY 251

reductions. The minimum economic batch size varied according to the jigs, tooling and fixtures required, the set-up time and the price of the product. Second, in our opinion the idea that FA brings greater flexibility and product diversity, which is often argued in the literature, can be accepted in the Mexican case only with certain qualifications. Although all firms had increased product diversity, most of the ‘new’ products were only variants of other products. This contrasts with the product diversity encountered in developed countries, where product technology is highly developed. Limited product development capabilities are a constraint for firms that wish to achieve higher prices by product differentiation. With a few exceptions, our firms reported that their customers exerted pressure on their profit margins. Thus, the production of small batches may not be as profitable for them as it would be for a similar firm with better product technology. In conclusion, this study shows the presence of both economies of scope and scale. Contentions that the flexibility derived from FA have displaced economies of scale, suggesting a reduction in the optimal capacity of plants and a trend towards smaller firms, have to be treated with great caution, at least in the case of Mexico. Nevertheless, it is important to emphasize the limits of our research. First, information was not always complete or adequate for the ten firms. Second, a sample of ten firms may be too small to reach definite conclusions about a very complex process. Third, we are dealing with the initial stages of the diffusion of FA. The results would be different if we extended our analysis over a longer period for the same sample. In short, much research is needed on this issue in order to arrive at more definite conclusions. Regarding the effects of FA on investment location, it has been argued that automation in developed countries may reduce the cost advantage of low-wage developing countries, leading to the withdrawal of multinational subsidiaries to their home countries. Our findings on the introduction of FA in joint-ventures are not consistent with these views. In our observations, foreign partners have not withdrawn. Rather, they have encouraged the introduction of FA in their subsidiaries to equip them to engage in export activities. This would confirm other studies of Mexican maquiladora industries. Multinational activities in Mexico are based on more than just low wages: other aspects, such as the local infrastructure, a trained labour force and geographical proximity, also favour location in Mexico. Additionally, one local firm in the sample had been supported in acquiring additional or newer machinery by its main customer, a multinational automobile producer. This suggests that some multinational export firms find it profitable to develop their local sourcing, where the technological capabilities are present. For these reasons, FA would be expected to increase in importance in the Mexican economy, given its importance in regional and global competition. In relation to investment location, it could be speculated that the de-skilling effect of FA would lower entry barriers for local firms in developing countries to compete with multinational producers. The empirical evidence of our sampled firms shows that FA has facilitated the production of a greater diversity of goods at competitive prices. The improvements in export performance of our sampled

252 THE COUNTRY STUDIES

firms are impressive. However, most of these exports are not complete products, but rather components. In other words, our sampled firms are second- or third-tier subcontractors in the international market. This may be explained by limitations which inhibit competition with multinational firms in the creation of new products. The de-skilling effect cannot compensate for technical constraints. Without the technological capabilities in product design which are characteristic of developing countries, the advantages of flexibility of automation may play a less important role in product differentiation. Not all of our sampled firms were engaged in product design, and those that were drew on very specific technological experience and exhibited a purposeful learning strategy. These features sustained a mechanism of cumulative change which, with the introduction of CAD and CNC machinery, allowed them to produce radically new products. In short, while FA is crucial in enabling firms to start production of new lines of technologically complex products more quickly, it is hardly the only factor. The complementary capabilities are quite common in developed countries, but this is not always the case in developing countries. Thus firms may not be able to protect their profit margins by taking advantage of the flexibility of FA for product differentiation. Another factor which perhaps works in the same direction is the lack of efficient distribution networks and marketing capabilities. New and differentiated products demand efficient marketing systems able to work on a global level. In spite of recent efforts, only two local firms had efficient distribution systems. This is a traditional problem for Mexican industry: production capabilities run ahead of marketing capabilities. Although FA theoretically permits local producers to export to developed countries, it hardly enables local firms to penetrate the niches of multinational corporations. There are important problems in developing countries that need to be addressed so that the advantages of FA can be fully reaped. Although considerable changes are taking place, some are proceeding more rapidly than others, and spill-overs are limited. Competitive forces by themselves are insufficient to resolve these problems, and the role of foreign firms in technological diffusion has been overestimated. This indicates the need for a coherent industrial policy supporting efforts towards increasing technological capability in the broadest sense, including process, engineering, design and marketing capabilities, as well as stimulating the development of subcontracting and distribution networks that are well integrated in the industrial chain. Notes 1 This interest has resulted in several papers focusing on the effects on labour relations and organizational changes. See Carrillo (1993) and Covarrubias (1992). 2 The authors found 286 CNC units, including welding, metal-forming, plastic injectors, and assembly robots.

FA AND THE MEXICAN ENGINEERING INDUSTRY 253

3 Car assemblers and engine producers were excluded in order to ensure a similar sample in each of the country studies. These firms undoubtedly would have been the most intensive users of new technologies. 4 Domínguez (1993) found only 38 machining centres in 1987, in a sample several times larger than that involved in this research. 5 The firms were not able to say what share of the fixed capital stock was invested in new technologies. The number of units may be a misleading measure, where the conventional machinery is old and its accounting value may be nil. 6 Domínguez (1993) found that whereas in the early 1980s local firms owned the majority of CNC machines, by 1987 foreign firms were responsible for 63 per cent of total acquisitions in her sample of users in local industry. 7 In Domínguez’s study (1993), autopart producers reported making several changes in the organization of production as a result of the introduction of new technologies; but machinery manufacturers did not report such changes. For additional recent evidence of the complementarity of new technologies and organizational changes in other sectors, see Pozas (1993). 8 This is similar to the work organization found in modern footwear processes. See Domínguez and Brown (1992). 9 In fact, many industrial exporters have moved to rural areas in northern Mexico for precisely this reason, creating in the process ‘new industrialization zones’. 10 For a detailed analysis of this issue see Carlsson, (1984:110). 11 The firm continued to produce carburettors with conventional machinery for the retail parts market. 12 The time elapsed from the placing of orders to delivery. 13 Three of the sample firms in our study were also covered in Domínguez (1993).

9 THE IMPACT OF FLEXIBLE AUTOMATION ON SCALE AND SCOPE IN THE VENEZUELAN ENGINEERING INDUSTRY Osvaldo Alonso, Francisco Tamayo and Vanessa Cartaya

1 Introduction: the political and economic context 1.1 Macroeconomic performance and development strategy in Venezuela The Venezuelan economy is based primarily on the production and exploitation of natural resources, of which the most significant is petroleum. In 1970 Venezuela was the world’s largest exporter of oil. By 1989 the country accounted for around 1 per cent of world gas output and 2.9 per cent of world oil output, and for 2.6 per cent and 6.4 per cent of the world’s demonstrated gas and oil reserves. More than 90 per cent of Venezuelan exports accounted for by oil products and more than 65 per cent of the country’s tax income comes from oil activity (Alonso 1993).1 Other important minerals are iron ore, nickel, coal and bauxite. Some macroeconomic indicators of trends in the Venezuelan economy are provided in Table 9.1. Venezuela’s development strategy since the late 1950s has been one of import substitution financed by a large foreign exchange windfall from the export of petroleum products. The availability of foreign exchange helped to build a strong capacity for capital accumulation and reinvestment in other local production activities. Policies to transfer oil earnings consisted mainly of subsidized credits, widespread tax exemptions and state subsidies on food—policies which were intended to keep purchasing power high despite low nominal wages. The government also heavily protected the local market from foreign competition and had the capacity to impose import quotas on items at the request of a potential producer, thus reserving the domestic market for local producers. In this context private firms’ main competitive strategy was to widen their product mix to capture markets previously served by imports. The first signs of the inadequacy of this model for development appeared in 1983 and continued in the following years as oil prices fell relentlessly and production ceilings were successively imposed by OPEC.2 The ability of the Venezuelan State to be self-sustaining without collecting taxes from the private

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 255

Table 9.1 Venezuela: average annual growth of GDP, exports and imports, and average export prices for Venezuelan oil industry, 1950–90

Source: Baptista (1991)

sector and to reduce labour costs by subsidizing the basic food products was increasingly eroded. By early 1989, unwillingness to address the country’s economic maladies had led to a large foreign trade deficit, the exhaustion of international reserves, increasing domestic inflation, and an excessive fiscal deficit. Income per capita had dropped by 20 per cent in a decade, and a large part of the population was living in poverty. Regulatory failures had allowed the financial sector to become severely undercapitalized, and threatened its stability. When elected to government in 1989 the second Pérez administration introduced a programme of fiscal adjustments by restraining public expenditure and progressively adjusting the prices of public services. This programme included the elimination of food subsidies, and an increase in interest rates to stimulate the level of domestic saving. The various exchange rates being used for trade became unified and the bolivar was allowed to be traded freely. Devaluation, and increasing financing, energy and telecommunications costs, resulted in a reduction of nearly 50 per cent in real wages. At the same time, a process of opening up the economy to international trade by eliminating nontariff barriers and substantially reducing import tariffs was initiated. By 1992, average tariffs had fallen to 12 per cent, from 35 per cent in 1988 (Naim 1993). Negotiations with foreign commercial banks led to a restructuring of the public debt and succeeded in lowering interest payments as a percentage of exports; the taxation system was overhauled and an extensive privatization programme was set in motion. The cost for the population was severe, with real incomes falling markedly and social and political upheavals culminating in two attempted coups in 1992. Nevertheless, the GDP began to grow, by 6.5 per cent in 1990, 10.4 per cent in 1991, and 7.3 per cent in 1992. It is interesting that the growth in 1992 was sustained by non-oil exports (oil prices having fallen in the aftermath of the Gulf War). Private sector exports responded particularly strongly to the adjustment programme, growing by 78 per cent in 1989 and 26 per cent in 1990 (Naim 1993). This growth slowed as the domestic market began to recover and distortions due to the sudden introduction of the programme were corrected or

256 THE COUNTRY STUDIES

Table 9.2 Venezuela: structure of manufacturing industry, 1968, 1974, 1984 and 1991 (in %)

Source: OCEIa.

worked through. Non-traditional exports, which had grown strongly in 1989 and 1990, even declined slightly in 1991 and 1992. 1.2 The Venezuelan manufacturing context Venezuelan industry has been built up as recently as the last 30–40 years and is relatively young compared with other Latin American countries. Until the 1950s, Venezuela had little manufacturing activity, except for processing agricultural products and oil. Industrial diversification then proceeded rapidly, initially via the establishment of the consumer goods and engineering industries, and then through investing in large-scale natural-resource-based processing projects such as iron and steel manufacturing, aluminium smelting and petrochemicals. Industrial production increased four-fold between 1950 and 1980, but nearly stagnated during the following nine years (1980–9), when it increased by only 0. 2 per cent (Baptista 1991). The number of industrial establishments remained at around 10,000 throughout the 1980s, by the end of which Venezuela was one of the least attractive sites for foreign investment (Naim 1993). Low output growth in the 1980s was reflected also in labour productivity, which was 28 per cent lower in 1991 than in 1975 (Alonso 1992). Manufacturing accounted for 18 per cent of output in 1964,16 per cent in 1974, 20 per cent in 1984, and 21 per cent in 1994 (Baptista 1991). In 1991, manufacturing accounted for 17 per cent of gross domestic product (GDP) employing about 1,105,000 workers, or 16 per cent of the workforce (BCV 1992) The average firm size was relatively small (forty workers per establishment) and the industrial structure was fragmented, with linkages between firms being absent or very weak. Workshops with less than five workers employed 45 per cent of the total workforce. The structure of production within manufacturing also changed, as shown in Table 9.2. By 1968, the chemical industry accounted for 6 per cent of industrial production, iron and steel for 3 per cent, the engineering industry for 13 per cent, refining for 19 per cent and food for 33 per cent. By 1991, the chemical industry had grown to account for 19 per cent, the engineering industry for 15 per cent,

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 257

and iron and steel for 12 per cent of industrial production (Baptista 1991). By 1991, the production of firms employing five or more workers had reached approximately US$25,912 million, of which 27 per cent was in the engineering industry, 19 per cent in the chemical and petrochemical industries, 24 per cent in the food-processing industry, 4 per cent in the textile, garment and footwear industries, and 16 per cent in other sectors (OCEIa). Since the liberalization of foreign trade, the manufacturing trade balance has become increasingly negative, with the deficit rising to US$3,866 million in 1992. Imports were two-and-a-half times higher than exports and amounted to US $6,271 (OCEIb; OCEIc). Some 29 per cent of inputs and most of the machinery and equipment used in manufacturing were imported. Unlike the import-substitution years, there is now no explicit or implicit industrial strategy to foster technical change or the development of any particular industry. The government is allowing firms and industries to cope by themselves with changes in relative prices and the new trade regime, and many firms are facing financial crisis, forcing them to take major decisions on product mix, technical and organizational change, and costs. Nonetheless, because industrialization in Venezuela came later than in other Latin American countries, it is facing these changes with a stock of industrial machinery that is much more modern than elsewhere in the region. 1.3 The engineering industry in Venezuela By 1991, the engineering industry, which according to local usage also includes the iron and steel industry, accounted for 27 per cent of Venezuelan manufacturing production, 28 per cent of manufacturing employment, 23 per cent of the manufacturing enterprises and 42 per cent of the total fixed capital in the manufacturing sector (OCEIa). Except for pipe production and car assembly, it is rare to find engineering firms in Venezuela employing more than 500 workers. Most establishments are small or medium-sized, with an average of 57 workers per establishment. For the firms with more than 100 workers, the average is 348 workers per establishment (OCEIa). Production is highly concentrated in the central region (Caracas-MaracayValencia), in the north-eastern state of Zulia, and around Puerto Ordaz in the Guayana region. Car production is located mainly in the central region, while equipment for the oil industry is manufactured in Zulia, and steel and aluminium are produced in Guayana. Most of Venezuela’s engineering firms are at least thirty years old. Because of the import-substitution strategy, many of them have focused since their inception almost exclusively on domestic markets, relying on supplementary imports to meet excess demand. Little attempt has been made to export. As a result, by 1991, imports of final goods were almost four times higher than exports, and accounted for 52 per cent of total domestic consumption in the engineering sector.3 Engineering firms have also traditionally depended on imports for their production equipment and inputs. In 1991, 37 per cent of raw material consumed

258 THE COUNTRY STUDIES

by the Venezuelan engineering industry was imported, a figure higher than the industrial average. Hence, engineering firms are heavy users of foreign exchange, and account for a large share of the manufacturing trade deficit. The focus on a relatively limited domestic market also meant that firms had continuously to introduce new products. Perhaps to a larger extent than other manufacturing firms, engineering firms maximized the utilization of installed capacity by producing for very dissimilar markets. As a result, a highly diversified product mix has become the norm. It is common to find one firm producing parts for the oil, car and iron and steel industries, etc., a fact which sometimes makes it difficult to classify firms into specific industries. Automotive firms are the only ones dedicated exclusively to manufacturing and assembling goods for one industry. The automotive industry accounts for 4 per cent of employment, 6 per cent of production and 3 per cent of fixed capital in manufacturing. It should be noted that its shares in total production and sales of the manufacturing sector have been constant since 1984. The average firm size is 77 workers. For the firms with more than 100 workers, the average size is 295 workers per establishment. Automotive firms are even more strongly oriented to the domestic market than is the Venezuelan engineering industry as a whole. In 1991, imports exceeded exports by thirteen times, accounting for 41 per cent of domestic consumption in this sector. In the same year, 53 per cent of the industry’s raw material, and all of its production equipment, were imported. 1.4 The diffusion of flexible automation within the engineering industry Generally speaking, the organization of the Venezuelan engineering industry is in workshops with high product diversification based on conventional machine tools. The diffusion of new production technologies (mainly equipment for machining metals) was initiated by the mid-1970s, with machines imported largely from the USA and Italy (Alonso 1989,1991; Iranzo 1988). Although there are no official records of what machines were installed during the 1970s, based on previous research we can assume that between 20 and 30 complete machines were installed. These were CNCs and numerically controlled machines with wired-in, rather than integrated, circuitry. In some cases automatic lathes were used, or numerical control was added to conventional machine tools. The initial diffusion of FA in Venezuela occurred within a context where foreign currency was readily available and imports were unrestricted, which favoured the introduction of FA equipment. Currency overvaluation, fuelled by a constant and increasing flow of US dollars from oil exports, favoured the import of capital equipment at relatively low prices. Capital equipment also was exempted from import tariffs and other taxes, and could be funded at preferential exchange and interest rates. Relatively cheap and subsidized capital equipment, together with the possibility of immediately capturing a large share of the domestic market at the expense of imports, and therefore quickly recovering

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 259

investment because of the high profit margins that ensued under these conditions, made the use of imported capital goods financially attractive. In addition to these financial considerations, there was a rising demand for products manufactured with advanced equipment from end-users such as oil companies and the State steel industry, which were more concerned with achieving specific technical standards than with any price consideration. Both financial and demand considerations meant that initial decisions on purchasing CNC machine tools were based on which machine was the latest, how easy it was to purchase, and how much trust it inspired in the buyer. Such decision making based on idiosyncratic elements was maintained until the end of the 1970s (Alonso 1991). The new equipment was imported mainly from the US and Italy (Alonso 1989, 1991; Iranzo 1988). In contrast to other Latin American countries such as Brazil, Venezuela did not introduce policies to promote local production of micro-electronics-based machinery. The new machine tools were installed directly by the users, since there were no local representatives of foreign suppliers to manage the transactions or provide aftersales services, such as technical assistance and training. The commercial links with US firms, the heavy cultural influence of the latter on the local governing class, and the absence of relative or opportunity cost considerations in purchase decisions seem to have favoured the introduction of US equipment. The fact that many firms were owned by Italians explains the introduction of Italian equipment. In the 1980s there was a switch towards machines supplied by Japanese firms. The growing shortage of foreign currency caused by the world oil crisis translated into a sharp devaluation of the bolivar in 1983, and made it necessary to improve plans for the introduction of equipment, through the assessment of different alternatives. The lower cost of Japanese machines, their increasing availability in the local market and their suitability to so-called ‘third world conditions’ became important considerations. Although these machines could not always guarantee quality (or precision) levels matching those of European machines, their performance was adequate. The users were particularly appreciative of the incorporation of systems to protect against power cuts, which are quite frequent in the industrial zones, as well as the robustness of the equipment and its ease of maintenance. Data provided by Watanabe (1993a:138) show that between 1980 and 1988 172 CNC metal-cutting machines were reported to have been exported to Venezuela by US, Japanese and German suppliers, compared to 1,389 machines to Mexico, 761 to Brazil and 310 to Argentina. The greater demand for automated equipment stimulated the establishment of local representatives for foreign equipment producers, who provide technical assistance and advice regarding equipment installation, operation and maintenance. Since 1989 the economic context has changed drastically to one of macroeconomic instability, acute external and internal adjustment processes and openness to foreign competition. The 1989 economic crisis had a deep recessionary impact on the engineering industry, from which engineering firms have not yet been able to recover, even today. Table 9.3 shows the performance of the main economic variables in the industry. Consumption was reduced by 84

260 THE COUNTRY STUDIES

Table 9.3 Venezuela: the engineering and autoparts industries: evolution of the main variables, 1988–91a (in millions of bolivars, 1984=100)

Sources: OCEIa, OCEIb. Note a The engineering industry includes the following branches of the International Uniform Industrial Classification: 381, 382, 383, 384, 385.

per cent between 1988 and 1989, was virtually unchanged in 1990, and recovered 78 per cent in 1991, but in the light of preliminary data now available it appears that this recovery was short-lived and demand fell again from the second semester of 1992. The recessionary environment particularly affected firms supplying equipment for Petróleos de Venezuela (PDVSA), which was able to reduce its orders by using previously acquired parts and reserve equipment more intensively. Output and imports also fell, by 4.5 per cent between 1988 and 1989, but the industry managed to increase its exports significantly during these years, particularly in 1989. The share of domestic production in total consumption increased from 48.9 per cent in 1988 to 63.8 per cent in 1990, but fell back to 49.3 per cent in 1991. The trends in the engineering industry as a whole were observed also in the automobile sector, but perhaps more sharply. Consumption and output fell by around 29.4 per cent between 1988 and 1991. The reduction of domestic demand and the increasing competition from imported goods have created a crisis that has threatened the ability of most local firms to survive in an open market. Some Venezuelan engineering and automotive firms have since increasingly adapted themselves to competitive strategies dictated by the world market. Demands for higher quality, shorter delivery times and lower prices and costs have become common internationally. These firms responded to the new pressures with organizational innovations (total quality control systems, just-intime approaches, and so on) and the incorporation of automated equipment to reduce inventories, transfer lots and machine downtime, while guaranteeing the best possible match, in quality and quantity, to market demand. Advanced machining equipment was installed, but the abundance of low-cost unskilled labour made it unnecessary to introduce automated devices for material handling, loading and unloading, except for those included in CNC machines such as machining centres.4 For the same reason, no robots were installed in the

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 261

Table 9.4 Venezuela: CNC equipment by type, 1992

Sources: Interview data obtained from Condibieca, Maquinarias Diekman, J.Petersen, and Fim Productividad, and own calculations.

country, except for some isolated cases. Almost everywhere, machines operate as stand-alone pieces of equipment. During the last few years some CAD equipment has been introduced to be used in redesigning and adapting, and the design of new products, but these also have no direct interconnections with CNCs. All this equipment was imported. By 1992, available sources had specific details of 375 machines operating in the country, of which 79 per cent are Japanese, 7 per cent American, 4 per cent Italian, and 10 per cent German. Two importers accounted for 74 per cent of all the machines installed. As to brands, 36 per cent were Yamazaki Mazak machines, 32 per cent from Mori Seiki, and Hitachi and Deckel both had 7 per cent, with the remainder being a mix of German, Italian, Japanese and American brands. The information concerning the industries in which the machines were installed is incomplete but, of 131 machines for which details were recorded, 15 per cent were installed in firms producing for oil and steel industries, 24 per cent in autopart firms, 27 per cent in firms producing capital goods and equipment, 12 per cent in workshops providing services to third parties, and the remainder in other processes including workshops producing for the textile and plastic industries. The equipment which has been installed comprises, in the first place, CNC lathes, which account for 67 per cent of the total (see Table 9.4); milling machines and machining centres with a capacity for 80 tool-bits, (for milling, drilling and thread-making) account for another 22 per cent; and the remainder was cutting, programming and simulation equipment. After being programmed and prepared, machining centres can perform the whole machining sequence, switching tools and controlling movements as required by the programme without manual handling of the workpiece. When the operation is over, machining centres can also remove the workpiece automatically. In a few cases, electronic controls had been adapted to run conventional machines.

262 THE COUNTRY STUDIES

1.5 Transfer mechanisms for flexible automation Transfer mechanisms have varied over time. Data on the availability of FA were collected throughout the 1970s and in the early 1980s from specialized magazines and through visits to international fairs (primarily Hanover) or direct contacts, aided by the fact that some of the owners of the firms were themselves foreigners. Up to the early 1980s, all equipment was purchased directly by the users. There were no local representatives of foreign suppliers. The lack of pre-sale and aftersales service in the country created delays in installing and setting-up the equipment, and also held up their repair, sometimes for several months (Alonso 1991). This caused frequent machine shut-downs, thus reducing their utilization rate. In the mid-1980s, local representatives of foreign suppliers, mainly those distributing Yamazaki Mazak and Mori Seiki equipment together with some German CNC representatives, began to operate in Venezuela, making it possible to rely on on-the-spot advice and the technical assistance needed for equipment installation, setting-up, operation, programming, maintenance and personnel training. User firms which had technology contracts with foreign firms (basically firms in the autoparts industry, but also some firms making oil and gas valves) could rely also on the technical assistance given by the technological partner. Together with the production licence, users normally received assistance in purchasing machines, and with direct training for technicians and engineers. The firms that we visited in the course of this study had recently automated, but did not exhibit any critical shortages of the skills required to set up, operate, programme and maintain equipment. Suppliers provide the technical assistance needed and also actively participate in the training of CNC operators and programmers. Some user firms have sent technicians and engineers abroad (mainly to the USA) for training purposes, so that they, in turn, can train the rest of the personnel. Suppliers, however, ask customers to use original circuit cards, over which they hold the design copyrights, and do not allow local repairs or the use of locally made printed circuit cards. This has limited their capacity to undertake CNC machine-tool maintenance locally, although according to the sampled firms maintenance is relatively simple. 2 Technical change in the engineering industry: the casestudy evidence 2.1 The firms studied We approached eleven firms as shown in Table 9.5. Based on the project guidelines we have classified the firms in our sample in two categories: firms

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 263

Table 9.5 Venezuela: sampled firms’ size (by number of employees), product and ownership type

Source: Interview data. Note a L=Local; J=Joint venture.

producing only autoparts; and other firms in the engineering industry. We tried to split each category into firms employing less than 100 workers, between 100 and 500, and more than 500 workers. But, as noted above, the atypical development of the Venezuelan engineering industry has meant that the average firm size is relatively small so that it was not possible to include firms with more than 500 workers. The ownership of a firm’s capital was also meant to be a criterion determining inclusion in the sample, but we were not able to find an automated firm which was foreign-owned. The subsidy and tax-exemption system maintained by the state until quite recently almost exclusively favoured firms with local or largely local capital. This has discouraged the establishment of foreign firms. Instead, we have included in our sample a group of firms which have technological and capital partnerships with leading foreign firms, although 51 per cent of capital ownership was local. These firms were found only in the autoparts industry. Table 9.5 shows the composition of the initial sample. Each firm is identified by a name describing its product rather than by its real name, to ensure confidentiality. Only Auto-axles and Auto-shock produce autoparts exclusively. Oil-valves manufactures valves for the petroleum industry, Gas-parts manufactures valves and connectors for household gas installations, and the remaining firms are highly diversified. Almost all the firms are located in the central region of the country, between Caracas and Valencia. One is based in Maracaibo. Firms with mixed-capital are identified by ‘J’ in the fourth column. Two of the selected firms (Firms 9 and 10) were shut down shortly before we approached them for an in-depth interview, being unable to face the decline in demand for their products. In a third case (Firm 11) the firm had sold all its CNC equipment,

264 THE COUNTRY STUDIES

considering that the reduction of the domestic market and its inability to export immediately would make it impossible to support the financial costs of the equipment installed. Therefore eight firms were finally considered for the study, as shown in Table 9.6. No firm in the group—not even the smaller ones—produces for only one customer. The majority sell their products to other firms (assembly plants, basic steel-processing plants, oil companies). The two firms that produce only autoparts (Auto-axles and Auto-shock), maintain technical assistance contracts with a leading US firm, which enables them to place their products in the US market. The firms that produce parts for oil companies (Machine-spares, Automachining), are assessed by Petróleos de Venezuela (PDVSA), as a requirement for qualifying as an industry supplier. However, they produce for export also. We evaluated the effect of automation by collecting data on a number of variables over a period beginning in each case one year before firms adopted automation and ending in 1993. Of the sampled firms, Machine-spares was first to adopt the automation process in 1986, and the remainder introduced FA after 1988 (Table 9.6). 2.2 The production process In general, engineering is defined as the industrial transformation of metals, in processes ranging from casting and forging to stamping, cutting and welding, to heat treatment, and the assembly of components and end-products. Given the size (small or medium) of most of the firms in our sample, they are involved only in machining, assembling and finishing operations; they outsource casting and forging operations to third parties. The only exception is Gas-parts, which could also extrude and forge brass. Before automation, their equipment consisted of conventional machine tools (lathes, drills, and milling and welding machines). Material and parts’ handling, as well as machine loading and unloading, were performed manually or by using cranes or hoists. Prior to the introduction of CNCs, some firms tried to automate conventional lathes by fitting them with automatic controls. The experience gained during the modification of lathes enabled these firms to build up experience for adaptation to local circumstances, thus eventually shortening the learning period when installing CNCs. However this experience was not widely diffused because of the lack of local capabilities in making this type of adaptation. Most firms had a functional process organization: materials and parts went through a series of sections, each specialized in one phase of the work and combining machines of the same kind (sections for drilling, milling, turning, etc.). Parts, pieces and components to be worked were carried from one machine to another and from one workshop to the next. The time spent waiting for, or handling, materials, tools and machines; in collecting and interpreting technical information; and in repairing equipment, was longer than the actual working

Source: interview data. Note a Bs=bolivars.

Table 9.6 Venezuela: main economic indicators in sampled firms

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 265

266 THE COUNTRY STUDIES

time. Several products were manufactured simultaneously, with tasks being assigned to available machines. Job functions were divided between the more specialized tasks for highly qualified workers (such as turning and milling operators) and more multifunctional tasks for the less-qualified workers, although this was limited to a horizontal accumulation of similar skills such as operating several machines of similar complexity. Whatever their qualifications, however, workers were limited to the operation and preparation of their respective equipment, but were dissociated from organizational and control tasks and from efforts at understanding and improving the whole production process. Production scheduling was based on forecasts of future sales, and setting the economical lot-size in terms of the process organization, setting-up times, etc. In periods of lower demand, the next production run would be held over until the orders on hand corresponded to the minimum lot-size. Quality control was undertaken by inspectors in a separate department, applying statistical methods to samples selected at the beginning, end, and at some intermediate point of the process. Maintenance was basically corrective, i.e. after a problem had arisen, but we found a few cases where preventive maintenance was carried out. Table 9.7 shows values for the main efficiency indexes before initiating the automation process in these firms. The firms suffered low machine efficiency, long delays in delivery, high material wastage and high rejection rates, as compared to the standards then required to compete globally, based on the new industrial paradigm of no delays, defects measured in parts per million, zero inventories, and so on. It is important to note that the only firm which had no delivery delays (Gas-parts) held stocks amounting to one year’ s production. 2.3 Modernization Modernization in the sampled firms implied changes both in hardware and in organization of resources, including production organization and human resource management. As far as hardware is concerned, automation was focused mainly in production. Some 40 CNC lathes, 12 CNC machining centres and 8 CNC milling machines were adopted by our eight firms. Although it is not clear how many conventional machines were replaced by CNCs—because many of the old machines are kept for auxiliary tasks—the firms reported that on average the 2–5 machining operations previously performed by individual machines were now performed by a single CNC. The level of automation or ‘density of diffusion’, measured in terms of the proportion of production accounted for by CNCs, varied between 40 and 100 per cent of all machining operations (Table 9.8). Most machines are Japanese (Mori Seiki and Mazak), although one firm had installed two German milling machines. The CNCs were provided by local representatives of foreign suppliers, except for two firms (Body-parts and Gasparts) that had imported them direct.

Source: Interview data.

Table 9.7 Venezuela: efficiency indexes before automation in sampled firms (% and days) FA AND THE VENEZUELAN ENGINEERING INDUSTRY 267

268 THE COUNTRY STUDIES

CAD equipment had also been introduced quite recently in two of the firms although there are fewer of these than there are CNCs, and they have not been connected to CNC machines. In these firms (Auto-shock and Oil-valves), work is carried out under foreign licence. In each case, it was pointed out that designing focused on minor product changes and adaptations to local conditions. In the case of Auto-shock, CAD had helped to improve the reaction speed for the production of shock absorbers, the firm’s main product. Most firms had also introduced digital measuring equipment for quality control, since faster production to more stringent standards of precision requires timely and more exact control. No new transfer equipment had been acquired, and the movement of parts and materials, and loading and unloading, were still done manually or with some mechanical assistance. Unskilled labour was considered low-cost and abundant, so that investments in automating these functions would not be justified. Changes in production hardware have been accompanied by extensive rearrangement of plant lay-out, grouping the machining process into cells in three firms (Auto-axles, Auto-shock and Oil-valves). Each cell includes a set of machines performing all the operations needed for manufacturing a part or a family of parts, so that the firms have moved from a functional to a productbased organization. The remaining firms have introduced only minor changes to the old production lines. Just-in-time practices aimed at reducing the work in process inventory had been adopted in two firms. Adopting JIT was eased by the facts that CNCs can be more quickly reprogrammed and prepared, that tool change-over also is more rapid and that CNC workstations are closer together, making it possible to transfer smaller lots without any loss of efficiency. Programmes to reduce the time taken for operations, to eliminate unnecessary operations, to organize and simplify activities and to reduce raw materials and work-in-progress inventories have been put in place. So, too, have continuous improvement approaches, based on the intensive utilization of problem-detecting and problem-solving techniques. Total quality control practices, aimed at reducing defects and reject products, which are based on preventative measures had been introduced by 7 of the 8 firms. These new practices involved the active participation of a greater proportion of the workforce in quality control activities and the application of simple statistical techniques to guarantee the early detection and prevention of defects. Some firms pointed out that it was necessary, if automation was to be articulated in organizational changes, to change human resource policy from the specialization of workers in limited and repetitive tasks, based on a narrow range of qualifications, which had characterized the old management styles. The wider participation of technicians, machine operators and other workers in the new forms of organization was necessary and led to the redefinition of many jobs as multi-skilled, with greater involvement and autonomy. The growing need for multi-skilled workers seems to support the view in the literature that the modernization process entails a change, rather than a reduction, in the basic qualifications which are required. We are dealing with vertically-integrated work

Source: Interview data. Note a The numbers in this row correspond to the number of machining operations, each of which would have been performed by a discrete machine of the older type, that 1 CNC machine now undertakes in the firm in question.

Table 9.8 Venezuela: flexible automation and organizational improvements in sampled firms

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 269

270 THE COUNTRY STUDIES

functions, where operational and technical tasks, such as quality control, settingup equipment, problem detection and problem solving, converge. Qualifications needed to be oriented not to dealing with individual machines, but rather to all aspects related to a certain sub-process or process. Indeed, in Auto-axles and Auto-shock, modernization had been accompanied by a far-reaching reformulation of human resource management. The traditional salary scale has been abolished and a new progressive seven-level categorization of skills has been created, corresponding to the various production lines. Workers are trained to advance through the various levels, and their wages are linked to the level they have achieved rather than to the specific task they are performing in the process. Strategic training programmes for each level have been designed and courses are open to all workers. Oil-valves and Water-valves are also advancing toward multi-skilled training, with wages linked to skills and performance. In the remaining firms, human resource management change does not seem to have been so far-reaching, so that more reliance is placed on automation to improve the efficiency and flexibility of the process. Body-parts is an extreme example: it has not been possible to improve labour productivity, and this has severely affected the firm’s performance. However, if one considers all the firms studied, the general pattern is one of adopting some innovation either in production organization or human resource management to take full advantage of the flexibility of automated equipment, suggesting a strong synergy between hardware and the organization of resources. 2.4 Selecting and adopting flexible automation: reasons and procedures The purchasing decision in all the sampled firms was made by upper management and/or the managing director, in consultation with technical or production management.6 The suggestions of foreign technology partners also were taken into account, for example in the case of two autopart producers (Autoaxles and Auto-shock) and two firms (Oil-valves and Water-valves) licensed to produce household gas valves and oil valves. Since 1989, higher market requirements in terms of quality, price, prompt delivery and service, together with sharp increases in real interest rates, have forced firms to plan investments in FA more carefully. Cost-of-equipment considerations have become highly relevant. All firms in the sample pointed to quality improvement and compliance with international quality standards for foreign markets as reasons for automation. The need to comply with quality standards imposed by PDVSA and the steel industry was the major reason for automation given by the firms producing capital equipment (see Table 9.9). In the case of the autopart manufacturers, assembly plant requirements regarding compliance with management and quality standards for autopart suppliers (such as Ford’s Q–101 standard), have been the main incentive for automation. Six of the eight firms mentioned flexibility as a

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 271

Table 9.9 Venezuela: reasons for adoption of flexible automation in sampled firms

Source: Interview data.

target of automation. However, in two cases, Machine-spares and Gas-parts, higher capacity and processing speed, and higher average lot-sizes, were specific aims in adopting FA. Reductions in machine downtime and the ability to compete with other automated firms, were two other targets mentioned by several firms. As observed by the studies on other countries, the reduction of labour costs is less important than are quality and flexibility considerations. But four firms (Auto-machining, Water-valves, Body-parts and Machine-spares) cited reduced labour costs and compensating for the insufficiently qualified workforce as reasons for automation. The drop in real wages, in dollar equivalents, during the 1980s has caused many highly qualified operators (especially Colombians who came to Venezuela 10–15 years ago) to return to their home countries. This has worsened the existing shortfall of traditionally skilled labour, which is an increasingly important motive for automation, particularly by some of the smaller firms. FA reduces the volume of labour required and changes the mix of required skills. However two firms (Body-parts and Auto-machining) suffered from adverse labour relations, with severe conflicts, during and after the automation process. Finally, many firms said that installing FA was part of a gradual plan of automation (which in two of the autopart producers meant automation of conventional equipment before installing CNCs). In Machine-spares, a small workshop, simpler CNCs were installed first, in order for workers to get used to

272 THE COUNTRY STUDIES

their operation, preparation and maintenance, and then more sophisticated machining centres were installed, based on this first experience. These results confirm what has been reported elsewhere (e.g. by Watanabe 1993a) regarding the significance of international competitive conditions for the decision to adopt new technology, the importance of foreign technology partners as a change-inducing factor, and the weight given to quality and flexibility as motivations for the investment. The recent opening-up of the market and the domestic crisis also forced firms to seek other markets for their products. The fact that six firms started to export after introducing FA further corroborates the decisive importance of these technologies in achieving international competitiveness. 2.5 Problems in adopting flexible automation Wherever the supplier had provided technical assistance and user firms had trained their personnel, the introduction of FA has encountered no major technical problems, although the lack of local capacity to provide components, particularly circuit cards, was indicated as a relevant restriction. In our view, the difficulties in implementing automation are not due to technical problems associated with the automated machine, but rather because the introduction of automated machinery is not accompanied by programmes to improve production scheduling and plant lay-out, and to reduce setting-up times and total waste.7 There is in Venezuala’s engineering industry no experience in managing integral transformations, since the way production is organized has not varied over the years. The nature of the ongoing changes, taken together with the particularities of each firm, require a depth of managerial ‘know-how’ that is not common. The low educational level of the workers, the lack of technical skills, and the obsolescence of traditional qualifications also make it difficult to optimize the automation process. 3 The impact of flexible automation on scale and scope 3.1 Changes in setting-up time and batch size Setting-up times (understood as the time-period between the production of the last good product of type A and the first good product of type B, when changing from A to B), have been sharply reduced for the whole sample by 50–92 per cent. In all but one case, the proportion of total machining time devoted to setting-up has also been reduced by 33–80 per cent (Table 9.10). The greatest reductions in the ratio of total machining time to setting-up time were obtained by the four smallest firms (Machine-spares, Gas-parts, Water-valves and Automachining), the first three of which have high product diversification. In Body-

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 273

parts, where this ratio increased by 12 per cent, severe labour conflicts are reported to have caused a three-fold increase in machine downtime over the study period. Two factors accounted for these reductions. First, computerized scheduling and the concentration in a single machining or turning centre operations which previously were scattered over a range of conventional lathes, milling and drilling machines allowed for considerable time saving in setting-up machines. Second, several firms (Auto-axles, Auto-shock, Oil-valves and Water-valves) also introduced organizational changes which eliminated unnecessary operations, and simplified and reduced the time taken for the necessary settingup operations. Auto-shock provides a particularly illustrative example of reduced setting-up times and the role of automation and organizational change. Apart from introducing several CNC machine tools and CAD, which resulted in around twothirds of the whole machining process being automated, the firm embarked on a programme called ‘single minute exchange die’ (SMED) which was based on the gradual improvement of production procedures. The programme seeks to distinguish between outside exchange die (OED) activities—which can be carried out without stopping the machine—and inside exchange die (IED) activities, which require stopping the machine. Based on this, three actions are taken: (1) OED and IED activities are separated (for instance, by not procuring tools and materials while machines are stopped); (2) OED activities are converted into IED (pre-heat moulds, standardize functions by standardizing tools and parts) and; (3) improve IED, for instance by changing from sequential to parallel activities, using quick change-over devices, or eliminating adjustments by replacing millimetric measures with functional clamps (Shingo 1986).8 The results were impressive. The time for dimensioning and setting parameters, as well as for adjusting and testing, was reduced from 288 minutes to 9 minutes, or 42 per cent of total setting-up time, largely as a result of automation. The time needed to prepare materials and tools at each change-over (including cleaning the machines), dropped from 432 minutes to 51 minutes, mainly as a result of the organizational improvements. Therefore, organizational improvements accounted for 58 per cent of the reduction in setting-up time, while automation accounted for the remaining 42 per cent. The experience of Auto-shock, and to some extent also of Machine-spares and Auto-machining, suggests the importance of accompanying automation with organizational change. This point must be stressed because the bulk of Venezuelan firms lack organizational and managerial skills. Disordered lay-outs, unbalanced production flows, disordered storage and retrieval of tools, parts and material, can all diminish the potential of automated machinery to improve efficiency. Automation programmes need to be complemented by actions to

Source: Interview data.

Table 9.10 Venezuela: setting-up times and average batch sizes in sampled firms

274 THE COUNTRY STUDIES

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 275

improve process organization and programming, and to eliminate excessive waste. Another, related, conclusion to be drawn from the experience of Autoshock is that firms require a minimum degree of process organization and have to be capable of introducing incremental improvements prior to automation. It was clear from our interview and visit that Auto-shock had long experience in dealing with incremental organizational change and was operating even before the new equipment with a well-organized and balanced production process. Yet firms that start below this threshold, and which learn quickly, can make enormous gains from FA, as compared to their starting-point. Thus, the starting-point and degree of maturity of firms seem to be important determinants of performance when adopting FA, and have to be taken into account when comparing the changes in efficiency of different firms. As regards batch sizes, our results were, at first sight, rather perplexing. Since reducing setting-up times diminishes the cost of switching production from one good to another, by reducing machine downtime, it should make it possible to reduce also the production batch size or product scale, so facilitating the flow of work-in-process and product diversification. But, although most firms recognized the need to improve their flexibility, only 4 had diminished their average batch size (Auto-shock, Auto-axles, Auto-machining and Oil-valves), while another 3 had increased the average batch size (Body-parts, Machinespares and Gas-parts), and the remaining firm (Water-valves) had held it constant, (see Table 9.10). We will examine the reasons for this result in the next section. Before moving on to the next section, one final point regarding yearly product volumes which are different from batch sizes. The production of the same product over a year may be done in several batches. The four firms that show reductions in average batch size (Auto-axles, Auto-shock, Auto-machining and Oil-valves) have also significantly increased total factory production, as we will see later. Firms did not declare how much of the increase in output was due to new products; but, given an increase in total production volume higher than the observed increase in the number of products, we can safely assume that an increase in yearly individual product volume also has occurred. 3.2 Scope and economies of scope Six of the firms had increases ranging over 100–1,188 per cent in the number of different products they produced before and after automation, and similar changes in the number of different parts machined (Table 9.11). Thus, at least for some firms, one of the effects of automation has been product diversification, an extension of the traditional strategy of engineering firms of increasing their product mix to cover different segments of a small market. In the case of autopart suppliers, diversification has been influenced also by the continuous introduction of new models in the assembly plants. In four firms (Auto-axles, Auto-shock, Auto-machining and Oil-valves), the increase in the product mix coincided with a reduction in batch size, and in three

276 THE COUNTRY STUDIES

of these (Auto-axles, Auto-shock and Auto-machining) also with significant reductions in delivery time (see Table 9.20), suggesting that firms were seeking to serve their wider ‘niche’ markets more rapidly and efficiently. The latter three firms also claimed to be eliminating work-in-process inventories, although no figures were provided. It is worth stressing that while automation has led to product diversification it has not done so in all the sampled firms. In two firms (Body-parts and Machinespares), the stagnation or reduction in the product mix coincided with increases in batch size, implying a behaviour that favours greater product scale economies rather than scope economies. In Machine-spares both batch size and the number of products were sharply reduced because of the need to achieve profitable minimum batches and an optimal number of products, since this firm used to produce uneconomical small lots or individual pieces. In the case of Gas-parts, a producer of household gas valves, which are a wholly standardized mass consumption product, the number of products doubled while the batch size also rose. On closer inspection it was found that, although there had been some diversification, the bulk of production was still concentrated on a limited ‘mix’ of a few mass products. According to the firm this was because it was facing a rising demand for its main products (Gas-parts also happens to hold a monopoly of the domestic market for gas valves), so that substantial diversification was not justified. Their main competitive advantage was still obtained from lower fixed costs due to the greater volume of production of their main products, produced in very large batches. Thus, the widespread idea that automation necessarily brings flexibility through product diversification needs to be challenged. Generally, the approach of Venezuelan firms after automation has been to achieve the optimal batch size and scope compatible with available process technology and market demand. If the initial batch size was too small (as it was for Machine-spares, for instance) it was increased, and if it was too big (as with Auto-axles, Auto-shock, Automachining and Oil-valves) it was reduced. Automation provides the potential to diversify, but whether firms actually do so depends on other considerations, notably efficiency, the potential reduction in unit costs and the type of demand being faced. Hence, flexibility should be understood not so much as the capacity to change product range as the ability to provide the market with the volume and variety of designs, end-products, and the mix of parts, pieces and materials demanded, at the time required and at minimal cost. 3.3 Changes in plant and firm scale With one exception, the physical volume of production increased, by amounts ranging over 89–525 per cent (see Table 9.12). Likewise, the total number of parts machined increased by 150–800 per cent, in 7 of the 8 firms. Body-parts, where the production volume was constant in terms of the number of parts, or was reduced if measured in tonnes, seems to have been less able than the other firms to face a sharp drop in market demand because of its inability to

Source: Interview data.

