Inward Investment, Technological Change and Growth The Impact of Multinational Corporations on the UK Economy
Edited by
Nigel Pain
INWARD INVESTMENT, TECHNOLOGICAL CHANGE AND GROWTH
This page intentionally left blank
Inward Investment, Technological Change and Growth The Impact of Multinational Corporations on the UK Economy Edited by
Nigel Pain Senior Research Fellow National Institute of Economic and Social Research
in association with NATIONAL INSTITUTE OF ECONOMIC AND SOCIAL RESEARCH
© National Institute of Economic and Social Research 2001 All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission. No paragraph of this publication may be reproduced, copied or transmitted save with written permission or in accordance with the provisions of the Copyright, Designs and Patents Act 1988, or under the terms of any licence permitting limited copying issued by the Copyright Licensing Agency, 90 Tottenham Court Road, London W1P0LP. Any person who does any unauthorised act in relation to this publication may be liable to criminal prosecution and civil claims for damages. The authors have asserted their rights to be identified as the authors of this work in accordance with the Copyright, Designs and Patents Act 1988. First published 2001 by PALGRAVE Houndmills, Basingstoke, Hampshire RG21 6XS and 175 Fifth Avenue, New York, N.Y. 10010 Companies and representatives throughout the world PALGRAVE is the new global academic imprint of St. Martin's Press LLC Scholarly and Reference Division and Palgrave Publishers Ltd (formerly Macmillan Press Ltd). ISBN 0-333-92536-X This book is printed on paper suitable for recycling and made from fully managed and sustained forest sources. A catalogue record for this book is available from the British Library. Library of Congress Cataloging-in-Publication Data Inward investment, technological change and growth : the impact of multinational corporations on the UK economy / edited by Nigel Pain ; in association with National Insitute of Economic and Social Research, p. cm Collection of papers presented at a conference held at the British Academy in Sept. 1999 Includes bibliographical references and index. ISBN 0-333-92536-X 1. Investments, Foreign—Great Britain. 2. Technology transfer—Great Britain. 3. Great Britain-Economic conditions—1997- I. Pain, Nigel. II. National Institute of Economic and Social Research HG5432 .1643 2000 330.94—dc21 10 10
09
9 08
8 7 6 5 4 3 2 07 06 05 04 03 02
Printed in Great Britain by Antony Rowe Ltd, Chippenham, Wiltshire
00-062601 1 01
Contents List of tables
vii
List of figures
x
Notes on the contributors
xi
1
2
The growth and impact of inward investment in the UK: introduction and overview Nigel Fain
1
The determinants of host country spillovers from foreign direct investment: a review and synthesis of the literature Magnus Blomstrom, Ari Kokko and Steven Globerman
34
3
Inward investment and technical progress in the UK Florence Hubert and Nigel Fain
4
Host country effects of FDI in the UK: recent evidence from firm data Sourafel Girma, David Greenaway, Katharine Wakelin and Nuno Sousa
5
Why do foreign-owned firms in the UK have higher labour productivity? Nicholas Oulton
66
104
122
6
Regional policy and the impact of FDI in the UK Nigel Driffield
7
A regional computable general equilibrium analysis of the demand and 'efficiency spillover' effects of foreign direct investment Gary Gillespie, Feter G. McGregor, J. Kim Swales and Ya Ping Yin
178
Sticky places in slippery space - the location of innovation by MNCs in the European regions John Cantwell, Simona lammarino and Camilla Noonan
210
8
162
vi
Contents
9
Foreign trade and FDI stocks in British, US and French industries: complements or substitutes? Lionel Fontagne and Michael Fajot
Index
240
265
List of tables 1.1 The global stock of foreign direct investment by recipient area 5 1.2 The composition of the UK inward FDI stock 9 1.3 Foreign-owned firms in the UK manufacturing sector 11 1.4 Business enterprise R&D in the UK 19 1.5 The UK's major exporters in 1997 22 2.1 The proximate determinants of spillovers 40 2.2 Variables related to FDI spillovers 42 3.1 Sectoral composition of employment 75 3.2 Labour productivity per employee hour 75 3.3 Industrial composition of inward FDI stock in the UK 77 3.4 Choosing between different panel estimators 79 3.5 Mean group estimates with endogenous technical change 80 3.6 The contribution of inward investment to UK output growth 83 3.7 Summary statistics from all firms' panel estimates 85 3.8 Share of foreign firms in manufacturing in 1992 86 3.9 Characteristics of foreign and domestic firms, 1983-92 87 3.10Inward FDI and technical progress in domestic firms 90 3.11 Pooled mean group estimates for eleven manufacturing industries 92 3.12Inward FDI and labour productivity 94 3.13Do spillovers vary according to nationality of investors? 95 4.1 Sample means (and standard deviations) 107 4.2 Differentials between domestic and foreign firms 108 4.3 Differentials by nationality of ownership 109 4.4 The impact of FDI on the productivity and wages of domestic firms 111 4.5 Dependent variable: EXPORTS - export decision 116 4.6 Dependent variable: EXPROPEN - propensity to export 117 5.1 Capital intensity (K/L) amongst 1973-93 survivors: manufacturing, by SIC80 class, 1993 128 5.2 Descriptive statistics in 1993 for 1973-93 survivors in manufacturing, by ownership 129 5.3 Growth rates of output, employment and capital, 1973-93: 1973-93 survivors in manufacturing, by ownership in 1993 130