Table 9.12 Venezuela: changes in production volumes (%) in sampled firms

Source: Interview data. Notes a Includes all the different generic parts which are machined. b Includes all varieties available, including products differing only in size. Source: Interview data.

Table 9.11 Venezuela: changes (as %s) in the scope of production in sampled firms

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 277

278 THE COUNTRY STUDIES

implement an integral change process and its recurrent labour conflicts. The four smaller firms (Auto-machining, Water-valves, Machine-spares and Gas-parts) registered the highest increases in production volume in terms of total machined parts and physical output. Given that these increases in output were accompanied by increases only in the number of hours worked in three firms, the effect on productivity for the sample as a whole has been very strong. 3.4 Factors affecting scale One factor influencing scale in some firms was the number of hours worked: in three firms (Auto-axles, Water-valves and Machine-spares) an increase in the number of hours worked resulted in higher capacity utilization and, hence, in an increase in output; on the other hand, in Oil-valves and Gas-parts there was no change in the number of hours worked; while in Auto-shock, Auto-machining and Body-parts the hours worked actually fell. Firms in the second and third groups stated that the main reasons for holding to or reducing the number of hours being worked were the severe competition from abroad and conflicts with workers and unions. Indeed, all firms said that the weakness of domestic demand and the growth in foreign competition were preventing them from operating in more than one or one-and-a-half shifts per day. Although automation and organizational improvement should help to open up new markets, these processes take years to come to fruition, since they require a learning process by the firm. In the short run, FA is helping some firms just to survive in a very competitive and depressed environment. The second factor is machining speed and efficiency. Because production volumes increased significantly more than did the total hours worked, some of the increase in output is due to changes in machining speed and efficiency. With one exception, the maximum machining speed increased in all firms, by between 15 per cent (Body-parts) and 454 per cent (Auto-machining) and in 7 of the 8 firms efficiency in the utilization of machines also improved (see Table 9.13). The only exception was Body-parts which, as already said, was facing an acute labour conflict and had made no significant change in management style with FA adoption. Of the firms that registered a greater increase in equipment utilization, 2 implemented a more mature organizational change (Auto-axles and Autoshock), and 2 were smaller firms (Machine-spares and Auto-machining). The very high improvement in the utilization indices after the modernization process had begun, particularly for Machine-spares and Auto-machining, is partly because they had started from a exceedingly disorganized and inefficient situation. The third factor affecting scale is the simplification and harmonization of the production flow, which helps to increase the rate of production per hour. Because the machining steps are reduced and concentrated, bringing workstations closer together, simplifying scheduling and rationalizing the production processes, workpieces can spend more time being machined, thus increasing machine productivity per unit of time and so contributing to the increase in the scale of

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 279

production. The space needed to operate a factory had also been reduced, by up to 50 per cent in Auto-axles, Auto-shock and Oil-valves, meaning that workpieces spend less time moving from one station to the next, so further increasing the time they are available for machining. 4 Changes in unit costs and profits 4.1 Changes in unit costs Before analysing the changes in unit costs it must be stressed that, together with modernization, firms in our sample have also been exposed to acute changes in factor and input prices arising from the new macroeconomic and policy environment. These changes have affected the level and structure of unit costs, and partially account for the very remarkable and diverse results that have taken place (Table 9.14). Five of the eight firms in the sample recorded reductions in unit costs ranging from 57–84 per cent, measured in current US dollars or in constant bolivars. The sharpest reductions in unit costs were registered in the four smaller firms, Automachining, Machine-spares, Gas-parts and Water-valves, although figures in current US dollars and constant bolivars are not strictly comparable. Automachining and Gas-parts recorded a reduction not only in unit costs but also in real costs. Auto-axles, Auto-shock and Body-parts recorded increases in unit costs. Table 9.15 shows that reductions in production unit costs were more significant than the changes in overheads for Oil-valves and Water-valves, while for Machine-spares overhead reductions were more important than productioncost reductions. In Auto-machining and Gas-parts there were equal reductions in production and overhead unit costs. Within production costs, the greatest savings were in labour unit costs, which fell in all but one of the firms. The reduction in labour costs is the single most important economic effect of automation and organizational change in our sample firms. Above-average reductions in input unit costs were recorded in five firms, and are another clear effect of automation. Within overhead unit costs, administration costs fell in most firms, but marketing costs increased. As far as overheads are concerned, the rise in marketing costs was the most significant impact of automation and organizational change. Capital unit costs Capital is one of the cost components which has changed most markedly during the period of study. The introduction of automated equipment costing, according to managers, 5–10 times the cost of conventional equipment meant a significant increase in investment in machinery and, as a result, in capital costs. Auto-axles, Auto-shock, Auto-machining, Oil-valves, Water-valves and Machine-spares had

Source: Interview data. Notes a On the basis of actual hours worked in the year. b Tonnage per hour if machines were operating continuously. c Hours machines are actually operating.

Table 9.13 Venezuela: utilization and efficiency indicators (as %s) in sampled firms

280 THE COUNTRY STUDIES

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 281

invested amounts of US$0.8–11.5 million in CNC equipment. As a result there has been a clear increase in capital intensity for all firms in the sample. In Autoshock, for instance, investment per worker increased 13 times. Firms stated that adapting their plant and premises to the new equipment adds 10–30 per cent to the cost of automation, particularly where the process is accompanied by a complete rearrangement of the plant lay-out. Office automation, to enable the firms to keep more up-to-date and precise records and control of production schedules, inventories, etc., has added to the cost of automation for most firms. In recent years there has been a dramatic increase in the cost of finance, which can now exceed 30 per cent in US dollars. Most firms had obtained loans to finance equipment purchases, and now have to face hefty interest payments. To deal with this problem, some firms were resorting to writing-off the cost of purchased equipment over the longest possible period, thus reducing the shortterm impact of depreciation on costs and avoiding drastic reductions in profitability.9 As was mentioned earlier, there have been significant reductions in the physical space used for production, but this has not necessarily translated into direct reductions in capital unit costs, because firms continue to manufacture in their old plants. The main benefit of space reductions has been to help increase machine productivity and reduce lead times. Labour unit cost The share of labour in total costs decreased in all the firms (Table 9.16). One factor accounting for this result has been a spectacular drop in real wages. In firms that provided figures, reductions amounted to 25–30 per cent for whitecollar workers, particularly for the technicians who programme, prepare and maintain CNCs, and 55–60 per cent for blue-collar workers. Such acute drops in real wages took place across the Venezuelan economy. A second factor explaining the lower share of labour in total unit costs was reduction in overall employment levels, although this did not take place in all firms. Auto-axles, Auto-shock, Auto-machining and Body-parts reduced their workforces by amounts ranging between 20 and 47 per cent. In Oil-valves, Water-valves, Machine-spares and Gas-parts, where employment increased, productivity gains were far lower than in the other firms, which suggests that there may be some room for employment reduction in these firms. The reductions in labour unit cost would have been larger had it not been for an increase in employment for engineers, technical personnel and skilled workers, and an increase also in labour training costs. In six of our sampled firms the proportions of qualified personnel rose, by 15–130 per cent, while the costs of the training required for FA operation grew by a factor of approximately four in all the firms that provided figures.

Source: Interviews.

Table 9.15 Venezuela: changes in unit cost (%) by source in sampled firms

Source: Interview data.

Table 9.14 Venezuela: changes in unit cost in sampled firms

282 THE COUNTRY STUDIES

Source: Interview data.

Table 9.17 Venezuela: changes (in % and number of days) in reject and waste rates and raw material inventories in sampled firms

Source: Interview data.

Table 9.16 Venezuela: share (%) of capital and labour in total unit costs in sampled firms

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 283

284 THE COUNTRY STUDIES

Input costs Reductions in input unit cost were the result of both lower prices and better utilization of raw materials. Sampled firms stated that they were benefiting from a decrease in the prices of casting and forging, in particular, arising from the more competitive environment. Reductions in tariff and non-tariff barriers also implied a wider availability of less expensive high-quality speciality steels which were increasingly preferred by our interviewed firms. The increasing use of foreign supplies was also helping to keep raw material costs down. Another source of input cost savings was the lower final product rejection rates and higher raw-material utilization rates made possible by the new equipment and quality control practices. All firms had been able to reduce their rejection rates, by 63–98 per cent (Table 9.17). In four firms (where this indicator is calculated), total waste had been reduced by 60–90 per cent, and it can be improved still further: one firm admitted to 15 per cent waste in material. Those which had achieved large improvements (60 per cent or more) had already had waste levels of under 10 per cent before adopting FA, while those which achieved little or no improvement had wastage rates of 10–65 per cent before automation. One reason for this pattern may be indicated by the extreme case, Water-valves, which had a wastage level of 65 per cent before automation, and did not improve this at all. This firm is a subcontractor whose raw materials are provided by the client. It has no incentive to reduce waste because any additional costs in this respect are charged directly to the customer. In addition to lower prices and better utilization there were dramatic reductions also in raw-material inventories, by up to 92 per cent in the case of Machine-spares, thereby reducing the costs of financing working capital. Autoshock, for example, had raw-material inventories for 150 days of production in 1986, which was equal to 14 per cent of the total value of its annual sales. In 1993, the firm reduced its inventories to 23 days of production, or 2.3 per cent of the total value of sales. If the firm had kept to its 1986 inventory level, with a market interest rate of 30 per cent on loans in dollars, by 1993 its production costs would have been 5 per cent higher from this factor alone. Higher credit costs were an added incentive to reduce working capital costs. Body-parts stated that one reason for keeping high raw-material inventories was that many of them were imported. Because of the lengthy bureaucratic procedures necessary to obtain foreign exchange and to clear customs, it was advisable to hold additional stocks to avoid potential delays. Indeed, delays quite often led to long lead times in the supply of raw materials to the factory, and were a constant headache to managers.10 Maintenance and energy costs Three of the four firms which provided data on maintenance costs indicated that this cost had not changed as a result of automation. The firms indicated (Table 9.18) that the use of CNCs (without using automation for material handling and the loading and unloading of machines) does not increase maintenance costs,

Source: Interview data.

Table 9.18 Venezuela: changes in energy, maintenance and repair components of unit cost in sampled firms (as % shares) FA AND THE VENEZUELAN ENGINEERING INDUSTRY 285

286 THE COUNTRY STUDIES

since technical assistance is within reach and does not lead to additional costs. This is probably in part because the fall in wages has driven down the local cost of technical services. As to energy, potential (and achievable) consumption savings from the use of CNCs have eluded the firms because of the marked increase in the cost of energy. As part of the government’s economic adjustment programme, the cost per kW/ hour increased by 62.5 per cent in real terms, between 1988 and 1993.11 As a result, four firms reported increases of 25–300 per cent in the proportion of total costs that were due to energy costs. Only Body-parts was able to keep energy costs constant, as a proportion of total costs. R&D, marketing and administrative costs The impact of automation on R&D activities has been uncertain. Three firms did not, and still do not, undertake any sort of R&D activity. Another three had focused mainly on making adaptations and design alterations to plans given them by their technological partners; and these firms continue to do so today. The other companies have always based their R&D activities on copying or redesigning parts given to them by customers. Of the five firms that gave R&D figures, the share of R&D expenditure in total unit cost fell in Auto-shock, rose in Bodyparts and Water-valves and remained the same in the others, suggesting no pattern. Furthermore, the changes in R&D costs which did occur were insignificant (see Table 9.19). It was expected, at the proposal stage of this project, that research and development expenditure would increase because of the more advanced knowledge required to develop new products and to use the new equipment. Handling the more sophisticated information would demand additional personnel and materials, increasing R&D fixed costs and thus stimulating further increases in production scale to cover these costs (Alcorta 1992a). Our results, however, suggest that adaptive copying or minor innovation-based product-process development strategies may continue to be relevant with FA, and that significant increases in R&D expenditure are not a necessary outcome of the adoption of FA. Marketing cost was the only overhead or indirect cost that increased, as a proportion of total unit cost, in each firm that had marketing activities in 1993. The need for foreign market research to increase exports, and the need to train personnel to market and sell more sophisticated products, have increased the importance of this item. Gas-parts not only increased expenditure on marketing but also created a marketing department as part of its modernization process. Oilvalves’ fivefold increase in marketing costs was mainly the result of searching for new clients abroad, particularly in Colombia, Trinidad and Curacao. Today the company exports 15 per cent of its total output. Finally, the share of administrative costs in total costs remained constant or declined, except at Auto-shock. One factor that may account for this is employment. In Auto-machining, Machine-spares and Gas-parts administrative employment fell significantly as a share of total employment. A reduction in administrative employment may have occurred also in Auto-axles and Oil-valves,

Source: Interview data.

Table 9.19 Venezuela: R&D, marketing and administration shares (%) in total unit costs in sampled firms

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 287

288 THE COUNTRY STUDIES

for which the share of overhead or indirect employment in total employment increased; but, judging from the changes in marketing expenditure, there may have been also a shift within overhead employment from administrative work to marketing. Other factors that may account for this result are: an increase in the productivity of administrative staff, as all firms said they had acquired computers for accountants and inventory purposes; higher productivity might have more than outweighed the effect of a lower than average fall in wages, particularly for lower-level administrative staff. More definitive conclusions in this respect would require more detailed information. Auto-shock’s experience reveals the potential cost implications of the advanced administrative modernization required to exploit to the full the new production equipment. Unlike the other firms, Auto-shock tackled modernization as a comprehensive production and administration process. The administrative objective was to have the most complete and rapid information possible about what was going on in the plant, particularly on production rates and inventory levels, so that cost accounting could be precise, overall control effective, and decisions appropriate. Although the company stopped short of full integration of administration and production, an effective information processing and communication system which eliminated most memoranda and reports was introduced. This system, however, involved major investments in computers and specialized software, which was the major cause of the increase in the share of administrative costs in total costs. 4.2 Changes in profits and delivery times Total profits have increased for 5 of the 8 firms, and have decreased but remained positive in the other 3, Auto-machining, Body-parts and Gas-parts (Table 9.20). Given the large reductions in price and unit profit, this must be put down largely to the drastic increase in output, which would not have been possible without FA and organizational change.12 With the exception of Bodyparts, whose output actually fell, total profitability of the other firms is increasingly dependent on the sheer volume of their output. Furthermore, given that there has been an overall fall in the output of the engineering and automobile industries, the firms in our sample must be accounting for an increasing share of total domestic production, although they were losing some market share to imports. It is clear that firms are experiencing substantial declines in profit for most of their products, ranging from 20 to 88 per cent. Intense domestic and foreign competition are increasingly driving the prices of goods down to the price of the least-cost producer internationally. Venezuelan firms that cannot keep up with the cost and price reductions, because of old technologies or their inability to use FA and to introduce organizational change effectively, will be forced to close. Indeed, a major restructuring process is already underway and should continue in the coming years.

Source: Interview data.

Table 9.20 Venezuela: prices, profits and delivery times in sampled firms (%)

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 289

290 THE COUNTRY STUDIES

While price and cost are clearly important determinants of profitability, so are non-price factors. Flexible automation, by lowering setting-up times, increasing machine speed and efficiency and shortening, simplifying and organizing the production process, also results in significant reductions in delivery times. Shorter delivery time is a key competitive advantage, particularly for firms producing for the oil industry in which a ready supply of spare parts is crucial, but increasingly for other industries as JIT is more widely adopted. Quick delivery can make the difference between obtaining a contract or not, thus affecting profitability. 5 Conclusion: flexible automation, scale and location of production in Venezuela Determinants of the study In the first place, the changes found in a particular firm depend considerably on the phase of transformation that firm was in when the study was carried out and on the moment at which FA was introduced. The impact of FA will be less visible in a firm which had already improved its flexibility and efficiency, and learned to change, before introducing CNCs. It would be helpful also, for future studies, to establish enterprise typologies defined in terms of extent and depth of modernization, combining the two dimensions of change. Fleury (1993) offers a useful classification of firm strategies in dealing with modernization —‘systemic’, ‘re-equipping’ and ‘conservative’. In the second place, it is very hard to differentiate between the effects of automation and those of organizational rationalization. Research to date, including this study, has not been able to show clearly the separate effects of the two components on productivity, quality and efficiency, for the simple reason that both aspects are part of the same logic of change. Thus a combined analysis of the whole process of change would be more relevant than an attempt to discriminate between automation and re-organization effects—a goal without theoretical or practical relevance since it ignores the integral logic of the process of change. Only the combination of both aspects of modernization in an integrated process allows firms to take full advantage of modernization (Bessant 1991; Kaplinsky 1987; Pérez 1986; Watanabe 1993a). In the third place, the study was carried out at a time when a fiscal and monetary adjustment programme had increased interest rates and energy costs significantly, while the reduction of barriers to foreign trade had led to the abrupt entry of competing imports. Financial strictures and foreign competition also had the effect of reducing aggregate demand, curbing the prices of the firms’ products while at the same time increasing production costs, thus creating adverse conditions for modernization. Indeed, despite attempts to introduce FA and to modernize, two firms in our original sample were forced to close down, while another had to sell all of its new equipment to survive. As noted by Fleury

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 291

(1993) in other similar contexts, an unfavourable economic environment makes it very difficult to profit from capital investments, suggesting that a sound and stable economy is a prerequisite for any industrial policy. One additional related remark is that the external and internal economic contexts, the introduction of automated processes and organizational rationalization, and the different initial situations of the firms in size and complexity, combine in such a way that they can form a complicated and often confusing picture that makes it difficult to credit automation, on its own, with the observed effects on scope and economies of scale, or on efficiency indicators. Integral change One of the key conclusions of the study is that automation and organizational change can be introduced independently of each other, as was the case in Bodyparts. Automation clearly could be adopted without operating just-in-time, and cellular organization could be implemented with conventional equipment. But the fact that, except for Body-parts, the firms were trying to varying degrees to combine automation and organization suggests that modernization presupposes the efficient articulation of technical change (basically process automation in machining, in our case) and organizational innovations based on the total quality approach (Kaplinsky 1987; Watanabe 1993a). It would appear that the goals of waste reduction (idle machines, inventories, delays and stoppages, time lost in the removal of workpieces and tools), curbing costs (higher worker productivity) and quicker customer response can be better pursued by linking new CNC equipment to the application of total quality control, the treatment of each production station as an ‘internal customer’, and experimentation with just-intime production. New approaches to labour, such as greater autonomy and better communication with assembly-line workers, and multi-skilling training, are further means of integrating human resource management with the modernization process and achieving even better results. If automation continues to be seen as separate from organizational change, we risk ending up with ‘waste automation’. Isolated automation is simply a luxury for organizations with scant resources and plenty of waste. Scope and product scale The study shows, first, that some firms diversified their product range and reduced their batch size considerably. In some cases the new products are of higher quality and firms have been able to capture new markets as a result. But doing so was not easy. Reaching the new markets has taken several years of learning and experimentation and has required a complete transformation of management practices, including a complete revision of marketing techniques and a comprehensive change in organization and working methods to take advantage of the potential of FA, initiatives which not all firms were prepared to venture.

292 THE COUNTRY STUDIES

It should be noted, also, that reductions in batch size or product scale are not the same as reductions in the annual scale of production of each individual good. Actually, the four firms which have reduced their average batch size seem also to have increased their total annual production tonnage significantly. Firms did not state how much of their production increases were due to new products, but where the increase in production volume is higher than the observed increase in the product range, we can safely assume that an increase in scale has occurred, together with the diversification process. Hence, product scale in this sense has increased, rather than decreased. More importantly than this however, and this is the second main conclusion regarding scope and product scale, is that in some firms batch sizes have not fallen but increased. Indeed, some of our sample firms began from an overdiversified situation and sought automation not for reasons of product diversification but to create product scale advantages through increased machining speed and reduced machine downtime and waste. Overtly high diversification has been a key feature of many small and medium-sized firms in Venezuela, bringing with it bottlenecks, product rejections, material and parts’ waste, and extreme under-utilization of machine capacity. A reduction in the product ‘mix’ clearly permitted one of the firms in the sample to increase the size of machining lots to levels compatible with the efficient use of CNCs. Thus, for a small country like Venezuela, what really needs to be achieved is not so much scope or product diversification as ‘flexibility’ —understood as responsiveness to change with maximum speed and minimal cost —independent of whether it involves an increase or decrease in product scope. Plant and firm scale and unit costs One clear conclusion of the study is that, together with an increase in product range and a reduction in batch size, the adoption of FA means an increase in that firm’s total production scale. The increase in physical product in the sampled firms was due not so much to the hours worked as to better capacity utilization and production organization, reductions in setting-up time, and higher machining speeds. In addition, modernization has reduced unit costs while changing the structure of production costs. Capital costs have increased as a proportion of total unit costs, due to the higher cost of CNCs and higher interest rates. Labour costs have decreased sharply due to lower labour requirements and a fall in real wages in the whole economy, which outweighed the higher nominal wages in jobs related to automation. Energy costs, which usually fall when FA is introduced, rose as a proportion of total costs because of an increase in electricity prices under the government’s fiscal adjustment programme. Finally, marketing costs increased because of the need to increase the qualifications of salesmen, to market more sophisticated products and to deal with the more demanding foreign markets.

FA AND THE VENEZUELAN ENGINEERING INDUSTRY 293

Entry barriers The study has two implications for the question of whether introducing FA reduces entry barriers. First, the higher cost of equipment, accentuated by the devaluation of the bolivar by 70 per cent, and interest rates of 20–30 per cent, in dollars, have become an important barrier to entry. In addition, low and decreasing labour costs seem to provide little incentive for investment in automation. There are already signs that after a steady period of growth in automation in the engineering industry, investment plans for the importation and installation of additional machines are being shelved. Second, the lack of the managerial ‘know-how’ needed to initiate and maintain continuous improvement programmes is perhaps a more important barrier to the adoption of FA. Management education and support by consultants with specialist knowledge of production process reorganization will be imperative if the modernization process is to go further. Even today there is a serious lack of process control, and plant disorganization and waste are rife. There is also a need for continuous training programmes to ensure a sustained supply of multi-skilled and knowledgeable workers, and it is not clear that management is willing to undertake this. There are numerous managers who continue to adhere to traditional approaches to human resource management. Increasingly, however, workers will need to have a deep understanding of the production process in which they are involved; and they will have to be capable of combining their specific technical functions with tasks such as early diagnosis of failures, quality assurance, and problem detection and solution. The situation of having each machine assigned to a particular individual, with a small group of technicians making technical decisions and solving all kinds of problems in the plant, is no longer appropriate. This in turn will also require a fundamental transformation of training systems within and outside of the firm. To sum up, FA is not a solution to scale problems in the smaller developing countries. Firms need to increase their overall scale to reduce costs and prices, if they are to be competitive. But FA provides the opportunity of capturing niche markets in which knowledge, design, prompt delivery, service, quality and nearness to the customer determine whether a sale will proceed: perhaps a sale for which higher prices can be charged. Therefore, FA creates also the possibility of catering simultaneously for mass and niche markets, a possibility that was not there before. Notes 1 The Venezuelan petroleum industry was nationalized in the mid-1970s. Petróleos de Venezuela, a state-owned oil company, has since been in charge of exploration, refining and marketing of oil, petrochemicals, and natural gas including the export of oil and derivatives. 2 The international price of Venezuelan oil dropped from US$38.2 per barrel in 1981 to $12.0 per barrel in March 1994. Per capita earnings from exports dropped from $1,410 in 1981 to $670 in 1994 (BCV).

294 THE COUNTRY STUDIES

3 For further information about the characteristics of Venezuelan firms, see Pirela et al. (1988). 4 The trend towards the installation of stand-alone computer-numerically-controlled (CNCs) machine-tools has been observed also in other Latin American countries (Fleury 1993; Domínguez, 1993). 5 Some firms provided data in dollars, and some in bolivars. The fact that there were multiple exchange rates in Venezuela until 1989 makes it inadvisable to translate bolivars into dollars, so the data for each firm is presented in the denomination used by the firm, although bolivar amounts have been adjusted for inflation. For reference, however, between 1986 and 1988 the official exchange rates were US $1=4.30, 6.00, 7.50 and 14.50 bolivars while the street exchange rates were US $1=21.50 bolivars (Bs) in 1986, 26.90 in 1987 and 32.00 in 1988. In the following year the exchange rates were freed so that US$1=39.10 Bs in 1989, 47.20 Bs in 1990, 56.93 Bs in 1991, 68.40 Bs in 1992, and 92.10 Bs in 1993. Inflation, which averaged around 13 per cent in the decade 1970–1980, reached 28 per cent in 1987 and 35 per cent in 1988, despite price controls, peaked at 80 per cent in 1989 when controls were removed, falling to 36.5 per cent in 1990, and stabilizing at about 31 per cent in 1991 and 1992. 6 In fact, in small firms, the roles of owner, manager and technician are usually not separated, and purchase decisions are made by those with technical knowledge. 7 In a study of the development of the electronics industry in Brazil and its relationship with users in the chemical firms, Quadros Carvalho (1992) highlights also as the main impediment to fully benefiting from the potential of FA the inability of these firms to introduce organizational change. 8 This procedure was initiated by Toyota in Japan. It was applied initially to reduce the set-up time of a pressing machine, but is now applied to similar machinery and equipment. 9 The firms reported a capital cost composed basically of annual depreciation and the cost of credit obtained to buy automated equipment. Although it is not possible to break down the total capital cost into these two items, the financing cost is estimated to account for two-thirds of the capital costs. 10 It should be noted that this problem started to abate in 1988, when the state stopped giving US dollars at preferential rates to importers. With the opening-up of the economy and the adjustment of market behaviour, input suppliers started to offer smaller lots, making it possible to keep smaller inventories. 11 Data provided by CAVEINEL (the Venezuelan Electrical Industry Association). 12 This obviously does not mean that every firm to adopt FA necessarily does well, as illustrated by firm 11, which was excluded from the sample because it had sold its CNC equipment.

10 THE IMPACT OF FLEXIBLE AUTOMATION ON SCALE AND SCOPE IN THE INDIAN ENGINEERING INDUSTRY Ghayur Alam

1 Introduction: industrial development in India 1.1 India’s industrial policies and development India’s industry and trade policies have undergone major changes during the last decade.1 For almost three decades after Independence India’s economic policies in general, and industry and trade policies in particular, were strongly influenced by a philosophy of self-reliance. There was widespread government intervention to guide investment to priority areas, protect local industries from imports and strengthen indigenous technological capabilities. The policy instruments used to achieve these objectives included: • Industrial licensing to control entry and expansion into various industrial activities, with the aim of directing investment into priority areas, establishing the public sector’s dominance in industry and reducing the presence of large private business groups. • Import controls to conserve foreign exchange and protect local infant industries. The controls severely restricted the import of capital goods, components and raw materials. • Restrictions on imports of foreign technology and capital through foreign collaborations and joint-ventures, to encourage and protect indigenous technology development. The policy was aimed also at bringing down the cost of imported technology and reducing restrictions placed by foreign owners on the use of imported technology (for example to sub-license or to export). Foreign investment also was discouraged and, with a few exceptions, foreign equity of more than 40 per cent was not allowed. • Promotion of the development of indigenous technological capabilities, largely by building an extensive scientific and technological infrastructure in the state sector. The role of the Council of Scientific and Industrial Research (CSIR), which was set up by the colonial government in 1942, was extended.

296 THE COUNTRY STUDIES

By 1963 the CSIR controlled about 30 research institutes, and its R&D expenditure increased from 28 million rupees (US$6 million) in 1958–9 to 136 million rupees (US$19 million) in 1971–2. India’s total R&D expenditure as a proportion of its GNP also underwent a slow but steady increase, from 0. 18 per cent in 1958–9 to 0.38 per cent in 1970–1. • Other initiatives included the establishment, in 1953, of the National Research and Development Corporation, to facilitate the transfer and commercial exploitation of know-how developed by various (mostly government) research organizations, and modifications to the patent laws to protect and encourage indigenous R&D. As a result of these policies, India succeeded in building a diverse industrial and technological infrastructure in a comparatively short time. An impressive number of core industries were set up, and dependence on imported capital goods was considerably reduced. Between 1950 and 1965 industry as a whole grew by 7.1 per cent, while the engineering industry grew by 14.9 per cent (Confederation of Engineering Industries 1987). Furthermore, with the establishment of domestic production facilities for a wide range of products and the introduction of extensive import restrictions, the relative importance of all kinds of imports in the economy declined sharply. This impressive performance began to decline from the mid-1960s. Industrial growth fell sharply, to 3.6 per cent per annum in 1965–75, and the engineering industry also experienced significantly lower growth during this period. Nevertheless, in the early 1970s, when the full impact of import substitution policies was evident, imports were less than 6 per cent of GNP. In 1973 imports fell to just 5.5 per cent of GDP, a figure lower than for any other country except the USSR (Wolf 1982). From the early 1970s there was mounting criticism of industrial inefficiency, much of it directed at India’s industry (including technology) and trade policies. These policies were held responsible for the perceived widespread inefficiencies and the sluggish growth of industry (see Ahluwalia 1985; Bhagwati and Srinivasan 1975; Desai 1981; and Lall 1982). The controls were said to be too complex and cumbersome, causing long delays, uncertainties and increased costs. The policies were also considered to be excessively protective. They discouraged competition (both domestic and from imports), leading to the establishment of inefficient production facilities. The restrictions on entry and expansion, and the controls on the import of technology and capital equipment, led to technological stagnation. In the absence of competitive pressure, industry had little incentive to improve its technological base, so it continued to produce outdated designs using outdated production technologies. As a result of the increased criticism, government policies have been considerably liberalized in recent years, beginning in the 1980s but accelerating during the 1990s. The changes in policy are aimed primarily at improving the efficiency and international competitiveness of industry by exposing it to more external and domestic competitive pressures, and by providing greater access to

FA AND THE INDIAN ENGINEERING INDUSTRY 297

technology, capital equipment and other inputs. Also there has been a greater emphasis on exports. The measures introduced to achieve these objectives include: • The relaxation of industrial licensing to remove constraints on entry and expansion. Many industries were exempted from industrial licensing, and firms were allowed to produce a number of related products without obtaining separate licences. By 1992, licensing had been abolished for all industries unless there were strong social, security or environmental arguments for continued control. • Controls on technology imports were relaxed. The number of industries in which technology imports were allowed was considerably increased and the regulations on technology payments were liberalized. For example, technology imports in high-priority areas, involving lump sum payments of up to 10 million rupees ($US0.41 million) and 5 per cent royalties were automatically permitted. • Foreign equity of up to 51 per cent was permitted in areas which required large investment and/or high technology. • Controls on the import of capital goods, including machine tools, were greatly relaxed. Almost all capital goods and raw materials can be imported, and the import duties on capital goods have been reduced to 20–40 per cent. Special concessions are available for the import of capital goods for export-oriented projects. As a result of liberalization, the number of firms operating in most industries increased sharply during the 1980s. The number of licences, for example, increased from 465 in 1975–6 to 2,454 during 1985–6. The 1980s saw also a revival of industrial growth, from 4.4 per cent during the 1970s to 7.8 per cent in the 1980s. The growth rate of manufacturing industries increased from 4.0 per cent in the 1970s to 7.5 per cent in the 1980s (Government of India (a)). But in 1991–2 and 1992–3, industrial growth was 0 per cent and 1.8 per cent, respectively. This has been attributed to a fall in effective demand due to a slowdown in the overall economic growth, a reduction in public investment in real terms, political disturbances, a shortage of imported inputs and the high cost of credit (Government of India (a)). The performance of the manufacturing sector was even worse: it experienced a negative growth of −1.5 per cent in 1991–2 and grew by only 1.5 per cent during 1992–3 (Government of India (a)). After growing by 17 per cent in dollar terms from mid-1986 to mid-1990, overall exports also experienced a sharp decline in the 1990s, to 9.1 per cent growth in 1990–1 and a decline of 1.55 per cent in 1991–2 (Government of India (a)). Engineering exports increased by 9.4 per cent in 1990–1 and by only 3.4 per cent in 1991–2. Low growth during the 1990s has been reflected in domestic demand. While a large number of new firms have begun to produce comparatively new products — some by assembling CKDs and SKDs—the demand has often not increased

298 THE COUNTRY STUDIES

Table 10.1 India: machine-tool industry, 1980–93

Source: IMTMA. Note e=estimate.

accordingly (e.g., for light commercial vehicles). As a result, plans to set up new production capacities have been delayed or cancelled. Industry’s reliance on foreign technology has increased, while its R&D and designing activities continue to be extremely limited. 1.2 India’s machine-tool industry Indian policy makers and planners placed particular importance on the development of the indigenous machine-tool industry. The first major move in this direction was made in the 1950s, when Hindustan Machine Tools Ltd was established. The production of machine tools has grown steadily since then, and India’s dependence on imported machine tools has declined sharply. In 1955 a little more than 10 per cent of the machine tools used by Indian industries were locally made, but by the 1980s the proportion had increased to about 70 per cent. In 1980 machine-tool production was valued at 1,859 million rupees (US$235 million). This accounted for 69 per cent of consumption (Table 10.1). Table 10.2 shows that production has been growing steadily, although the growth rate is declining. Estimates for 1993 (Table 10.1) show a marked decline in production and a collapse in export sales, accompanied by a surge in imports. Clearly, the industry has suffered from the stagnation of demand and the limitations of locally produced machines in comparison to imports. 1.3 Technology diffusion Technology imports and foreign collaborations increased sharply in the postliberalization period (see Tables 10.1 and 10.3). Foreign technology payments

FA AND THE INDIAN ENGINEERING INDUSTRY 299

Table 10.2 India: machine-tool industry performance in real terms, 1980, 1984, 1988 and 1992 (1980=100%)

Source: Figures from Table 10.1, above, corrected with separate price indexes for national production (all sectors), imports and exports, from Government of India (a). Table 10.3 India: foreign collaborations agreements, 1980–94

Source: Government of India (b).

also increased, from 0.15 per cent of production in 1981–2 to 0.55 per cent during 1986–7. The evidence suggests also that firms paid more for their technology imports during the second half of the 1980s. The average approved technology payment per collaboration in 1981 was only US$159,000. This increased to $467,000 in 1986 and $489,000 in 1989.2 In the 1980s there was a steady and significant increase in the proportion of these collaborations which involved foreign investment, from less than 14 per cent during the 1970s to 30 per cent in 1988. Although they spent more on technology imports, the R&D intensity of industrial firms increased only marginally in the 1980s, from a mere 0.93 per cent of value added in 1980–1 to 1.17 per cent in 1988–89. In the case of engineering industries, R&D expenditures grew from 0.44 per cent of production during 1980–1 to 0.53 per cent during 1990–1.

300 THE COUNTRY STUDIES

The production and use of CNC machines in India began in the late 1970s, when Hindustan Machine Tools began to produce CNC lathes. There are now more than 10 major producers of these machines. Most of the machines and control units are produced in collaboration with leading world producers. However, compared to developed countries and many newly industrialized countries (NICs), the use of CNC machines in India is limited. Although there is no official information, we have used data from two sources to estimate the total number of CNC machines in India, at the end of the 1993–4 financial year, at 5, 000.3 This can be compared with the Republic of Korea, which produced 3,675 CNC machines in 1991 alone, or with China and Brazil, which produced 1,360 and 1,052 CNCs, respectively, in 1989 (Watanabe 1993a: 5). Not only is the total number of CNC machines used in India small, the number of machines installed per year rose in 1990–1 and 1991–2, but fell by 33 per cent in 1992–3. The low rate of diffusion of CNC machines is due largely to the slow-down in industrial growth during the 1990s. 2 Diffusion of flexible automation in the Indian engineering industry 2.1 The sampled firms Of the 11 firms in the sample, 7 produce autoparts, 3 produce hydraulic valves, and the other is a producer of hydraulic pumps. By the criterion of number of employees (Table 10.4), 3 firms are large, 3 are medium-sized, and the remaining 5 are small.4 Four firms—three large and one medium-sized—have foreign equity, but this is less than 25 per cent in each case. The remaining firms are wholly Indian owned. Six firms have acquired technology through foreign collaborations. Another four firms carry out machining of components for a track manufacturer, from which they have acquired technology, mainly designs. The remaining firm, a medium-sized producer of industrial valves, relies on engineering consultants and in-house capabilities for technological inputs. The main characteristics of the eight autopart producers are described below, beginning with the largest firms, followed by the capital equipment producers. Tractors-India This firm belongs to a large engineering group and is one of the leading tractor producers in India. The firm was set up as a joint-venture with Ford in 1969, beginning with a single tractor model. It now produces three models and has a large R&D centre, which concentrates on indigenization efforts. As a result of these efforts, the imported content of production is less than 1 per cent. This firm was among the first automotive producers in India to use CNC machines: the first machine was introduced in 1982.

FA AND THE INDIAN ENGINEERING INDUSTRY 301

Steer-link This is one the oldest autopart producers in India. It was set up in 1961 for the production of steering linkages and tubes for trucks. Since then it has expanded its product range and is now a major supplier of these components to most automobile producers in India. Its market share of these components varies between 40 and 60 per cent. The firm developed appreciable R&D and design capabilities, which were used mainly for the indigenization of production, but are increasingly being used also to modify designs in response to the changing requirements of vehicle producers. Its introduction of new technology began during the second half of the 1980s. Maruti-steer This firm is a comparative newcomer. It was established in 1985 to supply steering components to India’s largest car producer, Maruti. The firm is in technical and financial collaboration with Koyo Seiko Ltd of Japan, which supplies steering components to Maruti’s collaborator (Suzuki) in Japan. The production methods employed by the firm are closely influenced by its Japanese collaborator and Maruti-Suzuki. It is keen to adopt both new production technology and organizational and quality control techniques. The introduction of new technology began two years after the firm was set up. Telco-tool, Telco-gears, Telco-diffs and Telco-shafts These four firms are similar in several respects: they are located in an industrial estate near TELCO and were set up during the early 1990s to produce components for TELCO. The firms still depend totally on TELCO for the market for their products: gear boxes, axles and ball bearings (Telco-tool); gear-box components (Telco-gears); differential cases and crown wheels (Telco-diffs); and drive shafts and gear boxes (Telco-shafts). In all these firms, production began with CNC machines. In fact, the firms would not be acceptable to TELCO as component suppliers without CNC machines. Hydro-valve Hydro-valve employs about 1,000 people. With about 50 per cent of the market, it is the largest producer of hydraulic valves in India. The firm belongs to one of India’s leading engineering groups. It produces both standard and custom-built valves and is involved in a number of technical collaborations. It was among the first firms in India to use CNC machines: the first machine was installed in 1978. Pumps-India The sole producer of pumps in the sample, Pumps-India, is a medium-sized firm and belongs to a prominent engineering group with 3,831 employees. Pumps-

302 THE COUNTRY STUDIES

India has 266 employees. It is one of India’s major pump producers and is in financial and technical collaboration with a UK firm. Its products include multistage pumps, split horizontal pumps and back pull-out pumps. The firm installed its first CNC machine in 1990. Sludge-valve The firm produces industrial valves for sewage- and slurry-handling plants. It was set up in 1978 in technical and financial collaboration with a UK firm, and employs 110 people. It specializes in the production of eccentric and gate valves. It introduced its first CNC turning centre in 1988. Indu-valve This is a small firm, with 58 employees, and owned by two engineers. It produces steel industrial valves for the petroleum, food and power industries. The firm does not have a foreign collaborator; the designs are provided by engineering consultants hired by the customers. At the time of our visit the firm was in the process of installing its first CNC machine. 2.2 Introduction of flexible automation CNC machine tools The first of the eleven sampled firms to have introduced a CNC machine was Hydro-valve, in 1978. This was, however, an exception. The great majority of the 91 CNC machines in the sampled firms have been introduced since 1985. The total includes 20 machining centres, 66 turning centres and 5 other CNC machines. As would be expected, the largest concentrations of CNC machines are to be found among the large producers in the automotive sector. The largest user is Tractors-India, which has 28 CNC machines, including 10 machining centres and 14 turning centres. The second largest user is Steer-link, which makes steering linkages and gears; it has a total of 14 CNC machines. The sampled firms did not provide data on the capital investment in CNC and conventional machines with which to estimate the degree of diffusion of CNCs. Instead, data on the number of employees operating CNC and conventional machines have been used to make indirect estimates of the density of diffusion of CNC machines (Table 10.4). The method begins with data on operator requirements of CNC and conventional machines. From this, we calculated the number of operators theoretically required if all the conventional machines were replaced by CNC machines. The actual number of CNC operators employed by the firm divided by this theoretical number is taken as the estimated degree of CNC diffusion.

FA AND THE INDIAN ENGINEERING INDUSTRY 303

Table 10.4 India: employment and diffusion in sampled firms

Source: Interviews.

According to our figures (Table 10.4), the four small producers of autocomponents had the highest density of CNC usage. Since these firms were set up specifically to produce components with CNC machines, this is not surprising. The density of CNC diffusion in the other firms is comparatively low, ranging from about 3 per cent in Pumps-India to 16 per cent in Tractors-India. The density of CNC diffusion does not appear to be related to either the size of the firm or its products. CAD and computers Only 5 of the 11 firms in our sample have installed CAD systems. Three of these are valve producers and the remaining two are producers of autoparts. Since most of the firms have foreign collaborators, which provide technical and design support, this is not surprising. The firms which do have CAD use it only for tool design and minor design modifications. This is particularly true of the three industrial valve producers, each of which has installed CAD. Many of the valves produced by these firms are custom-built. Of the remaining two firms with CAD (both autopart producers), one used it only for tool designing, and the other used it for modifying designs for steering assembly components to fit into various light commercial vehicles. Eight of the eleven firms have introduced computers to some extent, but the capabilities of the computer systems and the extent of their application were very limited. In most firms, computerization involved the use of less than ten standalone PCs for accounting and inventory. One firm had a mainframe computer with seventy terminals. None of the firms had linked CAD systems or computers to the CNC machines.

304 THE COUNTRY STUDIES

Table 10.5 India: reasons for the adoption of flexible automation in sampled firms

Source: Interviews. Note 1 = most important reason; 2 = second most important reason.