viii
List of tables
5.4 Comparison of foreign and UK-owned establishments in manufacturing: cross-section regressions, 1973-93 survivors in 1993 5.5 Regressions explaining productivity differences amongst survivors in manufacturing: dependent variable is log of value added per employee, 1989 and 1993 5.6 Contribution of measured inputs to explanation of productivity gap between foreign and UK-owned establishments: manufacturing in 1993 (1973-1993 survivors) 5.7 Labour productivity and its determinants by ownership type, 1995: distribution of within class means across SIC80 classes (UK independents = 100) 5.8 Ownership and the determinants of labour productivity in 1995 5.9 Effect of ownership on labour productivity and its determinants 5.10Effect of ownership on labour productivity (value added per employee): manufacturing versus non-manufacturing companies in 1995 5.11 Effect of recessions on capital stock growth in subsequent booms: manufacturing, 1973-93 survivors 6.1 Measure of'inward investment' 7.1 Comparison of the structural characteristics of the ownership-disaggregated database (1989) 7.2 Functional forms and key parameter values for foreign and UK-owned plants 7.3 The impact of our standard export FDI shock in terms of capital stock, exports, output and direct employment 7.4 The percentage change in key variables following a 20 per cent increase in FDI with 100 per cent export-intensity 7.5 The impact of efficiency gains to UK manufacturing from FDI using the export FDI shock as the base, with regional bargaining and migration 7.6 20 per cent increase in FDI with 100 per cent exportintensity and 0.8 per cent Harrod-neutral efficiency shock to UK manufacturing with regional bargaining and migration 7.7 Combined employment, value added and output multipliers from the inward investment expansion
132
134
136
139 140 141
143 147 172 185 186 187
190
194
197 199
List of tables 8.1 Distribution (D) and penetration (P) of foreign-owned patenting activity by location, 1969-95 8.2 Regional breakdown of USPTO patent grants to large firms, 1969-95 (percentage of each group's total grants); population and GDP by region (1995) 8.3 Average foreign-owned firm shares of US patents by macrotechnological sector and region, 1969-95 8.4 Regional dispersion of technological specialisation, 1969-95 8.5 Regression results 9.1 Export equations, United Kingdom 1987-96 9.2 Import equations, United Kingdom 1987-96 9.3 Export equations, France 1989-94 9.4 Import equations, France 1989-94 9.5 Export equations, United States 1989-94 9.6 Import equations, United States 1989-94
ix
216
217 220 222 224 250 251 253 254 256 257
List of figures 1.1 1.2 3.1 3.2 6.1 6.2 6.3 6.4 7.1
The stock of foreign direct investment in the UK, 1960-99 UK FDI inflows as a per cent of GDP Labour productivity per employee hour Implied technical progress functions Domestic and foreign fixed capital investment Share of foreign employment by industry Foreign fixed investment expenditure by UK region Foreign share of manufacturing employment by region Sectorally disaggregated employment change for a 100 per cent export FDI shock with regional bargaining and migration with LOOP approximation for foreign-owned sector 7.2 Build-up of efficiency spillover shocks - adjustment path implied by Barrell and Pain (1997) results 7.3 Aggregate employment gains from the efficiency spillover shock (using the results from the export FDI shock as the base), with regional bargaining and migration 7.4 The employment impact of the combined effects of FDI (20 per cent FDI, 100 per cent export-intensity plus build-up of efficiency spillover effects)
7 8 76 81 165 165 166 166
188 192
195
198
Notes on the contributors Magnus Blomstrom is Professor of Economics at the Stockholm School of Economics. He is also a Research Associate at the National Bureau of Economic Research in the US and a Research Fellow at the Centre for Economic Policy Research in the UK. John Cantwell is Professor of International Economics at the University of Reading. He has been a Visiting Professor of Economics at the University of Rome 'La Sapienza', the University of the Social Sciences, Toulouse, Rutgers University, New Jersey and the University of Economics and Business Administration, Vienna. His main research areas are the economics of technological change and multinational firms. Nigel Driffield is a Senior Lecturer in industrial economics at Birmingham Business School. His main research interests concern the impacts of foreign direct investment on host country firms. In this vein, recent papers include the impact of FDI on profitability, productivity, employment and wages in domestic firms. Lionel Fontagne is Professor of Economics at the University of Paris 1 Pantheon-Sorbonne and director of the Centre d'Etudes Prospectives et d'Informations Internationales (CEPII). He is also an adviser at the International