2.3 Reasons for the introduction of CNC machine tools Demand for quality has been the most important reason for the adoption of CNC machines by the firms in our sample. Ten of the firms said that the need to improve production quality was the most important reason for their decision to adopt CNC machines (see Table 10.5). With the liberalization of economic polices, the product groups included in this study (and many others not included) have seen a considerable upgrading of products. This has put strong pressure on firms to enhance their production capability. For example, a number of new vehicles based on comparatively advanced and complex designs have entered the automotive market in recent years. This has obliged the autopart firms to enhance their production capabilities to meet the high quality requirements. Similarly, many process plants established during the last decade incorporate technology which is more complex and sophisticated. The production of hydraulic valves and pumps for these plants requires superior manufacturing facilities which are capable of undertaking complex jobs more accurately than in the past. Flexibility was the second in importance as a reason for the adoption of CNC machines; seven firms gave it this ranking. Other factors have had no importance in the decision by 9 of the 11 firms to introduce CNC machines. None of the firms reported direct cost savings as the main reason for the introduction of CNC machines.5 Hydro-valve reported that the possibility of increasing exports played an important role in its decision. Hydro-valve is a large producer of industrial valves, making both large custom-built valves and small standard valves. All its production was based on conventional machines, until recently. However, it has

FA AND THE INDIAN ENGINEERING INDUSTRY 305

decided to increase its efforts to export gate valves, a market in which international demand has increased sharply in recent years. CNC machines were considered necessary for the production of valves for the export market, which requires high and consistent product quality. The firm has been successful in its export drive: about 63 per cent of the gate valves produced by the firm (in terms of value) were exported in 1993–4. It is notable that, while the autopart producers considered the flexibility provided by CNC machines to be important, the capital equipment producers considered it to be of little or—in three cases—no importance. The autopart producers’ reasons for using CNC machines were directly related to changes in the automobile industry which, since about 1985, have created a need for higher production quality and greater flexibility, both of which can be provided by CNC machines. For example, Steer-link, which is more than thirty years old, based its production of steering components on conventional machines until the mid-1980s. Since the vehicles made in India until that time had been based on older designs, components machined on conventional machines were acceptable to vehicle producers. With the large-scale entry of Japanese technology in the second half of the 1980s, the situation changed considerably. Vehicle manufacturers with Japanese collaborators demanded more accurately machined steering components, which could only be produced with CNC machines. Furthermore, while the number of vehicle manufacturers increased markedly during this period, the demand for vehicles did not increase correspondingly. To remain competitive, the component producers had to supply smaller volumes of components based on a greater diversity of designs. This persuaded Steer-link to introduce CNCs selectively, and it now has fourteen CNC machines. About 20 per cent of its current capital investment is in CNCs. The other steering-component producer, Maruti-steer, is a relative newcomer, having been set up in 1985 to produce steering components for India’s largest car producer, Maruti, which remains its only customer. Initially it used conventional machines to produce simple components, the more complex components being imported from Japan. Later, with an increase in demand and an escalation in the cost of imports (due to the appreciation of the yen), the production facilities were expanded. For reasons of quality and flexibility enhancement, it was decided to use CNC machines for the manufacture of some components from the very beginning. Maruti’s quality requirements are more stringent than those of other autocomponent producers in India. Many of its components can only be produced with CNC machines. Furthermore, the firm produces steering components for three car models produced by Maruti, and is expected to supply only two days’ requirements at a time. So it prefers to produce small batch sizes at a time, and finds CNC machines to be ideal. The firm’s decision to use CNC machines was influenced also by the fact that it follows a production system suggested by the foreign collaborator, which is characterized by low inventories and small production batches, although the firm has not yet fully achieved the expected level of JIT production. The changes in production technology adopted by Tractors-India for the production of cylinder blocks also highlight the dual importance of quality and

306 THE COUNTRY STUDIES

flexibility in the decision by autopart producers to use CNC machines. This firm began production of cylinder blocks in 1976 on special-purpose machines and general purpose machines (mostly radial drills). At that time the firm produced only the one tractor model, and the production volume was about 7,000 per year. In 1985, the firm decided to expand its production capacity and its range of tractors. By this time, some of the machines were old and were causing serious quality problems. The radial drills, in particular, were responsible for large quality variations and high rejection rates. The firm had a choice of adding new radial drills or acquiring CNC machining centres. It chose the latter, as the CNC machines provided it with the superior machining capability and greater flexibility necessary for the production of a larger range of engine blocks in smaller numbers. The four small producers of autocomponents for TELCO also use CNC machines for both quality and flexibility. In fact, TELCO agreed to subcontract the production of these components to these firms only because their use of CNC machines provided both the quality and the flexibility considered essential by TELCO. In contrast to autopart producers, the producers of industrial valves and pumps have always required flexibility in their production. Many of the valves and pumps produced by these firms are custom-built and the production volumes are often small. Conventional lathes, which are highly flexible, provide these firms with the flexibility required to produce a wide range of products in small volumes. Therefore they do not need to introduce CNC machines to add flexibility to the production facilities. Conventional lathes, however, did not provide the firms with the required degree of accuracy and repeatability. This posed serious problems when the firms had to machine components of complex shape, or to precise dimensions. Although these components could be machined on conventional lathes, the production required special care; in some instances components could only be produced in tool rooms. Consequently, the production volumes were small and the cost of production was high. The seriousness of this problem is illustrated by the experience of PumpsIndia, the pump manufacturer in our sample. Many of its pumps are custom-built for use in the power, sewage and fertilizer industries. The firm used only conventional machines until 1990. While these machines were adequate for machining most of the pump components, the firm faced serious problems in machining two components—chambers and impellers—which are among the most complex of the parts used in a multi-stage pump. Unless the machining is very accurate, there is a lack of concentricity in the pump, which affects its performance. To achieve the required level of accuracy, the parts were machined on two horizontal turret lathes in the tool room. But the rejection rate and the need for reworking were high, and production was slow. As a result, the firm could not meet its delivery schedules, and there was a large backlog. For example, in 1989–90 the firm had back orders for 300 multi-stage pumps, which it could not clear due to the bottleneck in the production of chambers and impellers. As the delivery schedules were becoming increasingly long (ten months), the firm was losing business. It was then, in 1990, that the firm decided

FA AND THE INDIAN ENGINEERING INDUSTRY 307

to replace the two turret machines with one CNC lathe for the production of these two components. With the help of this machine, the backlog was cleared in 6 months, and the delivery time has come down to 3 months. Quality was the most important motive also at Sludge-valve for the introduction of CNCs. This firm produces valves, in collaboration with a UK firm, for sewage, water-treatment and slurry plants. Until 1988 it had used only conventional machines, but as the quality requirements of the user industries became more stringent it felt unable to meet them without CNC machines. This, and the advice of the parent firm which uses CNCs extensively, were responsible for the firm’s decision to introduce new production technology, which has led to a significant improvement in quality. Rejection rates have fallen from about 2 per cent, using conventional machines, to almost zero, using CNC machines. The firm is planning to add several more CNCs during 1994–5, and aims to begin production of highly sophisticated valves (such as ball valves) which can be produced only with CNC machines. These are being imported at present. Two factors were reported by the firms as reasons for not introducing additional CNC machines: their small production volumes and the unavailability of suitable castings and forgings. In the circumstances, the firms were prepared to invest only in those machines which were considered essential for the production of crucial products. 2.4 Production process employed by sampled firms In order to appreciate the impact of CNC machines on the sampled firms, it is important to understand the main characteristics of the older technology and machines which have been replaced by CNC machines. The firms included in our sample are engaged primarily in the machining of forgings and castings to produce components of specific shapes and dimensions. The machining jobs commonly undertaken include turning, drilling, milling, welding and grinding. In the past, these operations were carried out by conventional engine lathes, milling and drilling machines, welding machines and grinders. The most flexible and common of the conventional machines are the engine lathes, which can be used to machine a wide range of shapes and dimensions. The repeatability and quality of these machines, however, are poor and the production quality depends greatly on the skills and experience of the machine operator. At the other extreme of the spectrum are special-purpose machines, which can perform only a predetermined set of operations, but have capabilities in respect of quality and repeatability which are superior to those of standard lathes and other similar machines. The producers of autoparts, by and large, used transfer lines and specialpurpose machines which have very little flexibility. The producers of valves and pumps, on the other hand, used engine lathes and other similar machines, which are highly flexible. Thus, even before the advent of CNC machines, the producers of valves and pumps had access to highly flexible technology. All of the sampled firms use both conventional and CNC machines. CNC turning lathes, turning centres and machining centres are installed alongside

308 THE COUNTRY STUDIES

chuckers, multiple spindles, engine lathes, and drill, spot facing, and special purpose machines. CNC machines are used only for jobs which are complex or require a high degree of accuracy. The combination of conventional machines and flexible automation can be illustrated by the case of Maruti-steer, which produces transmission and steering components, including differential castings. The castings are procured from subcontractors and are first pre-machined on engine lathes. This is followed by turning on two CNC turning centres. The remainder of the process is carried out on conventional machines: flange holes are drilled by multi-spindle machines; chamfering is carried out with a benchdrilling machine; finally, pin holes and cross holes are drilled, and boring and spherical-milling operations undertaken, on three special-purpose machines. This is followed by washing, assembly and heat treatment. Most components require heat treatment after machining, and this is undertaken in-house by most firms, as it is important for product quality. 2.5 Selection of CNC machine tools Most firms felt that they had experienced no particular difficulty in selecting the CNC machines suitable for their requirements. As would be expected, the large firms with strong technological capabilities were particularly well-informed about various types of CNC machine, and did not have any difficulty in making a choice. Only two firms felt that they had made wrong choices due to insufficient technical information. In both cases, the machines were chosen primarily on the criterion of cost. The machine in either case performed very poorly and required frequent and costly maintenance. Most firms approached major Indian machine-tool builders for information on the machines, their capabilities and prices. Trade fairs, both in India and abroad, also were said to be important sources of information. Foreign technology suppliers in some cases influenced firms to select a foreign machine (usually machines with which the technology supplier was already familiar). In most cases, the prices of the machines played an important role in the choice. Most firms preferred to select machines which can carry out the minimum functions required by them at the lowest cost. Consequently, most of the machines installed by the firms in our sample have limited capabilities. Only one firm has installed machining centres with pallets for loading and unloading the work pieces. Machines without pallet systems have to be stopped to load and unload the work-piece, so that their productivity is low. The country of origin of 77 of the machines was ascertained. Of these, 49 machines (64 per cent) were Indian, and all but one of the imported machines were from Japan. The firms’ views on the advantages and disadvantages varied greatly. By and large, Indian machines were preferred because the local machine producers were expected to provide better service. Those who preferred imported machines considered them to be more reliable and requiring fewer repairs. But, in either case, the experience of purchasers has been mixed, and it is difficult to draw any conclusion in this regard.

FA AND THE INDIAN ENGINEERING INDUSTRY 309

2.6 Choice of components/products to be produced on CNC machine tools Since most firms produce a variety of components or products, they may need to select items which are the most suitable for production on CNC machines. The technical complexity of the job and considerations of quality, were the factors which most influenced this choice. For example, Pumps-India’s decision to machine chambers and impellers with CNC machines was determined largely by the technical problems faced in machining these components using conventional machines. Similarly, Tractors-India’s use of CNC machines for the production of engine blocks was influenced by the degree of technical difficulty of machining engine blocks with radial drills. Another important consideration was the availability of castings and forgings of uniform quality, which are suitable for machining on CNC machines. For example, Pumps-India was able to use CNCs to machine impellers and chambers because the castings and forgings for those components are produced with superior technology. The castings and forgings for other components produced by the firm are made with manual moulds, and their quality is poor and uneven. Large production volume was another important consideration in the choice of these firms. This was particularly clear in the case of valve producers, who used CNC machines only in the high-volume production of small and medium-sized valves. Large valves were produced in small numbers using conventional machines. 2.7 Flexible automation and organizational changes The firms included in the sample had not made significant organizational changes as a result of the adoption of FA. This was largely because most firms produced only a small proportion of their output using CNC machines and other forms of new technology. The two important organizational changes which some of the firms have either made, or are in the process of introducing, relate to production lay-out and production schedules. The changes in lay-out appear to have depended on the extent of use of CNC machines in the production process. Where CNCs are used for many of the production operations, the lay-out is centred around particular products. For example, CNC machines play important roles in the production of differential casings by Maruti-steer and engine blocks by Tractors-India. In both cases all the machines involved in manufacturing these products are located together. The production lay-out for other components, which are produced with conventional machines, is functional. In other firms in our sample, where the role of CNC machines is small, the entire lay-out is functional. Some firms, especially those producing autoparts, have reported that they are in the process of rationalizing and reorganizing their production schedules. This is becoming necessary because vehicle manufacturers increasingly insist on

310 THE COUNTRY STUDIES

smaller sized order and tighter delivery schedules. This has put strong pressure on the firms to streamline all aspects of production (including material procurement). Although none of the firms has adopted the just-in-time system, it was mentioned as an aim by at least two of the firms. One of these, Maruti-steer, has close links with Japanese autocomponent producers and is under pressure from the Maruti car company, which buys most of its production. The other firm, Steer-link, is also a producer of autoparts. While some of the firms have reduced their product inventories significantly, they have been unable to reduce rawmaterial and component inventories because the suppliers—especially of castings and forgings —are not geared to the new system. They still use old production technology and are unable to supply material in small volumes at short notice. To avoid any disruption of production schedules, the firms in our sample maintain comparatively high inventories. Thus firms using FA are unable to take full advantage of them until both FA and production management practices have diffused through the whole chain of production. Although most firms said that quality improvement was an important goal, the concept and practice of quality control have not changed much in most firms. Quality control is still considered by many to be the same as product inspection, which is usually carried out during various stages of the production. Only one firm, Maruti-steer, reported taking steps to adopt TQM practices. The introduction of FA has affected employment and training structures although, since the density of diffusion is low, the impacts have thus far been limited. From a total of 5,477 employees in the sample firms, 3,341 (61 per cent) were directly involved with operating production machines, of whom only 171 (about 5 per cent) work on CNC machines. The general trend is to train the CNC operators only to set up and operate the machines. None of the firms has trained its CNC operators to program the machines, but some operators have learned to carry out minor adjustments to the programs. The older firms in our sample retrained their conventional machine operators as CNC operators; retraining took 3–6 months. The newer firms hired new operators with technical certificates from the Indian Institute of Technology and trained them for a similar period. This compares with the minimum of three years’ experience which is considered necessary for a conventional machine operator. Our case studies suggest that the introduction of CNC machines was accompanied by both de-skilling and the demand for certain new skills. The training for a CNC machine operator take one-sixth of the time taken to train conventional machine operators, which suggests deskilling as regards machine operators. The new skills which are associated with the use of CNC machines include software programming and maintenance of electronics and electrical machinery. However, most firms felt that the controls used in CNC machines were, by and large, user-friendly and that the programming did not require a high degree of skill. None of the firms had employed specialist programmers; most had trained some of their engineers for the task, and one small firm hired outside consultants for programming. Similarly, none of the firms had hired specialist maintenance personnel. The existing engineering and maintenance staff were

FA AND THE INDIAN ENGINEERING INDUSTRY 311

trained to look after minor problems, and all major problems are corrected by the machine suppliers or specialist maintenance companies. 3 Flexible automation and economies of scope and scale 3.1 Flexible automation and scope Setting-up times Setting-up time has an important bearing on the flexibility of production: the shorter the setting-up time the more economical becomes the production of small batches of diverse products. Table 10.6 presents figures for the changes in setting-up time, and for the production of the batch sizes mentioned in the case studies and in the section on batch size. The five newer firms in the sample had not used the older conventional technology, but the engineers in these firms had experience of producing the same products with conventional technology in other firms. They provided estimated times for use in comparisons. It can be seen that the use of CNC machines has led to a large decrease in the setting-up times of all the firms but one in our sample. The decreases, for 10 of the 11 products, ranged between 27 per cent and 86 per cent, with a typical value of around 75 per cent. Much of the reduction is as the process of setting-up is required less often with CNC machines because fewer machines are used and, in some instances, the number of operations required to complete a job are fewer. So where conventional machines might have to be set up six times in the course of a production process, we found that CNC machines would need to be set up only 2.5 times on average. Table 10.6 India: changes in setting-up times (in minutes) in sampled firms

Source: Interviews.

312 THE COUNTRY STUDIES

Table 10.7 India: delivery requirements and batch sizes in sampled firms

Source: Interviews.

Clearly, the greater flexibility of the CNC machines, as indicated by lower setting-up times, has enabled the user firms to produce smaller batches with less loss of time. There were significant savings also in machining and in loading and unloading times. Batch size According to data on the sampled firms’ 11 products shown in Table 10.6, the most common batch sizes are 251–500 units (for 5 products), and 100–250 units (for 3 products). Two products were produced in batches of less than 100, and one in batches of more than 500. Four products were produced in batches of 500. The time taken for the machining of typical batches of these products varied between 2 and 9 days; for 8 of the products, it is less than 4 days. The longest time (16 days of production) was for tractor engine blocks. As the tractor model concerned accounts for about 85 per cent of the firm’s tractor sales, the firm prefers to produce large batches (300) for this model. For other models, which account for a smaller proportion of sales, the batch sizes are smaller. For example, a typical batch size for a model which accounts for 5 per cent of sales is 100 units. A similar pattern was found in the valve and pump manufacturers, which produce large batches of small standard valves, and small batches of large or specialized valves. For example, one of the valve manufacturers typically produces small valves in batches of 300, and larger valves in batches of 50 or less. Another firm produces valves of up to 4 inches in diameter in batches of 50– 100, and larger valves in batches of 25–50. The batch size depends also on delivery schedules and the availability of castings and forgings. The batch sizes of the autopart producers were determined largely by the delivery schedules required by vehicle manufacturers. The latter have become increasingly keen on reducing inventory costs, and insist on receiving supplies equal to only one or two days’ requirements. Furthermore, their requirements change frequently. It must, however, be emphasized that in spite of various pressures to reduce batch sizes, firms try to maintain the minimum batch size which is considered to be economical. The relationships between delivery requirement and batch size, are some of the autopart producers, is shown in Table 10.7.

FA AND THE INDIAN ENGINEERING INDUSTRY 313

The availability of castings and forgings affects batch sizes. Most firms mentioned that, as good quality castings and forgings are often available only in small numbers, they have to keep this in mind when deciding on optimal batch sizes. Furthermore, the supply is erratic and the firms are sometimes forced to run smaller than usual batches. Since several of the firms in our sample were new, it is difficult to draw definite conclusions about the changes in batch sizes after the introduction of CNC machines. We have information for only five items, produced by four firms. The batch size has fallen for 2 products, increased for 2 products,6 and remained the same for the remaining product. Although we do not have data, the experience of the rest of the firms suggests that the flexibility provided by the CNC machines has allowed them to produce smaller batches. This is particularly true in the case of autopart producers (5 of the 7 firms for which data are not available). The shorter setting-up times associated with CNC machines were considered vital in enabling these firms to cope with frequent changes in demand, and short delivery times. Product diversity We have information on the diversity of products7 produced by ten firms. Of these, 4 produced less than 20 types, 3 produced 21–50 types, 2 produced 51– 100 types and 1 produced more than 100 types of product (see Table 10.8). Comparisons with the number of product types produced before adopting CNC are available for only 5 firms: 2 valve producers, 1 pump manufacturer, and 2 autopart producers. The introduction of CNC machines has led to an increase in the number of product types only in the two autopart producers, where the number of product types increased by 133 per cent and 50 per cent, respectively. Since the other autopart producers are newer firms, it is not possible to make a comparison. However they do produce a large range of products for the size of their capital investments. Two firms produce 7 and 14 products, respectively, and all the remaining firms produce more than 20 product types. The average is 37 types per firm. According to these firms, it would be impossible to produce such a diversity of products economically, with this amount of capital investment, using conventional technology. For a given amount of capital investment, CNC machines were found to provide far greater flexibility of production than was possible using conventional machines. The number of items produced by the valve and pump producers, on the other hand, was found not to have changed with the introduction of CNC machines. Clearly, there are other, and more important, factors which determine a firm’s product range in these industries. According to these firms, the need to specialize, access to technology and design support, and the growth of various user industries have influenced their strategies with regard to product range. It is important to note here that in most cases the CNC machines used by valve and pump producers have replaced conventional engine lathes, and not specialpurpose machines and transfer lines. In other words, even prior to the introduction of CNC machines, the firms were using highly flexible technology.

314 THE COUNTRY STUDIES

Table 10.8 India: changes in product diversity (types of product) in sampled firms

Source: Interviews. Notes a New firms. b The numbers refer to the tractor models produced by the firm. We do not have information on the number of types of product (component) produced by the firm.

The main objective of the introduction of CNC machines by these firms was to improve production quality, and not to gain flexibility. Does the use of CNC machines affect a firm’s decision to produce in-house or to subcontract? Since the use of CNC machines allows even small firms to produce complex components and parts, the potential for sub-contracting might be expected to increase. On the other hand, with the increase in flexibility, the users of CNC machines may find it economical to produce a greater diversity of components in-house (some of which may have been produced by specialist firms in the past). The evidence of our studies is mixed. One of the firms—Indu-valve — feels that the use of CNC machines will allow it to produce some of the valve components which in the past were sub-contracted, with greater control over quality and at more economical prices. But the evidence from other firms suggests a trend towards more subcontracting. For example, the four small autopart producers in our sample are in business because TELCO (India’s largest truck manufacturer) had adopted a policy of large-scale contracting-out of engine and gear box components to small firms with CNC machines. Similarly, Marutisteer is planning to expand its production capacity by contracting-out the production of some of the components to small firms with CNC machines. The subcontracting will allow it to expand production without making capital investments. These examples show that, in theory, CNC machines have widened the options available to firms. In practice, a firm’s choice will depend on a number of other factors, including the comparative cost of in-house production and subcontracting, capacity utilization and ability to invest capital.

FA AND THE INDIAN ENGINEERING INDUSTRY 315

Table 10.9 India: changes in machining times (in minutes) in sampled firms

Sources: Interviews.

3.2 FA and scale For a number of reasons we have not been able to examine the impact of FA on scale directly. Firms felt that CNC machines accounted for a very small proportion of their equipment, and their role in changes in output (at both product and plant levels) was small. The changes in the levels of output were due largely to changes in demand. However, we have sufficient indirect evidence to suggest that the CNC machines have increased the firms’ potential output. First, CNC machines operate at higher speeds, meaning shorter machining times and higher output. Table 10.9 compares the machining time required for different components. The figures clearly show that machining times with CNC machines are about 70 per cent shorter than with conventional machines. Furthermore, the time required to set up CNC machines is substantially shorter than for conventional machines (see Table 10.6). Second, CNC machines significantly reduce both rejection rates and the need for reworking. Rejection rates had fallen in the sampled firms by 50–70 per cent (Table 10.10). In some firms, Pumps-India for example, this led to a dramatic increase in production capacity. As mentioned earlier, Pumps-India had faced serious problems in machining two components for multi-stage pumps. These were machined on two horizontal turret lathes in the tool room, but rejection rates and re-working rates were still very high, and the firm could not meet its delivery schedules. In 1990, the firm replaced the two turret lathes with one CNC lathe which increased production capacity markedly and enabled the firm to clear its backlog within six months. Similarly, Tractors-India has been able to increase its engine-block production capacity by replacing radial drills with machining centres. The radial drills were old and were leading to high rejection

316 THE COUNTRY STUDIES

Table 10.10 India: changes in rejection rates (in %) in sampled firms

Source: Interviews. Table 10.11 India: changes in machine availability (in %) in sampled firms

Source: Interviews.

rates, thus slowing down production. With the addition of machining centres the firm was able to use the capacity of its engine-block production line more efficiently and to increase engine-block production. Third, in many cases the availability of CNC machines—the proportion of working hours for which they are actually able to produce—was higher than for conventional machines (Table 10.11). Of the 4 firms for which we have information, 3 reported increases in machine availability with CNCs, ranging from 13 to 28 per cent. This implies higher capacity utilization and a potential increase in production scale. The evidence presented in section 3.2, although not direct, is sufficiently strong to suggest that CNC machines have an up-scaling effect at both plant and product levels. 4 Production costs Because of the limited diffusion of CNC machines and the lack of information, we were unable to compare total unit costs for conventional and CNC machines. Instead, we have asked the firms to provide us with estimates of the costs of

FA AND THE INDIAN ENGINEERING INDUSTRY 317

equipment and labour required to produce the same volume of one of their products, using conventional machines and CNC machines. Capital cost Before arriving at estimates of capital and labour costs, it is necessary to examine the number of conventional machines replaced by CNC machines (for the same volume of production) and the labour required to operate the two sets of machines. This information was collected for the machines involved in the production of the components and products examined in detail for the study of setting-up and other times. According to this information, which was available for all 11 firms, 14 CNC machines have replaced 42 conventional machines, implying that 1 CNC has, on average, replaced 3 conventional machines (Table 10.12). Table 10.12 India: number of conventional machines replaced by CNC machines in sampled firms

Source: Interviews.

The data in Table 10.13 shows that the estimated cost of equipment required to produce the same output increased with the introduction of CNC machines in 9 of the 11 firms. In more than half the cases (six), capital requirements increased by more than 50 per cent. The total capital cost of the machines involved in the change from old to new technology increased from 44 million rupees (US$1.42 million) to 49.1 million rupees (US$1.58 million). Labour costs We found a universal decline in labour costs when CNC machines were used in place of conventional machines for the production of the same output. The 14 CNC machines referred to above required 32 operators (Table 10.14). If the same volume was produced on conventional machines, it was estimated that 124 operators would have been required. This implies that each CNC machine has

318 THE COUNTRY STUDIES

Table 10.13 India: cost of equipment in sampled firms

Source: Interviews. Table 10.14 India: changes in labour requirements of sampled firms

Source: Interviews.

displaced 6.6 employees, and that CNC machine operators are approximately four times as productive as conventional machine operators. Since these firms have a total of 88 CNC machines, and 171 CNC machine operators, the introduction of CNCs is estimated to have reduced labour requirements by more than 580 operators. We have used average salaries of CNC and conventional machine operators to examine changes in labour costs. As Table 10.15 shows, the decline in labour costs for the production of the same output ranges between 25 and 86 per cent. The total annual labour cost of the machines involved in the change from old to

FA AND THE INDIAN ENGINEERING INDUSTRY 319

Table 10.15 India: changes in annual labour cost in sampled firms

Source: Interviews.

new technology decreased from Rs. 4.40 million (US$0.14 million) to 1.34 million (US$0.04 million), a saving of Rs. 3.06 million (US$0.09 million). However, Indian firms are not permitted to dismiss excess labour, so they must redeploy the excess workers in other productive jobs. This would lead to a substantial increase in capacity, thereby further increasing the scaling-up effect of FA. And, in a situation of stagnating demand, firms will be unable to benefit from the labour-saving potential of the new technology. Inventory It is generally agreed that the increased production flexibility provided by CNC machines and other forms of new technology can enable firms to reduce their raw-material and production inventories considerably. We have already mentioned that some of the firms in our sample, particularly autopart producers, have reduced production batch sizes. This has reduced the need to maintain large material and production inventories, leading to substantial savings in costs. Table 10.16 presents data on the size of material inventories for ten firms. Of these, four autopart producers are new firms without a history of using conventional technology, so that no comparison with previous inventory levels is possible. However their inventory levels are clearly very low; three days of inventory for both materials and output. The firms said that it would have been impossible to maintain such low inventories had conventional technology been used. This would be in accordance with the suggestion, for which we could not obtain confirmatory data, that batch sizes have fallen. Of the remaining 6 firms for which data are available, 4 have reduced material inventories by 40 per cent or more. Only 2 firms reported no change in inventory levels. These, and some other firms, said that CNC machines offer potential reductions in inventory, but that they had been unable to take full advantage of

320 THE COUNTRY STUDIES

Table 10.16 India: changes in material inventories (number of days) in sampled firms

Source: Interviews.

these for various reasons. For example, Maruti-steer, which was in the process of adopting a number of advanced production management techniques, such as JIT delivery and TQM, was finding it difficult to reduce its material inventory below 15 days’ because the suppliers of castings and forgings were unable to supply small batches of material at short intervals. Moreover, the suppliers are located at some distance and the cost of transportation increases significantly for small lots. Similar difficulties were mentioned by other firms. Most felt that unless the firms producing the castings and forgings update their production techniques and adopt more flexible technologies, further reduction in inventories will be difficult. Space The fact that one CNC machine replaces three conventional machines on average means that the space requirements of CNC machines are much smaller. No data were available on the cost implications of the space savings, but they would appear to be significant. Table 10.17 shows that the reduction has been substantial for the five firms on which we have data. Maintenance Most engineering firms have in-house facilities to undertake minor repairs of conventional machines. In fact, large engineering firms often have fullyequipped maintenance departments, which look after most of their needs. As a result, the cost of maintenance of conventional machines is comparatively small. The maintenance of CNC machines, on the other hand, requires specialized skills which are often not available in-house. Firms using these machines usually employ outside experts (machine suppliers or firms/consultants specializing in maintenance service) to undertake maintenance. This service is comparatively costly. Unfortunately, we have data on maintenance costs for only two firms, Steerlink and Telco-diffs. One of these firms is large and the other is small: both are

FA AND THE INDIAN ENGINEERING INDUSTRY 321

Table 10.17 India: changes in space requirementsa in sampled firms

Source: Interviews. Note a Square metres.

producers of autoparts. In spite of the differences in size and technical competence between the two firms, both reported higher maintenance costs for CNC machines. Each firm had spent 0.5 per cent of the cost of conventional machines on maintenance, and now spends 1.5 per cent of the cost of CNC machines on their maintenance. Software Since CNC machines have computerized controls, they have to be programmed. The firms had found that the controls were user-friendly and did not require highlevel skills. None of the firms had hired specialist programmers, and the training in programming given to engineers generally took less than one month. This training was provided by the machine suppliers. Only Telco-diffs, one of the small producers of autoparts, employs a part-time consultant for programming support. All the firms felt that the cost of programming was negligible. R&D and technology imports Foreign collaboration is the main source of technology for six of the firms in our sample. These comprised 2 valve producers (Hydro-valve and Sludge-valve), 1 pump producer (Pumps-India) and 3 autopart producers (Tractors-India, Steerlink and Maruti-steer). Only the autopart firms undertake sizeable R&D activities. Their basic product designs and process details are provided by foreign collaborators, and they concentrate on the adaptation of technology for Indian conditions, and on indigenization. The valve and pump producers have small design departments which carry out minor product modifications to meet customer requirements. When major modifications in products are required, the problem is referred to foreign collaborators. Of the 5 firms which do not have foreign collaborations, 4 are small producers of autoparts. They rely completely on TELCO for product designs and technical support, and do not have R&D or design departments. The fifth firm, Indu-valve, gets its basic designs from engineering consultants, who are hired by the clients.

322 THE COUNTRY STUDIES

They also do not have in-house R&D capabilities, but have a small design department with CAD facilities. None of the firms in our sample felt that the costs of technology imports and R&D have been affected by the use of new technology. Since they all produce well-proven products, the use of CNC machines does not pose serious technical difficulties which would require an increase in R&D efforts or higher expenditure on technology imports. Also, none of the firms has undertaken major modifications to their products to make them more suitable for production on CNC machines. It is safe to conclude that the use of FA has not led to an increase in the costs of R&D or technology imports in the sampled firms. Marketing and administration The costs of marketing and administration also were found to be unaffected by the use of FA. These firms are not producing ‘new’ products the marketing of which would require greater technical training and knowledge than in past. The use of computers was reported to have reduced the cost of administration to some degree, but none of the firms had figures on this. Unit costs, summary Thus, FA was found to be associated with an increase in the costs of purchasing and maintaining capital equipment. On the other hand, the cost of labour had declined. The costs of technology acquisition (both R&D and technology import) and marketing, which are generally believed to increase with the use of new technology, were found to be unaffected, and programming had not proved to be a significant cost. 5 Conclusions India’s industrial and trade policies have undergone major changes in recent years. An emphasis on planning and protection has been replaced by greater reliance on market forces. The policies which restricted entry and expansion, foreign investment and imports of technology and capital goods, in particular, have been liberalized. The removal of entry barriers has encouraged a large number of firms to enter markets with comparatively new products, with the help of foreign collaborations. However, the growth of industry (including the engineering industry) during the 1990s has been extremely low. Although the production and use of CNC machines in India began comparatively early on, their diffusion has been slow, particularly in the 1990s. This can be attributed to the low level of industrial growth during this period. By and large, the sampled firms confirm this trend: except for the four smaller firms which were set up specifically to produce automobile components with CNC machines, the density of diffusion of CNC machines and other computerized

FA AND THE INDIAN ENGINEERING INDUSTRY 323

technologies is low. It is therefore important to emphasize that our conclusions with regard to the impact of these technologies on scale and scope are based on rather limited evidence. The main motives for the introduction of CNC machines were the need to enhance machining capabilities to produce comparatively new products, and, in the case of the autopart producers, to meet flexibility requirements. On the other hand, the producers of valves and pumps, which were already using flexible production equipment, did not consider flexibility to be an important motive for the introduction of CNC machines. The use of CNC machines has led some firms to make limited organizational changes, such as partially switching from a functional to a product-based production lay-out. Steps to reduce inventory size were also reported by some firms. However, the extent of these changes is severely limited by the low diffusion of CNC machines. With regard to the scope of production, we found that the effect on product diversity is closely related to the nature of the technologies used by the firms prior to the introduction of CNC machines. Where the firms were already using flexible technologies (as in the case of pump and valve producers), the use of CNC machines has not led to an increase in product diversity. But firms that had been using less flexible equipment (for example, the transfer lines used by autopart producers) have increased their product diversity, and their production flexibility, considerably. However, even in these firms, the increased flexibility has not led to increases in economies of scope. This is mainly because the demand is not sufficient to expand output appreciably and because the predominance of conventional machines does not permit the use of CNC machines on a three-shift basis. Only the four small producers of autoparts (which have sufficiently large demand, and where the diffusion of CNC machines is high) are able to operate the machines in three shifts; they have used the increased flexibility to gain increased economies of scope. Although we could not examine the impact of FA on scale directly, we found sufficient indirect evidence to suggest that the use of CNC machines by the sampled firms is associated with potentially higher output, because of lower setting-up and machining times, and lower rejection rates. It is therefore clear that the use of CNC machines has not reduced economies of scale. The scaling-up effect of FA is extended by an increase in the capital cost of CNC equipment, as compared to conventional machines, for the same output. Although CNC machines led to a significant decline in labour costs, Indian firms have had to redeploy excess labour to other production activities, which has amplified the scaling-up effect of FA. The R&D and designing activities of these firms are minor, and have not increased in size or importance with the introduction of CNC machines. The costs of marketing also have been unaffected, and programming has not proven to be a significant cost. Since the optimal use of CNC machines was found to require higher output, it is clear that some of the old entry barriers related to scale continue to be important. At the same time, it must be pointed out that these machines have

324 THE COUNTRY STUDIES

reduced some other important entry barriers. They allow firms in developing countries to enhance their production capabilities and remove specific bottlenecks in a comparatively short time. It is this aspect of CNC machines, rather than flexibility, which affects the location of production. FA allows firms in developing countries to undertake the manufacture of complex products of high quality, which could not have been produced using conventional machines by operators possessing the current levels of skills. The positive impact of FA on the location of production in small firms is even more obvious. The experience of the four autopart producers highlights this point. It was the capabilities offered by CNC machines which led the vehicle manufacturer (TELCO) to consider subcontracting its engine and gear-box component production to small firms and these same capabilities that enable such small firms to produce such complex components. To conclude, the use of CNC machines is associated with higher output and increased capital costs, suggesting an increase in scale-related entry barriers. Higher output appears to be necessary to derive full benefit from lower labour costs. At the same time, a number of production and technical bottlenecks related to inferior machines and limited operator skills have been reduced by the enhanced capabilities characteristic of these machines. It is this aspect of CNC machines in particular, and of new technology in general, that is likely to increase the possibilities of locating production in developing countries. Notes 1 The section dealing with policy changes is largely based on Jacobsson and Alam (1994). See also Mohan (1993). 2 Calculated from Government of India (1991). 3 According to a census of machine tools carried out by the Central Machine Tools Institute (now the Central Manufacturing Technology Institute), there were 1,182 CNC machines installed in India by 1986 (IMTMA). The Indian Machine Tool Manufacturing Association has provided us with data (which do not include the number of CNC machines imported in 1989–90 and 1993–4) showing that 3,688 machines were installed between 1987–8 and 1993–4. Thus, there is evidence that at least 4,870 CNC machines have been installed, and taking into account the number of machines imported in the two years for which data are not available, the total number of CNC machines in India is about 5,000. 4 Firms with more than 500 employees are classified as large, those with between 100 and 500 employees are medium-sized, and those with less than 100 employees are small. 5 Many firms in our sample have experienced reductions in production cost due to reduced rejection rates resulting from the use of CNC machines. However, the primary objective of these firms was to enhance machining capability, leading to quality improvement. None of the firms was motivated by the possibility of cost savings through higher machining rates, savings in space, etc. 6 These products (chambers and impellers of multi-stage pumps) are produced by the same firm. When they were produced with conventional machines, the tools had to

FA AND THE INDIAN ENGINEERING INDUSTRY 325

be replaced after machining 10–20 chambers and 5–10 impellers. Since the machines then had to be set-up again, these numbers were commonly used as batch sizes. With the use of a CNC turning centre, the tools need to be replaced less frequently, and the firm is able to fix the batch size at twenty-five. 7 Product type takes into account differences arising from size, shape, and technical variations. For example, Pumps-India produces 20 product types. These include multistage, back pull-out and split horizontal pumps, in various sizes and for various applications. Steer-link produces more than 80 product types, including 6 types of steering linkage for 14 vehicles.

11 THE IMPACT OF FLEXIBLE AUTOMATION ON SCALE AND SCOPE IN THE TURKISH ENGINEERING INDUSTRY Hacer Ansal

1 The Turkish economy and the engineering industry 1.1 Historical developments Turkey adopted an import substitution industrialization (ISI) strategy in the 1950s. As in many other developing countries, a major consideration was to safeguard foreign exchange by accelerating the national industrialization process. ‘Encouragement schemes’ were prepared by the government to create an attractive climate for domestic and foreign investment in the industrial sector. Entrepreneurs were protected against foreign competition by high tariff barriers, and encouraged with tax exemptions and preferential exchange rates for technology imports. Producing for the highly protected internal market, however, greatly limited the firms’ technological efforts to adopt flexible automation (Ansal 1990). Following the new constitution adopted after the 1960 coup, the State Planning Organization was established to run the economic activities with the aid of five-year plans. During the planned-economy period, until the onset of crisis in 1977, the economy grew at 6–7 per cent. This is Turkey’s longest period of sustained high growth. The industrial sector was the fastest-growing sector, with rates of 9–11 per cent. Inability to obtain adequate local supplies of production inputs in the face of the expanding domestic market, however, resulted in a mounting demand for foreign exchange. Towards the end of the 1970s, firms’ allocations for imported production inputs were restricted, due to the foreign exchange shortage, and this forced cutbacks in production. The economic crisis that commenced in 1977 brought a huge external debt, a large foreign trade deficit, and accelerating inflation. The dramatic increase in oil prices and the decreasing remittances from Turkish workers abroad further restricted imports of necessary inputs in the industrial sector and resulted in substantial falls in production levels.

FA AND THE TURKISH ENGINEERING INDUSTRY 327

Table 11.1 Turkey: economic trends, 1977–92

Sources: SIS (a) and SPO (1993).

As a solution to the problem of production under the strategy of ISI, an exportoriented industrial strategy was adopted in 1980. One of the aims of this strategy was to restructure Turkish industry so that it could be competitive in the world market. The measures adopted included the gradual removal of restrictions on imports of manufactured goods and a gradual reduction in tariffs. Manufacturers were encouraged to export their products through government incentives and subsidies. The austerity programme which was initiated with the aid of the 1980 coup, together with the adoption of some macroeconomic and fiscal stabilization policies, lowered inflation to 37 per cent (from a record high of 104 per cent in 1980) and led to renewed economic growth, which averaged 4.5 per cent in the 1981–4 period. In 1985–8, rationalization of the incentive system through trade liberalization, the freeing of domestic financial institutions, and the restructuring of public spending led to an increased average growth rate of 6.1 per cent. In the 1989–92 period, following the liberalization of capital flows in external accounts, the growth rate slowed to 4.4 per cent and inflation rose to 60 per cent (Celasun 1991). The trends in the Turkish economy shown in Table 11.1 indicate that exports have shown a remarkable improvement since 1980, which has helped to overcome the foreign currency shortage. However these increases in exports were associated with excessive export subsidies, managed exchange rate policies and a significant contraction of domestic demand (Boratav 1990). The ratio of exports to imports increased to 81.4 per cent in 1988, but after the further liberation of some manufactured imports, it dropped to 58.1 in 1990 and was still only 64.3 per cent in 1992. The sharp increase in the ratio of exports to GNP, from 2.8 per cent in 1977 to 12.8 per cent in 1988, also slowed in 1990 and 1992. Despite the remarkable growth of exports, gross fixed investments have not increased significantly, although there was some growth after 1988. The ratio of

328 THE COUNTRY STUDIES

investments to GNP reflects a stagnation in investments. Since the growth in exports does not seem to have been related to investments, it cannot be the result of the successful transformation of industrial production by the adoption of FA. Foreign capital investment during the ISI era was relatively low in comparison with levels in some developing countries. Turkey’s unattractiveness to foreign capital was related to relatively high wages and an overvalued local currency during this period, which decreased profitability for foreigners. Although foreign capital investments increased significantly in the 1980s, total foreign investment remained low, except in joint-ventures with Turkish capital in some key sectors such as the automobile, pharmaceutical, petro-chemical and rubber industries. The populist economic policy adopted by the newly elected government in 1989 aimed at high economic growth. The huge budget deficits it generated were covered by keeping real interest rates high to attract ‘hot money’, and by internal and short-term external borrowing. Economic growth increased to 9.2 per cent in 1990, fell back to 5.9 per cent in 1992 (as can be seen in Table 11.1) and reached 7.9 per cent in 1993. This resulted in a dramatic increase in internal demand in the early 1990s, which greatly influenced the investment decisions of our sample firms. The economic conditions generated by the government’s policy led to a major economic crisis in 1990–3. This mood of crisis was apparent in the plant visits and interviews carried out for this study. Most of the managers were in the crisis-management mode, holding long meetings to work out a strategy to safeguard their firms, particularly by trying to increase exports. In the last week of February 1994 the price of the US dollar increased by 250 per cent. Economic uncertainty paralysed the manufacturing sector. A stabilization programme was announced by the government in April 1994. The severe fall in demand had led to production cuts in the engineering industry and large-scale redundancies, and GNP decreased by 6 per cent in 1994. 1.2 Developments in Turkish manufacturing The rise in manufactured exports in the 1980s was remarkable (see Table 11.2). The ratio of manufactured exports to total exports rose from 33 per cent in 1977 to 83 per cent in 1992. This export success, however, again appeared to be unrelated to investments made in the industry, since the proportion of total fixed capital investments going to the manufacturing industry decreased from 31 per cent in 1977 to 15 per cent in 1988, although it then rose slightly to 17 per cent in 1992. Due to the privatization of state enterprises in the 1980s, fixed capital investments in the public manufacturing sector were almost completely cut. Private sector investments reveal a similar trend, although the proportion of private fixed capital investments going to the manufacturing industry was higher than the figure for state and private manufacturing together. This pattern of stagnation in manufacturing investments again indicates that export growth in manufactured goods has resulted, not from investments in production capacity, but from increased capacity utilization rates. According to

FA AND THE TURKISH ENGINEERING INDUSTRY 329

Table 11.2 Turkey: trends in manufacturing, 1977–92

Sources: SIS (a) and SPO (1993). Table 11.3 Turkey: real wage index in manufacturing, 1980–91

Source: Yeldan 1994:62.

the Istanbul Chamber of Industry, the rate of utilization of productive capacities had reached 90 per cent by the end of 1988 (Duruiz and Yentürk 1992). Production increased by 95.9 per cent over 1981–91, according to the production indexes calculated by the National Productivity Centre (NPC) for the manufacturing industry. The centre’s employment indexes show a slowly rising trend, which reversed after 1988 although the volume of production continued to increase. Labour productivity therefore increased by 90.7 per cent in the 1981–91 period, which is only a little less than the increase in production. The productivity rise was particularly remarkable in the 1988–91 period, when it rose by 23.8 per cent while production increased by only 13.4 per cent (Kaya 1992). Real wages in the 1980s fell drastically until 1988, but then recovered rapidly, by 97 per cent from 1988 to 1991 (Table 11.3). This increase in wages underlies the drop in employment in manufacturing in recent years and may also have been a motive for investments in labour-saving FA. This wage increase had significant effects for most of our sample firms. 1.3 The Turkish engineering industry The engineering industry (ISIC 38) in Turkey is not a well-developed part of the manufacturing sector (ISIC 3), although its share in manufacturing value added rose slightly in the 1980s, from 18.1 to 19.5 per cent (see Table 11.4). This share remains rather low in comparison to that of some of the other developing countries in this study, such as Brazil, Mexico, India and Thailand.1 The output, employment and fixed capital investment shares show slight improvements in the early 1990s, but unfortunately there are no data available for more recent years to give some indication of the possible impact of investments in FA.

330 THE COUNTRY STUDIES

Table 11.4 Turkey: share (%) of the engineering industry in total manufacturing value added, 1980–90

Source: SIS (b).

Small firms (with less than 100 employees) constituted 70 per cent of the total number of firms in the engineering industry in 1987, and 67 per cent in 1990. In the same period, the proportion of medium-sized firms (with 100–499 employees) increased from 24 per cent to 26.6 per cent, and the proportion of large firms (with 500 and more employees) rose slightly, from 6.1 per cent to 6.4 per cent (Kisaer 1993:80). Large firms accounted for 47 per cent of employment in engineering. This share was increasing slightly, while the employment shares of both small and medium-sized firms had fallen in the years leading up to the study. The data do not indicate any trend in Turkish engineering firms to ‘downsize’ in terms of numbers of employees. Investments in fixed assets seem to have been made predominantly by large firms, especially in the 1987–90 period, when their share of total investments in fixed assets in the engineering industry rose from 43 to 65 per cent. Mediumsized firms’ share in investments fell from 50 to 25 per cent. Small firms, which dominate the sector in terms of firm numbers, were not investing much in fixed capital. Their investment share ranged over 7–14 per cent, without showing any growth trend (Kisaer 1993:99). This may suggest that small and medium-sized firms in Turkey cannot afford to invest in FA. Small firms, which constituted 69 per cent of the total number of firms in the engineering industry, produced about 11 per cent of the total value added in the industry. Large firms, on the other hand, created around 60 per cent of the total value added. There does not seem to have been any change in this share, despite their increased share of fixed investments in 1990 (Kisaer 1993:112). Thus the measures of firm numbers, investments, and value added all show that, in Turkey, small engineering firms’ viability has not increased. The shares of the various engineering subsectors in engineering value added are shown in Table 11.5. The transport equipment and electrical machinery subsectors are the most important, followed by non-electrical machinery and fabricated metal products. The share of precision instruments is very low, which may reflect the underdevelopment of the Turkish engineering industry, although there is a growth trend after 1987. When we examine the value added per employee in the engineering industry and its subsectors (at 1987 prices), we see (in Table 11.6) that there was a significant increase in 1990 in all subsectors, although it is difficult to relate this development to the diffusion of FA since there has been no apparent increase in fixed capital investments in the engineering industry. It may, however, be

FA AND THE TURKISH ENGINEERING INDUSTRY 331

Table 11.5 Turkey: structure of the engineering industry, by subsector, 1983–9

Notes ISIC 381=Manufacture of fabricated metal products ISIC 382=Manufacture of machinery except electrical ISIC 383=Manufacture of electrical machinery apparatus, appliances and supplies ISIC 384=Manufacture of transport equipment ISIC 385=Manufacture of professional and scientific, and measuring and controlling equipment not elsewhere classified. Table 11.6 Turkey: value added per employee in the engineering industry, 1987–90

Source: Kisaer 1993:162.

associated with the substantial drop in the number of employees, from 191,316 in 1986 to 128,094 in 1990 and just 112,507 in 1992 (SIS (b, c)). The non-electrical machinery subsector (ISIC 382) accounted for the largest share of imports for the engineering sector, with 16–20 per cent of the total in the 1988–91 period, followed by the electrical machinery subsector (ISIC 383) with 9–11.5 per cent of engineering imports. The engineering industry’s share in total manufactured exports was insignificant. The leading subsector, electrical machinery, accounted for only 2–4 per cent of manufactured exports, again showing the low level of development of the Turkish engineering industry (Kisaer 1993:9). Despite its poor export performance, production in the engineering industry increased in the 1986–91 period by 37–55 per cent in all subsectors except for ISIC 385 (precision instruments), due to increased internal demand. This increase was reflected in the production, employment and labour productivity indexes as shown in Table 11.7. The rise in production, however, was not

332 THE COUNTRY STUDIES

Table 11.7 Turkey: production, employment and productivity indexes in engineering industry subsectors, 1987–91 (1986=100)

Source: Aydin (1993:92, 96, 100, 104, 108).

accompanied by a rise in employment levels in subsectors 381 and 382, where labour productivity increased by 75 and 54 per cent, respectively, in the same period. In subsectors 383 and 384, employment rose by around 10 per cent while production increased by 55 and 60 per cent, respectively. The productivity increases in these subsectors were significant, at 39 and 35 per cent. From the indexes shown in Table 11.7, it would appear that increased production for the internal market has been accompanied by increased efficiency in recent years, except in the precision instruments subsector, which seems to be the weakest subsector of Turkey’s engineering industry. 1.4 Diffusion of flexible automation in the Turkish engineering industry Local production of new technology in the engineering industry is recorded, as a statistical category, under the manufacture of metal- and wood-working machines (ISIC 3823) and of office, computing and accounting machinery (ISIC 3825). The value added in these subsectors fell as a proportion of value added in the non-electrical machinery industry (ISIC 382) between 1983 and 1989, from 6 to 4 per cent for ISIC 3823 and from 0.7 to 0.6 per cent for ISIC 3825. The levels of value added in these sectors were too low for them to be regarded as having

FA AND THE TURKISH ENGINEERING INDUSTRY 333

contributed significantly to the diffusion of FA in the Turkish engineering industry. The two most important machine-tool producers in the Turkish engineering industry, namely Taksan and Tezsan, have been producing CNC machine tools since 1989 and 1990, respectively. Taksan, a Taiwanese-licensed company, exported all of its CNC production to the USA in 1989 and 1990. The data available for these firms indicate that they produced only 9 CNCs in 1991, 27 in 1992 and 58 in 1993. Thus the Turkish engineering industry’s demand for new technology has been met mainly by imports. Changes in the import regime in 1990 are claimed to have encouraged the rise in demand for FA. Besides receiving domestic and/or foreign government credits at convenient terms, investors in machine tools have benefited from reductions in ‘import funds’— additional taxes imposed on top of import duties—and have been completely exempted from customs duties (Eser 1990:9). One of the biggest importers is Yamazaki Mazak (YM). Its sales in Turkey rose from 11 CNCs in 1989 to 57 CNCs in 1993. The manager of YM’s distributor identified automotive assemblers and autopart suppliers as their most important customers for CNC machines. This was related mainly to the remarkable increase in production in the Turkish automotive industry, together with the increased labour costs after 1989. The convenient credit terms recently provided by foreign suppliers have also accelerated the diffusion of FA. 2 The diffusion of flexible automation at firm level 2.1 The case-study firms The empirical investigation of the impact of FA on firms’ performance, focusing on the implications for scale and scope, was carried out in twelve firms which had recently adopted FA, mainly in the form of CNC machine tools, to significant extents. We aimed at achieving a diverse sample in terms of firm size, capital ownership and product range. The sampled firms were selected after a series of interviews with foreign new technology suppliers (such as Yamazaki Mazak and Hitachi Seiki), local CNC manufacturers (such as Taksan and Tezsan), the Istanbul Small and Medium-Scale Industry Development Organization, the Automotive Parts Manufacturers’ Association, Automotive Manufacturers’ Association and a number of engineers from the sector. Autopart manufacturers were preferred to auto-assembly firms due to the simpler production processes involved, so the firms’ performances should more clearly reflect the impact of FA. In addition to the 6 autopart suppliers, 1 general parts’ producer, 2 capital equipment manufacturers and 3 customized-product (mouldmaking) firms were chosen. Although the production process in mould-making is ‘unit production’, in contrast to the batch production in the other sampled firms, the mould-making sector is known to have been adopting FA widely. Of the 12

334 THE COUNTRY STUDIES

firms, 4 are small, 6 are medium-sized, 1 is medium-to-large and 1 is large. Only one firm is foreign-owned, because foreign direct investment in the form of establishing subsidiaries has not been very common in the Turkish engineering industry (Table 11.8). The field study consisted of several visits to firms and long interviews with managers, engineers and in some cases with machine operators. A questionnaire was used for the collection of quantitative data, in addition to the firms’ internal documents, records and annual reports. Comparative data were collected on the firms’ output and production performances, with conventional machine tools and after the adoption of FA (1993, for all the firms), in an effort to assess the impact of FA on product and plant scales. For Family-parts and Vari-moulds there are no comparative data on production performance since both firms were established with CNCs. The main characteristics of the firms are summarized in Table 11.8. They have been grouped according to the segment of the engineering industry in which they are active, with the larger firms in each segment listed before the smaller ones. 2.2 Adoption of flexible automation in sampled firms Autoparts firms There appears to have been little pressure on autopart suppliers to reduce their costs through more efficient production organization until the 1994 economic crisis. Although the effective protection rates for the Turkish automotive industry were gradually being decreased, in the second half of the 1980s they were still around 39 per cent in the cars sector and up to 52 per cent for commercial vehicles, even in 1993. Production costs could be passed on in prices on the protected internal market, and profit margins in this industry were known to be the highest in the Turkish manufacturing sector. Only 2 of the 6 autopart firms started their plant modernization process in the first half of the 1980s, and one of these firms, Auto-casting, does casting and machining for a wide range of parts and components for both the auto industry and the engineering industry. This medium-to-large firm was one of the first in Turkey to incorporate CNCs in its production processes. Its main goal was to increase efficiency in its highly diversified production (around 200 different parts) and to reduce wastage and scrap. After getting acquainted with its first machines, the modernization process in Auto-casting continued steadily from the mid-1980s with more CNCs, including CNC machining centres, and computers for administration, accountancy, purchasing, stock control, etc. However, production lay-out is still along typical functional lines, although 50 per cent of parts are now machined in CNC machines. At the time of the study visits, management had just become interested in learning more about Japanese organizational techniques, including cellular production organization.

Table 11.8 Turkey: main characteristics of sampled firms

FA AND THE TURKISH ENGINEERING INDUSTRY 335

336 THE COUNTRY STUDIES

Source: Firm interviews.

FA AND THE TURKISH ENGINEERING INDUSTRY 337

338 THE COUNTRY STUDIES

Uni-trans was forced to adopt FA in the early 1980s, in order to diversify its product range and increase quality in response to new orders of transmission shafts for the vehicles which were then beginning to be manufactured in the country. As the twenty new products were included in the firm’s product range in the late 1980s, customers’ demands for increased precision and reduced lead times required the firm to purchase more CNCs and to adopt some organizational innovations such as total quality control (TQC) and the just-in-time (JIT) inventory system. These organizational changes, however, seem to have been far from systemic, and its functional plant lay-out has not been changed. The other four autopart firms started to adopt FA in the late 1980s, in rather similar circumstances. After the drastic drop in output during the 1978–80 period of economic crisis, the automotive industry’s output equalled its previous highest level of 141,000 units in 1986. In the period 1986–92, the total output of the industry increased by 143 per cent, reaching 343,000 units in 1992 (AMS 1988, 1993). The dramatic rise in demand for motor vehicles after 1990 was generally said to have forced the autocomponent firms to increase their production capacity. This demand, together with increased labour costs and higher quality requirements for new car models in the late 1980s, led these firms to become the most important buyers of FA. In the face of increasing demand, Trans-axle expanded from a small workshop to a medium-sized plant in 1989. Four Mazak CNC lathes were purchased in 1990, and the satisfactory results obtained led the management to continue the adoption of FA. Although the firm has a relatively large stock of CNCs, this has not been accompanied by any organizational changes. At the time of our firm visits the management was just becoming interested in studying Japanese organizational techniques, as clients’ demands for shorter delivery times were increasing. The modernization processes in firms Multi-liners, Brake-shock and Familybrakes have involved some organizational changes, such as TQC and JIT inventory systems, mostly because of clients’ demands for higher quality and reduced lead times. It was said in the interviews that TQC, which involves delegating quality control to production workers, had been introduced following the adoption of CNCs. The adoption of computers made it possible to apply the JIT inventory system, by greatly facilitating the organization of production. Systemic organizational changes, including cellular plant lay-out, however, seemed to be beyond the reach of these firms, except for Multi-liners, which is the largest firm in our sample. Family-brakes started to incorporate FA by purchasing a CNC lathe in 1987, which greatly improved quality and production capacity. The capability to produce a wider range of brake drums for different models of commercial vehicle led to increased demand from foreign customers. In 1992, in an effort to meet foreign clients’ demands for better quality, the casting section was totally modernized and all twenty conventional machine tools in the machining section were replaced by CNC machines. This also helped to reduce labour costs. The firm’s 100 per cent diffusion density is by far the highest rate in our sample.

FA AND THE TURKISH ENGINEERING INDUSTRY 339

Multi-liners, however, has the largest stock of CNC machines in the sample. All of them are devoted to the production of cylinder liners. The modernization process, carried out in 1990, increased the percentage of machining using CNC machines in the production of cylinder liners to 60 per cent. The production of both cylinder liners and piston rings also was reorganized on cellular lines. This organizational change was said to have reduced the lead time in piston-ring production from 78 days to 44 days, and the lead time for cylinder liners fell by 50 per cent. These changes were accompanied by wage increases and a campaign aimed at obtaining workers’ agreement to reduced employment levels and to ensure a positive attitude on the part of the remaining workers towards change and innovation. TQC was adopted, with the responsibility for achieving better quality delegated to the production workers, who were motivated to participate in quality circles and to ensure JIT delivery to the firm’s customers. These changes helped in reducing lead times. Most autopart suppliers in the sample were not involved in product development and did not participate in their clients’ product design. They work rather from detailed product drawings supplied by customers. Although Brakeshock produces shock absorbers under licence from a German company, it is exceptional in having its own R&D department, which was said to work intensively on product development, although at the time it did not have a CAD/ CAM system. The R&D department was established as part of a significant modernization process in 1991, aimed at increasing the firm’s production capacity and extending its product line. After incorporating FA, the firm started producing very different products such as washing-drying machine dampers, sintered components and complete brake systems. Its sales doubled in a year. Capital equipment firms The rising demand in the market motivated Boiler-pumps and Local-valves to increase production capacity in the mid-1980s. In Local-valves, new and more complex valves and a higher value added product line of spherical valves had created a production bottleneck, because of significantly longer machining times using conventional machine tools. The need to reduce machining times and to increase quality were major incentives for Local-valves to incorporate more CNCs in 1991 and 1993. Since the early 1990s, the firm has increased its product development activities. A CAD/CAM system has enabled it to increase its product diversity and complexity. In fact product development is an important activity for the capital equipment firms. Both Boiler-pumps and Local-valves closely followed the international fairs in their branch of engineering; both firms try to reproduce product developments by backward engineering, using CAD/ CAM systems. High labour costs were said to be an additional factor motivating Boilerpumps to adopt more CNCs in the early 1990s. In a further effort to reduce the cost of labour, Boiler-pumps started to subcontract to some ex-employees some of the simpler production stages in order to reduce production costs, since these

340 THE COUNTRY STUDIES

small subcontracting firms usually do not carry the burden of social security and other social taxes. The adoption of CNC machines within the plants has, however, not been accompanied by any organizational innovations in Boilerpumps and Local-valves. Local-valves did not seem to feel the same pressure to reduce costs as did Boiler-pumps, perhaps because its market share was around 70 per cent. Competition from imported valves was not a major problem because this product line requires suppliers to hold big stocks for immediate delivery, and also because the various import tariffs and customs duties raised the price of imported valves. Family-parts is a small workshop which had started its production activities in 1990 with FA, so that no comparison between production with conventional and CNC machine tools was possible. Its successful performance with FA, however, was thought to warrant special attention, hence its inclusion in our sample. Thirty per cent of the firm’s production capacity has been utilized to produce various spare parts for the Bobbinvar manufacturing company, which is owned by the same family, and the remaining capacity is contracted-out. The firm produces hundreds of different parts and components with just twelve employees, thus fully benefiting from the flexibility of the firm’s CNCs. One technician, who had previously learned to operate and programme CNCs in a medium-sized firm prior to joining Family-parts, seemed to be managing the entire production. CNCs were utilized for 125 hours a week in three shifts, creating 25–30 per cent profitability. According to the company accounts, the break-even point for CNCs is 100 hours’ work per week. The company started to receive orders from export markets in 1992 and now exports more than half of its production. Customized product firms In the customized-products industry, CAD/CAM systems and CNC machine tools have significantly affected the production process in terms of both operation times and precision. Both Glass-moulds and Tin-moulds undertook plant modernization with FA as soon as they thought that they could afford the investment. The customized-products industry has boomed in Turkey since the mid-1980s. With the adoption of FA in the sector, the moulds that used to be manufactured in countries such as Portugal have begun to be produced locally. Glass-moulds, which had more than 500 employees before 1990 but was now only a medium-sized firm, emphasized the importance of the twenty-four-hour machining operations that can be achieved with CNCs. Very small tolerances are required in mould making, and, with conventional machine tools, it was found that it was impossible for workers and supervisors to maintain sufficient concentration in the second and third shifts. Glass-moulds needed to expand its production capacity at the beginning of the 1980s and, instead of 13 conventional lathes and 5 milling machines, it decided to purchase 3 CNC lathes and 2 CNC milling machines. As the incorporation of FA accelerated in the 1989–93 period, Glass-moulds’s production extended beyond the glass moulds and spare parts required by the domestic conglomerate of which the firm is a part. More sophisticated and higher quality products have been produced using

FA AND THE TURKISH ENGINEERING INDUSTRY 341

CAD/CAM systems in the three units of the plant—the glass-mould, spare-part and machinery-production units—which produce also military products, indicating that the firm has met the requirements of AQAP-4 (the NATO Inspection System Requirements for Industry). Tin-moulds, a small mould-making firm, claimed to have greatly improved its quality and reduced manufacturing time by incorporating FA, first in 1988 and more intensively in 1991. However functional lay-outs had been preserved in both Glass-moulds and Tin-moulds, and it was said that Japanese organizational practices are ‘not quite applicable’ within the sector, since production in the mouldmaking industry is ‘unit production’. The third mould-making firm, Vari-moulds, started manufacturing activities using CAD/CAM systems and one CNC machine tool in 1992. Although the firm lacks production experience with conventional machines, its interesting and successful performance with 5 engineers, 2 technicians and 2 workers led us to include it in our case-study firms. After six months of successful production using FA, it encountered a bottleneck in the CNC operations for manufacturing moulds; but, instead of purchasing more CNC machines, the firm started subcontracting work to other small workshops which could not fully utilize their CNC production capacities. The firm’s performance, which it claimed was comparable with a medium-sized mould-making company with 150–200 employees in terms of the quality and complexity of the moulds they were manufacturing, seems to be outstanding in the booming Turkish mould-making sector. The firm also provides an interesting case in terms of the impact of FA on scale, as we will see below. 2.3 Other factors related to the diffusion of flexible automation In adopting FA, all the firms in our sample have benefited from government incentives for investment, such as import tariff reductions, exemption from customs duties, and the availability of domestic and/or foreign credit at convenient terms. These incentives however, have apparently not been very effective in encouraging firms to invest, due largely to the bureaucratic difficulties involved in obtaining the incentives. Information on FA, and more particularly on CNCs, was gathered mostly through visits to trade fairs. All the firms seemed to be very knowledgeable about the latest developments and the advantages offered by FA. Some of the medium-sized and large firms, have sent engineers to other foreign firms which have adopted FA, to see the equipment in operation and fully understand the gains. In general, the decision on the type of FA to buy had been reached by senior managers and technical personnel after careful consideration of the firm’s production process requirements and the technical characteristics of the available makes and options. Back-up service, the reliability of the supplier firm and the price also influenced the selection process. In both the capital equipment firms and the autopart suppliers in the sample, the rate of adoption of FA increased in the 1990–93 period. The FA purchased seems to have been selectively

342 THE COUNTRY STUDIES

incorporated for certain operations or certain products, mainly to launch new products or to increase production capacity and quality. As a result the percentage of operations carried out with CNCs within a firm generally differs significantly from one product to another. In every case, the FA equipment was installed by the supplier firm, and the operation was said to have gone smoothly. One week of training is commonly provided by FA suppliers, mostly to the engineers, technicians and foremen. The training aimed to develop requisite competences in both programming and operating the new equipment; and there was general agreement that the learning outcomes had been adequate on the operating side but only partially satisfactory on programming, although the latter was said to be ‘not too difficult to learn’. Some firms, usually the small firms, had to seek further help from local computer software consultants on programming for production of sophisticated parts. Engineers and foremen who had been trained by the supplier in turn trained the workers to operate the new machines. None of the firms permitted workers to interfere with the programmes, and breaches of the rule would be punished, for instance by dismissal. Evidence in the literature shows that there has been a significant reduction in the skills per worker required to operate the new machine tools, as well as in the number of people required to learn these skills, due to the labour-saving effect of FA. All the firms in our study said that the use of CNC machine tools requires from the workers new attitudes and a technical orientation very different from the traditional manual skills. However, it was also claimed in the interviews that workers experienced in using conventional machines could generally be trained to operate CNCs without much difficulty. Older firms adopting FA do not wish to create resentment among their personnel by firing existing workers and hiring new ones. The firms that did have to hire new workers do not seem to have faced any problems in finding suitably qualified personnel. 3 The impact of flexible automation on scale In accordance with the generally accepted definition of economies of scale (EOS) as ‘reductions in average costs attributable to increases in scale’ (Pratten 1971:3), EOS will be analysed as a relationship between output and costs. This section deals with the production performance—and hence output—side of the equation; the next section deals with the effect of FA on the cost element of this relationship. Since plant output is identical with firm output in our study, we will not consider firm scale.

FA AND THE TURKISH ENGINEERING INDUSTRY 343

3.1 Product scale Machine adjustment times In the interviews, the flexibility of CNC machines—the ease with which the machines can be reset in terms of tools and fixtures, and the positioning of different parts, was mentioned as the most significant advantage of FA. Better quality and reduced machining time were mentioned as the other main advantages. The loading and unloading of workpieces in the machining process, on the other hand, were said to be not much different from using conventional machines, although firms generally had no records of times spent in loading and unloading. The time for setting up for a new product batch varies according to whether the machine-cutting tools or fixtures need to be changed for the new part, and how many such changes need to be made. To make a comparison, machine-tool adjustment times were obtained for each firm for one typical part or production process. These were named in Table 11.8. Table 11.9 shows that, in the first eight firms (which had batch-production processes), resetting times with CNCs were substantially less than with conventional machine tools, the reductions ranging from 53 to 98 per cent. Since CNCs can carry out the numerous separate operations that used to be done by different conventional machines, the number of machines that need to be adjusted is reduced, while the machine adjustment activity itself is much simpler and easier with CNC machines in batch production. Although there is not a single case in the literature of an increase in setting-up times with FA, our two mould-making firms—Tin-moulds and Glass-moulds— reported that machine adjustments take more time with CNCs. Resetting for typical mould-making processes for a tin box at Tin-moulds took 50 per cent more time with CNCs. Glass-moulds reported that resetting times varied significantly from one workpiece to another but, on average, resetting CNCs takes 20 per cent more time than resetting conventional machines. For the firm’s typical product, the glass moulds known as ‘ebisor’, machine adjustment time is 30 per cent more with CNCs. This increase in setting-up times in the mould-making firms seems to be specific to the resetting process in mould-making, in which tolerances are very small so that programming involves very accurate measurements. After the programming has been done in the office, it is first run once with the newly attached cutting tools to check for the required tolerances and to correct faults. With conventional machine tools, the setting-up process ends with the attachment of the cutting tool. The rest is over to the manual skill of the operator, who can correct the tolerances at any time during the machining process. In mould making, the change-over process usually means changing a whole magazine of perhaps ten cutting tools, because the products are completely different from each other, requiring different programming and cutting tools, and different tolerances. In batch production the same cutting tools are often left in the

344 THE COUNTRY STUDIES

Table 11.9 Turkey: changes in machine-tool adjustment time in sampled firms

Source: Interview data

magazine, or two or three may be changed. In spite of this increase in setting-up times, the adoption of CNCs in the mould-making firms was very important because of the high precision obtained by CNCs as well as the reduced machining times, as we will see later in this section. Batch sizes Most of our case-study firms which manufacture in batches work according to the size of the orders they receive, rather than beginning with calculations of economic batch quantities. The same product may be produced in very different batch sizes according to the size of the orders firms receive. Hundreds of differently sized batches are handled monthly, often in a production schedule which cannot be predetermined. In most of our cases, batch sizes are related to the amount which can be produced in one working day. The firms preferred to adjust and programme machines in the morning, and to manufacture the same product for the rest of the day in order to eliminate a batch change during the second or third shift, for those firms working in shifts. Excluding the 3 mouldmaking firms, which produce individual moulds, 5 of the 8 firms said in interviews that CNCs had not had any direct impact on batch sizes. When reminded of FA’s flexibility, they agreed that CNCs greatly facilitate smaller batch size where that is necessary, because of substantial reductions in machine adjustment times. For four firms—Auto-casting, Trans-axle, Family-brakes and Local-valves—batch-size reductions seem to be a potential effect of CNCs, rather than an actual effect. Even Family-brakes, which has a density of FA diffusion of 100 per cent, has not yet made a deliberate effort to benefit from this potential, although, since they do not have records of their batch sizes, there is no way of checking this. The other four firms claimed to have deliberately tried to reduce batch sizes in parallel with increased product diversity, with the aim of keeping inventories as low as possible to improve cash flow. The batch-size records of Uni-trans, Brakeshock and Boiler-pumps in Table 11.10 clearly show that more batches of less

FA AND THE TURKISH ENGINEERING INDUSTRY 345

Table 11.10 Turkey: batch sizes (units and %) beforea and after flexible automation adoption (1993 for all firms)

Source: Interview data. Notes a The before FA adoption date for each firm is given in column 1. b Average batch size for Multi-liners in 1989 was 2,871 units; and in 1993 this was reduced to 1,890 units.

than 100 items were produced after the adoption of CNCs, and less batches of more than 500 items were produced. In Multi-liners, the average batch size in cylinder liner production—for which CNCs were used—has been reduced from 2,871 units to 1,890. This has contributed greatly to the 50 per cent decrease in the lead time for cylinder liner production after the adoption of FA. Most of the batch-production firms in the sample, particularly Brake-shock, Uni-trans and Multi-liners, said that their clients have required increasingly shorter delivery times as a result of their adoption of the JIT inventory system, and this also has forced the sampled firms to reduce their batch sizes. In spite of these pressures however, the firms have not tried to determine optimal economic batch sizes for their products. This may be due to the limited adoption of FA and the difficulties involved in calculating costs. Product diversity and product scale Flexible automation has clearly had a positive impact on the product diversity of the sampled firms, in addition to the substantial increase in product quality. Firms were able to extend their product range to more sophisticated goods and to parts requiring greater accuracy, as well as to new parts for clients’ new products. For some products that has meant vertical integration since it was not technically and/or economically feasible to manufacture some components in-house with conventional machine tools. Moreover, the adoption of CNC machines,

346 THE COUNTRY STUDIES

Table 11.11 Turkey: percentage increases in product diversity

Source: Interview data.

sometimes combined with CAD/CAM systems, enabled the sampled firms to manufacture some components which previously were imported. Trans-axle said that it was possible to manufacture some new products, such as the differential housings, with conventional machines, but that production became practicable only after the adoption of CNCs, due to the high labour costs and poor quality with old technologies. The production of brake discs at Autocasting and spherical valves at Local-valves, became possible only after the adoption of CNC machines. It is difficult to quantify the increase in product diversity, due in some cases to inadequate firm records and in other cases to the difficulty involved in identifying the relevant figures in the available records. Table 11.11 has been compiled from the lists of parts and products provided by the firms and from firm records. In this table, the number of parts machined includes all parts of whatever type, whereas the number of different parts shows only how many different types of component the firm manufactures. ‘Products’ indicates the main types of assembled products, such as oil pumps, hydraulic pumps and submersible pumps (for Boiler-pumps). ‘Product families’ differentiate between completely different types of product, such as rear axles, brakes, valves, shock absorbers, piston rings, cylinder liners, etc. Table 11.11 shows that both the total number of parts machined and the number of different parts produced increased, by between 33 and 650 per cent, respectively, after the firms adopted FA. New products, each comprising numerous different parts, have been added to the product range, as in the case of spherical valves at Local-valves. The number of different products increased in all firms, by amounts ranging between 18 and 233 per cent. Some firms had also begun to manufacture different product families, like washing-machine dampers and sintered parts in addition to shock absorbers at Brake-shock, and differential housings at Trans-axle, showing that these firms have exploited the flexibility provided by FA to switch production between products.

FA AND THE TURKISH ENGINEERING INDUSTRY 347

Table 11.12 Turkey: export performance (% of sales) of sampled firms before and after flexible automation

Source: Interview data.

Although we do not have before and after comparisons for Family-parts and Vari-moulds, they have benefited fully from the flexibility provided by FA. Family-parts was said to have machined approximately 5,000 different parts over the past five years. Four firms had investment plans to extend their product ranges. An increase in product diversity in the other firms was technologically possible and desired by their managements, but its realization was said to depend on demand. At the time of our visits, when the Turkish economy was in the midst of a crisis, the usual local orders had been cut sharply in all of the firms. They were making great efforts to increase their capacity utilization by establishing contacts with foreign firms, and to increase their export manufacturing. As Table 11.12 shows, most firms had increased export sales since adopting FA. Establishing relationships with foreign automotive and capital goods manufacturers from Europe, the Middle East and North Africa has forced our firms to increase their ability to switch production to new products economically and quickly. Since there are no available output data for individual products manufactured by the sampled firms before and after the adoption of FA, it is not possible to systematically analyse the impact of FA on product scale. However, it was said in the interviews that the output of the firms’ main products has increased (or, in the case of piston-ring production in Multi-liners, to have remained unchanged) with the adoption of FA, despite increased product diversity.

348 THE COUNTRY STUDIES

Table 11.13 Turkey: cumulative machining times for typical batch sizes in sampled firms

Source: Interview data.

3.2 Plant scale Machining times Reduced machining time has been one of the most important impacts of CNCs. Separate operations are grouped together in one CNC machine. For example, in Trans-axle, machining front spindles used to involve 3 operations taking 6 minutes on 5 conventional lathes, whereas it now involves just one CNC machine and takes 1.5 minutes of operating time. In Auto-casting, the machining of bus gear boxes required 4 operations taking about 90 minutes. One horizontal CNC machining centre now carries out the whole machining process in one operation that takes only 28 minutes. Cylinder liner machining in Multi-liners has been reduced from 6 to 3 operations, in 2 CNC machines operated by a single worker. The total machining times in the course of a typical manufacturing process for a typical batch size in the sampled batch-production firms (and for typical moulds for Tin-moulds and Glass-moulds) are shown in Table 11.13. According to these figures, machine times for specific machining processes have fallen by 30–88 per cent, depending on the number of operations which still have to be carried out by conventional machines. The reduced machining times also contribute substantially to reducing lead times. Additional data were provided by Glass-moulds, in the form of a list comparing the machining times for operations in the manufacture of 22 standard parts, using CNC or conventional machine tools. The savings in machining time using CNCs for these parts ranged from 46 to 86 per cent, and averaged 68.6 per cent. Thus, in the mould-making sector CNCs offer advantages in machining times in addition to improved quality.

FA AND THE TURKISH ENGINEERING INDUSTRY 349

Table 11.14 Turkey: changes in plant output, production capacity and sales in sampled firms

Source: Interview data

Plant output The increase in machine productivity associated with reduced resetting and machining times led to substantial growth in output. Outputs, measured by the volume of machined castings in tonnes, and/or by the number of parts machined (total number of individual items) increased by varying amounts, depending on the density of diffusion, as can be seen in Table 11.14. Production capacities also increased. It should be noted, however that the increased output and production capacity are related also to other factors, such as specific microelectronics-based equipment other than CNCs (as in Brake-shock) and organizational improvements such as TQC, JIT inventory and cellular/group manufacturing techniques. Multi-liners, for example, increased its production of cylinder liners from 0.63 million units in 1989 to 1.31 million units in 1993, due both to the introduction of CNC machines and the shift to cellular production lay-out, TQC and the JIT system. On the other hand, the organizational changes adopted for piston ring production have not resulted in any rise in output, but they have contributed significantly to reduced lead times, and reduced stocks of semifinished products, which fell from 3.4 million units in 1989 to 1.9 million in 1993, so improving the firm’s cash flow (see Table 11.14). The percentage changes in sales in Table 11.14 are substantially higher than the changes in output volume, except for those at Brake-shock. This exception is explained by the fact that the unit prices of Brake-shock’s new product families, such as washing machine dampers and sintered parts, are much lower than the unit prices of shock absorbers, and the latter have also been under pressure from the firm’ s main customer, a car manufacturer. Competition from imports is felt

350 THE COUNTRY STUDIES

Table 11.15 Turkey: changes in employment and labour productivity in sampled firms

Source: Interview data Notes a=units. b=tonnes. c=hours

more in car manufacturing than in the commercial vehicle sector, where the protection rate is higher. The greater increase in sales in the rest of the firms indicates that the average unit prices for their product ranges are higher since the arrival of FA, because the new products have higher quality, sophistication and precision. For Multi-liners, the discrepancy between the percentage changes in output (4.3 per cent) and sales (55.4 per cent) is related to a change in the product mix. The total output of piston rings has not changed, but the output of cylinder liners, which are manufactured using CNCs, and whose unit prices are higher, increased by 108 per cent. Moreover, the unit prices of the new cylinder liners, which were added to the firm’s product range after the adoption of CNCs, were even higher, and 40 per cent of the increase in cylinder liner production was due to these new cylinder liners. A similar pattern of higher quality and price is reflected in Auto-casting’s new gear boxes, Trans-axle’s differential housings, and Local-valves’s new spherical valves weighing up to 200 kg. Other factors related to plant scale The literature indicates that new plants equipped with FA have smaller scales, although scaling-up is taking place in some sectors, as discussed in detail by Alcorta (1992a). The data from the Turkish case-study firms showed that the FA machines adopted have a much greater capacity than the conventional machines they replace, so that the plant scale of all our sample firms increased. As a result, these firms have been forced to expand their product ranges in order to produce

FA AND THE TURKISH ENGINEERING INDUSTRY 351

Table 11.16 Turkey: capacity utilization rate in sampled firms, 1993

Source: Interview data

more. Employment levels have fallen, however, since each CNC machine replaces 3–4 conventional machine tools, each of them needing an operator. Where cellular production has been introduced, one operator quite commonly operates two CNC machines. The employment data before and after FA in Table 11.15 shows the labour-saving effect of FA in most of our firms. The increase in total employment in four of the firms was due mainly to plant expansions. Trans-axle and Tin-moulds had made investments, in each case to expand production to a larger plant, including a larger building in a different location. Therefore their rise in employment is not associated with the adoption of FA. Firms Uni-trans and Glass-moulds also have expanded their total employment, by 7 and 11 per cent, respectively, due to a general expansion of the production units. Again, this does not reflect increased employment requirements with FA. The adoption of FA has substantially reduced the number of different machines required in the course of a process, so that total machine adjustment times and processing times were significantly reduced and production capacity per working day was much higher with FA, leading to increased labour productivity. The labour productivity in terms of output/worker and sales/worker (Table 11.15) shows clearly that labour productivity has substantially increased in all of our firms, including those with increased employment levels. The high cost of FA equipment pressured the firms to work longer hours, with more shifts, to fully utilize their production capacity. The exception was Localvalves, which reasoned that as CNCs were very expensive machines, only one operator should be responsible for using and maintaining each one. Shiftwork was therefore regarded as undesirable for the FA sections. The firm was fortunate in not having to face competition from imported valves. Table 11.16 shows the reported capacity utilization rates of FA during the hours worked by the sampled firms in 1993. From these figures we have calculated the number of hours that CNC machines were actually utilized in a

352 THE COUNTRY STUDIES

day, and the capacity utilization rates in relation to a plant running three shifts for full utilization. According to the figures in Table 11.16, the firms generally have not been successful in fully utilizing the adopted FA. Low utilization rates can be ascribed to operational problems, bottlenecks in the parts of the production process still using old technology and, more significantly, insufficient market demand. The most common operational problem related to the repair and maintenance of the FA, which has become very complex. Although the FA supplier is generally responsible for repair and maintenance, problems often could not be solved by the local representative. More complex repairs have meant calling in an expert from the supplier’s headquarters in Europe, which involves high cost and a loss of time. The firms that purchased locally produced CNCs complained about the irresponsible attitude of the local suppliers in making repairs, and about the lack of any backup service. Due to these difficulties, the firms seem to have been accumulating the technical capability and experience to solve their own maintenance problems. Some of them contracted the maintenance work to a local computer/software company, and were satisfied with the service provided. Bottlenecks arose because of the limited production capacity of other conventional or special-purpose machines used for particular operations to varying degrees in the firms. If the machine configuration is not optimal for balanced plant capacity, in the light of increased production flows and greater product variety, the firm may be unable to utilize fully the capacity of its FA. Although one of the most frequently mentioned motives for adopting FA was to increase production capacity, the firms appear to have ended up with more capacity than they can utilize, given the level of demand. Managerial and marketing skills are therefore the most important capabilities needed to utilize the FA capacity to the full. Limited marketing capabilities, particularly when orders from their normal customers fall, limit firms’ abilities to overcome capacity underutilization problems. 4 Changes in unit costs 4.1 Changes in the level of unit costs It was not easy to obtain figures on the level and structure of unit costs from the firms, and in two cases no data were made available. Despite our impression that the cost figures that we did obtain were estimates rather than figures determined from an accurate calculation of all the elements of manufacturing costs, and despite the firms’ irritation and reluctance to discuss these figures, our checks showed that the figures were generally reasonable and reliable. The significant changes in prices in the 1980s make the cost analysis even more difficult, especially for the firms that started their modernization process earlier so that there is a long break between the cost figures before and after FA.

FA AND THE TURKISH ENGINEERING INDUSTRY 353

Table 11.17 Turkey: changes in unit cost in sampled firms

Source: Interview data.