Trade Center. His research interests cover European integration, international trade theory and trade liberalisation in developing countries. He has written recently on foreign direct investment and the prospects for tax co-ordination in Europe, exchange rate strategies in the competition for attracting FDI and endogenous symmetry of shocks in a monetary union. Gary Gillespie is affiliated to the Fraser of Allander Institute, Department of Economics, University of Strathclyde. His research interests include foreign direct investment, regional economic modelling and regional policy. Sourafel Girma is a Research Fellow in the Centre for Globalisation and Labour Markets at the University of Nottingham. He has worked at the Business School at the University of Nottingham and in the Economics Department at the University of Manchester where he completed his XI
xii
Notes on the contributors
PhD. His research interests are primarily in panel data modelling and productivity analysis. Steven Globerman is the Director of the Center for International Business at Western Washington University. He has published extensively on a range of issues related to international trade and foreign direct investment, and he has also consulted for businesses, governments and international organisations on those matters. David Greenaway has been a member of staff at the University of Nottingham since 1987. He is a Professor of Economics and has been a Pro-Vice-Chancellor since 1994. His research interests lie primarily in the fields of international trade policy, economic development and European integration. Florence Hubert is a Research Officer at the National Institute of Economic and Social Research, where she has been investigating topics such as foreign direct investment, European economic integration and the location of multinational production. She is also involved in the European forecasting exercise carried out by the NIESR. Simona lammarino is currently working as a researcher at the Italian Institute of Statistics and as a senior economist a the Institute for International Affairs in Rome. Her main research interests are in the fields of international trade, economic integration and regional and international aspects of technological change. She collaborates with the Department of Economics of the University of Reading in the field of multinational corporations, technology and regional systems of innovation in the European Union. Ari Kokko is Professor of International Business at Abo Akademi University, Turku, Finland, and research associate at the European Institute of Japanese Studies in Stockholm. His research deals with foreign direct investment, trade and economic integration. Peter G. McGregor is affiliated to the Fraser of Allander Institute, Department of Economics, University of Strathclyde. His research interests include regional economics, regional policy and applied general equilibrium analysis.
Notes on the contributors
xiii
Camilla Noonan is a lecturer in International Business at the Michael Smurfit Graduate School of Business, University College Dublin. Her research interests include innovation and technological change, multinational corporations and the global business environment. Nicholas Oulton has been a full-time consultant to the Bank of England since 1998. Before that he was a Senior Research Fellow at the NIESR. His research interests centre round the study of productivity at the level of both the company and the industry as well as the whole economy. He is also interested in structural change and how this influences aggregate productivity performance. Nigel Pain is a Senior Research Fellow at the National Institute of Economic and Social Research. Before joining NIESR he worked in the Treasury. He has published widely on the determinants of location choice by multinational corporations and on their impact on the growth prospects and industrial structure of host and home economies. Michael Pajot is affiliated to TEAM (Theorie et Applications en Microeconomie et Macroeconomie) at the University of Paris 1 Pantheon-Sorbonne, where he researches into foreign direct investment and international trade. Nuno Sousa is a PhD student in the Centre for Globalisation and Labour Markets at the University of Nottingham. His PhD is on the effects of multinational companies' activities in host economies. J. Kim Swales is affiliated to the Fraser of Allander Institute, Depart of Economics, University of Strathclyde. His research interests include regional economics, regional policy and applied general equilibrium analysis. Katharine Wakelin is a Research Fellow in the Centre for Globalisation and Labour Markets at the University of Nottingham. She has worked at MERIT in Maastricht and at the National Institute of Economic and Social Research in London. Her research interests include international economics, the economics of technical change and foreign direct investment. Ya Ping Yin is affiliated to ESST, University of Hertfordshire Business
xiv
Notes on the contributors
School. His research interests cover imperfect competition and open economy macroeconomics, applied general equilibrium analysis and trade and development.