The average unit costs shown in Table 11.17 are calculated from total costs divided by total output, either in tonnes of machined castings or in numbers of parts machined (total number of individual items). Since the total number of parts machined includes many very different items with different unit costs, the change in the average unit cost after the adoption of FA provides only an approximate indication of the impact of FA on EOS. It would be better to use the separate unit cost of each product, or of a typical product or part, but this information could be obtained only from Trans-axle, Multi-liners and Tinmoulds. Table 11.17 shows that the average unit cost fell only in Trans-axle, Brake-shock and Boiler-pumps. This may be taken as evidence of increased EOS, since output increased after the adoption of FA. A check of these figures against figures for output, measured in the volume and number of parts machined, confirmed the trend. Whether these reductions in unit cost indicate increased plant or firm economies of scale would require further analysis of the shares of manufacturing and managerial costs in total costs. Trans-axle does just 15–40 per cent of its machining with CNC, so it is difficult to attribute the firm’s unit-cost reduction entirely to the adoption of FA. The difference between the unit-cost reductions for rear axles (1) and front spindles (2) could, however, be related to the fact that 25 per cent more CNC machining time is involved in rear-axle production. Moreover, the capacity utilization rate of FA in Trans-axle (Table 11.16) is one of the highest (65 per cent) of those of our sample firms, which may indicate that FA has had a positive impact on unit cost reduction. The density of diffusion is low also in Brake-shock (25–30 per cent), but the capacity utilization of FA, at 70 per cent, is the second-highest of those of the sampled firms, and output has increased by 141 per cent. The organizational

354 THE COUNTRY STUDIES

changes mentioned earlier also may have contributed to the cost reduction, by reducing inventories. The third firm which achieved scale economies was Boiler-pumps. Again, the density of diffusion is low (10–35 per cent), but the capacity utilization for the FA was 86 per cent. This resulted in a 20 per cent rise in output. Moreover the firms has faced competitive price pressure from imported pumps, and responded by subcontracting some of the simpler production stages, reducing batch sizes, lead times and inventory costs. Unit costs increased in six other firms, and no cost data could be obtained from Uni-trans. At Multi-liners, the unit cost of piston rings (1) increased by 50 per cent (Table 11.17), due, according to the firm, to the increased labour costs and constant production levels, despite the firm having adopted a cellular production lay-out. But the unit cost of cylinder liners, also produced with CNCs, fell by 34. 4 per cent. In this case output increased by 108 per cent, outweighing the increase in labour costs. Auto-casting, which had the largest stock of CNC machines in our sample as well as a 50 per cent CNC density in machining operations, increased its output by 67.5 per cent. However the capacity utilization of its CNCs was only 53.6 per cent. Although no detailed cost structure could be obtained from the firm, it was clear from the interviews that high depreciation costs explained the rise in unit costs after FA. Family-brakes had replaced all its conventional machines with CNCs. Its output increased by 51.5 per cent in terms of the number of parts machined, but the capacity utilization rate was still only 42.9 per cent. Thus, the high cost of FA may be the reason for the 15.3 per cent increase in unit cost. However, the greatly improved quality made it possible for the firm to pass on the increased unit cost and still achieve a 198 per cent increase in sales. Tin-moulds supplied details on the unit cost of one of its typical products, the bottom mould of a 5-kilogram tin box, before and after the adoption of FA. The unit cost of this typical mould fell by 45.4 per cent, but this is not representative of the whole product range since the average unit cost over the total output increased by 68.9 per cent. This suggests that the products that were included in the product range after FA was introduced have much higher unit costs than this typical product (see Table 11.17). Since output increased by 351.9 per cent after the adoption of FA, but the number of different parts increased by 650 per cent (Table 11.11), Tin-moulds had significantly increased its product diversity. Difficulties in manufacturing a new mould for the first time were said to have increased costs. However, the firm’s capacity utilization rate was only 28.6 per cent, so that high equipment costs may have significantly contributed to the increased unit cost. Glass-moulds had a high density of FA diffusion (75–85 per cent), resulting in a 55 per cent increase in output and a 271 per cent increase in sales. Its capacity utilization rate was the highest in the sample, at 95.1 per cent. In spite of these positive outcomes, the average unit cost, in cost per hour worked, increased by 57.4 per cent. This increase, however, seems to be related to substantially

FA AND THE TURKISH ENGINEERING INDUSTRY 355

Table 11.18 Turkey: change in the shares (%) of manufacturing and overhead costs in unit costs in sampled firms

Source: Interview data.

increased overhead costs, as will be seen from the discussion of unit-cost structures below. 4.2 Changes in the structure of unit costs From the structure of unit costs shown in Table 11.18, we can identify the impact of FA on total manufacturing and overhead costs—the latter covering administration, marketing and R&D costs. Although marketing expenses had become more important with the diversification and differentiation of products and intensified export activities, the high share of overhead costs in the total unit cost for most firms was related to increased administration costs. Apparently it is a general practice in the 1990s to pay high fees to the major shareholders, owners, executive board members, etc., since the tax is then lower than the tax on company dividends. These fees appear as administrative costs. The overhead costs of Trans-axle and Tin-moulds were low as a proportion of overhead costs, but these firms have only a small number of owners who are paid wages, which are then accounted as part of the labour costs. Depending on the level of investment in FA, the increased capital cost (as shown in Table 11.19) arising from the high cost and hence high depreciation of FA tools, especially when combined with low utilization rates, has been one of the main factors behind increased production costs. Except for Local-valves and Glass-moulds where there were decreases, capital costs have risen in all the firms by 1.4–66.7 per cent. The decreased unit capital cost at Local-valves was difficult to explain, since the firm works only one shift with a capacity utilization rate of a mere 33 per cent. It appears that the very substantial rise in overhead costs (54.5 per cent), related to high payments to shareholders, has had the effect of lowering the share of manufacturing costs in the unit cost. In the case of Glass-moulds, capital cost fell due to the 95 per cent utilization rate of CNCs,

356 THE COUNTRY STUDIES

Table 11.19 Turkey: change in the shares (%) of capital and labour costs in unit costs in sampled firms

Source: Interview data.

working in three shifts. Moreover, the significantly increased share of administration costs (up by 105 per cent), which related to the increase from 34 administrative personnel in 1983 to 52 in 1993 (excluding high-level managers), affected the share of capital costs. The increase in administrative personnel was said to be related to the intensification of export activities, which have not been accounted as marketing costs by the firm. The share of labour in the unit costs of the sampled firms fell at Brake-shock, Family-brakes, Glass-moulds and Tin-moulds. At Trans-axle the share remained constant, (although it was very high to begin with). This firm had expanded its plant and employment significantly at the same time as it had adopted FA; moreover, Trans-axle paid high wages to its owners, so that the before and after comparison is hardly meaningful. Labour costs increased as a share of unit costs at Multi-liners, Uni-trans, Boiler-pumps and Local-valves. Boiler-pumps and Multi-liners, each of which had reduced employment levels (see Table 11.15) but had seen a rise in the share of labour costs, ascribed this to the ‘astronomical’ wage increases obtained through collective bargaining by trade unions in 1989 and 1990, which increased wages by 150–250 per cent. As was shown in Table 11.3, real wages in the Turkish manufacturing industry increased by 25.2 per cent in 1989, 18.5 per cent in 1990 and 32.8 per cent in 1991. From 1989, when many of the case-study firms began investing in FA, to 1991, real wages increased by 57.3 per cent, explaining the rise in labour unit costs. These wage increases were a major factor in firms’ decisions to invest in FA— because its operation requires fewer workers. Two other factors behind the rising labour costs were mentioned by the firms. One was the higher wages of CNC operators (despite the arguments in the literature that less skills are required per operator for the new machine tools). Newly employed CNC operators were both better qualified educationally (mostly technical high-school graduates) and better paid. The other factor related to the substantial increase in training and educational costs, particularly in those firms that had combined organizational innovations

FA AND THE TURKISH ENGINEERING INDUSTRY 357

Table 11.20 Turkey: changes in the raw materials’ share in unit costs and in scrap rates in sampled firms

Source: Interview data. Notes a=piston rings. b=cylinder liners.

with the adoption of CNCs. For example, according to the firm records of Multiliners, training programmes averaged 3.4 hours per employee in 1991 and 90.8 hours per employee in 1993. These systematic training programmes were essential for the success of organizational innovations such as TQC, JIT, quality circles and cellular production lay-out. Raw-material costs, as a share of unit costs, have fallen in all the sampled firms for which comparative data were obtained, by amounts varying from 0.4 to 52 per cent (Table 11.20). One major factor was said to be the significant falls in scrap rates, ranging from 11 to 96 per cent, due to the more precise machining capabilities of FA and hence substantially reduced rejection and wastage rates. Table 11.21 shows the scrap rates before and after FA for the firms which had kept records. Another factor contributing to the fall in raw-material costs as a share of unit costs was the fall in the real prices of raw materials, the most important of which were those for cast grey and modular iron and steel, from the early 1980s to 1993. The fall in the share of raw materials is less for the firms which adopted FA in the late 1980s or later. Energy costs, as a share of unit costs, increased slightly in 3 of our firms and fell in another 5, as shown in Table 11.21. One of the main problems in using FA mentioned in the interviews was the cost (including the time lost) of repairing the new machine tools (Table 11.21), as discussed in section 2. The maintenance and repair service given by the suppliers was generally found to be unsatisfactory. Repair costs, as a share of unit costs, had increased substantially, and it was often necessary to bring in an expensive expert from Germany, which involved costly delays. The increase in repair costs was highest at Brake-shock, because it is located 250 kilometres from Istanbul.

358 THE COUNTRY STUDIES

Table 11.21 Turkey: changes in the share (%) of energy and repairs in unit costs in sampled firms

Source: Interview data.

5 Conclusions An analysis of the data on production performance in the case study firms confirms the findings in the literature concerning the advantages of CNCs over the conventional machine tools which they replace. Since CNCs can perform a series of operations by automatically changing tools in the spindle, instead of shifting the part or product from one conventional machine tool to another, they have greatly increased machine productivity. Other advantages of CNCs include reductions in setting-up time and machining time, quality improvement, savings in labour and floor space, increased product mix capacity, and reduction of rejects and waste. Since the FA machines used by the firms have a much greater capacity, due to the greater machine productivity, than the conventional machines that they replace, our sample firms have been forced to produce more. Therefore there has been a definite increase in output in all of our sample firms. They have been trying to utilize their increased production capacity by expanding their product diversification and their intensified export activities, linked to greatly improved product quality. The advantages of FA, as manifested by the firms’ improved production performances and hence increased output, do not seem to have enabled these firms to realize large economies of scale. Yet lower unit costs in four firms show that they have been able to generate economies with increased scale. It is noteworthy that, with the exception of Glass-moulds; the firms in which unit costs had increased markedly all had low utilization rates. It should also be kept in mind that the unit-cost figures we obtained are evidently estimates of average unit costs over many different items. The significant changes in relative prices in the Turkish economy in the 1980s also make the cost analysis more difficult.

FA AND THE TURKISH ENGINEERING INDUSTRY 359

Three reasons emerged for underutilization. The first and most important relates to insufficient domestic demand, which meant that it was not possible to utilize the FA to its full capacity by working three shifts. The second relates to the repair and maintenance of the new equipment, but this is becoming less important as the firms accumulate experience and the technical capability to solve many repair problems, and as supplier firms improve their back-up service. The third factor relates to the ‘selective’ incorporation of FA: the firms have not gone through comprehensive modernization in which all the elements of the production process are in balance. Old machinery used in carrying out certain operations can create bottlenecks preventing the full utilization of FA. These problems of underutilization and production bottlenecks suggest that new small firms with a balanced array of machines, including CNCs, may have a lower optimal scale. The two small firms have in fact had outstanding physical and economic performances, with a high utilization ratio for their FA machines. But since we do not have comparative cost figures for these two firms, we cannot use them as evidence of changed EOS as a result of the adoption of FA. Further research which fully explores the cost and output characteristics of new plants and established old plants in a particular sector would increase our understanding of the impact of FA on EOS and optimal scales. In the literature it has been stated that there are organic links between the diffusion of microelectronics-based FA and new organizational practices. The adoption of numerically controlled (NC) technology without appropriate organizational changes may not have significant effects on economies of scale. But in our sample only one large firm (Multi-liners) had tried systematically to adopt organizational innovations together with NC technology. It seems that organizational changes are still beyond the small and medium-sized firms, although the firms’ managements were becoming increasingly conscious of these innovations. Thus there may well be room for further improvements as the level of FA diffusion increases. Although the literature suggests that international competitiveness for developing-country firms depends on the scaling-down effect of FA, our findings indicate that firms have achieved substantially better competitiveness in spite of increased scales, even with the partial incorporation of FA. The advantages provided by FA enabled them to improve their prospects in export markets, which in turn helped to increase their FA utilization rates. The adoption of FA was therefore essential to achieve competitiveness in order, in turn, to meet the quality requirements of highly competitive export markets. This suggests that developing countries may achieve international competitiveness by adopting FA as they increase their technological and managerial capabilities and accumulate international market experience. This may lead to the establishment of more industry, provided of course that the economy is stable and macroeconomic conditions favour investments in FA. The resulting acceleration of FA diffusion may in turn lead to a change in the pattern of location in some industries that have been dominated by firms from the more industrialized countries.

360 THE COUNTRY STUDIES

Note 1 The output of the engineering industry as a share of total manufacturing industry output is about the same as its value added share, and is also low compared to other developing countries.

12 THE IMPACT OF FLEXIBLE AUTOMATION ON SCALE AND SCOPE IN THE THAI ENGINEERING INDUSTRY Peter Brimble1

1 Introduction: industrial development in Thailand 1.1 The industrialization process In the early 1960s, when Thailand began the industrialization process, industrial policy emphasized import substitution of consumer goods such as textiles, automobiles and household appliances. In the late 1960s and early 1970s, in the face of balance of payments’ crises, the government shifted towards an export promotion strategy. This has essentially continued to this day, although recent moves do encourage production of parts and components for the local market. The industrial sector grew extremely rapidly throughout the 1960s and 1970s and greatly increased its share in GDP. The world recession of the early 1980s slowed Thailand’s industrial growth, but the macroeconomic reforms that were implemented to combat severe macroeconomic imbalances set the stage for the economic boom period of 1987–1993. Driven to a large extent by foreign investment and supported by rapid export growth, the economy grew by over 10 per cent per year from 1987 to 1991, and by over 8 per cent in the three years to 1994. The share of the industrial sector has grown by around 1 percentage point per year for the past six years, to around 30 per cent of GDP in 1994. On the down side, the extremely rapid growth has created tremendous problems of infrastructure supply, manpower shortages, environmental degradation, and income distribution. These in turn have led to increasing costs of production across the board. The key challenge to Thai producers in the mid-1990s is to enhance production capabilities and move up the value-added tree, as the country faces competition from newly emerging countries, especially China, India, Indonesia and Indochina. The economy can no longer rely on cheap supplies of labour or land, and producers must find ways to address the shortages of skilled labour.

362 THE COUNTRY STUDIES

1.2 The machine-tool industry In the 1960s and early 1970s Thailand imported machinery parts and other equipment to produce consumer goods, initially for domestic consumption and later for export. This path of industrial development led to a high dependence on foreign technology and an unfavourable balance of overall trade, compounded by high imports of capital equipment. From the Fifth National Development Plan period (1982–86), the Thai government therefore shifted emphasis towards supporting the development of capital goods industries, including the machinery industry, to strengthen the industrial structure and bolster linkages with other sectors of the economy. Machinery development has been recognized as a crucial component in Thailand’s plan to cultivate industrial linkages, absorb technology and exploit technical skill spillovers from foreign-owned to Thaiowned enterprises. The machine-tool industry is regarded as the key industry in the capital goods sector. This emphasis on support for the engineering sector continued into the Sixth and Seventh Plans, although it is not always evident that the policy intentions expressed in the Plans actually filter down into effective policy measures. Since the early 1970s, a number of Thai firms have started manufacturing simple machine tools such as lathes, presses and powered hacksaws through reverse engineering, while importing vital parts. All machines are for local applications, and the product range and quality have changed very little since the 1970s. In the 1980s, the rapidly growing local demand for machine tools to support the tremendous growth in the industrial sector had a minimal impact on the growth of the domestic machine-tool industry, since the local supply could not deliver the quality required. Machine-tool imports rose from 791 million baht (US $39 million) in 1980 to 2,177 million baht ($80 million) in 1985, an annual growth (in baht) of 22.5 per cent. The high-growth items were metal-cutting machines, planing and shaping machines, lathes and grinding machines—most equipped with NC or CNC features. In contrast, exports of machine tools dropped from 36 million baht (US$1.76 million) in 1980 to 16 million baht ($.6 million) in 1985. From 1986 to 1991, a number of new foreign and Thai firms started manufacturing machine tools in Thailand while the older Thai manufacturers continued to stagnate. Most of the new firms aimed at manufacturing computercontrolled and higher-end products for export, although government policy allowed a certain percentage to be supplied to the domestic market. In October 1990 the government reduced the import duty on most machinery to 5 per cent. Although firms promoted by the government had previously received duty exemptions on imports of machinery, this new policy extended the low import duties on machinery to all manufacturers, something which has doubtless permitted many small and medium-sized suppliers to upgrade their capital stock. Most machine tools used in Thailand continue to be imported.2 Imports of machine tools jumped rapidly from 6.8 billion baht (US$269 million) in 1988 to

FA AND THE THAI ENGINEERING INDUSTRY 363

17.2 billion baht ($677 million) in 1992, showing a much greater growth rate than in the first half of the 1980s. Driven by the handful of foreign producers, exports also grew over the same period, from 481 million baht (US$19 million) to 1,102 million baht ($43 million). These exports included computer-controlled and higher-end machine tools such as electrical discharge machines (EDMs), grinding machines and machining centres. Two features of the local machine-tool industry warrant mention: first, the machine-tool industry has not grown in line with other industrial sectors over the last two decades; and, second, the number of local manufacturers, along with their production capacities and product ranges, have shrank over the years. It would appear that the underlying problem is technological rather than economic. Local machine-tool makers were not able to make the transition from conventional and simple machine tools to the CNC and sophisticated machine tools for which there has been increasing demand in the 1980s and 1990s as Thailand’s industrial structure became more sophisticated. This was due to weak technological capabilities among Thai tool manufacturers, especially in terms of metallurgical knowledge, process technology and design capability. These are precisely the areas in which Thai industry needs to develop in order to compete in the future. 1.3 The policy climate and institutional support for flexible automation3 The Seventh Plan (1992–6) differed from the previous six Plans by shifting the emphasis of scientific and technological (S&T) development from the public sector to the private sector. This Plan indicated that the three successful outcomes of past S&T efforts were: the creation of an environment supportive of scientific and technological development; the promotion of research and development in the public and private sectors; and the production of more S&T personnel. The R&D expenditure target was increased to 0.75 per cent of GDP: 0. 5 per cent from the government and 0.25 per cent from the private sector. The Plan targeted certain industries seen to be crucial for Thailand’s development in the next five years, including metal-working and machinery and information technology. There are no significant government restrictions on the import and application of CNC machine-tools or computers. Tariff rates for the final product stand at 5 per cent, although tariffs on spare parts are 35 per cent and on some parts and components may be as high as 40 per cent. On the investment side, the Thai Board of Investment (TBOI) has long provided investment incentives to firms producing machine-tools, including a number of CNC manufacturers. In May 1994, BOI identified ten activities which directly influence the development of supporting industries in Thailand, and which would receive increased investment privileges. These activities include cutting tools, grinding tools, machining centres, and other engineering and metalworking activities. Companies granted TBOI privileges in these areas receive an

364 THE COUNTRY STUDIES

eight-year corporate tax exemption and import tax reductions depending on company location (areas further away from Bangkok receiving greater privileges). A key characteristic of the policy environment is the weakness of the support afforded by institutional mechanisms to science and technology activities in general and the introduction of flexible automation in particular.4 While the institutional base supporting science and technology development has been strengthened in recent years, Brimble and Sripaipan (1994) conclude that: The S&T system in Thailand is characterized by no clear determination of responsibilities, resulting in duplication of effort and low effectiveness in implementation. Moreover, the formal system, represented primarily by the Ministry of Science, Technology and Environment (MOSTE), focuses too much on domestic R&D, as opposed to the broader elements of technology policy, including acquiring foreign technology, using and diffusing technology efficiently, and developing an appropriate technical human capital base. To a large extent this conclusion applies also to the institutional support for the application of FA. Although a number of agencies, notably the National Electronics and Computer Technology Centre (NECTEC) under MOSTE, the Department of Industrial Promotion under the Ministry of Industry, the Federation of Thai Industries, and a number of technical universities5 have provided financial and technical support to firms implementing FA, the resources have been extremely small in comparison to the demand and need. In sum, government support to FA development has not been very effective to date. The policy measures have been fragmented and have not been co-ordinated with the needs of the production sector.6 As Brimble and Sripaipan (1994) point out, for overall science and technology policy the government’s new technology support programmes need to be incorporated into a coherent and well-articulated strategy to improve the technological level of the manufacturing base. 1.4 Applications and diffusion of flexible automation in manufacturing Since no systematic analytical research on the use of FA has been carried out, this section draws primarily on interviews with suppliers7 and informed individuals in the various institutions concerned. It also utilizes information from two product studies (IMRS 1993, 1994) carried out by Industrial Market Research Services, a private market research company, and a sector overview by TDRI (1992). CNC machines have been used in some companies in Thailand for many years, especially by autocomponent manufacturers which have bought used machines from Japan. They began to be introduced in significant numbers in 1988, and

FA AND THE THAI ENGINEERING INDUSTRY 365

there was a huge jump in the number of CNC machines in use in Thailand in 1989–90. In 1993 the market for CNC metal-cutting machine-tools in Thailand was around US$23.4 million, with an average annual real growth rate of approximately 6 per cent. Imports accounted for 85 per cent of supply, with the remaining 15 per cent supplied by domestic manufacturers—principally EDMs made by the Japanese firm Sodick, and Okamoto grinding equipment. Japanese CNC metal machine-tool manufacturers hold a 70 per cent share of the market, with Taiwanese manufacturers a distant second with 7.36 per cent. This is because the majority of the manufacturing investment in Thailand is of Japanese and Taiwanese origin. CNC metal cutting machine-tools are concentrated in the automotive and motorcycle parts and the mould and die industries. In both of these sectors, the largest firms are Japanese or Thai-Japanese joint-ventures. In addition, many companies previously used Japanese manual machine-tools. US and European brands are less well-established, partly because of higher prices and the perception that the aftersales service and spare parts availability are not as good as for Japanese brands because they have to come a greater distance. However Maho Decker (a European brand) was said by users to be competitive. Machine tools are sold through general trading companies or specialist machine-tool representatives. General trading companies have machine-tool divisions within their organizations, and these represent several manufacturers. Suppliers provide installation and training on how to operate, program and do simple maintenance on machines. These services are included in the sale price. If training is not needed, the price may be reduced slightly. Suppliers and users mentioned Okuma as having the biggest market share in Thailand. Marazaki also is big: its Japanese trading company, Yamazen, was the first into the Thai market. Mitsubishi has been successful in the EDM field, and will start promoting CNCs and machining centres through other agents in late 1995. Yamazaki Mazak, a latecomer to the market, is probably in fourth place in the Thai CNC market. MSM Co. Ltd is the local Yamazaki distributor for government purchases, while Numac Machinery Ltd handles commercial business. Although users’ experience shows that the level and depth of service for CNC machine-tools leave a lot to be desired, local servicing and spare-parts support has grown. Marazaki, Hitachi, Mitsubishi and Mazak all have exclusive servicing representatives in Thailand. Some of them also offer service and maintenance contracts after the warranty period. Robots are very new to production in Thailand.8 However, as companies begin to utilize more CNC machines, processing heavier metal and more sophisticated products, more firms are considering the potential of robots. A study on industrial robotics in Thailand conducted in 1993 found that 21 out of 50 selected potential users, mostly in the automotive and autoparts industry, were using robots in production. Seven companies use robots in arc-welding applications, and only three companies use robots in material handling. This is mainly because using robots in welding can help improve both quality and working conditions, while using robots in material handling can help only in improving efficiency. The

366 THE COUNTRY STUDIES

main reason given by these companies for introducing robots into their production process was to improve quality. Secondary reasons were to upgrade technology and improve productivity. However many companies remain hesitant over purchasing a robot because of the high investment and uncertainty about the service and advisory capabilities of local suppliers.9 The relatively low labour costs in Thailand are another factor explaining the low use of robots. Robot use will be encouraged by improved education in the use of FA, increasing labour costs, increasing sensitivity to working conditions, and greater quality requirements. There is a trend for manufacturers in Thailand to use more robots in arcwelding and painting work, especially in the automotive and autoparts industry. However, the fact that several high-profile users of robots have had less than successful experiences will to some extent act as a brake on robot use. 1.5 Factors affecting diffusion The principal motives for adopting FA were quality and efficiency considerations, the shortage of traditionally skilled labour, a desire for greater product diversity and flexibility, and cost considerations. Machine suppliers stressed the importance of Japanese companies’ insistence that parts subcontractors should use CNC machines. CNC machines are seen increasingly as a necessary membership-card to the parts and components market, principally for quality reasons. Their superior precision was widely agreed by users. Producers spoke of selectively adopting CNCs in their production processes, starting with the products with the highest quality specifications, or with the most urgent operations. Most, if not all, continued to operate both older and new technologies concurrently in the production process. Users reported that skilled manufacturing labour was becoming difficult to find in Thailand, and that CNC equipment could be used to overcome this shortage. CNC machines are quite simple to operate and require less skill and training than do manual machines. Workers can be trained on the job and do not need specialized prior knowledge. They usually have only to push a start button and load or unload work pieces. They do not have to know how to program.10 Policy factors also have played a role, and could play a greater role. The government has recently established a National Supplier Development Programme (NSDP), which develops and extends the backward linkage development programme (BUILD) of the TBOI and related programmes of other public and private agencies. The NSDP draws on experience from other countries and has strong support from the government. Under the NSDP, the small and medium-sized enterprises (SMEs) sector is seen as the roots of Thailand’s industrial tree, and a number of the NSDP programmes relate to new technologies in the SME sector. These include capability-building programmes such as the promotion and rationalization of training facilities, especially in areas such as precision machining and statistical quality-control techniques, assistance

FA AND THE THAI ENGINEERING INDUSTRY 367

in achieving the ISO 9000 standard, and the provision of experts for technical assistance programmes (Brimble and Sripaipan 1994: Annex 7). One consideration against adopting CNCs, mentioned by both the machine suppliers and the manufacturers, is the high production volume required to cover the high capital investment. Users identified lower machine prices as the factor that would most encourage the diffusion of CNC machines. 2 Evidence on diffusion from the case studies 2.1 The firms studied11 The research guidelines indicated that the companies to be interviewed should be those which had introduced FA into their production processes in the not-toodistant past, having previously used conventional technologies. Given the relatively low degree of sophistication of Thai firms and the fact that many foreign investors in recent years have started their production activities using FA, the search for sample companies proved rather difficult but also informative. Starting from a rough selection of 50–60 companies in the automotive and metalworking sector12 that were judged to be ‘modern’ and likely to utilize FA (i.e. CNC machine-tools), 30 firms were approached to participate in the survey exercise. Taking into account the 4 firms that refused to participate in the survey, the selection process yielded 11 companies for subsequent interview. An overview of the selection process is presented in Table 12.1. The selection process revealed that many medium-sized and large companies used CNC machine-tools primarily for producing or maintaining moulds and dies. Fewer companies used the machines in the production process itself. It was difficult to identify small companies (employing less than 100 workers) to be interviewed, although machine suppliers indicated that numerous very small firms had bought one or two CNC machine-tools. The market is very segmented: CNC machine-tools are used either by medium-sized to large modern firms or by very small machine shops.13 Many of the larger companies that were using CNC machines for production had started production using CNC machine-tools, so no comparison between before and after FA adoption would have been possible. Even the companies which had shifted from traditional to flexible automation had done so gradually, again making before-and-after comparisons rather difficult. Some industry-specific characteristics also became evident during the selection process. Companies producing valves and pistons do not use CNC machines in production. They use only semi-automatic and automatic machines, since the various products produced are broadly similar except for the dimensions. Companies producing engine blocks are large and have used CNC machines for certain processes from the beginning, because of quality requirements.

Source: Survey questionnaire.

Table 12.1 Thailand: summary information on companies approached for interview

368 THE COUNTRYSTUDIES

FA AND THE THAI ENGINEERING INDUSTRY 369

Profiles of sampled companies The eleven companies surveyed represent a range of product groups and production processes. The main characteristics of the firms are shown in Table 12.2. General sample characteristics Eight companies are in automotive parts manufacturing, two companies manufacture machinery parts, and one company makes sanitary fittings. By ownership, 6 companies are 100 per cent Thai and 5 companies are jointventures with Japanese partners. No interviewed companies were 100 per cent foreign, since none of the foreign companies approached had used manual machines prior to introducing CNC machine-tools. One company is small (up to 100 employees), five companies are medium-sized (100–500 employees), and five companies are large (500 and more employees). Nearly all the companies produce for the original equipment manufacturing (OEM) market in Thailand. Of the 11 companies interviewed, 6 use aluminium diecast or cast-iron products as a raw material, 2 companies use both castings and metal rods or sheets as raw materials, and 3 companies use only metal rods or sheets. Four companies have their own foundry shops, while another four subcontract all casting work and focus on machining operations. One company subcontracts significant amounts of metal-plating work. Only 4 companies (Part-castings, Thai-brakes, Gear-shafts and Cylinder-liners) are engaged exclusively in machining. The remainder have many other production operations, such as painting and the assembly of final products. Most factories are laid-out by product lines or by cells, but Thai-brakes, Plumbingparts, Thai-pumps and Gear-shafts arrange their production lines by process. In most firms the quality-control process consists of two steps: workers’ rough inspection of appearance, and a more systematic dimension check. Most companies have quality-control both during the processing stages and at the final stage. The most-cited process-related problems of the interviewed firms concerned the low quality of raw materials or other subcontracted work, and the general shortages and high turnover of skilled labour. 2.2 The adoption of flexible automation: motivation, selection, integration and level of automation Motivation When asked about the main reason(s) for adopting FA (see Table 12.3), 6 of the 11 firms cited greater productivity and higher quality as equally important primary reasons. One company reported that a CNC production line was installed

370 THE COUNTRY STUDIES

purely to produce high-precision parts (i.e. quality), while three others reportedthat the prime factor was simply productivity and improved competitiveness. Onefirm (Thai-pumps) producing products in an uncomfortable climate cited the needto improve working conditions as the primary reason. Nearly all companies, especially those supplying Japanese assemblers, reported that the market increasingly demands precision and quality that can be achieved only by utilizing CNC machines in certain processes.14 This was especially the case for export products (both direct and indirect), where CNC machining centres were often employed to carry out special operations, and also for some automotive companies looking to move up-market or to supply to more demanding customers.15 Most of the companies facing problems with hiring and keeping skilled labour did find that the introduction of CNC machine-tools reduced the need for skilled workers. However, the determining factor was labour availability rather than cost. Selecting flexible automation CNC machines are much faster than traditional machines for complex processes, since the CNC machine can perform many operations in one load. CNC machinetools are least likely to be used in short and simple operations such as drilling, since it is cheaper and faster to use manual or automatic machines. This is especially the case for those products that require the same operation with only dimensional changes, such as pistons and valves, where automatic machines are used more effectively. Manual machines can also be used for simple and low quality products. CNC machines are more flexible than automatic machines, but if volumes are very small the manual machines may be more flexible. While automatic machines can be effective in carrying out some specific operations for specific types of product, they are much less flexible and less amenable to switching to different products or processes. The CNC machines most used in the production processes are lathes, milling machines and machine centres. Other varieties of CNC machine are used in the mould and die industry, such as more sophisticated machine centres, EDM and wire-cut machines. Since the majority of the automotive and motorcycle assemblers in Thailand are Japanese-owned or Japanese-influenced, Japanese customers, partners or advisers were the major source of recommendations of or requirements for the use of CNC machines. These same companies were often the main source of information on which CNC machines to install, but the suppliers of machine-tools were seen increasingly as a major source of advice in selecting the appropriate machine and even in advising on a long-term strategy for the gradual introduction of FA into the production process. Joint-venture companies usually choose the same model of CNC machine as is used by affiliates abroad, especially when there is a local agent for that brand in Thailand to provide service and spare parts. If there is no local agent for that brand, some companies might change to other equally good brands that have a local agent. Smaller local companies usually buy from local agents, whereas joint-venture companies usually import directly.

Table 12.2 Thailand: characteristics of sampled firms

FA AND THE THAI ENGINEERING INDUSTRY 371

Source: Survey questionnaire. Note a In a company preceding Part-castings

372 THE COUNTRY STUDIES

Source: Survey questionnaire data. Notes a Levels of automation as defined in Kaplinsky (1984), and discussed in the next section. b Density of automation is a rough estimate of the share of CNCs in total production capacity.

Table 12.3 Thailand: reasons for introducing flexible automation in sampled firms

FA AND THE THAI ENGINEERING INDUSTRY 373

374 THE COUNTRY STUDIES

The criteria used in selecting CNC machines include the intended operation, and the accuracy, speed and output volumes required. Each CNC brand has a different speed and level of accuracy. Production work where accuracy is not that important usually requires fast machines. Mould and die making require greater accuracy and heavy-duty machinery, because the materials used are very hard. Price, local availability of spare parts and aftersales training and service are important factors in selecting brands. Machines with local agents are preferred because they are convenient sources of spare parts and service. Distributors will service CNC machines of other brands where possible, but will not guarantee the results. Integration in the production process Most companies did not scrap their traditional or automatic machines, if they were in working order, when they introduced CNC machine-tools. In general, they would move the old machines to a low-volume or low-precision production line, and install the new machines in the places vacated. In some cases they continued to operate old machines in parallel with a CNC machine-tool, creating a split-production line. Staff training and support were provided by the suppliers. Competition among CNC suppliers in the Thai market is sharper now, and suppliers know that they must maintain good relationships with customers as most will buy one or two CNC machines at first and then repeat the order, to gradually replace their conventional machines with CNCs (although some companies do buy larger numbers of CNC machines on the first order, especially where the company’s management has visited other companies abroad). As a result of the competition, Thai suppliers offer a wider range of support services than do suppliers in some other countries. One supplier noted that in other countries it is usual for suppliers to give a once-only one-week training course to teach existing operators to work the new CNC machinery, but in Thailand they have to repeat the course several times. This is because the staff turnover in Thailand is high, so there are always new employees to be trained, and because it takes some time for these workers to switch from conventional machine-tools to CNCs. From observation, it appeared that more women were being used for simple CNC machine-tending. Programmers were generally recruited, although some firms said they provided programmer training as a way to upgrade existing operators’ skills and motivation. In most companies, production engineers were able to propose modifications of the production process using more CNC machines, with approval from the managing director and advice from foreign advisers. But the semi-enforced installation of CNC machines in part of the production process often led to problems resulting from the operators’ lack of understanding of the machine and the way it was inserted into the production process.

FA AND THE THAI ENGINEERING INDUSTRY 375

Level of automation Only 2 companies were using robots in production. Omni-parts used a robot in arc-welding applications, while Part-castings was using a robot for material handling and transferring workpieces between workstations. Part-castings purchased its robot almost two years prior to the study, but were not fully satisfied with the experience at the time of the interview. At first the firm used the robot simply to educate its engineers about modern technology and automation, assigning the engineers to program and observe the robot at work. Then the robot was moved to serve as a raw-material feeder to the machining process. The factory manager noted that Part-castings had plans to buy an application program from the supplier in the near future in order to use the robot more effectively. One company (Gear-shafts) has machines with automatic loaders, and the rest use manual raw-material feeders. No company runs its machines, in any shift, completely unmanned. Companies in Thailand have not implemented complete automation systems since labour is relatively cheap and investment costs are significant. In certain cases, CNC machines have been developed to automate some operations in the machining process. Six companies (Omni-parts, Part-castings, Thai-brakes, Electrical-auto, Plumbing-parts and Thai-pumps) have been using CAD/CAM. Only three companies (Motor-parts, Thai-brakes, Cylinder-liners) had retrofitted programmable logic controllers (PLCs) in old machines. Thai-diesel does use automatic machines equipped with PLCs, but these are purchased as such rather than being retrofitted. The automation in the Thai firms interviewed remains essentially at the level of intra-activity automation, the first level in Kaplinsky’s three-level structure (1984). Intra-activity automation occurs when new technologies are introduced in individual activities within one of the spheres of production. In manufacturing, for example, this could involve the introduction of a CNC lathe in the forming activity; in design, the introduction of a CAD program in the basic design phase. Kaplinsky’s second level of automation, the intra-sphere level, involves new technologies which have links with other activities within the same sphere. In manufacturing, for example, this occurs when the raw-material feed process into a CNC machine is carried out by a robot; in co-ordination, it might involve the introduction of a local area network which links together the purchasing and the sales departments to permit better planning of inventories. The third level of automation, the inter-sphere level, occurs when new technologies are utilized to co-ordinate activities within different spheres of the production process. Examples would include integrated CAD/CAM, involving direct links between design and manufacturing, or computer links between sales (customer demands) and the specifications of the individual products being produced on the production line (features in an automobile). Thai firms have yet to introduce innovations that link together distinct activities within each production sphere. The use of CAD in the design process marks the beginning of the second level of automation in several companies.

376 THE COUNTRY STUDIES

Some preliminary attempts at CAD/CAM integration, which does involve an element of inter-sphere automation, have been observed, but this is generally limited to the production of moulds and dies rather than being applied to the manufacturing process itself. CAD/CAM integration in the manufacturing process was being considered by some firms.16 Two companies, Gear-shafts with its automatic loaders17 and Part-castings with its robot machine-tender, have introduced FA to link together the rawmaterial feed and the forming/machining activities within the manufacturing sphere. As such, they have moved to the second, or intra-sphere, level of automation. And many of the firms interviewed were in the process of introducing computer systems that may eventually link together several activities within the co-ordination sphere of production. An interesting story of a near purchase of a robot, and a potential move to the second level of automation was found in Thai-brakes. The company recently introduced a new production line to machine brake drums for the truck-assembly sector. The firm purchased four CNC machining centres to perform the various steps of the machining process. Since the casted part is rather heavy, over 40 kilograms, the factory manager considered the use of a robot as a machine server to take the non-machined part from the trolley, pass it from machine to machine for machining, and then return it to the trolley. This would have reduced the manual elements of transporting the heavy part to and from the production line and within the production line itself. Apparently, the Japanese principal was not confident that Thai-brakes would be able to handle the complexities of introducing the robot into the process and insisted on the manual loading of the part into and out of the mini-production line and the use of rollers to transfer the semi-machined part between the CNC machine-tools in the production line, even though the robot would have been a cheaper solution and the factory-owner himself is a dealer for a well-known brand of robots. 2.3 Problems encountered with flexible automation The most serious complaint voiced with regard to the introduction of CNC machine-tools concerns the problems of servicing and maintaining the machines. Very simple maintenance of the CNC machines can be done in-house but serious problems and spare parts have to be dealt with or supplied by suppliers. Since CNC machines tend to be more sensitive to misuse or improper operation than traditional machines, problems often resulted from carelessness or inexperience. Few firms had developed the in-house capability to repair CNC machines, especially the electronic elements, and therefore had to rely on outside suppliers and engineering service companies. In many cases, obtaining key spare parts took up to a week, which became a more serious problem as the machines got older and required servicing more frequently. Part-castings reported that this lack of internal maintenance and repair capability put them at a competitive disadvantage vis-à-vis Japanese firms which can get the machines fixed in a day

FA AND THE THAI ENGINEERING INDUSTRY 377

or less. They have been developing an internal training programme and requiring more training from their machine suppliers as part of the purchase package. Several companies also stated that the CNC machines they purchased did not live up to their expectations. Part-castings’s manager felt that this was partly because the expectations were based on the experience of Japanese manufacturers who work very closely with the predominantly Japanese machinetool makers to produce machines that fit the specific needs of the manufacturer. Another reason is that Thai users do not have the knowledge to put the FA to the best use, either as stand-alone machines or as part of a production line. Thai firms lack the experience and technical support to push suppliers to adapt the machines to their specific requirements or to provide the service necessary to push the CNC machines to their maximum effectiveness. This lack of coordinated efforts, between suppliers and users in Thailand, to make the continual incremental improvements needed to maximize the effectiveness and returns of FA is a key constraint. Despite these unmet expectations, most companies noted that CNC machines are generally easier for workers to operate, and they did not express serious concerns about finding operators. The lower skill requirements for operating CNC machines had partly solved previous skilled technical manpower shortages. Indeed, Omni-parts’s manager felt that since CNC machines are relatively easy to operate, they are not very conducive to staff development. He expressed concern that people might leave for more interesting or challenging jobs. However, several firms did report difficulty in recruiting good programmers to program the CNC machines and in retaining them. Incremental programming improvements have great potential for improving machine utilization. To solve this problem some firms had internal training programmes to bring some skilled technicians up to the programmer level, but such workers are in high demand and firms had to keep several programmers on their staff in case of defection. The final major problem encountered by CNC machine users resulted from the irregular power supply found in some areas. Gear-shafts reported that the AC servo drives of their CNC machines are liable to break down as a result of power failures. Part-castings found that all CNC machines have to be reset from scratch if they were in the middle of a process when power was cut. This happens at least once a week, sometimes twice, in the rainy season. 2.4 Flexible automation and organizational change Flexible automation clearly has an important role to play in stimulating innovative changes in the organization and strategy of industrial companies. Organizational changes include both changes on the factory-floor and those which take place on the organiza-tional chart. At present, ‘re-engineering’ is extremely popular in Thailand and many firms are seeking to redefine their objectives and strategies in line with the new principles, and certainly with the assistance of FA. However, a number of characteristics temper the impact of the introduction of FA on the organization and strategy of the sampled firms.

378 THE COUNTRY STUDIES

In the first place, most firms in the sample, and in Thailand’s industrial sector as a whole, are not very old. Seven of the eleven firms interviewed were established in 1974–80, two were pre-1974 and two were post-1980. There is therefore not the depth of industrial experience that is observed among older firms in developed countries. Second, the ownership and control of the firms: 5 of the 11 are Japanese jointventures, the mother companies of which maintain considerable control over both organization and strategy. Plumbing-parts is another Japanese joint-venture, but one which has developed more independence from its Japanese partner than have the other five. Its product is a little simpler than those of the other jointventures; and it has had more success in developing a Thai-designed product and marketing it as a Thai brand, whereas the other firms are almost entirely dependent on the Japanese-controlled automotive sector for sales. The five remaining firms are not joint-ventures, but rely heavily on Japanese-controlled sectors for a large part of their output. Four of the six Thai firms are owned by family concerns which, in the past, took a rather traditional approach to firmlevel management, control, technology strategy, and innovation. Third, FA has been introduced in the sampled firms relatively recently and the machines are not yet sufficiently entrenched to have had a significant impact on organization—either of the factory or of the company. In most cases, the CNC machines have been simply inserted into the production line or into the batchproduction process in place of traditional machines. And in most cases this was required by outside agents rather than being instigated from within the firm itself as part of a coherent technology or marketing strategy. The result is that the impact of the CNC machines, particularly among the Japanese joint-ventures, has been essentially to improve quality, to increase the number of models, or to increase productivity. There has been little impact to date on significant organizational issues. Nevertheless, in some cases FA has had a significant impact on firm strategy: Cylinder-liners was able to broaden its product line significantly as a result of the introduction of CNC machines, and is now searching for new markets; Thaibrakes has moved into an important new area as a subcontractor to a major Thai automotive firm, with the addition of a CNC production line in its existing factory to produce heavy machined metal brake drums; and Omni-parts had moved out of general automotive parts and into higher value-added and higher precision parts. Some of the main car assemblers have reported that they are subcontracting processes previously performed in-house (as in the Thai-brakes’ example), and others said that they would like to be able to do so. Whether the net effect on market structure will be to increase the market share of small firms is difficult to say, but it is quite possible that small firms will play an increasing role in some niche markets as technology advances and the entry price (i.e. cost of machinery) goes down. One would expect to see them becoming more important in subcontracting for the production of small quantities of machining parts and operations that can be easily separated from the production process.

FA AND THE THAI ENGINEERING INDUSTRY 379

Three key trends appear likely to accelerate the impact of FA on the organizational and strategic elements of Thai-based firms. First, growing competition in world markets, coupled with rising labour costs in Thailand and greater demands for precision, will force Thai firms to explore the strategic possibilities of applying FA to develop new product lines, enter niche markets, and remain competitive. The process will be expedited by reductions in the cost of the computers and machines. Second, as computers and information technology become increasingly integrated into the production process, along with CNC and related machines, Thai firms will find that flows of information and increasing levels of automation will redefine the ways in which the factory must be laid out and the ways in which the various spheres of production need to be organized to support the new emphasis on information, as discussed in TDRI (1992 and 1993). Third, as Thailand’s industrial structure continues its shift towards more sophisticated products while increasing the depth of its supporting industries, and as the world market becomes more integrated, new production and subcontracting relationships will become inevitable and FA will begin to link together firms as well as spheres of activitiy within firms. In this climate, strategic and organizational innovations will be inescapable and FA will be at the heart of these. 3 Firm-level impacts of flexible automation In our discussions, Dr Pansak Siriratchtapong of the National Electronics and Computer Technology Centre (NECTEC) identified two principle impacts of the introduction of FA. The first is on product scale and scope: FA are seen to enable more efficient and stable production, to support a broader product range, and to increase potential capacity utilization. The second is on unit costs: FA increase the cost of machinery (by approximately 5 times for medium-size machinery), require less skilled labour to control machinery, involve higher maintenance and repair costs, and have higher R&D and engineering costs, and overhead costs, at the beginning of operation. 3.1 Product scale Setting-up time Most of the firms interviewed reported that machine set-up times were substantially reduced with the introduction of CNC machine-tools (see Table 12.4), although Omni-parts reported that setting up takes much longer if the CNC machines have suffered alignment problems. However, these time savings did depend to some extent on the type of machine being replaced. Both Gear-shafts and Plumbing-parts reported that the time taken to set up a manual turret lathe was little more than a few minutes,

380 THE COUNTRY STUDIES

Table 12.4 Thailand: effects of flexible automation on set-up times in sampled firms

Source: Interviews.

while the set-up time for a CNC machine-tool was considerably longer. Gearshafts, which had to meet rather high standards, found that the set-up time for the CNC machines was similar to that for automatic machines but 100 per cent longer than for manual machines. Plumbing-parts’ production manager claimed that for small production runs and in certain processes, a skilled operator on a good manual machine-tool can be more cost effective than automatic or CNC

FA AND THE THAI ENGINEERING INDUSTRY 381

machines. However, he had serious difficulty in finding and keeping such skilled machine operators. The set-up time comparison also depended on the type of machine-tool. In general, CNC lathes and bending machines were found to be easier to set up than CNC machining centres, because of the greater time required to install the tooling and jigs, to adjust the machine calibration and settings, and to produce and test a few samples before beginning real production. Since no companies had at the time of the interviews installed anything approaching a fully-automated production line, the comparisons of setting-up times refer exclusively to machine, and not production-line, set-up times. In those companies producing with product lines, as distinct from those carrying out purely process operations, it is the machine with the slowest set-up time that determines the overall set-up time. Batch sizes As Table 12.5 shows, batch sizes ranged from 120 (in the case of Electrical-auto) to over 10,000 (in the case of Omni-parts)18 and differed significantly within the same firm depending on the nature of the product and the size of the order. Given the greater flexibility of CNC machines and the ease with which they can be re-programmed, one might expect the potential economical batch size, and the hence product scale, to decrease. In theory, this was recognized by most of the interviewees.19 However virtually all the firms found that the introduction of CNC machine-tools had little direct effect on the actual (as opposed to the economically feasible) batch size. The key factor influencing the batch size was the size of the orders received from clients, largely because most of the interviewed firms produced customized products to order rather than standardized products to be sold as pure commodities. Even for Thai-diesel, which produces engines, and Thai-electric, which produces electrical motors, final demand was the key factor that influenced batch size, although Thai-diesel ran significantly the larger batches through its non-CNC machine-tool production line due to the higher set-up costs.20 Smaller jobs and those with higher precision requirements were processed through the CNC machine lines. In fact, several companies (those for which productivity was a major reason) introduced CNC machine-tools specifically in order to meet increased greater demand. These firms tended to try to maximize batch size to reap the benefits of economies of scale and of the increased productivity and speed of the new machine-tools. 3.2 Product diversity (scope) All firms interviewed reported that the introduction of CNC machines greatly facilitated the introduction of new products or variants (in size and dimension) of

382 THE COUNTRY STUDIES

Table 12.5 Thailand: effects of flexible automation on batch size and processing time in sampled firms

Source: Interviews.

the same product. The greater flexibility of CNC machines was seen as a key element in adapting more rapidly to product demand changes in terms of both quality requirements and product range. Cylinder-liners was able to increase its product range four-fold after introducing FA—extending its product line from cylinder liners to connecting rods, drum brake centres, and disc brakes—while Motor-parts and Thai-electric more than doubled their product offerings. Some of the autocomponent part producers (Omni-parts, Part-castings, Thaidiesel, Gear-shafts) were able to shift progressively into product ranges requiring higher levels of precision. In the case of Omni-parts, there was effectively a

FA AND THE THAI ENGINEERING INDUSTRY 383

product shift rather than an increase in product range, since the firm now produces precision metal parts almost exclusively and has stopped producing certain simple electrical products. Thai-diesel was able to progressively reintegrate its production by producing certain precision components in-house which previously it had imported or outsourced. Part-castings found that CNC machines enabled it to adapt and produce a much wider range of products at higher levels of quality. In sum, the ability to react more quickly to product demand shifts was seen as one of the greatest benefits of having introduced FA. In most cases the increased product range and flexibility were manifested in extremely high annual growth in company revenues. 4 Plant scale and unit costs Since no company interviewed had kept systematic records of actual performance changes after the introduction of CNC machines, most of the responses about impacts on cost and plant scale are somewhat qualitative, being based on the impressions of the factory managers or engineers. Even where it is possible to identify the specific times at which CNC machines were introduced, the accounting and financial data provide extremely unreliable information about the effects, because the data reflect many other factors influencing the firms’ performances and, in these firms, were rarely broken down by product group or production line. Multiple book-keeping practices are also known to be relatively common in Thailand. 4.1 Plant scale21 The Thai economy and, in particular, the manufacturing sector have grown very rapidly over the past 5–8 years, and all the firms surveyed had also grown substantially over the past 5 years or so. For the companies for which time series data were available, the growth rates in the nominal value of sales were as follows: Omni-parts by 4 times (1987–91); Part-castings by 12 times (1987–92); Thaidiesel by 3 times (1988–92); Thai-brakes by 6 times (1987–93); Thai-electric by 4 times (1987–92); and Plumbing-parts by 4 times (1989–93).22 Electrical-auto and Thai-pumps grew more slowly, Electrical-auto by 25 percent in 1989–93 and Thai-pumps by 60 per cent in 1991–94. From indications such as machine purchases it would appear that the other three firms also experienced rapid growth. Most firms claimed that they would not have been able to grow without flexible automation, either because they would not have been able to keep up in terms of product quality or range or because they would not have been able to meet the rapidly growing demand without the increased productivity of FA. Several reported introducing FA specifically to enable them to gear up their

384 THE COUNTRY STUDIES

Table 12.6 Thailand: effects of flexible automation on product diversity in sampled firms

Source: Interviews.

production, while the higher costs of FA machines made it imperative to increase the firm’s scale of production in order to cover these higher capital costs. Thus it is difficult to differentiate between the impact of FA in increasing firm scale and the effect of the tremendous growth of the economy and increasing depth of industrialization.