1
The growth and impact of inward investment in the UK: introduction and overview Nigel Pain1
This volume contains a collection of papers from a conference held at the British Academy in September 1999. The objective of the conference was to help identify the channels through which inward investment can affect host economies and provide some quantitative evidence on the impact of foreign-owned firms on the size and industrial structure of the UK and other developed economies over the last decade. There is considerable interest in government and academic circles in the extent to which international transfers of technology and knowledge by multinational companies may affect the performance of host economies. The location of economic activity could be an important endogenous influence on the size of host economies. Many new theoretical models view the creation and exploitation of knowledge as two of the key factors driving the process of economic growth. The Competitiveness White Paper issued by the UK government in 1998 argued that foreign direct investment is one of the main transmission mechanisms behind the diffusion of knowledge, both codified and tacit, across national borders (DTI, 1998). This appears to be supported by empirical evidence that international openness raises economic growth in the UK and other Western European economies (Proudman and Redding, 1998; Barrell and Pain, 1997, 1999b; Hoeller et al., 1998). Foreign investments by multinational companies can provide a channel through which new ideas, working practices and technologies can arrive in host economies, as well as a means by which indigenous companies are exposed to greater competitive pressures. Previous competitiveness White Papers had also suggested that the high level of inward investment into the UK during the 1980s played an important role in the transformation of the production process (Eltis and Higham, 1995; Eltis, 1996).
1
2
Inward investment, technological change and growth
The desire to attract inward investment is one of the few industrial policies pursued consistently by successive UK governments over the past 25 years. Similar interest in the behaviour of multinational corporations is apparent in other advanced economies, with intergovernmental agencies such as the OECD, the World Trade Organisation and the Bank for International Settlements all having undertaken studies exploring the impact of foreign direct investment on trade and employment in host and home economies. The collection of papers in this volume will help in the evaluation of the net benefits of inward investment for host economies and the design of appropriate national policies and institutions to maximise the potential size of those benefits. There is a particular focus on three key issues: whether foreign-owned firms have specific, ownership advantages compared to indigenous companies, the extent to which such advantages subsequently become available to improve the performance of indigenous companies, and whether there are spillovers across industries and regions as well as within the industry and region in which investment takes place. The majority of papers focus in detail on the quantitative impact of foreignowned companies in the British economy, which has long been the primary location for inward investment within Europe. However the methodologies employed can readily be applied to other host economies, and the empirical papers are complemented by a synoptic paper from Magnus Blomstrom, Ari Kokko and Steven Globerman which provides a comprehensive overview of the existing literature on spillovers into productivity and employment in developed economies and highlights a number of issues worthy of further research. Comparatively little is known about the quantitative importance of spillovers from inward investment in industrialised economies, despite the interest of policymakers in this issue. This is especially true of Britain, which even lacks a comprehensive official data source on the activities of foreign-owned firms throughout the economy as a whole. There is some indication from qualitative survey evidence that new technologies and standards have been adopted by UK producers as a result of inward investment. In a study of the impact of technology transfer by US multinational companies, Mansfield and Romeo (1980) found that over half of the UK firms in their survey had introduced new products or processes more quickly because of a transfer of a new product or process by a US-based firm to its overseas subsidiary, with around
The growth and impact of inward investment in the UK
3
two-thirds of the UK firms indicating that their technological capabilities had been raised by such transfers. However there were doubts about the strength and durability of the linkages established by many overseas investors in the UK at that time, particularly in some regional economies (Young etal., 1988). More recent evidence suggests that inward investment in the UK has helped to bring about a significant improvement in the product quality of suppliers (Dunning, 1988; PACEC, 1995). The quantitative importance of such findings is not clear. The detailed statistical and econometric analyses in the papers in this volume help to fill this gap by providing a comprehensive overview of the relative performance of foreign-owned firms in the UK and the extent to which their presence has had a sustained impact on the performance of UKowned firms over time. In order to understand the growing impact of foreign-owned firms and the importance attached to inward investment by the policy community it is useful to review briefly the factors behind the global growth of inward investment and the particular characteristics of inward investment in the UK. This serves to highlight some of the most important channels through which inward investors might influence the performance of indigenous companies.