FA AND THE THAI ENGINEERING INDUSTRY 385

4.2 Costs of production Table 12.7 summarizes the evidence available with regard to the effects of FA on the various component costs of production. Final product reject rates have fallen in most companies over the past five years or so. In addition to the impact of CNC machines, most companies also attributed improvements which occurred to better working practices, improved raw-materials and better quality-control. Plumbing-parts raised the interesting point that following the introduction of CNC machines in its factory, some months had very high final reject rates. This was because an incorrect set-up for the CNC machine-tool could result in many hundreds or thousands of rejects before testing identified the problem. This occurs less now, but it does occur, especially during the night-shift when few of the most highly skilled technical workers are at the factory. Several firms had good quality-control systems in place throughout the production process before installing FA, and had not experienced significant reductions in final reject rates. The impact of CNC machines on reject rates was most visible in reductions in intermediate reject rates as a result of better machining, and in lower levels of rejected partly processed products and components. Thai-brakes reported that internal reject rates fell by 10–15 per cent. Most firms found that defect rates of partly processed products fell substantially. Although Plumbing-parts reported that certain high quality automatic machines (such as a three-axis rim machine that the firm had recently purchased) were around the same price as a CNC machine, CNC machine-tools were generally more expensive to buy and maintain than traditional machines. Several firms reported that tool life is longer with CNC machines, due to better coolant systems and better control of the depth of cut. However, this did not have a significant cost impact since CNC tooling itself was more expensive than tooling for traditional machines. One firm (Part-castings) reported that its CNC machines were more energy efficient than traditional machines. The higher costs of the machines meant that most firms also attempted to operate the CNC machines at utilization rates higher than they had previously attained with traditional machines. Several firms, particularly those producing along process (or batch) lines, operated their CNC machines for three shifts, while traditional machines were run for only one or two shifts. Several firms also reported significant reductions in the processing time for each piece. Electricalauto reported reductions of up to 80 seconds per pipe bending process, while Thai-brakes and Cylinder-liners found that processing time for smaller batch-size machined parts was considerably lower. Many companies were able to save factory space through the introduction of CNC machines, and those operating in Bangkok where land prices are high were able to save considerably by increasing output with little requirement for additional land. All firms reported reductions in labour cost as a result of introducing CNC machines. However, this was not seen to be as important a consideration as the

386 THE COUNTRY STUDIES

fact that the number of highly skilled machinists (who are in short supply) required was reduced. The unit costs of most companies fell, since the reductions in labour and rawmaterial costs (due to fewer defects), combined with greatly increased productivity, outweighed the higher machine-investment and maintenance costs. It was generally agreed that profitability increased following the introduction of FA, due in part to reductions in unit cost and in part to the better quality and greater consistency (and hence higher prices). Better control of production, leading to more reliable delivery, also helped. Research and development activities Since most interviewed companies produce for the OEM market, they had not developed a significant internal research, development and engineering capability. In most cases, the product specifications and designs are provided by the customers and sometimes by mother companies abroad. Modifications are generally not permitted without rigorous control by the customer. Certain firms had built up product development teams, notably Part-castings which focused on material testing and product development, Thai-diesel which modified and improved the efficiency of its engines for specific applications in the Thai market, and Plumbing-parts, which did some simple product development with a focus on appearance. It did not seem that the introduction of CNC machines had had any major impact on product development work, although CNC machines made it easier to adapt new products. Of the 11 companies interviewed, 6 are using integrated CAD/CAM, principally for the production of prototypes or moulds and dies. The data are transferred directly from the design to the machining operations.23 Those companies making moulds and dies showed a greater tendency to develop design capabilities. 5 Summary and conclusions This section synthesizes the principal conclusions that can be derived from the qualitative information gathered, as well as providing some indications on where future research efforts would be most productive. There is no doubt that the introduction of FA into Thailand’s industrial structure has had generally positive impacts on the productivity and quality of output and is a critical component of the country’s continued competitiveness. However, both the sampled firms and the economy in general have been enjoying tremendous growth over the past decade. The resulting ‘noise’ in the system, the fact that none of the firms had carried out any systematic analysis of their CNC machine-tools, and that most would not or could not provide any detailed statistical information, made it difficult to identify with any precision the exact impacts of the introduction of FA in Thailand.

FA AND THE THAI ENGINEERING INDUSTRY 387

Table 12.7 Thailand: effects of flexible automation on unit costs in sampled firms

388 THE COUNTRY STUDIES

Source: Interviews.

Extent of flexible automation diffusion Thai companies are now familiar with FA. All firms contacted had at least considered FA introduction, and a very high percentage of medium- to largescale producers have already installed some CNC machines. Virtually all have used computers to some degree in the production process (broadly defined). And many small machine shops have some CNC machine-tools. The fragmented way in which most CNC machines are utilized, and weaknesses in the links between the various spheres of the production process, indicate that there is significant scope at all levels for more efficient and effective use of FA. In other words, FA is quite well diffused in quantitative terms, but not in qualitative terms. The Thai situation is basically one of significant automation at the intraactivity level, some limited automation at the intra-sphere level, and very little real automation at the inter-sphere level. It seems likely that the introduction of FA will generate significantly greater effects in terms both of productivity and quality and also with regard to the organizational structures of the firm, when the level of automation reaches further into the second and third levels. Advances in information technology and the increasing use of computers are likely to speed this process substantially in the next few years.24 Firm-level impacts: flexibility, scale, scope and competitiveness At the firm level, the introduction of CNC machine-tools had generally positive impacts. Firms were able to improve quality and productivity levels significantly, as well as producing much larger quantities of higher quality products. In many cases, FA has enabled firms to maintain their positions in the market or to enter new market segments. These quality and productivity effects are attributable to CNC machines. Plant scale also increased dramatically, but it is

FA AND THE THAI ENGINEERING INDUSTRY 389

not clear that this can be only attributed to the introduction of CNCs: it appeared that plant scale and the selection of technologies to achieve the required scale were determined also by demand. CNC machine-tools have permitted greater flexibility and thus wider product ranges. While the flexibility of FA could support smaller batch sizes, most firms, faced with growing demands, used FA with bigger batches to increase output. The reductions in setting-up time experienced by all the firms surveyed fed directly into increased productivity and enhanced rates of machine utilization. Although more expensive to purchase and maintain, FA has resulted in lower costs for labour, land and energy, and in much greater productivity and higher quality (and hence higher prices). Unit-production costs fell substantially, and profitability increased across the board. These results point to four conclusions in relation to the discussion in the literature on ‘de-scaling’ and the likely impact of FA on the developing world: (1) They suggest that the introduction of FA has not resulted in ‘de-scaling’ at either the product level or the plant (firm) level. (2) The capital cost of the FA machines has proved to be substantially greater than for more traditional technologies (although several interviewees felt that falling prices for electronic components will bring the costs down eventually, and reduce the cost-differential between new and traditional technologies). (3) Although the effects are not yet fully evident, FA has enabled firms to produce more sophisticated and higher quality products than was possible with traditional technologies. This will eventually permit more subcontracting of smaller quantities of quality products. While the firms surveyed were generally of larger size, there is evidence that smaller producers with one or two CNC machines are beginning to enter the parts and components market. (4) The evidence from the sampled firms is that the scale of production of the final product must rise in order to reap the benefits of economies of scale, especially in light of the higher costs of FA. However, significant niche markets for the production of higher quality parts and components do exist and can be exploited by innovative entrepreneurs with the capability to apply FA. Suggestions for future research Few if any firms in Thailand appear to have made serious attempts to systematically evaluate the effects of CNC machines on their production processes. Moreover, most science and technology institutions still have some way to go if they are to support the inevitable shift towards increased automation. Thus two lines of future research are indicated:

390 THE COUNTRY STUDIES

(1) At the firm level, this study has examined 11 companies which had introduced FA into their production processes. This permitted a broad assessment of Thai firms’ experiences with FA, but the problems in obtaining meaningful quantitative data, described above, made it impossible to make in-depth analyses of the impacts at the firm level. To do so would require considerably greater effort devoted to studying a smaller sample of firms in more detail and over a longer period.25 It is important also to devote more attention to the numerous smaller workshops and machine shops that use only one or a few CNC machine-tools. They probably face different types of problem requiring different solutions or forms of assistance. Indeed, they may well benefit from structured programmes to increase their linkages with some of the larger firms with longer experience in using FA. (2) With regard to the contributions of supporting institutions, the various technology-related agencies in Thailand should play an increasing role in promoting the understanding of FA and assisting manufacturers to assimilate these technologies into their production processes, especially in light of the sophistication of FA and the speed with which it is changing. It appears to be relatively easy to use flexible automation, but much harder to develop the capabilities required to use it really well and to carry out activities which lead to incremental improvements in FA use. The specific areas where such agencies can assist require identification, then refinement. Such areas could range from technical and information support to financial and management assistance.26 Finally, the key to the success of this two-level approach is co-ordination between the public and private sectors in the process of adopting FA and exploiting the benefits it offers. The public sector must take a long-term view of the strategic development of FA while keeping a focus on the demands of the productive sector. The private sector, in turn, must work to understand its own needs and requirements in relation to the introduction of flexible automation, and to communicate its understanding to the relevant government agencies. Notes 1 The author is grateful to the following for support in the preparation of this paper: Pattanun Woodtikarn, Wilasinee Bintasan, Pitchayapong Charnkla and Kathryn Scott assisted in surveying users and suppliers of new technologies and provided additional research support; Sheldon Jacobs, Scott Rosenberg and Lilantha Siriwardene provided effective editorial contributions; and participants in a UNU/ INTECH workshop in Maastricht provided most useful insights. Ludovico Alcorta provided invaluable advice and constant encouragement, not to mention harassment; and Sen McGlinn’s conceptual and textual editing greatly improved the final product. Lastly, the valuable time and opinions provided by the managers of the companies surveyed must be gratefully acknowledged.

FA AND THE THAI ENGINEERING INDUSTRY 391

2 See TBOI (1993) for a complete breakdown of machine-tool imports and exports and a description of the sector and investment opportunities. 3 Brimble and Sripaipan (1994: Annex 3) present a detailed summary of the S&T institutional infrastructure in Thailand. 4 See Brimble and Sripaipan (1994) for a comprehensive summary and evaluation of the major institutions involved in both S&T support and new technologies’ support. Brimble and Sripaipan consider the present status of these S&T factors in Thailand, identify the major weaknesses, and recommend policy solutions. Their paper develops and updates the arguments of Dahlman and Brimble (1991), which also emphasized the importance of S&T development and policies in Thailand’s industrialization process. 5 Including Kasetsart University; the three King Mongkut’s Institutes of Technology in Lad Krabang, Thonburi and North Bangkok; Chulalongkorn University; and the Department of Vocational Education. 6 None of the firms interviewed reported receiving any significant support from public sector agencies. 7 The suppliers contacted were Champion Machine Tools (Thailand), Cosmos Machinery (Thailand), Dawa Machinery, Maruka Machinery (Thailand), Numac Machinery (Thailand). 8 The following observations on the nascent robotics sector in Thailand are derived from a survey of robot users and potential robot users carried out by SEAMICO (1993) for a private sector client. Many of the observations mirror those made about CNC machines although generally to a greater degree. 9 Even more so than in the case of CNC machine-tools. 10 Although there are few programmers available, and an increasing shortage in recent years as the information revolution takes hold in Thailand, companies need only one or two programmers unless they plan to go into detailed programming (e.g. CAD/CAM) for themselves, which most have not yet done. The dearth of programmers does affect the extent to which firms can carry out so-called ‘incremental’ improvements to the often pre-packaged or simply modified programs supplied by the machine supplier or sometimes the customer. 11 Considerable efforts were made to identify appropriate interview targets. Those selected were subsequently interviewed by at least two interviewers (generally an engineer and an economist), with the interviews lasting between half-a-day and a full day. The interviews followed an outline, although some flexibility had to be accepted depending on the interviewees’ willingness, time, and/or ability to reply. 12 The list of firms was felt to be reasonably representative, based on previous research exercises and firm-level surveys such as the SEAMICO (1993) and work carried out by the researchers for the TBOI Unit for Industrial Linkage Development (BUILD). 13 These small companies, which do machining under subcontract, are difficult to contact; it is even more difficult to obtain information from them. 14 This confirms the observation of suppliers that use of CNC machines is frequently a ticket to membership of the component-suppliers’ club: an indication that a certain quality standard has been met. 15 An interesting example of government-induced adoption of new technologies was found in Thai-diesel where the local content requirements for engine manufacturing required the firm to build a new production line to make some of the precision parts

392 THE COUNTRY STUDIES

16

17

18

19

20

21 22 23

24 25

26

that they had previously imported. In fact, they would have been satisfied to subcontract the parts in question, but were reluctant to trust an outsider to meet the quality requirements. A number of firms have used CAD/CAM in the production of prototypes or moulds and dies at a very basic level, but none has yet made the full link between design and production that would constitute a move to the third, or inter-sphere, level of automation. Interestingly, the engineer at Gear-shafts reported that the automatic loaders are actually slower since they must do a dimension check before unloading the workpiece from the CNC machine-tool. If the dimensions are incorrect, the automatic arm will place the defective workpiece in a reject bucket, thus resulting in more consistent output. In some cases (e.g. Part-castings) the average batch size was rather difficult to define since some lines/machines processed products more or less continuously, sometimes performing the same activity on many different products/batches. However, for small batches, manual machines would be preferable, while at very large batch sizes, automatic machine processes are more cost effective. These observations are effectively dictated by the technical characteristics of the three different configurations—manual, CNC, and automatic. Some of the processes at Thai-diesel were more suitable for the automatic machines which they had kept in their production lines even after introducing CNC machines. All the firms were operating entirely or largely from one factory at the time of interview, so that plant size and firm size are the same. Inflation rates over this period averaged well under 5 per cent per year. Part-castings is probably the most sophisticated of the interviewed companies in this regard. They use the Unigraphic CAD program in designing parts, dies and also tooling, and are looking at linking the CAD system with the CNC machines for direct transmission of the program to the manufacturing process. See TDRI (1993) for a likely scenario. It is probable that several of the companies surveyed would be amenable to such an exercise if they felt that the results would provide them with information which would help them to plan and implement their shifts towards increased levels of automation. A better understanding of new technologies would greatly benefit the National Supplier Development Programme as it gets under way.

APPENDICES: GLOBAL DATA

Table A1 Total metal-forming and metal-cutting machine-tool production, 1973–94 (US $m.)

Source: American Machinist, various years. Note a 1985 and 1994 are estimated figures.

394 APPENDICES: GLOBAL DATA

Table A2 World metal-cutting machine-tool production, 1973–94 (US$m.)

Source: American Machinist, various years.

APPENDICES: GLOBAL DATA 395

396 APPENDICES: GLOBAL DATA

Table A3 CNC and non-CNC lathe production, 1975–94 (units)

Source: Our own elaboration on the basis of data provided by CECIMO (1992) and Jacobsson (1986) for years prior to 1980.

APPENDICES: GLOBAL DATA 397

398 APPENDICES: GLOBAL DATA

Table A4 CNC and non-CNC lathe production, 1975–94 (US$m.)

Source: Our own elaboration on the basis of data provided by CECIMO (1992) and Jacobsson (1986) for years prior to 1980.

APPENDICES: GLOBAL DATA 399

Source: Our own elaboration on the basis of data provided by CECIMO (1992).

Table A5 Production of machining centres, 1986–94 (units)

400 APPENDICES: GLOBAL DATA

Source: Our own elaboration on the basis of data provided by CECIMO (1992).

Table A6 Production of machining centres, 1986–94 (US$m.)

APPENDICES: GLOBAL DATA 401

Table A7 Total machine-tool importsa, 1962–94 (US$m.)

402 APPENDICES: GLOBAL DATA

APPENDICES: GLOBAL DATA 403

404 APPENDICES: GLOBAL DATA

Source: Our own elaboration on the basis of United Nations (1996) and official government statistics. a SITC 715.1 and 719.54, Rev.1.

APPENDICES: GLOBAL DATA 405

Table A8 Total machine-tool exports,a 1962–94 (US$m.)

406 APPENDICES: GLOBAL DATA

APPENDICES: GLOBAL DATA 407

408 APPENDICES: GLOBAL DATA

Source: Our own elaboration on the basis of United Nations (1996) and official government statistics. Note aSITC 715.1 and 719.54, Rev. 1.

APPENDICES: GLOBAL DATA 409

Table A9 Total metal-cutting machine-tool exports,a 1978-94 (US$m.)

410 APPENDICES: GLOBAL DATA

Source: Our own elaboration on the basis of United Nations (1996) and official government statistics. Note aSITC 736.1 Rev.2.

APPENDICES: GLOBAL DATA 411

Table A10 Total metal-cutting machine-tool imports,a 1978–94 (US$m.)

412 APPENDICES: GLOBAL DATA

Source: Our own elaboration on the basis of United Nations (1996) and official government statistics. Note a SITC 736.1 Rev.2.

APPENDICES: GLOBAL DATA 413

Table A11 Total lathe exports,a 1978–94 (US$m.)

414 APPENDICES: GLOBAL DATA

Source: Our own elaboration on the basis of United Nations (1996) and official government statistics. Note a SITC 736.13 Rev.2.

APPENDICES: GLOBAL DATA 415

Table A12 Total lathe imports,a 1978–94 (US$m.)

416 APPENDICES: GLOBAL DATA

Source: Our own elaboration on the basis of United Nations (1996) and official government statistics. Note aSITC 736.13 Rev.2.

APPENDICES: GLOBAL DATA 417

418 APPENDICES: GLOBAL DATA

Table A13 Total CNC lathe exports,a 1984–94 (US$m.)

Source: Our own elaboration on the basis of United Nations (1996) and official government statistics. Note a SITC 731.31 and 731.35 Rev. 3.

APPENDICES: GLOBAL DATA 419

Table A14 Total CNC lathe imports,a 1989–94 (US$m.)

Source: Our own elaboration on the basis of United Nations (1996) and official government statistics. Note a SITC 731.31 and 731.35 Rev. 3.

420 APPENDICES: GLOBAL DATA

Table A15 Total machining centre exports,a 1989–94 (US$m.)

Source: Our own elaboration on the basis of United Nations (1996) and official government statistics. Note a SITC 731.21 Rev.3.

APPENDICES: GLOBAL DATA 421

Table A16 Total machining centre imports,a 1989–94 (US$m.)

Source: Our own elaboration on the basis of United Nations (1996) and official government statistics. Note aSITC 731.21 Rev.3.

422 APPENDICES: GLOBAL DATA

Table A17 Other CNC metal-cutting machine-tool exports,a 1989–94 (US$m.)

Source: Our own elaboration on the basis of United Nations (1996) and official government statistics. Note a SITC 731.42; 731.44; 731.51; 731.53; 731.61; 731.63 and 731.65 Rev.3.

APPENDICES: GLOBAL DATA 423

Table A18 Other CNC metal-cutting machine-tool imports,a 1989–94 (US$m.)

Source: Our own elaboration on the basis of (1996) and official government statistics. Note a SITC 731.42; 731.44; 731.51; 731.53; 731.61; 731.63 and 731.65 Rev.3.

424 APPENDICES: GLOBAL DATA

Table A19 World machine-tool consumption, 1973–94 (US$m.)

Source: American Machinist, various years.

APPENDICES: GLOBAL DATA 425

BIBLIOGRAPHY

Abernathy, W.J. (1978) The Productivity Dilemma: Roadblock to Innovation in the Automobile Industry, Johns Hopkins University Press, Baltimore/London. ABIMAQ/SINDIMAQ (Associação Brasileira da Indústria de Máquinas e Equipmentos/ Sindicato Nacional da Indústria de Máquinas) Relatório Anual, various years. Acs, Z.J. and Audretsch, D.B. (eds) (1990) The Economics of Small Firms: A European Challenge, Dordrecht: Kluwer. ——(1990a) Innovation and Small Firms, Cambridge, MA: MIT Press. Massachusetts. ——(1990b) ‘Small Firms in the 1990s’, in Acs and Audretsch (1990). ——(eds) (1991) Innovation and Technological Change: An International Comparison, New York: Harvester Wheatsheaf. ——(1991a) ‘Innovation and Technological Change: An Overview’, in Acs and Audretsch (1991). ——(1991b) ‘R&D, Firm Size and Innovative Activity’, in Acs and Audretsch (1991). Acs, Z.J., Audretsch, D.B. and Carlsson, B. (1990) ‘Flexibility, Plant Size and Industrial Restructuring’, in Acs and Audretsch (1990). Ahluwalia, I.J. (1985) Industrial Growth in India: Stagnation since the Mid-Sixties, New Delhi: Oxford University Press. Alänge, S., Jacobsson, S. and Jarnehammar, A. (1995) ‘Some Aspects of an Analytical Framework for Studying the Diffusion of Organizational Innovations’, in Jarnehammar (1995). Alcorta, L. (1989) ‘La Dinámica del Capital Financiero: Un Estudio Comparativo de Gran Bretaña y Japón’, Apuntes 24. ——(1992a) ‘The Impact of New Technologies on Scale in Manufacturing Industry: Issues and Evidence’, UNU/INTECH Working Paper 5, Maastricht: UNU/INTECH. ——(1992b) El Nuevo Capital Financiero: Grupos Financieros y Ganancias Sistémicas en el Perú, Lima: Fundación Friedrich Ebert. ——(1993a) ‘Are Economies of Scope Replacing Economies of Scale? Implications for Developing Countries’, Paper presented at the First UNU/INTECH Conference on Technological Change in Developing Countries, June, Maastricht. ——(1993b) Flexible automation, economies of scale and scope, and location of production in developing countries: a research proposal (Mimeo), Maastricht: UNU/ INTECH. ——(1994) ‘The Impact of New Technologies on Scale in Manufacturing Industry: Issues and Evidence’, World Development 22 (5). ——(1995a) ‘The Impact of Industrial Automation on Industrial Organisation: Implications for Developing Countries’ Competitiveness’, Discussion Paper #9508, December, Maastricht: UNU/INTECH. ——(1995b) ‘New Technologies, Scale and Scope, and Location of Production in Developing Countries’, Discussion Paper #9502, March, Maastricht: UNU/INTECH.

BIBLIOGRAPHY 427

Alonso, O. (1989) ‘La Automatización Industrial en Venezuela: El Caso de la Industria Productora de Bienes de Capital’, III Seminario Latinoamericano de Gestión Tecnológica, ALTEC, Buenos Aires. ——(1991) Automatización y Estrategia Sindical, Editorial Nueva Sociedad, Caracas: Instituto Latinoamericano de Investigaciones Sociales. ——(1992) ‘Indicadores de Productividad Nacional y Sectorial’, Caracas: FIM Productividad. ——(1993) Competitividad y Sindicatos, Editorial Nueva Sociedad, Caracas: Instituto Latinoamericano de Investigaciones Sociales (ILDIS). Altshuler, A., Anderson, M., Jones, D., Ross, D. and Womack, J. (1984) The Future of the Automobile: The Report of MIT’s International Automobile Programme, London: Allen and Unwin. American Machinist, (1974–95) ‘Machine Tool Ouput’. AMS (Automotive Manufacturers Association) (1984, 1988, 1993) General and Statistical Information Bulletin of Turkish Automotive Manufacturers, Istanbul: AMS. Amsden, A.H. (1977) ‘The Division of Labour is Limited by the Type of Market: the Case of the Taiwanese Machine Tool Industry’, World Development 5 (3). ——(1985) ‘The Division of Labour is Limited by the Rate of Growth of the Market: The Taiwan Machine Tool Industry in the 1970s’, Cambridge Journal of Economics 9. Amstead. B.H., Ostwald, P.F. and Begeman, M.L. (1986) Manufacturing Processes, New York: John Wiley & Sons. Andersen, E.S. (1992) ‘Approaching National Systems of Innovation’, in Lundvall (1992). Ansal, H. (1990) ‘Technical Change and Industrial Policy: the Case of Truck Manufacturing Industry in Turkey’, World Development 18 (11). Arrow, K.J. (1962) ‘The Economic Implications of Learning by Doing’, Review of Economic Studies 29:155–73. Association for Manufacturing Technology (1989) 1989–90 The Economic Handbook of the Machine Tool Industry, The Association for Manufacturing Technology, McLean, VA: Statistical Department. ——(1993) 1993–94 The Economic Handbook of the Machine Tool Industry, McLean, VA: Association for Manufacturing Technology, Statistical Department. Åstebro, T. (1992) ‘The International Diffusion of Computer Aided Design’, in Ayres et al. (1992c). Atallah, A.K. (1966) ‘The Economies of Scale in Theory and Reality’, Dissertation, The Hague: University of Rotterdam. Audretsch, D.B. (1995) Innovation and Industry Evolution, Cambridge, MA: MIT Press. Audretsch, D.B. and Yamawaki, H. (1991) ‘Sub-Optimal Scale Plants and Compensating Factor Differentials in US and Japanese Manufacturing’, Discussion Paper, No. FS IV 91–21 (June), Berlin: Research Unit Market Processes and Corporate Development, Wissenschaftszentrum Berlin für Sozialforschung. Auty, R. (1992) Emerging competitiveness of newly industrializing countries in heavy and chemical industry: effects of the product cycle and technological change (Mimeo), Lancaster: Lancaster University. Aydin, A. (1993) ‘Productivity, Production, Employment Indicators in Manufacturing Industry and Sub-Sectors’, Publication 506, Ankara: National Productivity Centre Publications.

428 BIBLIOGRAPHY

Ayres, R. (1984) The Next Industrial Revolution: Reviving Industry Through Innovation, Cambridge, MA: Ballinger Publishing Company. ——(1991) Computer Integrated Manufacturing, Volume I: Revolution in Progress, London: Chapman & Hall. Ayres, R., Dobrinsky, R., Haywood, W., Uno, K. and Zuscovitch, E. (eds.) (1992a) Computer Integrated Manufacturing, Volume IV: Economics and Social Impacts, London: Chapman & Hall. Ayres, R., Haywood, W., Merchant, M.E., Ranta, J. and Warnecke, H.-J. (eds) (1992b) Computer Integrated Manufacturing, Volume II: The Past, the Present and the Future, London: Chapman & Hall. Ayres, R., Haywood, W. and Tchijov, I. (1992c) Computer Integrated Manufacturing, Volume III: Models, Case Studies, and Forecasts of Diffusion, London: Chapman & Hall. Ayres, R. and Miller, S. (1983) Robotics, Cambridge, MA: Ballinger Publishing Company. Babbage, C. (1832) On the Economy of Machinery and Manufactures, London, Knight. Bagchi, A. (1986) ‘The Impact Of Microelectronics-based Technologies: the Case of India’, World Employment Programme Research Working Paper WEP 2–22/wp 169, Geneva: ILO. Bailey, E.E. and Friedlaender, A.F. (1982) ‘Market Structure and Multiproduct Industries’, Journal of Economic Literature XX: September. Bain, J.S. (1954) ‘Economies of Scale, Concentration, and the Condition of Entry in Twenty Manufacturing Industries’, The American Economic Review XLIV (1). ——(1956) Barriers to New Competition, Cambridge, MA: Harvard University Press. Baloff, N. (1965) ‘The Learning Curve—Some Controversial Issues’, The Journal of Industrial Economics 14. Banco Central de Venezuela (BCV) (1989–92) Informe Económico. Baptista, A. (1991) Bases Cuantitativas de la Economía Venezolana, 1830–1989, Caracas: Ediciones María Di Mase. Baumol, W.J. (1987) ‘Indivisibility’, in Eatwell et al. (1987). Baumol, W.J., Panzar, J. and Willig, R. (1988) Contestable Markets and the Theory of Industry Structure, San Diego, CA: Harcourt, Brace, Jovanovich Publishers. Bedworth, D.D., Henderson M.R. and Wolfe, P.M. (eds) (1991) Computer-Integrated Design and Manufacturing, New York: McGraw-Hill Inc. Bell, R.M. (1972) Changing Technology and Manpower Requirements in the Engineering Industry; Research Report No.3, Brighton: Sussex University Press. ——(1984) ‘Learning and the Accumulation of Industrial Technological Capacity in Developing Countries’, in Fransman and King (1984). Bell, M., Ross, B. and Westphal, L.E. (1984) ‘Assessing the Performance of Infant Industries’, Journal of Development Economics 16:101–28. Belussi, F. (1987) ‘Benetton: Information Technology in Production and Distribution. A Case Study of the Innovative Potential of Traditional Sectors’, SPRU Occasional Paper Series 25. ——(1992) ‘Benetton Italy: Beyond Fordism and Flexible Specialization. The Evolution of the Network Firm Model’, in Mitter (1992). Benson, R. (1989) ‘Process Systems Engineering: Past, Present and a Personal View of the Future’, Computers & Chemical Engineering 13 (11/12).

BIBLIOGRAPHY 429

Benson, R. and Ponton, J. (1991) Process miniaturization—a route to total environmental acceptability? (Mimeo) March, Imperial Chemical Industries/University of Edinburgh. Bernardes, R. and Lemos, P. (1995) Produtividade e Ajuste na Industria Paulista: 1986/ 1994, São Paulo: Fundacao SEADE, DAPD/GISE. Bessant, J.R. (1991) Managing Advanced Manufacturing Technology: The Challenge of the Fifth Wave, Manchester: NCC Blackwell. Bessant, J.R., Bowen, J.A., Dickson K.E. and Marsh, J. (1981) The Impact of Microelectronics: A Review of the Literature, New York: Pica Press. Bessant, J.R. and Kaplinsky, R. (1995) ‘Industrial Restructuring: Facilitating Organizational Change at the Firm Level’, World Development 23 (1). Best, M. (1990) The New Competition: Institutions of Industrial Restructuring, Cambridge: Polity Press. Beswick, P.F. and Owen, A.E. (1981) ‘The Impact on Production Management’, in Twiss (1981). Bhagwati, J. and Srinivasan, T.N. (1975) Foreign Trade Regimes and Economic Development: India, New York: National Bureau of Economic Research. Bhalla, A. and James, D. (1988) ‘Some Conceptual and Policy Issues’, in A. Bhalla and D. James (eds); New Technologies and Development: Experiences in Technology Blending, London: Lynne Rienner Publishers. Biemans, F.P.M. and Vissers, C.A. (1991) ‘A Systems Theoretic View of Computer Integrated Manufacturing’, International Journal of Production Research 29 (5). Blair, J.M. (1948) ‘Does Large-scale Enterprise Result in Lower Costs? Technology and Size’, American Economic Review 48 (2). Bolwijn, P.T., Boorsma, J., van Breukelen, Q., Brinkman, S. and Kumpe, T. (1986) Flexible Manufacturing, Amsterdam: Elsevier. Boon, G.K. (1985) Computer Based Techniques: A Technology/Policy Assessment in a North-South Perspective, Discussion Paper TSF, 85–1, Noordwijk: Technology Scientific Foundation. Boon, G.K. and Mercado, A. (eds) (1990) Automatización Flexible en la Industria: Difusión y Producción de Máquinas-Herramienta de Control Numérico en America Latina, Mexico: Editorial Limusa. Boratav, K. (1990) Stabilization and Adjustment Policies and Programmes. Country Study: Turkey, Helsinki: UNU/WIDER. Boston Consulting Group (1991) ‘The Competitive Challenge facing the EC Automotive Components Industry’, Executive Summary prepared for the Commission of the European Community, London: BCG. Braun, E. (1980) ‘From Transistor to Microprocessor’, in Forrester (1980). Brimble, P. and Sripaipan, C. (1994) ‘Science and Technology Issues in Thailand’s Industrial Sector: the Key to the Future’, Report prepared for the Asian Development Bank, Bangkok: ADB. Brown, F. and Domínguez, L. (1991) ‘Nuevas Tecnologías en la Industria Maquiladora’, in Ramírez(1991). ——(1994) ‘The Dynamics of Productivity in Mexican Manufacturing: 1984–1990’, The Developing Economies: September. Brown, F. and Mertens, L. (1994) Estrategias de competitividad, productividad, recursos humanos y empleo en los noventa (Mimeo).

430 BIBLIOGRAPHY

Brozen, Y. (1957) ‘Economics and Changing Technology, the Economics of Automation’, The American Economic Review XLVII (2). Bureau of Industry Economics (1988a) ‘The Impact of Microelectronics on Scale and Competitiveness in Australian Industry: Case Studies of Automotive Product Industries’, Research Report 27, Canberra: BIE. ——(1988b) ‘The Impact of Microelectronics on Scale and Competitiveness in Australian Industry: Case Studies of Pulp and Paper Product Industries’, Research Report 26, Canberra: BIE. Calero, F. (ed.) (1988) Nuevos Retos del Sindicalismo, Editorial Nueva Sociedad, Caracas: Instituto Latinoamericano de Investigaciones Sociales. Carlsson, B. (1984) ‘The Development and Use of Machine Tools in Historical Perspective’, Journal of Economic Behaviour and Organization 5. ——(1989a) ‘Flexibility and Theory of the Firm’, International Journal of Industrial Organization 7:179–203. ——(1989b) ‘The Evolution of Manufacturing Technology and Its Impact on Industrial Structure: An International Study’, Small Business Economics 1 (1). ——(1991a) ‘Factory Automation and Economic Performance: a Micro-to-Macro Analysis’, Paper presented at The Wallenberg Symposium on Economics of Technology, Marstrand, Sweden, August 18–22. Preliminary draft. ——(1991b) ‘Flexible Manufacturing and US Trade Performance’, Weltwirtschaftliches Archiv 127 (2). Carlsson, B. and Jacobsson, S. (1991) ‘What Makes the Automation Industry Strategic?’, Economics of Innovation and New Technology 1:257–69. ——(1993) ‘Technological Systems and Economic Policy: the Diffusion of Factory Automation in Sweden’, Research Policy 23. Carlton, D.W. and Perloff, J.M. (1994) Modern Industrial Organisation, New York: HarperCollins College Publishers. Carrie, A. (1988) Simulation of Manufacturing Systems, Chichester: John Wiley & Sons. Carrillo, J. (1993) ‘La Ford en Mexico: Restructuración Industrial y Cambio en las Relaciones Sociales’, Tesis de Doctorado en Ciencia Social, El Colegio de Mexico. CECIMO (Comité Européen de Cooperation des Industries de la Machine Outil) (1995) International Statistics of Machine Tools 1981–1995, Brussels: CECIMO. Celasun, M. (1991) ‘Trade and Industrialization in Turkey’, Paper presented at The UNU/ WIDER Conference on Trade and Industrialization Reconsidered, Paris. Central Intelligence Agency (1992) Handbook of International Economic Statistics 1992, Washington, DC: Directorate of Intelligence. Chandler, A.D. (1990) Scale and Scope, Cambridge, MA: The Belknap Press of Harvard University Press. Chapman, K. and Walker, D.F. (1990) Industrial Location, Principles and Policies, Oxford: Blackwell Publishers. Chenery, H. (1949) ‘Engineering Production Functions’, Quarterly Journal of Economics 63. Cheng, C. (1972) The Machine-Building Industry in Communist China, Edinburgh: Edinburgh University Press. Cheng, T.C.E. and Podolsky, S. (1993) Just-in-Time Manufacturing: An Introduction, London: Chapman & Hall.

BIBLIOGRAPHY 431

Chudnovsky, D. (1988) ‘The Diffusion and Production of Numerically Controlled Machine Tools with Special Reference to Argentina’, World Development 16 (6): 723–32. Chudnovsky, D. and Nagao, M. (1983) Capital Goods Production in the Third World: An Economic Study of Technology Acquisition, London: Frances Pinter Publishers. Clark, K.B. and Fujimoto, T. (1991) Product Development Performance, Strategy, Organization and Management in the World Auto Industry, Boston, MA: Harvard Business School Press. Clarke, R. (1985) Industrial Economics, Oxford: Blackwell. Cody, J., Hughes, H. and Wall, D. (eds) (1980) Policies for Industrial Progress in Developing Countries, Oxford: Oxford University Press. Colman, D. and Nixson, F. (1978) Economics of Change in Less Developed Countries, Oxford: Philip Allan Publishers. Commission of the European Communities (1988) ‘Studies on the Economics of Integration: Research on the “Cost of Non-Europe”’, Basic Findings, vol. 2, Brussels: Commission of the European Communities. Confederation of Engineering Industries (1987) Handbook of Statistics, New Delhi: CEI. Cooper, C. (1971) ‘Science, Technology and Development’, Economic and Social Review II (2). Coriat, B. (1992) Pensar al Revés, Mexico City: Siglo XXI Editores. Corsi, M. (1991) Division of Labour, Technical Change and Economic Growth, Aldershot: Avebury. Coutinho, L. and Ferraz, J.C. (eds) (1994) Estudo da Competitividade da Industria Brasileira, Campinas: Papirus e MCT. Covarrubias, A. (1992) La Flexibilidad Laboral en Sonora, Mexico City: Fundación Friedrich Ebert. Covarrubias, A. and Lara, B. (eds) (1993) Relaciones Industriales y Productividad en el Norte de Mexico: Tendencias y Problemas, Mexico City: Fundacion Friedrich Ebert. Cusumano, M. (1988) ‘Manufacturing Innovation: Lessons from the Japanese Auto Industry’, Sloan Management Review: Fall. Dahlman, C.J. and Brimble, P. (1991) ‘Technology Strategy and Policy for Industrial Competitiveness: A Case Study of Thailand’, in World Bank (1991). Dahlman, C.J., Ross-Larson, B. and Westphal, L.E. (1987) ‘Managing Technological Development: Lessons from the Newly Industrializing Countries’, World Development 15 (6): 759–75. Dalton, D. (1981) ‘Electronics, Microelectronics and Microprocessors: What They Are’, in Twiss (1981). Davies, S., Lyons, B., Dixon, H. and Geroski, P. (eds) (1988) Economics of Industrial Organisation, London: Longman. De Garmo, E.P., Temple Black, J. and Kosher, R.A. (1984) Materials and Processes in Manufacturing, New York: Macmillan Publishing Company. De Meyer, A., Nakane, J., Miller, J., and Ferdows, K. (1989) ‘Flexibility: the Next Competitive Battle. The Manufacturing Futures Survey’, Strategic Management Journal 10 (2). De Toni, A. and Zipponi, L. (1991) ‘Product Standardization and Process Similitude’, Integrated Manufacturing Systems 2 (1). Department of Scientific and Industrial Research (1991) Research and Development Statistics, New Delhi.

432 BIBLIOGRAPHY

Dertouzos, M., Lester, R., and Solow, R. (1990) Made in America: Regaining the Productive Edge, Cambridge, MA: The MIT Press. Desai, A.V. (1981) Technology import policy in the Sixties and Seventies: changes and their consequences (Mimeo), New Dehli: National Council of Applied Economic Research. Dijk, M.P. van, Asselbergs, C. and Sleuwaegen, L. (1992) ‘Vertical Adjustments in the European Automobile Production’, A Contribution to the ECOZOEKDAG, June 3. Doll, W. and Vonderembse, M. (1987) ‘Forging a Partnership to Achieve Competitive Advantage: The CIM Challenge’, MIS Quarterly (June). Domínguez, L. (1993) ‘Mexico’, in Watanabe (1993). Domínguez, L. and Brown, F. (1992) ‘Employment and Income Effects of Structural and Technological Changes in Footwear Manufacturing in Mexico’, World Employment Programme, WEP 2–22/wp.224 (February), Geneva: ILO. ——(1995) La naturaleza micro y macroeconómica del cambio técnico: un caso de la industria metalmecánica (Mimeo), November. ——(1996) La naturaleza micro y macroeconómica del cambio técnico: un caso de la industria de autopartes (Mimeo), March. Dosi, G. (1984) Technical Change and Industrial Transformation: The Theory and an Application to the Semiconductor Industry, London: Macmillan Press. ——(1988) ‘Sources, Procedures and Microeconomic Effects of Innovation’, Journal of Economic Literature XXVI (3). Dosi, G., Freeman, C, Nelson, R., Silverberg, G. and Soete, L. (eds) (1988) Technical Change and Economic Theory, London: Frances Pinter Publishers. Dosi, G., Pavitt, K., and Soete, L. (1990) The Economics of Technical Change and International Trade, New York: Harvester Wheatsheaf. Dosi, G. and Soete, L. (1988) ‘Technical Change and International Trade’, in Dosi et al. (1988). Dostal, W. (1988) ‘New Technology and Development in Employment. ScenarioOriented Forecast of Long-Term Development in Employment, with Particular Attention to Expert Systems, Technologies of Light and Biotechnology’, FAST Series, no. 22, Brussels: Commission of the European Communities. Douglas, S. and Craig, C.S. (1989) ‘Evolution of Global Marketing Strategy: Scale, Scope and Synergy’, Columbia Journal of World Business (Fall). Dudley, L. (1972) ‘Learning and Productivity Change in Metal Products’, The American Economic Review LXII (4): 662–9. Duruiz, L. and Yentürk, N. (1992) Facing the Challenge: Turkish Automobile, Steel and Clothing Industries’ Responses to the Post-Fordist Restructuring, Istanbul: Ford Foundation. Eatwell, J., Milgate, M. and Newman, P. (eds) (1987) The New Palgrave: A Dictionary of Economics, London: Macmillan Press. ECE (Economic Commission for Europe) (1985) ‘Production and Use of Industrial Robots’, ECE/ENG.AUT/15, Geneva: United Nations. ——(1986) ‘Recent Trends in Flexible Manufacturing’, ECE/ENG.AUT/22, Geneva: United Nations. ——(1988) ‘Engineering Industries, Dynamics of the Eighties: the Long-Term Assessment of the Role and Place of Engineering Industries in National and World Economies, 1979–1986’, ECE/ENG.AUT/35, Geneva: United Nations.

BIBLIOGRAPHY 433

——(1992a) ‘East-West Trade in Machine Tools 1970–1990’, Occasional Paper, Geneva: United Nations. ——(1992b) ‘World Trade in Machine Tools 1970–1990, Analysis of Changing Trade Patterns among the Seven Largest Exporting Countries’, Occasional Paper, Geneva: United Nations. ——(1992c) ‘Annual Review of Engineering Industries and Automation 1990’, ECE/ ENG.AUT/47, vols I and II, Geneva: United Nations. ——(1994) ‘World Engineering Industries and Automation, Performance and Prospects, 1992–1994’, ECE/ENG.AUT/52, Geneva: United Nations. ——(1995) ‘World Engineering Industries and Automation, Performance and Prospects, 1993–1995’, ECE/ENG.AUT/56, Geneva: United Nations. ——(1996) ‘World Engineering Industries and Automation, Performance and Prospects 1994–1996’, ECE/ENG.AUT/57, Geneva: United Nations. Edquist, C. (1992) ‘Technological and Organizational Innovations, Productivity and Employment’, World Employment Programme, WEP 2–22/wep.233 (July), Geneva: ILO. Edquist, C. and Jacobsson, S. (1988) Flexible Automation. The Global Diffusion of New Technology in the Engineering Industry, Oxford: Blackwell. Edwards, C. (1955) ‘Conglomerate Bigness as a Source of Power’, in National Bureau of Economic Research (1955). ——(1956) Big Business and the Policy of Competition, Cleveland, OH: The Press of Western Reserve University. Ehrnberg, E. and Jacobsson, S. (1993) ‘Technological Discontinuity and Comparative Strategy-Revival through FMS for the European Machine Tool Industry?’, Technological Forecasting and Social Change 44. Ellis, S. and Fellner, W. (1953) ‘External Economies and Diseconomies’, in Stigler and Boulding (1953). Enos, J.L. and Park, W.-H. (1988) The Adoption and Diffusion of Imported Technology: The Case of Korea, London: Routledge. Ernst, D. and O’Connor, D. (1992) Competing in the electronics industry—the experience of newly industrializing economies, (Mimeo), Paris: OECD Development Centre. Eser, K. (1990) Takim Tezgahlari Sanayi, Ankara: T. Is Bakasi Iktisadi Arastirmalar, Müdürügü. European Commission (1994) ‘The European Report on Science and Technology Indicators 1994’, EUR 15897 EN (October), Brussels: European Commission. Evenson, R. and Ranis, G. (eds) (1990) Science and Technology, London: Intermediate Technology Publications. Fagerberg, J. (1992) ‘The Home Market Hypothesis Reexamined: the Impact of Domestic User-Producer Interaction on Export Specialization’, in Lundvall (1992). Ferguson, C. (1988) ‘From the People Who Brought You Voodoo Economics’, Harvard Business Review 3: May-June. Ferraz, J.C., Rush, H. and Miles, I. (1991) Development, Technology and Flexibility: Brazil Faces the Industrial Divide, London: Routledge. Financial Times (1993) ‘Survey on World Automotive Suppliers’, Monday 28 June. ——(1995) ‘Survey on Automotive Components’, Thursday 29 June. Fleury, A.C.C. (1988) ‘The Impacts of Microelectronics on Employment and Income in the Brazilian Metal Engineering Industry’, World Employment Programme Working Paper, WEP2–22/wp.188, Geneva: ILO.

434 BIBLIOGRAPHY

——(1993) ‘Brazil’, in Watanabe (1993) ——(1995) ‘Quality and Productivity in the Competitive Strategies of Brazilian Industrial Enterprises’, World Development 23 (1). Fleury, A.C.C. and Humphrey, J. (1993) ‘Human Resources and the Diffusion and Adaptation of New Quality Methods in Brazilian Manufacturing’, Research Report 24, Brighton: Institute of Development Studies, University of Sussex. Forrester T. (ed.) (1980) The Microelectronics Revolution, Oxford: Basil Blackwell. ——(1990) The Information Technology Revolution, Oxford: Blackwell. Foxley, A., McPherson, M.S. and O’Donnell, G. (eds) (1986) Development, Democracy, and the Art of Trespassing: Essays in Honor of Albert O.Hirschman, Notre Dame: University of Notre Dame Press. Fransman, M. (1986a) ‘International Competitiveness, Technical Change and the State: the Machine Tool Industry in Japan’, World Development 14 (12). ——(ed.) (1986b) Machinery and Economic Development, London: Macmillan Press. ——(1991) ‘Biotechnology-Generation, Diffusion and Policy: An Interpretive Survey’, UNU/INTECH, Working Paper 1 (June), Maastricht: UNU/INTECH. Fransman, M. and King, K. (eds) (1984) Technological Capability in the Third World, London: MacMillan Press. Freeman, C. (1982) The Economics of Industrial Innovation, London: Frances Pinter. ——(ed.) (1990) The Economics of Innovation, Aldershot: Edward Elgar Publishing Co. ——(1992) The Economics of Hope: Essays on Technical Change, Economic Growth and the Environment, London: Pinter Publishers. Freeman, C. and Perez, C. (1988) ‘Structural Crisis of Adjustment, Business Cycles and Investment Behaviour’, in Dosi et al. (1988). Freeman, C. and Soete, L. (eds) (1987) Technical Change and Full Employment, Oxford: Blackwell. Geroski, P.A. and Schwalbach, J. (eds) (1991) Entry and Market Contestability: An International Comparison, Oxford: Blackwell. Ghoshal, S. (1987) ‘Global Strategy: an Organizing Framework’, Strategic Management Journal 8. Gilder, G. (1988) ‘The Revitalization of Everything: the Law of the Microcosm’, Harvard Business Review 2: March-April. Gitahy, L. and Rabelo, F. (1991) ‘Educação e Desenvolvimento Tecnológico: O Caso da Indústria de Autopeças’, Textos para Discussâo 11, Campinas: DPCT/IG/ UNICAMP. Gold, B. (1981) ‘Changing Perspectives on Size, Scale, and Returns: an Interpretive Survey’, Journal of Economic Literature XIX: March. Goldhar, J. and Schlie, T. (1991a) ‘Computer Technology and International Competition: Part 1: the Factory of the Future’, Integrated Manufacturing Systems 2(1). ——(1991b) ‘Computer Technology and International Competition: Part 2: Managing the “Factory of the Future” to Achieve Competitive Advantage’, Integrated Manufacturing Systems 2 (2). González, B. and Ramírez, J.C. (eds) (1990) Subcontratación y Empresas Transnacionales, Mexico City: Colegio de la Frontera Norte and Fundación Friedrich Ebert. Government of India (a), Economic Surveys, 1990–1994, New Delhi: Ministry of Finance. ——(b) Foreign Collaborations, various years, New Delhi: Department of Scientific and Industrial Research.

BIBLIOGRAPHY 435

——(1991) Research and Development Statistics, New Delhi: Department of Scientific and Industrial Research. Groover, M.P. (1987) Automation, Production Systems and Computer Integrated Manufacturing, London; Prentice-Hall International. ——(1996) Fundamentals of Modern Manufacturing: Materials, Processes, and Systems, Upper Saddle River: Prentice Hall. Gu, S. (1994a) ‘Spin-Off Enterprises in China: Channelling the Components of R&D Institutions into Innovative Businesses’, Working Paper 16 (December), Maastricht: UNU/INTECH. ——(1994b) ‘The Emergence of New Technology Enterprises in China: A Study of Endogenous Capability Building via Restructuring’, Journal of Development Studies 32 (4). ——(1995) ‘A Review of Reform Policy for the S&T System in China: From Paid Transactions for Technology to Organizational Restructuring’, Working Paper 17 (January), Maastricht: UNU/INTECH. Gu, S. and Steinmueller, W.E. (1997) China’s national innovation system approach to participation in information technology: the innovative recombination of technological capability, (Mimeo), February. Hall, P. (1994) Innovation, Economics and Evolution: Theoretical Perspectives on Changing Technology in Economic Systems, London: Harvester Wheatsheaf. Harrison, B. (1994) Lean and Mean: The Changing Landscape of Corporate Power in the Age of Flexibility, New York: Basic Books. Harvard Business Review (1991) The New Manufacturing, Boston, MA: Harvard Business Review Paperback. Hay, D.A. and Morris, D.J. (1991) Industrial Economics and Organization: Theory and Evidence, Oxford: Oxford University Press. Hayes, R., Wheelright, S. and Clark, K. (1988) Dynamic Manufacturing, New York and London: The Free Press. Haywood, B., Vuorinen, P., and Toth-Hizsnyik, E. (1991) ‘Automation and Developing Countries: With a Specific Focus on Africa, and the Textile, Clothing and Footwear Industries’, Final Report, Laxenburg and Vienna: International Institute for Applied Systems Analysis/UNIDO, August. Hilbrink, J.O. (1989) ‘Economic Impact and Technical Change’, IEEE Transactions on Engineering Management 36 (1). Hodgson G. and Screpanti, E. (eds) (1991) Rethinking Economics, Aldershot: Edward Elgar. Hoffman, K. (1989a) ‘Technological Advance and Organizational Innovation in the Engineering Industry’, World Bank Industry and Energy Department Working Paper, Industry Series Paper 4, Washington, DC: World Bank. ——(1989b) ‘Technological and Organizational Change in the Global Textile—Clothing Industry. Implications for Industrial Policy in Developing Countries’, in UNIDO (1989). Hoffman, K. and Kaplinsky, R. (1988) Driving Force: The Global Restructuring of Technology, Labor and Investment in the Automobile and Components Industries, Boulder, CO: Westview Press. Holland, M. (1989) When the Machine Stopped: A Cautionary Tale from Industrial America, Boston, MA: Harvard Business School Press.

436 BIBLIOGRAPHY

Humphrey, J. (1994) ‘A Gestao de Mao-de-Obra e os Sistemas de Producao no Terceiro Mundo’, Estudios Avancados 8 (21), São Paulo: Instituto de Estudos Avancados, University of São Paulo. ——(1995) ‘Industrial Reorganization in Developing Countries: From Models to Trajectories’, World Development 23 (1). IBGE (Instituto Brasileiro de Geografia e Estatística) Industrial Surveys, various years, Rio de Janeiro: IBGE. IMRS (Industrial Market Research Services Co. Ltd) (1993) CNC Metal Cutting Machine Tools (September), Bangkok: IMRS. ——(1994) CIM Manufacturing Systems, (March), Bangkok: IMRS. IMTMA (Indian Machine Tool Manufacturers’ Association) (1993) 46th Annual Report 1991–92, New Delhi: IMTMA. Indian Machine Tool Manufacturers’ Association Bulletin, various years, New Delhi. INEGI (Instituto Nacional de Geografía y Estadística) (1994) National Survey on Employment, Wages, Technology and Qualifications, Mexico City: INEGI. Inohana, H. (1993) Desarrollo de los RRHH en las Compañías Japonesas, Caracas: FIM Productividad. Iranzo, C. (1988) ‘Cambio Tecnológico y Trabajo’, in Calero (1988). Jacobsson, S. (1980) ‘The Use and Production of Numerically Controlled Machine Tools in Argentina’, Economic Commission for Latin America, Working Paper 37, (December), Buenos Aires: ECLA. ——(1982) ‘Technical Change and Technology Policy: The Case of Numerically Controlled Lathes in Argentina’, Economic Commission for Latin America, Working Paper 44, (March) Buenos Aires: ECLA. ——(1986) Electronics and Industrial Policy: The Case of Computer Controlled Lathes, London: Allen & Unwin. ——(1989) ‘Technological Change in the Machine Tool Industry: Implications for Industrial Policy in Developing Countries’, in UNIDO (1989). ——(1991) ‘Government Policy and Performance of the Indian Engineering Industry’, Research Policy 20. ——(1993) ‘The Length of the Infant Industry Period: Evidence from the Engineering Industry in South Korea’, World Development 21 (3): 407–19. Jacobsson, S. and Alam, G. (1994) Liberalization and industrial development in the Third World: a comparison of the Indian and South Korean engineering industries (Mimeo), Göteborg and New Delhi: Chalmers University of Technology/Centre for Technology Studies. Jaikumar, R. (1986) ‘Postindustrial Manufacturing’, Harvard Business Review 5: November/December. Jain, A., Kharbanda, V.P., Cheng-Ben, Q. and Ai-Lan, H. (1995) Industrialisation in India and China: A Comparative Study of Machine Building and Electronic Industries, New Delhi: Har-Anand Publications. Jalan, B. (ed.) (1993) The Indian Economy: Problems and Prospects, New Delhi: Penguin Books. James, D.D. (1991) ‘Capital Goods Production and Technological Learning: the Case of Mexico’, Journal of Economic Issues XXV (4). James, J. (1991) ‘Microelectronics and the Third World: an Integrative Survey of Literature’, Working Paper 2, Maastricht: UNU/INTECH.

BIBLIOGRAPHY 437

James, J. and Bhalla, A. (1991) ‘Microelectronics, Flexible Specialization and Small-scale Industrialization in the Third World’, World Employment Programme, (August), Geneva: ILO. Japan External Trade Organization (1990) Your Market in Japan: Machine Tools, Tokyo: JETRO. Japan Machine Tool Builders’ Association (1992) Guide to Japanese Machine Tools 92– 93, Tokyo: JMTBA. ——(1993) Machine Tool Industry in Japan, Tokyo: JMTBA. Jarnehammar, A. (1995) Towards a Framework for Analysing the Diffusion of Organisational Innovations, Göteborg: Chalmers University of Technology. Jelinek, M. (1987) ‘Production Innovation and Economies of Scope: Beyond the “Technological Fix”’, Engineering Costs and Production Economics 12. Jenkins, R. (1993) ‘International Competitiveness in the Automotive Industry’, Report prepared for the Regional and Country Studies Branch (March), Vienna, UNIDO. Jones, D.T. (1984) Automotive components industry study: conclusions, (Mimeo), SPRU, Brighton: University of Sussex. Jones, D.T. and Womack, J.P. (1984) ‘Developing Countries and the Future of the Automobile Industry’, World Development 13 (3): 393–407. Kaeten, M. (1993) ‘The Age of the Specialized Generalist’, Training 30 (12): 48–53. Kang, S.M. and Steinmueller, W.E. (1991) ‘Applications of Modern Industrial Technology in Developing Nations: Government Policy and its Consequences in Brazil’, Publication 266, Stanford, CA: Center for Economic Policy Research. Kaplinsky, R. (1981) ‘Radical Technical Change and Export-oriented Industrialization: the Impact of Microelectronics’, Viertel Jahres Berichte 83: March. ——(1982) Computer-Aided Design: Electronics, Comparative Advantage and Development, New York: Macmillan Publishing Co. ——(1984) Automation: The Technology and Society, Harlow: Longman. ——(1987) Micro-Electronics and Employment Revisited: A Review, World Employment Programme, Geneva: ILO. ——(1989) ‘Technological Revolution and the International Division of Labour in Manufacturing: a Place for the Third World?’, European Journal of Development Research 1 (1): 5–37. ——(1990a) The Economies of Small: Appropriate Technology in a Changing World, London; Intermediate Technology Publications. ——(1990b) ‘Flexible Specialisation: Productivity Policies and Institutional Aspects’, Report of UNIDO Mission to Jamaica, July-August 1990 (August). ——(1991) From mass production to flexible specialization: micro-level restructuring in a British engineering firm, (Mimeo) (April), Brighton: Institute of Development Studies, University of Sussex. ——(1994) Easternisation: The Spread of Japanese Management Techniques to Developing Countries, Ilford: Frank Cass Ltd. ——(1995) ‘Technique and System: The Spread of Japanese Management Techniques to Developing Countries’, World Development 23 (1). Karshenas, M. and Stoneman, P. (1995) ‘Technological Diffusion’, in Stoneman (1995). Katz, J. (1982) ‘Cambio Técnico en la Industria Metalmecánica Latinoamericana: Resultado de un Programa de Estudios de Casos’, Programa sobre Investigaciones sobre Desarrollo Científico y Tecnológico en America Latina, Monografía 51, Buenos Aires: IADB/ECLAC/CIIU/UNDP and ECLAC.

438 BIBLIOGRAPHY

——(1983) ‘Cambio Tecnológico en la Industria Metalmecánica Latinoamericana’, Revista de la CEPAL XX: April. ——(ed.) (1987) Technology Generation in Latin American Industries, London: Macmillan Press. Katz, J. and Vera, H. (1995) Evolución histórica de una planta metalmecánica Chilena, (Mimeo), Santiago, ECLAC. Kay, J. (1993) Foundations of Corporate Success: How Business Strategies Add Value, New York: Oxford University Press. Kay, N. (1988) ‘The R&D Function: Corporate Strategy and Structure’, in Dosi et al. (1988). Kaya, E. (1992) ‘Productivity, Production, Employment Indicators in Manufacturing Industry and Sub-Sectors’, Publication 459, Ankara: National Productivity Centre Publications. Kehoe, T.J. (1987) ‘Comparative Statics’, in Eatwell et al. (1987), vol. III: 151–52. Kelley, M. and Brooks, H. (1992) ‘‘Diffusion of NC and CNC Machine Tool Technologies in Large and Small Firms’, in Ayres et al. (1992). Kimberly, J.R. (1987) ‘Organizational and Contextual Influences on the Diffusion of Technological Innovation’, in Pennings and Buitendam (1987). Kirkpatrick, C.H., Lee, N. and Nixson, F.I. (1984) Industrial Structure and Policy in Less Developed Countries, London: Allen & Unwin. Kisaer, H. (1993) ‘Manufacture of Fabricated Metal Products, Machinery and Equipment, Transport, Professional and Measuring and Controlling Equipment Not Elsewhere Classified’, Publication 500, Ankara: National Productivity Centre Publications. Kline, S. and Rosenberg, N. (1986) ‘An Overview of Innovation’, in Landau and Rosenberg (1986). Korea Machine Tool Manufacturers’ Association (1993) Machine Tool Industry: Korea 1993, Seoul: KMTMA. Koutsoyiannis, A. (1980) Modern Microeconomics, London: Macmillan Press. ——(1982) Non-Price Decisions: The Firm in a Modern Context, London: Macmillan Press. Krafeik, J. (1988) ‘Triumph of the Lean Production System’, Sloan Management Review: Fall. Krinsky, I. and Miltenburg, J. (1991) ‘Alternate Method for the Justification of Advanced Manufacturing Technologies’, International Journal of Production Research 29 (5). Kumar, N. (1995) ‘International Linkages, Technology and Exports of Developing Countries: Trends and Policy Implications’, Discussion Paper #9507, (August), Maastricht: UNU/INTECH. Kumpe, T. and Bolwijn, P.T. (1988) ‘Manufacturing: the New Case for Vertical Integration’, Harvard Business Review 2: March-April. Lall, S. (1982) Indian Government’s Trade and S&T Policies that May Affect the Acquisition of Technological Capabilities, New Delhi: Indian Council of International Economic Relations. Landau, R. and Rosenberg, N. (eds) (1986) The Positive Sum Strategy, Washington DC: National Academy Press. Landes, D.S. (1969) The Unbound Prometheus: Technological Change and Industrial Development in Western Europe from 1750 to the Present, Cambridge: Cambridge University Press.

BIBLIOGRAPHY 439

Lauridsen, L.S. (1995) ‘Liberalization and Industrial Development in the Third World: A Comparison of the Indian and South Korean Engineering Industries, book review of Jacobsson, S. and G. Alam’, The European Journal of Development Research 7 (2): 438–40. Lee, J. (1996) ‘Technology Imports and R&D Efforts of Korean Manufacturing Firms’, Journal of Development Economics 50. Lieberman, M.B. (1984) ‘The Learning Curve and Pricing in the Chemical Processing Industries’, Rand Journal of Economics 15 (2). Little, I.M.D., Mazumdar, D. and Page, J. (1987) Small Manufacturing Enterprises: A Comparative Analysis of India and Other Economies, New York: Oxford University Press. Lorentzen, A. (1984) ‘The Capital Goods Sector in Mexico: Significance and Conditions of Development’, Project Paper A.84.3, Copenhagen: Centre for Development Research. Lucke, M. (1987) ‘Scale Efficiency in the Brazilian Motor Vehicle Industry: an International Comparison’, Kiel Working Paper No.298 (August), Kiel Institute of World Economics. Lundvall, B-Å. (ed.) (1992) National Systems of Innovation. Towards a Theory of Innovation and Interactive Learning, London: Pinter Publishers. Lyons, B. (1988) ‘Barriers to Entry’, in Davies et al. (1988). Machines Production (1993) ‘Edition spéciale: Guide des Machines-Outils a CNC et Manuelles’, Machines Production 604: September. Machline, C., Udry, O. and Rattner, H. (1990) ‘Brasil: Difusión de Máquinas-Herramienta de Control Numérico y Política Gubernamental’, in Boon and Mercado (1990). Mair, A. (1991) ‘The Japanization of Volkswagen?’, Coventry: Local Government Centre, Warwick Business School, University of Warwick. Mansell, R. and Steinmueller, W.E. (1996) ‘The Way Forward: Socio-Economic and Policy Issues & Advanced Communication Technologies and Services’, A Report for the ACTS FAIR Project AC093, Brighton and Maastricht: SPRU and MERIT. Mansfield, E. (1988) Microeconomics: Theory/Applications, New York: W.W. Norton & Co. Markowski, S. and Jubb, C. (1989) ‘The Impact of Microelectronics on Scale in Manufacturing Industry’, Australian Journal of Management 14 (2). Marshall, A. (1890) Principles of Economics, London: Macmillan Press. Maxwell, P. (1981) ‘Technology Policy and Firm Learning Efforts in Less-Developed Countries: A Case Study of the Experience of the Argentine Steel Firm, Acindar SA’, PhD thesis, Brighton: University of Sussex. McNeil, I. (1990) ‘Basic Tools, Devices and Mechanisms’, in I.McNeil (ed.), An Encyclopedia of the History of Technology, London: Routledge. Mercado, A. (1990a) ‘‘Mexico: Repercusiones Económicas de la Difusión de MHCNC y Política de Automatización’, in Boon, and Mercado (1990). ——(1990b) ‘La Tecnología Asistida por Computadora en Mexico y sus Implicaciones Laborales y Educativas’, Informe de Investigación 88, Paris: Instituto Internacional de Planeamiento de La Educación. Metalworking Engineering and Marketing (1993) ‘Supply and Demand in Main Sectors of FA Industry’, MEM 15 (5). Meyer-Stamer, J. (1995) ‘Micro-level Innovations and Competitiveness’, World Development 23 (1).

440 BIBLIOGRAPHY

Meyer-Stamer, J., Rauh, C., Riad, H., Schmitt, S. and Welte, T. (1991) Comprehensive Modernization on the Shop Floor: A Case Study in the Brazilian Machinery Industry, Berlin: German Development Institute. Micheli, J. (1994) ‘Nueva Manufactura, Globalización y Produción de Automóviles en Mexico, Facultad de Economía, Mexico City: Universidad Nacional Autonoma de Mexico. Milgrom, P. and Roberts, J. (1990) ‘The Economics of Modern Manufacturing Technology: Strategy and Organization’, The American Economic Review 80 (3). Milgrom, P., Qian Y., and Roberts, J. (1991) ‘Complementarities, Momentum, and the Revolution of Modern Manufacturing’, American Economic Association Papers and Proceedings, The American Economic Review 81 (2). Mital, A. (1992) ‘Economics of Flexible Assembly Automation: Influence of Production and Market Factors’, in Parsaei and Mital (1992). Mitter, S., (ed.) (1992) Computer-Aided Manufacturing and Women’s Employment: The Clothing Industry in Four European Community Countries, London: SpringerVerlag. Mody, A. (1991) Learning through alliances (Mimeo), World Bank, September. Mody, A. and Dahlman, C. (1992) ‘Performance and Potential of Information Technology: An International Perspective’, World Development 20 (12). Mody, A., Suri, R. and Sanders, J. (1992) ‘Keeping Pace with Change: Organizational and Technological Imperatives’, World Development 20 (12). Mody, A., Suri, R., Sanders, J. and Van Zoest, D. (1991a) International competition in the footwear industry: keeping pace with technological change (Mimeo), World Bank, June. Mody, A., Suri, R., Sanders, J. and Tatikonda, M. (1991b) International competition in printed circuit board assembly: keeping pace with technological change (Mimeo) World Bank, , July. Mody, A., Suri, R., Sanders, J. and Thompson, E. (1991c) International trends in steel mini-mills: keeping pace with technological change (Mimeo), World Bank, July. Mody, A., Suri, R., Sanders, J., Rao, C. and Contreras, F. (1991d) International competition in the bicycle industry: keeping pace with technological change (Mimeo), World Bank, July. Mody, A. and Wheeler, D. (1990) Automation and World Competition, London: Macmillan Press. Mohan, R. (1993) ‘Industrial Policy and Controls’, in Jalan (1993). Mohr, L.B. (1987) ‘Innovation Theory: An Assessment from the Vantage Point of New Electronic Technology in Organizations’, in Pennings and Buitendam (1987). Morroni, M. (1991) ‘Production Flexibility’, in Hodgson and Screpanti (1991). ——(1992) Production Process and Technical Change, Cambridge: Cambridge University Press. Mueller, D.C. (1991) ‘Entry, Exit, and the Competitive Process’, in Geroski and Schwalbach (1991). Mytelka, L.K. (1993) Rethinking development: a role for innovation networking in the ‘other-two-thirds’ (Mimeo). NAFIN (Nacional Financiera) (1985) ‘La Industria de Bienes Capital’, Mexico City: NAFIN. ——(1987) ‘Estudio de Capacidad Instalada, Potencial Tecnológico y Ventajas Comparativas de la Industria de Bienes de Capital’, Mexico City: NAFIN.

BIBLIOGRAPHY 441

Naim, M. (1993) Paper Tigers and Minotaurs: The Politics of Venezuela’s Economic Reforms, Washington, DC: Carnegie Endowment for International Peace. National Bureau of Economic Research (1955) Business Concentration and Price Policy, Princeton, NJ: Princeton University Press. Nayyar, D. (1978) ‘Industrial Development in India: Some Reflections on Growth Stagnation’, Economic and Political Weekly (Special Issue). Nelson, R.R. (1991) ‘Why Do Firms Differ, and How Does it Matter?’, Strategic Management Journal 12:61–74. ——(1994) ‘The Co-evolution of Technology, Industrial Structure, and Supporting Institutions’, Industrial and Corporate Change 3 (1). Nelson, R.R. and Winter, S. (1982) An Evolutionary Theory of Economic Change, Cambridge, MA: The Belknap Press of Harvard University Press. Noori, H. (1990) ‘Economies of Integration: a New Manufacturing Focus’, International Journal of Technology Management 5 (5). OCEI (a) Encuesta Industrial, 1989–92, Caracas: Oficina Central de Estadística e Informática. ——(b) Estadísticas de Comercio Exterior, 1989–92, Caracas: OCEI. ——(c) Anuario del Comercio Exterior de Venezuela, 1989–91, Caracas: OCEI. O’Connor, D.C. (1989) ‘Microelectronics-based Innovations: Strategic Implications for Selected Industries in the Second-Tier Newly Industrializing Countries (NICs) of Southeast Asia’, Final version of a report prepared for the OECD Development Centre Research Project on ‘Technological Change and the Electronics Sector— Perspectives and Policy Options for Newly Industrializing Countries (NICs)’, November. OECD (Organization for Economic Cooperation and Development) (1988) New Technologies in the 1990s, Paris: OECD. ——(1989) Government Policies and the Diffusion of Microelectronics, Paris: OECD. ——(1991a) Managing Manpower for Advanced Manufacturing Technology, Paris: OECD. ——(1991b) Technology and the Economy: The Key Relationships, Paris: OECD. ——(1991c) Technology and Productivity, Paris: OECD. Owen, N. (1983) Economies of Scale, Competitiveness, and Trade Patterns within the European Community, Oxford: Clarendon Press. Paéz. T., Gómez, L. and Raydan, E. (1992) Nueva Gestión de Recursos Humanos, Caracas: FIM Productividad. Parsaei, H. and Mital, A. (eds) (1992) Economics of Advanced Manufacturing Systems, London; Chapman & Hall. Patel, P. and Pavitt, K. (1991) ‘Europe’s Technological Performance’, in Freeman et al. (1991). Pavitt, K. (1984) ‘Sectoral Patterns of Technical Change: Towards a Taxonomy and a Theory’, Research Policy 13 (6). Pennings, J.M. (1987a) ‘On the Nature of New Technology as Organizational Innovation’, in Pennings and Buitendam (1987). ——(1987b) ‘Technological Innovations in Manufacturing’, in Pennings and Buitendam (1987). Pennings, J.M. and Buitendam, A. (eds) (1987) New Technology as Organizational Innovation: The Development and Diffusion of Microelectronics, Cambridge, MA: Ballinger Publishing Company.

442 BIBLIOGRAPHY

Pérez, C. (1985) ‘Microelectronics, Long Waves and World Structural Change: New Perspectives for Developing Countries’, World Development 13 (3). ——(1986) ‘Las Nuevas Tecnologías: Una Vision de Conjunto’, Anuario RIAL-CEPAL, pp. 43–89. ——(1989) ‘Technical Change, Competitive Restructuring, and Institutional Reform in Developing Countries, Strategic Planning and Review’, Discussion Paper 4 (December), Washington, DC: World Bank. Pérez, C. and Soete, L. (1988) ‘Catching Up in Technology: Entry Barriers and Windows of Opportunity’, in Dosi et al. (1988). Pindyck, R.S. and Rubinfeld, D.L. (1989) Microeconomics, New York: Macmillan Publishing Co. Piore, M.J. and Sabel, C.F. (1984) The Second Industrial Divide: Possibilities for Prosperity, New York: Basic Books Publishers Inc. Pirela, A., Arvanitis, R., Mercado, A. and Rengifo, R. (1988) ‘La Conducta Empresarial ante el Hecho Tecnológico’, Caracas: CENDES and UCV. Porteous, M. (1992) ‘Revolution in a Recession? Advanced Technologies and Brazil’s Machine Tool Sector in the Crisis’, in Schmitz and Cassiolato (1992). Porter, M.E. (1980) Competitive Strategy. Techniques For Analyzing Industries and Competitors, New York: The Free Press. Posthuma, A.C. (1991) ‘Changing Production Practices and Competitive Strategies in the Brazilian Auto Components Industry’, PhD thesis, Brighton: University of Sussex. ——(1994) ‘The Next Hurdle: Restructuring and Changing Market Conditions in the Brazilian Auto Components Industry’ (Mimeo), Paper prepared for ECLAC, United Nations, São Paulo. ——(1995) ‘Japanese Techniques in Africa? Human Resources and Industrial Restructuring in Zimbabwe’, World Development 23 (1). Pozas, M.A. (1990) ‘Estrategias Empresariales ante la Apertura Externa’, in Gonzáles and Ramírez (1990). ——(1993) ‘Mecanismos de Innovación y Transferencia Tecnológica y sus Efectos Sobre la Organización del Trabajo’, in Covarrubias and Lara (1993). Prado, A.J.C. (1989) ‘A Difusão da Automação Flexível na Indústria Brasileira de Autopeças’, MA thesis, Instituto de Economia, Campinas: University of Campinas. Pratten, C.F. (1971) Economies of Scale in Manufacturing Industry, Cambridge: Cambridge University Press. ——(1975) ‘Economies of Scale in Manufacturing Industry’, University of Cambridge, Department of Applied Economics, Occasional Paper 28, Cambridge: Cambridge University Press. ——(1988) ‘A Survey of the Economies of Scale’, in Commission of the European Communities (1988). ——(1991a) ‘The Competitiveness of Small Firms’, University of Cambridge, Department of Applied Economics, Occasional Paper 57, Cambridge: Cambridge University Press. ——(1991b) ‘Economies of Scale and Latin American Exports’, in Schwartz (1991). Primrose, P.L. (1991) Investment in Manufacturing Technology, London: Chapman & Hall. ——(1992) The economics of batch manufacture (Mimeo), Total Technology Dept., Manchester: UMIST.

BIBLIOGRAPHY 443

Quadros Carvalho, R. (1992) ‘Why the Market Reserve Is Not Enough: Lessons from the Diffusion of Automation Technology in Brazilian Process Industries’, in Schmitz and Cassiolato (1992) ——(1993) ‘Programmable Automation and Employment Practices in Brazilian Industry’, PhD thesis, Brighton: University of Sussex. Quadros Carvalho, R. and Schmitz, H. (1989) ‘Fordism is alive in Brazil’, IDS Bulletin 20 (4) (Portuguese version published in Novos Estudios CEBRAP 27, São Paulo, 1990). Rada, J.F. (1980) The Impact of Micro-Electronics, Geneva: ILO. ——(1982) The Impact of Micro-Electronics and Information Technology: Case-studies in Latin America, Paris: UNESCO Ramírez, J.C. (ed.) (1991) Subcontración y Empresas Transnacionales, Tijuana: El Colegio de la Frontera Norte y Fundación Friedrich Ebert. ——(1993) ‘Recent Transformations in the Mexican Motor Industry’, IDS Bulletin 24 (2). Ramírez, J.C. et al. (1988) ‘Los Recientes Cambios Internactionales y sus Efectos en el Proceso Maquilador: Las Ramas Electrónica y de Autopartes en Tijuana’, Final Report, Departamento de Estudios Economicos del Colegio de la Frontera, Tijuana: Colegio de la Frontera Norte. Ramtin, R. (1991) Capitalism and Automation: Revolution in Technology and Capitalist Breakdown, London: Pluto Press. Reger, G. and Kungl, H. (1993) Sector Study: Mechanical Engineering in the EC, Karlsruhe: Fraunhofer Institute of Systems and Innovation Research. Reintjes, F. (1991) Numerical Control: Making of New Technology, New York and Oxford: Oxford University Press. Rhys, D.G. (1992) Economic changes in the motor industry: reality and myth (Mimeo), Cardiff: University of Wales Business School. ——(1993) The West’s response to Japanese competition and the boundaries of the effects of lean production (Mimeo), Cardiff: University of Wales Business School. Romer, P.M. (1986) ‘Increasing Returns and Long-run Growth’, Journal of Political Economy 94 (5). ——(1990) ‘Are Nonconvexities Important for Understanding Growth?’, AEA Papers and Proceedings 80 (2). ——(1994) ‘The Origins of Endogenous Growth’, Journal of Economic Perspectives 8 (1). Rosegger, G. (1986) The Economics of Production and Innovation: An Industrial Perspective, Oxford: Pergamon Press. Rosenberg, N. (1976) Perspectives on Technology, Cambridge: Cambridge University Press. ——(1982) Inside the Black Box: Technology and Economics, Cambridge: Cambridge University Press. ——(1988) ‘New Technologies and Old Debates’, in Bhalla and James (1988). Rosenberg, N. and Frischtak, C. (eds) (1985) International Technology Transfer: Concepts, Measures and Comparisons, New York: Praeger Special Studies. Sabel, C. (1986) ‘Changing Models of Economic Efficiency and Their Implications for Industrialization in the Third World’, in Foxley et al. (1986). Sambasiva Rao, K.V. and Deshmukh, S.G. (1993) ‘Strategic Framework for Implementing Flexible Manufacturing Systems in India’, International Journal of Operations & Production Management 14 (4): 50–63.

444 BIBLIOGRAPHY

Sandoval, S. (1986) ‘Los Enlaces Económicos y Políticos en la Ford Motor Company en Hermosillo Sonora: Internacionalización Productiva y Nuevas Tecnologías’, graduation paper, Hermosillo: El Colegio de Sonora. Saunders, R.J., Warford, J.J. and Wellenius, B. (1994) Telecommunications & Economic Development, Washington, DC: World Bank. Sayer, A. (1988) Post-Fordism in question (Mimeo), University of Sussex, July. Sayer, A. and Walker, R. (1992) The New Social Economy: Reworking the Division of Labour, Oxford: Blackwell. Scherer, F.M. (1980) Industrial Market Structure and Economic Performance, Boston: MA, Houghton Mifflin Company. ——(1991) ‘Changing Perspectives on the Firm Size Problem’, in Acs and Audretsch (1991). Scherer, F.M., Beckenstein, A., Kaufer, E. and Murphy, R.D. (1975) The Economics of Multi-Plant Operation, Cambridge, MA: Harvard University Press. Scherer, F.M. and Ross, D. (1990) Industrial Market Structure and Economic Performance, Boston, MA: Houghton Mifflin Company. Schmalensee, R. (1988) ‘Industrial Economics: an Overview’, The Economic Journal 98: September. Schmenner, R. (1988) ‘The Merit of Making Things Fast’, Sloan Management Review: Fall. Schmitz, H. and Cassiolato, J. (1992) Hi-Tech for Industrial Development: Lessons From the Brazilian Experience in Electronics and Automation, London: Routledge. Schonberger, R. (1987) ‘Frugal Manufacturing’, Harvard Business Review: SeptemberOctober. Schwalbach, J. (1988) ‘Economies of Scale and Intra-community Trade’, in Commission of the European Communities (1988). Schwartz, H.H. (ed.) (1991) Supply and Marketing Constraints on Latin American Manufacturing Exports, Washington, DC: Inter-American Development Bank. Scriberras, E. and Payne, B.D. (1985) Machine Tool Industry: Technical Change and International Competitiveness, London: The Technical Change Centre, Longman House. SEAMICO (Southeast Asia Management and Investment Company) (1993) The Status of and Potential for Industrial Robotics in Thailand, Bangkok: SEAMICO Business Information & Research. Shaiken, H. and Herzenberg, S. (1987) Automation and Global Production: Automobile Engine Production in Mexico, the United States, and Canada, San Diego, CA: Center for US-Mexican Studies. Shapiro, H. (1994) Engines of Growth: The State and Transnational Auto Companies in Brazil, Cambridge: Cambridge University Press. Shepherd, W.G. (1990) The Economics of Industrial Organization, London: Prentice-Hall International. Shingo, S. (1986) Study of Toyota Production System from Industrial Engineering Viewpoint, Tokyo: Japan Management Association. Silberston, A. (1972) ‘Economies of Scale in Theory and Practice’, The Economic Journal, 82 (Special Issue). Silva, B.E. (1992) ‘Automation and Work Organization: Contrasts between Brazilian and British Car Factories’, in Schmitz and Cassiolato (1992).

BIBLIOGRAPHY 445

Silverberg, G. and Soete, L. (eds) (1994) The Economics of Growth and Technical Change: Technologies, Nations, Agents, Aldershot: Edward Elgar Publishing Ltd. SINDIPECAS (Sindicato Nacional da Indústria de Componentes para Veículos Automotores) Desempenho do Setor de Autopecas: Estadistica Anual, various years, São Paulo. Skinner, A. (1974) ‘Introduction’, in Smith (1974). Smith, A., (1974) The Wealth of Nations [1776], New York: Penguin Books. SOBRACON (Sociedade Brasileira de Comando Numérico) (1992) Automatização Industrial e Computação Gráfica—Guia Brasileira de Automatização Industrial, São Paulo: SOBRACON. Soete, L. (1985) ‘International Diffusion of Technology, Industrial Development and Technological Leapfrogging’, World Development 13 (3): 409–22. Soete, L. and Dosi, G. (1983) Technology and Employment in the Electronics Industry, London: Frances Pinter Publishers. State Planning Organization (1993) Economic and Social Indicators (1950–1992), Ankara: SPO. Stal, E. (1993) ‘Metal Leve S/A’, CYTED Subprograma XVI, Cadernos de Gestão Tecnológica, December. Stalk, G. (1991) ‘Time—The Next Source of Competitive Advantage’, Harvard Business Review. State Institute of Statistics (a) Statistical Yearbook, 1985, 1993, Ankara: SIS. ——Annual Manufacturing Industry Statistics, 1985, 1987, 1989, Ankara: SIS. ——(c) Quarterly Manufacturing Industry Statistics, 1993, Ankara: SIS. ——(d) Foreign Trade Statistics, 1981–1991, Ankara: SIS. Steudel, H.J. and Desruelle, P. (1992) Manufacturing in the Nineties: How to Become a Mean, Lean, World-Class Competitor, New York: Van Nostrand Reinhold. Stevenson, J. and Walker, P. (1987) ‘Power Plant Economy of Scale—Revisited’, Transactions of the Annual Meeting of the American Association of Cost Engineers, vol. 31, no. J5. Stevenson, R. (1980) ‘Measuring Technological Bias’, The American Economic Review 70 (5). Stewart F. and James, J. (eds) (1982) The Economics of New Technology in Developing Countries, London: Frances Pinter Publishers. Stigler, G.J. (1939) ‘Production and Distribution in the Short Run’, Journal of Political Economy 47. ——(1951) ‘The Division of Labour is Limited by the Extent of the Market’, Journal of Political Economy LIX (3). ——(1968) The Organization of Industry, Chicago, IL: University of Chicago Press. Stigler, G.J. and Boulding, K.E. (eds) (1953) Readings in Price Theory, London: Allan & Unwin. Stiglitz, J. (1987) ‘On the Microeconomics of Technical Progress’, in Katz (1987). Stoneman, P. (1983) The Economic Analysis of Technological Change, New York: Oxford University Press. ——(1995) ‘Introduction’, in P.Stoneman, (ed.), Handbook of the Economics of Innovation and Technological Change, Oxford: Blackwell. Storper, M. (1991) Industrialization, Economic Development and the Regional Question in the Third World: From Import Substitution to Flexible Production, London: Pion Limited.