T H E G R O W T H OF GLOBAL F O R E I G N DIRECT INVESTMENT The rapid growth of foreign direct investment over the last 25 years has made an important contribution to the ongoing globalisation of economic activities. Of course by some measures, such as the ratio of total trade and capital flows to income, it might be said that many industrialised economies are no more open at present than they were a hundred years ago. However changes in the feasible span of managerial control and the nature of technical progress mean that the implications of, and the motivations for, dispersed production are quite different now. New business advances are increasingly knowledge-based, arising from specific ideas and organisational innovations, both tacit and codified. The potentially infinite expansibility of knowledge means that many of these assets can be utilised simultaneously across multiple establishments under common ownership, generating economies of scale for the firm as a whole.
4
Inward investment, technological change and growth
Empirical evidence for countries such as the United States, the UK and Germany suggests that outward investment is more likely to occur in industries with a high level of R&D expenditure in the home country (Barrell and Pain, 1999b; Pain, 1997; Hubert and Pain, 1999), suggesting that new ideas and technologies are being utilised in host economies. Foreign direct investments are no longer dominated by the establishment of free-standing subsidiaries operating mining, agriculture and transportation facilities in emerging markets or producing finished goods at sub-optimal scale in host economies with significant barriers to external trade. Whilst there continues to be a net capital flow from the industrialised economies to the developing ones, postwar capital market integration has been characterised by the high level of gross flows between the industrialised economies. Business strategies have changed, with many multinational firms choosing to reorganise their activities on a regional or global basis. Many leading brand names now have global recognition. Investments in non-manufacturing sectors such as financial services and public utilities have expanded rapidly, helped by widespread use of privatisation policies by many governments and the need for a local presence in foreign markets in activities which rely heavily on professional expertise, personal contacts and international reputation as well as price competitiveness. In recent years national and regional governments and local development agencies have actively competed to attract inward investment through policy inducements and promotional campaigns in so-called 'location tournaments'. If product markets were perfectly competitive such policies would be unlikely to have permanent effects. Location patterns would be driven by geographical endowments, transport costs and production costs in different regions. Investment incentives might succeed in attracting new industries, but these would soon depart if subsidies were withdrawn (Wheeler and Mody, 1992; Barrell and Pain, 1999b). Developments in the theories of international trade and economic geography under imperfect competition have quite different implications for the organisation of firms, as Lionel Fontagne and Michael Pa jot discuss in their paper. Temporary differences in national or regional characteristics, such as investment incentives, can have permanent effects on the location of activities if firms are drawn to particular regions by the availability of agglomeration economies (Fujita etal., 1999). Such economies arise from any location-bound economic activity in an area
The growth and impact of inward investment in the UK
Table 1.1. The global stock of foreign direct investment by recipient area
World ($ billion) of which (%): Developed countries Western Europe UK
1960
1973
54.5
166.7
67.3 22.9 9.2
72.9 36.5 8.9
1985
1990
1998
782.3 1,768.5 4,088.1 69.7 32.4 8.2
78.8 44.4 12.4
68.1 38.4 8.0
Sources: Dunning (1988, Table 3.2) and UNCTAD (1999, Annex Table B3).
that generates positive externalities for nearby firms. If the new technologies and ideas brought by foreign firms do generate spillovers for host economies, there is a possibility of cumulative causation, with agglomerations attracting new investments which then influence the growth process. This suggests that the relative size of host locations is endogenous, rather than fixed by current factor endowments and access to common technologies. Several recent studies have suggested that agglomeration effects are an important determinant of investment decisions by multinational firms. Wheeler and Mody (1992) and Mody and Srinivasan (1998) find that the global location of foreign direct investment by US and Japanese multinationals is positively related to variables reflecting the total stock of past inward investment in the host economy. In two detailed studies using plant-level data, Head et al. (1995,1999) report that the location of new Japanese investments in the United States is closely related both to the location of existing investments in the industry in which investment takes place and to the location of investments by other Japanese companies. Devereux and Griffith (1998) and Barrell and Pain (1999b) find that agglomeration effects also help to determine the location choice of US multinationals in Europe. Within the UK the concentration of financial services within the City of London is an obvious example of a self-reinforcing agglomeration. The long-term trends in the location of direct investment are shown in Table 1.1. In the first half of the 20th century some two-thirds of all investments were located in the developing economies. Over the last 40 years the picture has been quite different. Over two-thirds of all inward investments are now held within the developed economies. The rate of growth of the nominal investment stock has accelerated from an average 9 per cent per annum between 1960 and 1973, to around 133/4 per
6
Inward investment, technological change and growth
cent per annum between 1973 and 1998. Controlling for movements in the global price level during these periods, as proxied by the GDP deflator for the OECD economies, indicates that in real terms the growth of the global FDI stock has risen from AVA per cent per annum during 1960-73 to 7 per cent per annum since that time. The UK has been relatively successful in attracting inward direct investment, accounting for 8-9 per cent of the global stock throughout most of the period. Inward investment grew especially rapidly in the latter half of the 1980s, and by 1990 the stock of inward FDI in the UK is estimated to have been almost \2xh per cent of the global total. Even though the UK share is estimated to have slipped to 8 per cent by 1998, this remains approximately 2Vi times the share of UK output in global GDP measured on a purchasing power parity basis. 2 The growth of inward investment in the UK needs to be viewed not only in terms of the national policies and institutions that have served to make the UK a desirable location for investors, but also in the context of the policies pursued collectively by all European governments and the European Commission. Western Europe has now become the dominant host region for investment amongst the developed economies. The amount of investment held in Europe has risen especially rapidly since 1985. This period has coincided with moves towards greater European integration. The implementation of the measures to remove capital controls and non-tariff barriers to trade in the Single Market Programme (SMP) has clearly stimulated the mobility of capital and generated a significant change in investment patterns and levels of foreign involvement in production. Work undertaken for the European Commission by Arrowsmith etal. (1997) and EAG (1998) on European capital markets and on foreign direct investment shows clear positive effects from the SMP on the level of inward direct investment in both service and manufacturing activities in the UK. At the same time, the external barriers to trading with the European Union faced by producers located outside Europe have encouraged additional flows of investment into the EU, with production bases now having guaranteed access to a Europe-wide market. This has been particularly important for many Asian companies (Barrell and Pain, 1999a; Belderbos and Sleuwaegen, 1998). A low level of labour costs per unit of output compared to that in many other EU member states has helped the UK to attract many of the resulting investment projects.
The growth and impact of inward investment in the UK
7
I N W A R D I N V E S T M E N T IN T H E UK The long-term trend in the stock of inward foreign direct investment in the UK is shown in Figure 1.1. By the end of 1999 the stock was equivalent to 27 per cent of GDP, compared to approximately 6V2 per cent of GDP in 1960. This is considerably larger than the share of inward FDI in most other large industrialised economies, although comparable to that seen in a number of other small, open European economies such as the Netherlands. There are two periods in which the UK inward stock has risen especially rapidly in real terms, the latter half of the 1980s and, more recently, since 1997. This is particularly clear from the trends in the flow of new investments summarised in Figure 1.2. Inflows averaged 21/4 per cent of GDP per annum over 1985-9 and just under 3Vi per cent per annum over 1995-9, compared to a long-run average of 1.6 per cent per annum since 1960. Indeed the first official estimates suggest that 1999 was an exceptional year, with FDI inflows of £50.8 billion, equivalent to 5.7 per cent of GDP. The impact of the growth of inward investment in the 1980s is studied in greater detail in the papers by Florence Hubert and Nigel Pain, and by Nigel Driffield. The impact of the high level of inward investment in the late 1990s has yet to be examined. 30