446 BIBLIOGRAPHY

Sutcliffe, R.B. (1971) Industry and Underdevelopment, London: Addison-Wesley Publishing. Sylos-Labini, P. (1962) Oligopoly and Technical Progress, Boston, MA: Harvard University Press. Talaysum, A.T., Zia Hassan, M. and Goldhar, J. (1987) ‘Uncertainty Reduction Through Flexible Manufacturing’, IEEE Transactions on Engineering Management EM-34 (2). Tamayo, F. (1991) Gestión de Manufactura: Nuevos Enfoques de Racionalización, Caracas: FIM Productividad. Tayler, P. (1985) ‘The Development of the Theory of Economies of Scale’, AstonWarwick Scale Project Working Paper WP-06, University of Warwick. TBOI (Thai Board of Investment) (1993) Investment Opportunities Study: Metal-Working and Machine Tool Industries in Thailand, Bangkok: Office of the Board of Investment, Office of the Prime Minister. TRDI (1992) ‘Development of the Machinery and the Equipment for Information Industries in Thailand’, Final Report to the National Institute for Research Advancement, Bangkok: Thailand Development Research Institute. (1993) ‘The Role of Information Technology in the Information Society in the Year 2010’, Report submitted to the Canadian International Development Agency, Bangkok: Thailand Development Research Institute. TECNOS, FEDEMETAL, FNCI (1990) La Automatización Programable en la Metalmecánica Colombiana, Bogota: Editorial Gente Nueva. Teece, D.J. (1980) ‘Economies of Scope and the Scope of the Enterprise’, Journal of Economic Behavior and Organization 1. Teubal, M., Halevi, N. and Tsiddon, D. (1986) ‘Learning and the Rise of Israel’s Exports of Sophisticated Products’, World Development 14 (12): 1,397–410. Tidd, J. (1991) Flexible Manufacturing Technologies and International Competitiveness, London: Frances Pinter Publishers. Tribe, M.A. and Alpine, R.L.W. (1986) ‘Scale Economics and the “0.6 Rule”’, Engineering Costs and Production Economics 10. Twiss, B.C. (1981) ‘Microelectronics—the Managerial Dilemma’, in B.C.Twiss (ed.), The Managerial Implications of Microelectronics, London: Macmillan Press. United Nations (1996) COMTRADE Statistics (data provided in electronic form), Geneva: United Nations. UNCTAD (United Nations Conference on Trade and Development) (1985a) ‘The Capital Goods Sector in Developing Countries: Technology Issues and Policy Options’, Study by the UNCTAD Secretariat, Geneva: United Nations. (1985b) ‘The Diffusion of Electronics Technology in the Capital Goods Sector in the Industrialized Countries’, UNCTAD/TT/65, Geneva: United Nations. UNIDO (United Nations Industrial Development Organization) (1989) ‘New Technologies and Global Industrialization’, Regional and Country Studies Branch, Industrial Policy and Perspectives Division, PPD.141, Vienna: United Nations. (1990a) ‘La Machine-Outil en Amérique Latine’, ID/WG.508/3 (Spec.), Vienna: United Nations. (1990b) ‘Co-operation in Industrial Automation between Argentina and Brazil’, ID/WG. 508/5 (Spec.), Vienna: United Nations. (1990c) ‘Machine Tools in Latin America’, Corrigendum, ID/WG.508/3 Corr.1, Vienna: United Nations.

BIBLIOGRAPHY 447

(1991a) ‘The Machine Tools Industry Worldwide’, Fourth Session on Capital Goods, ID/ WG.514/4, Prague: United Nations. (1991b) ‘Study on Trends in Technological Development in the Petrochemical Industry’, UNIDO, ID/WG.522/1 (Spec.), Vienna: United Nations. (1991c) ‘Estudio Regional sobre la Industria de Máquinas Herramienta en Asia: el Caso de la India’, ID/WG.514/1, Vienna: United Nations. (1991d) ‘La Industria Mundial de Máquinas Herramientas’, ID/WG.514/4, Vienna: United Nations. ——(1991e) ‘The Machine Tool Industry in the Republic of Korea’, ID/WG.514/5, Vienna: United Nations. ——(1993a) ‘Industrial Automation: Priorities and Programmes’, Regional and Country Studies Branch, PPD.269 (Spec.), Vienna: United Nations. ——(1993b) ‘Industrial Automation: an Introduction’, Regional and Country Studies Branch, PPD.270 (Spec.), Vienna: United Nations. ——(1993c) ‘Industry and Development’, Global Report 1993–94, Vienna: United Nations. ——(1995) Industrial Statistics (data provided in electronic form), Vienna: United Nations. United States’ Department of Commerce (1988) Manufacturing Technology 1988, Washington, DC: US Bureau of the Census. ——(1991) Manufacturing Technology: Factors Affecting Adoption 1991, Washington, DC: US Bureau of the Census. Vassilakis, S. (1987) ‘Learning-by-Doing’, in Eatwell et al. (1987), vol. I: 517–21. Vermulm, R. (1994) ‘Estratégias Empresariais no Setor de Máquinas-ferramenta no Brasil na década de 80’, Dissertation, São Paulo, University of São Paulo. Viner, J. (1953) ‘Cost Curves and Supply Curves’, in Stigler and Boulding (1953). Vuorinen, P. (1992) ‘Flexible Automation in Less Developed Countries’, in Ayres et al. (1992a). Wangwe, S.M., (ed.) (1995) Exporting Africa: Technology, Trade and Industrialization in Sub-Saharan Africa, London: Routledge and UNU Press. Watanabe, S. (1987) ‘Introduction’, in S.Watanabe, (ed.), Microelectronics, Automation and Employment in the Automobile Industry, Chichester: John Wiley & Sons. ——(1993a) ‘Microelectronics and Third World Industries: an Overview’, in S. Watanabe (ed.), Microelectronics and Third World Industries, Basingstoke: Macmillan Press. ——(1993b) Implications of Microelectronic Factory Automation Technology for Employment, Trade and Industrialisation of Developing Economies, Tokyo: Japan Machine Tool Builders’ Association. Weiss, J. (1988) Industry in Developing Countries: Theory, Policy and Evidence, London: Croom Helm. Westphal, L., Kritayakirana, K.Petchsuwan, K., Sutabutr, H. and Yuthavong, Y. (1990) ‘The Development of Technological Capability In Manufacturing: a Macroscopic Approach To Policy Research’, in Evenson and Ranis (1990). Westphal, L. and Rhee, Y. (1983) The methodology of investment analysis for a nonprocess industry (Mimeo), Washington, DC: World Bank. Whitney, D. (1991) ‘Manufacturing by Design’, in Harvard Business Review (1991). Williams, K., Cutler, T., Williams, J., and Haslam, C. (1987) ‘The End of Mass Production?’, Economy and Society 16 (3). Wolf, M. (1982) India’s Exports, New York:Oxford University Press.

448 BIBLIOGRAPHY

Womack, J.P., Jones, D. and Ross, D. (1990) The Machine That Changed the World, New York: Rawson Associates. Woodward, J. (1965) Industrial Organization: Theory and Practice, Oxford: Oxford University Press. World Bank (1980) ‘Development of the Engineering Industries in Thailand’, Report No. 2647-TH, Washington, DC: World Bank. ——(1991) Decision and Change in Thailand: Three Studies in Support of the Seventh Plan, Part II, Washington, DC: World Bank. Yeldan, E. (1994) ‘Kamu Finans Krizinin Ardindaki Etkenlerin Bölüsüm Göstergeleri Açisindan Irdelenmesi’, Isletme ve Finans 9 (100): 56–73. Yin, R.K. (1984) Case Studies Research: Design and Methods, Applied Social Science Research Methods Series, Vol. 5, London: Sage Publications.

INDEX

Abernathy, W.J. 119–20 academic conferences 68 accuracy of CNC machine tools 102, 128, 376, 378 Acs, Z.J. 28 adaptability 17–20 adaptation 71–3 administration: changes in 244 administration costs 65; India 326; Mexico 250; Turkey 359, 360; Venezuela 290–2 adoption of FA 48–9, 171; Brazil 187–98; India 306–13; Mexico 223–8; model case study 53–6; reasons for see motives for diffusion; selection see selection of FA equipment; Thailand 373–80; Turkey 338–45; Venezuela 268–76 Africa 81 age of firm 227, 382 Alänge, S. 70 Alcorta, L. 28, 29, 67, 82, 208, 290, 354 Alonso, O. 258, 260, 263, 266 American Machinist 50–1 Ansal, H. 330 approval of selection/investment 107 Asia 71, 72, 73, 81 assembly operations 44, 186 assessment of equipment see selection of FA equipment

Audretsch, D.B. 28 automated guided vehicles (AGV) 25 automated storage and retrieval (AS/R) 25 automatic tool changing 128 automation, level of see density of automation automobile industry 16, 24–5, 100, 152–6 autopart firms 42, 99, 100–2, 102–3, 116, 154–6; Brazil 180–2, 191–5, 197–8; India 307, 309–10; Mexico 226, 233, 234–6, 236–7, 240–1, 246; Turkey 338–43; Venezuela 262, 264 average incremental cost (AIC) 22 Ayres, R. 27, 135, 146 Babbage, C. 13 Bailey, E.E. 20–3 Bain, J.S. 3 Baptista, A. 260, 261 barriers to entry 150–6, 297, 328–9 batch production 125–7, 145 batch sizes 15–16, 113–16; Brazil 201–2; India 316–17; Mexico 234–6; model case study 57–8; Thailand 385, 386; Turkey 348–9; Venezuela 276–9, 295–6 Baumol, W.J. 4, 20–3, 125 BCV 260 Bell, M. 148, 149 Belussi, F. 31, 32 Benetton 32 449

450 INDEX

Bernardes, R. 178 Bessant, J.R. 26 Best, M. 120–1 Bhalla, A. 225 bicycle industry 27 Boon, G.K. 40, 161 Boratav, K. 331 Boston Consulting Group 153, 154, 156 Brazil 46, 81, 108–9, 168, 177–216, 304; adoption of FA 187–98; diffusion of FA 87–93 passim, 182–4; diffusion of FA in an international perspective 82–5 passim; economic and technical change 177– 84; engineering industry 178–82; factors underlying diffusion of FA 99– 104 passim; industrial restructuring and prospects for location of production 214–16; intra-firm diffusion process 104–8 passim; manufacturing 177–8; plant scale 203–7; product scale and scope 198–203; sampled firms 184–7; selecting and adopting FA 197–8; structure of unit costs and firm scales 210–14; total unit costs 207–10 Brimble, P. 368, 371 Brown, F. 48, 221 Bureau of Industry Economics 146 capacity: plant capacity in Brazil 203–6; plant capacity in Turkey 353–4, 356; productive and divisibility 131–2; utilization see utilization rates capital costs 2, 28–9; Brazil 210–11; India 321–2; Mexico 247–8; model case study 62–3;

and plant scale 133–7; Thailand 389, 393; Turkey 359–60; Venezuela 284–6, 287 capital equipment: indivisibilities 15; size 2, 27–8 capital goods firms 42, 116; Brazil 179–80, 187–91, 197–8; Mexico 226–7, 233–7 passim, 241, 246; Turkey 343–4 capital saving 137–8, 161 car industry 16, 24–5, 100, 152–6; see also autopart firms Carrillo, J. 220 Carlsson, B. 27, 28, 225 Carlton, D.W. 21 Carter, C.F. 127 case studies approach 39–40 castings 184, 311–12, 313, 317, 324–5 catalytic cracking 145 CECIMO (European Committee for Cooperation of the Machine Tool Industries) 50–1 Celasun, M. 331 cellular manufacturing (CM) 26, 45, 94–9 passim, 129; Mexico 228–30; Venezuela 272 Chapman, K. 144, 157, 159 chemical industry 35, 260–1 chemical processes 186 Chenery, H. 3 Cheng, T.C.E. 129 China 71, 73, 81, 82, 168, 304; supply and demand 76–7 Colombia 104 comparative statics 40 competitive advantages 241–4 competitive strategies 147–8 competitiveness 392–3; international 252, 276, 363–4, 383 complementarities 26, 97–8 component manufacturing suppliers 153–5 components, choice to produce on CNC machine tools 313

INDEX 451

computer aided design (CAD) 182–3, 224, 272, 307–8 computer aided design/manufacturing (CAD/CAM) 25, 44–5, 87, 88, 379–80, 391 computer integrated manufacture (CIM) 25 computer numerical control (CNC) machine tools 25, 44, 45, 108–9, 146; Brazil 182–3, 210–11; costs 133–7, 210–11; diffusion in surveyed countries 82–5; divisibility 131–2; factors underlying diffusion 99–104; firm-level diffusion 87–93; global CNC lathe production 400–3; global exports 422, 424, 426; global imports 80, 81, 423, 427; global production of CNC machine tools 77–81; impact on scale 30, 113–16; impact on scope 116–21; India 304, 306–7, 308–13; international diffusion 77–82; intra-firm diffusion 104–8; lathes 422, 423; Mexico 220–1, 223–5; price and efficiency effects 137–9; smallness 131; stocks of 83, 84; Thailand 368–9, 376–8; Turkey 337, 338–45; Venezuela 262–6, 270–4 computers 307–8 concentration, industrial 151 Confederation of Engineering Industries 300 consultants 48, 67 consumption, global for machine tools 71, 72, 428–9 continuous process industries 16–17, 145 Coriat, B. 228 cost leadership strategy 147–8 cost structure: Brazil 210–14; India 321–7; Mexico 247–50; Turkey 359–62; Venezuela 284–92

costs 30–2, 50, 137–9; Brazil 207–14; cost conditions in engineering 145–50; India 321–7; Mexico 226, 239, 244–53; model case study 61–5, 137; new organizational techniques and 97; reduction as motive for adoption of FA 101, 104; scale and 12–14, 122–3; scope and 17–23; technical change, scale, scope and 32– 6; Thailand 388–91; Turkey 356–62; Venezuela 284–92, 296, 297; see also under individual types of cost Council of Scientific and Industrial Research (CSIR) 299–300 countries, selection of 46–7 customized products 42, 91, 92, 102, 116; Brazilian firms 187–9; Turkish firms 344–5 cylinder blocks 310 Daewoo 148, 154 Dahlman, C.J. 107 data requirements 47–51 De Garmo, E.P. 113, 126, 127 deficiencies, freedom from 130, 146 delivery requirements 317 delivery schedules 115 delivery times 241, 244, 292–4 demand 77, 115; increased 100; local and location of production 161–3 density of automation 90, 91; India 307; and new organizational techniques 94, 95, 98–9; and plant scale 121–2, 123; and product scale 114; and product scope 116, 117; Thailand 379–80; Venezuela 270–2 Department of Industrial Promotion 368 depreciation 247

452 INDEX

de-scaling 1, 28–9, 34–5, 167, 393 de-skilling 255–6, 315 Desruelle, P. 130 development strategy: Brazil 177–8; India 299–302; Mexico 217–18; Thailand 365; Turkey 330–2; Venezuela 258–60 differentiation strategy 147–8 diffusion of flexible automation 5, 70–111; Brazil 182–4; CNC machine-tool diffusion in an international perspective 82–5; extent and motives 168–9; factors underlying 99–104; India 304–15; intra-firm diffusion 104–8; metal-cutting machine tools 71–7; Mexico 219–21; and new organizational techniques 94– 9; numerically and computer-numerically controlled machine tools 77–82; in selected firms 85–93; Thailand 368–83, 392; Turkey 336–46; Venezuela 262–6 Dijk, M.P. van 120–1 dimensional effects 15 discrete-good industries 16–17 distribution 256; costs see marketing costs divisibility 18–19 division of labour 12–13 Domínguez, L. 46, 48, 220, 221, 226, 248 Dosi, G. 138, 161 downsizing 192 Dudley, L. 149 Duruiz, L. 333 dynamic specialization gains 14–15 ECE 46, 86; ECE/OECD databank 50–1 economic batch/lot-size model 16 economic context:

Brazil 177–84; India 299–302; Mexico 217–18; Thailand 365–8; Turkey 330–2; Venezuela 258–60, 294–5 economic factors: underlying diffusion of FA 99–100, 104; underlying increases in plant scale 133– 7 economic indivisibilities 15 economic learning 148 economic theory 4 economies of scale 21–3, 125; flexible automation and 23–32; origins of economies of scale theory 12–14; scale, optimal scale and 14–17; see also scale economies of scope 17–23, 125; flexible automation and 23–32; Venezuela 279–81; see also scope Edquist, C. 30, 40, 138, 139, 164, 184; diffusion of machine tools 83, 85; skill-saving 160–1 education 174; see also training efficiency 1, 137–9, 172; efficiency indexes in Venezuela 270, 271; gains from introduction of FA 127–31; range of efficient points 19–20, 36 Ehrnberg, E. 82 electric arc furnace 34–5 electricity 159 electronic autocomponents 156 electronic components 107–8 employment 64, 138, 172–3; India 304, 307, 314; Mexico 248, 249; Turkey 333, 334, 335, 336, 354, 355; Venezuela 286 energy costs 65, 213, 288–90, 361–2 engine lathes 312 engineering industry 41–2, 43, 103; Brazil 178–82;

INDEX 453

FA and production and cost conditions 145–50; India 300; Mexico 218–19; Turkey 333–7; Venezuela 260–1, 261–2, 264 Enos, J.L. 148 entry: late entrant firms 149–50; scale, scope and firm entry 150–6 entry barriers 150–6, 297, 328–9 ‘envelope’ curve 14 Eser, K. 337 Europe 31 expectations, unmet 381 export promotion strategy 217–18, 331–2, 365 exports: global lathe exports 418–19; global machine-tool exports 410–13; global machining centre exports 424; global metal-cutting machine-tool exports 414–15; global CNC lathe exports 422; India 309; Mexico 218; other global CNC metal-cutting machine-tool exports 426; Turkey 351 external factors 50 factor prices and biases 159–61; see also capital costs; labour costs Fayol, H. 24 Federation of Thai Industries 368 Ferraz, J.C. 46 final assemblers 153–4 final products, transport costs of 157 finance, policy and 173–4 financial capacity 171–2 financial costs 250, 286 firm entry see entry, entry barriers firm scale 17, 112; Brazil 213–14; changes in 121–5; Mexico 238–44;

Thailand 392–3; Venezuela 281–4, 296 firm size 41, 171; diffusion of FA 90–2; and entry 155–6; Mexico 220–1; and new organizational techniques 96, 99; Turkey 334 firms: age of firm 227, 382; Brazilian sample 184–7; characteristics and adoption of FA 171, 294; diffusion of FA in 87–93; history 48; identification of 67–8; Indian sample 304–6, 307; methodology and 40–1; Mexican sample 221–3; Thailand’s sample 371–3, 374–5; Turkish sample 337–8, 339–41; Venezuelan sample 267–8, 269 first-tier component suppliers 154 fixtures: jigs and 119 Fleury, A.C.C. 46, 94, 183, 294, 295 flexibility 18–20, 281, 392–3; motivation for adoption of FA 101, 102–3, 226–7, 275, 308–11 flexible automation (FA) 6; and economies of scale and scope 23– 32 flexible manufacturing systems (FMS) 25, 45, 53, 55, 87, 88, 146, 224 flow-based production processes 16–17, 145 focus strategy 147 food industry 260–1 Ford 16, 24–5, 103, 119–20, 305 Ford Mazda 228 foreign capital 227; India 299, 301, 303–4; Turkey 332 foreign collaborations 303–4, 326 foreign-owned firms 41, 91, 92–3, 99 foreign participation 41, 91, 93 foreign technology 77, 164, 266, 303–4

454 INDEX

forgings 184, 311–12, 313, 317, 324–5 France 76 freedom from deficiencies 130, 146 Freeman, C. 145, 161 Friedlaender, A.F. 20–3 functional flexibility 19 gearbox casings 56–7, 60, 61 General Motors 228 Germany 76, 77, 85, 133, 134, 135–6 Gilder, G. 27–8 Groover, M.P. 42–4, 126, 128, 130 group technology (GT) 129 Harrison, B. 162 Hay, D.A. 151 Hayes, R. 49 Herzenberg, S. 220 heterogeneity of products 29 Hindustan Machine Tools 302, 304 Hitachi 47 Hoffman, K. 27, 28, 158 homogeneity of production processes 145– 6 hours of working 60, 123–4, 282, 283 human resource constraints 94–7 human resource management 272–4 Humphrey, J. 94 Hydraul (model case study) 51–65, 120, 127; adoption of FA 53–6; the firm 51–2; old technology 53; plant scale 59–61; product scale and scope 56–9; production process 52; selection of technology 55–6, 107; technical change 52–6; unit costs 61–5, 137; utilization rate 60, 123–4 Hyundai 154

idle capacity 204–6 import controls 299, 301 import substitution industrialization (ISI) 217, 258, 330–1

imports 73–5; costs of technology imports 303–4, 326; global CNC lathe imports 423; global CNC machine-tool imports 80, 81; global imports of other CNC metalcutting machines 427; global lathe imports 420–1; global machine-tool imports 406–9; global machining centre imports 425; global metal-cutting machine-tool imports 75–6, 416–17; India 300; Thailand 366–7; Venezuela 261–2, 263, 265 in-house production vs subcontracting 319 incentives: investment 345; work 231 India 47, 108–9, 168, 299–329; choice of components/products to be produced with FA 313; CNC machine-tool prices 133, 134, 135–6; costs 321–7; diffusion of FA in an international perspective 82–5 passim; diffusion of FA in studied firms 87–93 passim, 304–15; factors underlying diffusion of FA 99– 104 passim; industrial policies and development 299–302; intra-firm diffusion 104–8 passim; introduction of FA 306–8; machine-tool industry 81, 302–3, 304; organizational changes 94, 99, 314–15; production processes 311–12; reasons for introduction of FA 308–11; sampled firms 304–6, 307; scale 319–21; scope 315–19; selection of FA 312–13; technology diffusion 303–4 indirect costs 247, 248–50;

INDEX 455

see also under individual types of cost indivisibilities 15; see also economic indivisibilities; technical indivisibilities industrial licensing 299, 301 Industrial Market Research Services (IMRS) 368 industrialization 3; Brazil 177–8; India 299–302; Mexico 217–18; Thailand 365; Turkey 330–2; Venezuela 258–60 industry: concentration 151; diffusion of FA by type of 92; motivation for adoption of FA and 100– 4; new organizational techniques and 99; product variety 116; sampling 41–4 industry journals 50–1, 65–6 INEGI 219–20 information: policy and provision of 173; sources of 105–6, 197–8, 376–8 information systems 195 information technology 25–6; in project and management 190 infrastructure 97, 156–9, 174 innovation 116–21, 156 input unit costs 212–13, 288 input utilization rate 128–9, 212, 287, 288; see also rejection rates inspection and testing 44, 129–30 installation 107–8, 346 institutional support 367–8, 394 integral change 295 integration: of operations 127–8; in production process 378 interest rates 100, 173 international competitiveness 252, 276, 363–4, 383 inter-sphere automation 379–80 interviewing 47–8, 68 intra-activity automation 379–80

intra-firm diffusion 104–8 intra-sphere automation 379–80 inventories 314, 324–5 investment 247; incentives 345; minimum investment requirements 150–6; Turkey 332, 334 iron and steel 260–1 Italy 76, 133, 134, 263 Jacobsson, S. 40, 82, 86, 105, 139, 164, 184; capital costs 138; CNC machine-tool diffusion 83, 85; India 47; learning-period duration 148, 149; productive capacity 132; scaling-up 30; skill-saving 160–1 Jain, A. 81 James, D. 225 Japan 81, 85; CNC machine exports to Thailand 369; CNC machine exports to Venezuela 263; prices of CNC equipment 133, 134, 135–6; R&D expenditure 31 Jenkins, R. 100 jigs, fixtures and 119 job redefinition 272–4 job-shop production 86, 125–7, 145 joint-production 20–3 journals, industry 50–1, 65–6 Jubb, C. 33 just-in-time production (JIT) 26, 45, 129, 156–9; FA and 94–9 passim; Mexico 228–30; Venezuela 272, 273 Kaplinsky, R. 27, 29, 32, 99, 158, 379 Karshenas, M. 104, 107 Kaya, E. 333 Kehoe, T.J. 40

456 INDEX

Kia 154 Kirkpatrick, C.H. 151 Kisaer, H. 334, 335 Kline, S. 32 Korea, South 73, 82, 168, 304; CNC machine prices 133, 134, 135–6; supply and demand 76–7 Koutsoyiannis, A. 17 Koyo Seiko Ltd 305 labour: lack of skilled labour 101, 103–4, 370, 381; Mexico 230–3; reliability 226; saving 138 labour costs: Brazil 184, 211–12; India 322–4; and location of production 159–60; Mexico 247–8; model case study 63–4; motivation for adoption of FA 101, 103–4; Thailand 389; Turkey 360–1; Venezuela 275, 284, 286, 287 labour productivity 131, 146; Brazil 211–12; Mexico 241, 242–3; Turkey 333, 336, 354, 355 labour-related indivisibilities 15 large developing countries 164–5 large firms 90, 155–6, 163–4, 171; see also firm size large local markets 161–3 late entrant firms 149–50 lathes 312; global CNC lathe exports 422; global CNC lathe imports 423; global production 400–3; prices of CNC lathes 133–5, 136, 137; total global exports 418–19; total global imports 420–1 Latin America 46, 81; see also Brazil; Mexico

lay-out 314, 383; cellular see cellular manufacturing Laysperes method 49 lead times 101, 102, 103, 127, 131; Mexico 240–1, 242–3 learning period duration 148–50 Lemos, P. 178 level of automation see density of automation licensees 106 licensing, industrial 299, 301 Lieberman, M.B. 149 light commercial vehicles (LCVs) 158 linkages 383 local affiliates 106 local demand 161–3 local machine tool industry 71–3; Brazil 180, 183; India 81, 302–3, 304; Thailand 366–7; Turkey 336–7 local sales representatives/distributors 105– 6, 266 locally-owned firms 41, 91, 93 location of production 5–6, 144–66, 170, 215–16, 255–6, 328–9; FA scale and in Venezuela 294–7; factor prices and biases 159–61; necessity of large local demand 161–3; production and cost conditions in engineering 145–50; scale and scope and firm entry 150–6; transport costs, JIT and infrastructure 156–9 long-run 13–14 lot sizes see batch sizes Lucas LFS 29 Maastricht workshop, September 1994 68 machine adjustment times see setting-up times machine availability 320–1 Machine Production 132, 135 machine productivity 101, 102, 103 machine-tool industry, local see local machine-tool industry

INDEX 457

machine-tool manufacturers’ documentation 50–1 machine tools: conventional machine tools replaced in India 321–2; global consumption 71, 72, 428–9; global exports 410–13; global imports 406–9; price comparison of CNC and conventional machine tools 135–7; see also computer numerical control (CNC) machine tools; metal-cutting machine tools; numerically controlled (NC) machine tools machining centres 44, 146, 224, 265–6; global exports 424; global imports 425; global production 404–5; prices 135–7 machining operations 44, 184–6 machining service firms 92, 187, 195–7, 198 machining speed 282, 283 machining times 60, 128; Brazil 198–200; India 319, 320; Turkey 352 macroeconomic conditions 100; see also economic context maintenance see repairs and maintenance Mair, A. 158 management costs 17 management education 297 managers, interviews with 47 manufacturing operations 152 manufacturing sector: Brazil 177–8; India 301; Mexico 219–21; Thailand 368–70; Turkey 332–3; Venezuela 260–1 maquiladora industries 221, 255 Marazaki 369 market 245; extent of 12; need for large local market 161–3

marketing 256 marketing costs 2, 17, 31–2, 65; Brazil 213–14; India 326; Mexico 250; Venezuela 290–2 Markowski, S. 33 Marshall, A. 4, 13 mass production 24–5, 85–7, 125–7, 145 material handling and storage operations 44 mechanization 13 Mercado, A. 40, 220, 221 metal-cutting machine tools 44; diffusion of 71–7; global exports 414–15; global imports 75–6, 416–17; global production 73, 74, 397–9 metalworking industry 27, 28 methodology 38–69; approach, method and unit of analysis 39–41; data requirements 47–51; implementation of the study 65–8; model case study 51–65; sampling 41–7 Mexico 46, 108–9, 168, 217–57; cost structure 247–50; diffusion of FA 87–93 passim, 219–21; diffusion of FA in an international perspective 82–5 passim; engineering sector 218–19; factors affecting scale 239–41; factors underlying diffusion of FA 99– 104 passim; findings of the study 253–6; intra-firm diffusion 104–8 passim; macroeconomic trends and policy issues 217–18; organizational change 228–33; other competitive advantages 241–4; process of adoption of FA 227–8; product diversity 236–8, 255; production processes 223–5;

458 INDEX

production volumes and capacity utilization 238–9; profit margins 250–3; reasons for adoption of FA 225–7; sampled firms 221–3; scope of production 233–8, 254; unit cost and prices 245–6 Meyer-Stamer, J. 94 microelectronics, advances in 25–6 Milgrom, P. 26 Miller, S. 27 minimum efficient scale (MES) 150–6 minimum investment requirements (MIR) 150–6 Ministry of Science, Technology and Environment (MOSTE) 368 Mital, A. 160 Mitsubishi 369 model case study see Hydraul ‘modern technology’ literature 23–32 modernization: Venezuela 270–4, 292 Mody, A. 27, 28, 30, 98 Mori Seiki 265, 266 Morris, D.J. 151 Morroni, M. 15, 18–19 motivation, workers’ 207 motives for diffusion of FA 99–104, 109, 168–9; India 308–11; Mexico 225–7; Thailand 373–6, 377; Venezuela 274–6 motor power 132 motor vehicle components see autopart firms motor vehicle industry 16, 24–5, 100, 152– 6 mould-making firms 92, 344–5, 347–8 Mueller, D.C. 150 multinational corporations 106, 164 multi-product firms 20–3, 32–3 multi-skilling 272–4 NAFIN 225, 250 Naim, M. 259–60

National Electronics and Computer Technology Centre (NECTEC) 368 National Productivity Centre (NPC) 333 National Research and Development Corporation 300 National Supplier Development Programme (NSDP) 370–1 national surveys 170–1 neo-classical theory 4 new organizational techniques 26, 45–6, 70, 94–9, 109; see also organizational changes new products 116–21, 156 Nissan 28, 29 normalized density of diffusion 83 North America 71, 72 North America Free Trade Agreement (NAFTA) 100, 218 0.6 rule 15 numerical flexibility 18–19 numerically controlled (NC) machine tools: diffusion 77–82; see also computer numerical control (CNC) machine tools OCEI 261 O’Connor, D.C. 47 OECD 31, 50–1 OEMs 153 off-line feeding 128 Okuma 369 OPEC 259 optimal scale 14–17; see also scale organizational changes: Brazil 190–1, 194–5, 197; India 94, 99, 314–15; integral change 295; Mexico 228–33; Thailand 94, 381–3; Turkey 98, 338–45 passim, 363; Venezuela 272–4, 278–9, 294, 295 organizational techniques, new 26, 45–6, 70, 94–9, 109 overall cost leadership strategy 147–8

INDEX 459

overhead costs: Brazil 213–14; India 326–7; Turkey 359; Venezuela 284, 285, 290–2; see also marketing costs; research and development costs ownership 41; diffusion of FA 90, 92–3, 220–1; Thai firms 382 Paasche method 49 Panzar, J. 4 Park, W.-H. 148 Perloff, J.M. 21 Petrobrás 189 Petróleos de Venezuela (PDVSA) 92, 264, 268, 274 petroleum industry 145 pilot study 67 pin factory 12 Piore, M.J. 25, 120–1 plant scale 2, 16–17, 27–30, 112; Brazil 203–7; changes in 121–5; economic factors underlying increases 133–7; methodology 59–61; Mexico 238–44; technical factors underlying increases 125–32; Thailand 388; Turkey 352–6; Venezuela 281–4, 296 Podolsky, S. 129 policy: Brazil 177–8; India 299–302; Mexico 217–18; recommendations 173–4; Thailand 365–8, 370–1; Turkey 330–2; Venezuela 258–60 Porteous, M. 46 Porter, M.E. 147, 150 Posthuma, A.C. 182, 183, 191 power failures 381

Prado, A.J.C. 183 Pratten, C.F. 3, 99, 146, 155, 346 precision 102, 128, 376, 378 prices 133–9; CNC machine tools 133–7; factor prices and location of production 159–61; Mexico 245–6, 250–3; Turkey 354; Venezuela 292–4; see also costs primary data 47–50 printed circuit boards 107–8 printing industry 16 process control 44 process control computers (PCCs) 25, 45, 87, 88 processing operations 42–4 processing time 206, 386, 389 product differentiation 256 product diversity 116–21; Brazil 187, 191–2, 202–3; India 317–19; Mexico 236–8, 255; model case study 54–5, 58–9; Thailand 385–7; Turkey 349–51; Venezuela 281, 296 product features (quality) 130, 146 product scale 15–16, 49–50, 112; Brazil 198–203; changes in 113–16, 139–40; India 315–17; Mexico 233–6; model case study 56–9; Thailand 383–5; Turkey 347–51; Venezuela 276–9, 295–6 product scope see scope product-specific scale economies 22–3 production capabilities 328–9 production capacity see capacity production conditions, engineering 145–50 production costs: Brazil 210–13; India 321–5; Mexico 247–8;

460 INDEX

reduction as motive for adoption of FA 101, 104; Thailand 388–91; Turkey 359–62; Venezuela 284–90; see also under individual types of cost production flow 282–4 production lay-out see lay-out; cellular manufacturing production management and control 130–1 production management and engineering 60–1 production process control computers (PCCs) 25, 45, 87, 88 production processes 125–7; Brazil 184–7; characteristics 48–9; homogeneity 145–6; India 311–12; integration in Thailand 378; Mexico 223–5; model case study 52; Thailand 376–80; Turkey 338–45; Venezuela 268–70 production run see batch sizes production schedules 314 production volumes: Mexico 238–9; Turkey 336, 353–4; Venezuela 279, 280, 281–2, 296 productivity 64; labour see labour productivity; machine 101, 102, 103; Mexico 219, 241, 242–3; motive for adoption of FA in Thailand 373–6 products: choice for production on CNC machines 313; heterogeneity of 29; new 116–21, 156 profit margins/profits 146–7; Mexico 250–3; Thailand 389; Venezuela 292–4 programmable automation 183

programme logic controllers (PLCs) 25, 44, 224 programming 326, 381 Proton 154 pumps 310–11 Quadros Carvalho, R. 46, 183, 184 quality: Brazil 186–7, 189, 192; of castings and forgings 313; and increases in plant scale 129–30; India 308–11, 314; Mexico 226, 231, 232, 242–3, 244; motive for diffusion of FA 100–2; product quality 130, 146; Thailand 370, 373–6, 393; TQM 45, 94–9 passim, 129–30; Turkey 354; Venezuela 272, 273, 274–5 quality accreditation 45–6, 94–9 passim questionnaires 47–50 Ramírez, J.C. 220, 221 raw materials: costs 64, 247, 287, 288, 361; efficient use 128–9, 212, 287, 288; inventories 287, 288, 324; transport costs 157 ray average costs (RAC) 22 ray economies of scale 21–3 rejection rates 128–9, 207, 287, 288, 319– 20, 361, 388–9 repairs and maintenance: costs 65, 288–90, 325, 362; problems with 107–8, 266, 356, 380–1 research: suggestions for future research 170–3, 393–4 research and development (R&D) costs 2, 17, 30–1; Brazil 213–14; India 326; Mexico 250; Thailand 391; Venezuela 290–2

INDEX 461

research institutes 66 researchers, contacting 66, 67 resistance, worker 107 results, presentation and discussion of 68 retrofitting 225 Rhys, D.G. 146, 158, 162 Roberts, J. 26 robots 44, 87, 88; Brazil 182–3; Thailand 369–70, 379 Romi 210 Rosenberg, N. 32, 71 Ross, D. 15, 17, 151 Sabel, C.F. 25, 120–1 sales 353–4 sampling 41–7 Sandoval, S. 220 Saunders, R.J. 159 Sayer, A. 85–6 scale 1–3, 5, 11–37, 112–43, 167; Brazil 198–207, 214–16; data requirements 49–50; de-scaling 1, 28–9, 34–5, 167, 393; economies of scale and optimal scale 14–17; FA and economies of scale and scope 23–32; firm see firm scale; impact of FA 169; India 319–21; Mexico 238–44, 254; origins of economies of scale theory 12–14; plant see plant scale; product see product scale; scaling up 30–2, 35, 167; scope, firm entry and 150–6; technical change, costs, scope and 32– 6; Thailand 383–5, 388, 392–3; Turkey 346–56; Venezuela 276–84, 294–7 schedules: delivery 115; production 314 Scherer, F.M. 3, 15, 17, 151

Schmitz, H. 183 scientific and technological (S&T) development 367–8 scope 2–3, 5, 11–37, 49–50; Brazil 198–203; changes in product variety/scope 112, 116–21, 140; data requirements 49–50; and economies of scope 17–23; FA and economies of scale and scope 23–32; FA impact 169; India 315–19; Mexico 233–8, 254; model case study 54–5, 56–9; scale, firm entry and 150–6; technical change, costs, scale and 32–6; Thailand 385–7, 392–3; Turkey 349–51; Venezuela 279–81, 295–6 scrap rates 128–9, 207, 287, 288, 319–20, 361, 388–9 second-tier component suppliers 154–5 secondary data 50–1 selection of FA equipment 106–7; Brazil 197–8; India 312–13; model case study 55–6, 107; Thailand 376–8; Turkey 345–6; Venezuela 274–6 servicing see repairs and maintenance setting-up times 15–16; Brazil 198–200, 206–7; impact of FA 26–7, 113–16; India 315–16; Mexico 233–4; methodology 56–7; Thailand 383–5; Turkey 347–8; Venezuela 276–9, 295–6 Shaiken, H. 220 Shapiro, H. 162 shifts 60, 123–4, 282, 283 Shingo, S. 278 short-run 13–14 Silberston, A. 15 Silva, B.E. 46

462 INDEX

single minute exchange die (SMED) 207, 278 Siriratchtapong, P. 383 size of firm see firm size skills 138–9, 160–1, 315, 346; de-skilling 255–6, 315; lack of skilled labour 101, 103–4, 370, 381; multi-skilling 272–4; see also training small firms 171, 382–3; diffusion of FA 90–2; entry 155–6; Turkey 334; see also firm size small and medium-sized (SME) enterprise sector 370–1 Smith, A. 1, 4, 12, 14 Soete, L. 138, 161, 245 software costs 326 space 325, 389 span of control 98 spare parts 378 special-purpose machines 312 specialization 12–13, 14–15 Sripaipan, C. 368, 371 stability, macroeconomic 100 Stal, E. 147 Stalk, G. 31 start-up 107–8; see also installation state 77; see also policy State Planning Organization 330 static specialization gains 14 steel mini-mills 30, 34–5 steering components 309–10 Steudel, H.J. 130 Stevenson, R. 33 Stigler, G.J. 4, 17–20, 36 Stiglitz, J. 33, 35 stocks of CNC machine tools 83, 84 Stoneman, P. 104, 107 Storper, M. 85 subcontracting 120–1, 240, 319, 382–3 subsidiaries 106 supervisors 98, 231 suppliers of FA equipment:

source of information 55, 66–7, 105–6; support services 378 supply and demand 76–7 support services 378 surveys 39 Sutcliffe, R.B. 157 Sweden 85, 138 Taiwan 73, 82, 168, 369; prices of CNC machine tools 133, 134, 135–6; supply and demand 76–7 Taksan 337 Tayler, P. 13 Taylor, F.W. 24 TDRI 368 teams 230–1, 232 technical change: Brazil 177–98; costs, scale, scope and 32–6; India 306–15; Mexico 221–33; model case study 52–6; Thailand 373–83; Turkey 338–45; Venezuela 267–76 technical complexity 313 technical factors: underlying diffusion of FA 100–4; underlying increases in plant scale 125– 32 technical indivisibilities 15 technological blending 225 technological capabilities 171–2 technology transfer 77, 164, 266, 303–4 Teece, D.J. 21 TELCO 90–2, 305, 310, 319, 328 telecommunications infrastructure 159 terms of lending 173–4 testing and inspection 44, 129–30 Tezsan 337 Thai Board of Investment (TBOI) 367–8 Thailand 47, 108–9, 168, 365–96; applications and diffusion of FA 368– 70; diffusion of FA 87–93 passim, 368–83, 392;

INDEX 463

extent of FA diffusion 392; FA diffusion in an international perspective 82–5 passim; factors affecting diffusion of FA 99– 104 passim, 370–1; firm-level impacts 392–3; industrialization process 365; integration in production process 378; intra-firm diffusion 104–8 passim, level of automation 379–80; machine-tool industry 366–7; motives for adopting FA 373–6, 377; organizational change 94, 381–3; plant scale 388; policy climate and institutional support for FA 367–8; problems with FA 380–1; product diversity 385–7; product scale 383–5; production costs 388–91; sampled firms 371–3, 374–5; selection of FA 376–8; suggestions for future research 393–4 tool costs 65, 213 total quality management (TQM) 45, 94–9 passim, 129–30 Toyota 27, 158 trade 172; liberalization 99–100; see also exports; imports trade fairs 105 training 107, 266, 346; costs 138–9, 361; India 314–15; Mexico 231, 232; policy recommendations 174, 297; Thailand 378 transfer lines 86 transfer mechanisms 266; see also technology transfer transitional economies 76 transnational corporations 106, 164 transport costs 156–9 transray concave cost functions 23 transray convex cost functions 23

Turkey 46, 108–9, 168, 330–64; adoption of FA 338–45; development of the economy 330–2; diffusion of FA 87–93 passim, 336–46; engineering industry 333–7; FA diffusion in international perspective 82–5 passim; factors underlying diffusion of FA 99– 104 passim, 345–6; intra-firm diffusion 104–8 passim; manufacturing sector 332–3; organizational change 98, 338–45 passim, 363; plant scale 352–6; product scale 347–51; sampled firms 337–8, 339–41; unit costs 356–62 Twiss, B.C. 26 UNIDO 46, 47, 145, 146 unit costs 122–3; Brazil 207–14; India 321–7; Mexico 245–6; model case study 61–5, 137; Thailand 388–91; Turkey 356–62; Venezuela 284–92, 296; see also costs United Nations (UN) COMTRADE database 50–1 United States (US) 76, 151; diffusion of CNC machine tools 77–81, 82; exports of CNC equipment to Venezuela 263; prices of CNC machine tools 133, 134, 135–6; R&D expenditure 31; workers per CNC machine tool 85 utilization rates 60, 122–4, 131, 146; India 320–1; Mexico 238–9;

464 INDEX

Thailand 389; Turkey 333, 355–6; Venezuela 282, 283 value added 334–5 valves 311 variety, product see product diversity Venezuela 46, 108–9, 168, 258–98; determinants of the study 294–5; diffusion of FA 87–93 passim, 262–6; engineering industry 260–1, 261–2, 264; entry barriers 297; FA diffusion in an international perspective 82–5 passim; factors underlying diffusion of FA 99– 104 passim; integral change 295; intra-firm diffusion 104–8 passim; macroeconomic performance and development strategy 258–60; manufacturing context 260–1; modernization 270–4, 292; plant and firm scale 281–4, 296; problems in adopting FA 276; production process 268–70; profits and delivery times 292–4; sampled firms 267–8, 269; scope and economies of scope 279–81, 295–6; selection and adoption of FA 274–6; setting-up time and batch size 276–9, 295–6; transfer mechanisms for FA 266; unit costs 284–92, 296 Vermulm, R. 81, 180 vertical integration 120–1, 150–6 Viner, J. 4, 13–14 visits 55, 106 volume production see mass production Volvo 30 wage structures 231, 273, 274

wages: Mexico 247–8; Turkey 333, 360; see also labour costs Walker, D.F. 144, 157, 159 Walker, R. 85–6 Wangwe, S.M. 67 Watanabe, S. 40, 47, 263, 304 water supply 159 Williams, K. 85–6 Willig, R. 4 Wolf, M. 300 Womack, J.P. 29, 31 Woodward, J. 126 work-in-progress (WIP) 129 worker resistance 107 workers per CNC machine tool 83–5, 87– 90, 322–3 workers’ motivation 207 working hours 60, 123–4, 282, 283 working practices reorganization 26, 45, 94–9 passim; Mexico 230–1, 232; Venezuela 272–4 Yamazaki Mazak (YM) 265, 266, 337, 369 Yentürk, N. 333