r
25
a
o 20
S 8 15 10
o ^sD ON
m
^D ON
^D \D ON
ON ^O ON
*o IN ON
oo IN ON
r^
OO ON
^t" 00 ON
Noo ON
O ON ON
ro ON ON
\D ON ON
ON ON ON
Figure 1.1. The stock of foreign direct investment in the UK, 1960-99
8
Inward investment, technological change and growth
1.5 1 0.5 0 ^ ^ ^ O
O N > D ^ O N
^ O
^ £
f >
[
O
N N
O
t N
N N
\ \ O
^ O N
J O
O
O
O O
N O
^ N
O O O O N O N O N O N
O ,
O O O
Figure 1.2. UK FDI inflows as a per cent of GDP (annual
N N N N
averages)
In the early 1980s there was some concern that inward investment was imposing a net financial cost on the UK economy, with the value of the post-tax earnings remitted overseas by inward investors exceeding the value of new capital inflows financed from sources other than unremitted profits between 1980 and 1 9 8 5 . However since that time this has happened only once, in 1994. During 1 9 8 5 - 9 8 , the post-tax earnings of inward investors averaged lVi per cent of GDP per a n n u m , compared to capital inflows of 2VA per cent of GDP per a n n u m . Reinvestment through unremitted profits averaged just over Vi per cent of GDP per a n n u m . The composition of the inward FDI stock is summarised in Table 1.2. Looking firstly at the data by sector, the long-term decline in the relative importance of the manufacturing sector is clear, although the share of this sector does appear to have stabilised in recent years at about onethird. Inward investors remain relatively concentrated in manufacturing activities, since the manufacturing sector n o w accounts for a b o u t onefifth of total UK GDP. The importance of the energy sector is also clear, with significant investments having taken place in the N o r t h Sea and, in recent years, in the privatised public utilities. Nonetheless, the relative level of investment in this sector has begun to decline, coinciding with the decline in the relative share of oil production in total UK GDP.
The growth and impact of inward investment in the UK
9
Table 1.2. The composition of the UK inward FDI stock (per cent) By sector: 1968 1978 1988 1998
Manufacturing
Energy
58.6 46.3 32.8 31.8
29.7 29.4 27.2 15.6
By source: United States 1968 1978 1988 1998
66.8 59.5 39.4 49.2
Other nonmanufacturing
Western Europe 21.8 29.7 39.8 35.7
11.7 24.3 40.0 52.6 Asia 0.1 1.2 5.5 5.1
Rest of the world 11.3 9.6 15.3 10.0
Sources: Business Monitors M04 (1978, 1984), MA4 (1988, 1998). Notes: Energy investments include foreign investments in electrictity, gas and water supply companies. Ownership figures for 1968 and 1978 are calculated using data excluding oil, banks and insurance company investments. In recent years there has been strong growth in investments in nonmanufacturing activities such as financial services, as shown in greater detail by Hubert and Pain. The ownership data highlight the longstanding importance of inward investment from the United States. For all the attention paid to inward investment from Japan, Korea and other Asian economies, it continues to represent a small minority of total inward investment.3 It is clear that the rapid growth in inward investment in the 1980s was driven by the growth of inward investment from Western Europe and Asian investors. In contrast, investment from US-owned companies has been the most important force behind the upturn in the late 1990s. It may be that the data for the stock of inward FDI are becoming a less useful guide to the overall impact of foreign-owned firms, although in some industries, and some countries, they are the only data available if applied research is to be undertaken. Recent data reveal as much about developments in international capital markets as about the underlying locational attractiveness of the UK, with a handful of large mergers and acquisitions dominating the level of investment. A more fundamental issue is that fluctuations in the level of a financial concept, such as foreign direct investment, may differ from changes
10
Inward investment, technological change and growth
in the activities of foreign-owned firms in the host country. Technically foreign direct investment is just one means of financing the activities of inward investors. Once foreign investors have established a presence in the host country, tacit and codified firm-specific knowledge can be transferred continually from parent companies independently of other financial transactions. Equally, productive facilities can be established and expanded using capital raised outside the home country of the parent firm. Such capital will not be included in the direct investment statistics. Some simple calculations suggest that the fixed capital assets of foreign affiliates in the UK are likely to differ from the book value of financial transfers from their parent companies. Over the period 198695, which included a period of rapid growth in the late 1980s and a recession in the early 1990s, the annual net fixed capital expenditure of foreign-owned firms in the UK manufacturing sector averaged 15 per cent of their value added output. Using this to approximate the steadystate ratio of investment to output, and assuming an annual depreciation rate of 7Vi per cent (in line with the assumptions made in the paper by Nicholas Oulton) and a real growth rate of 2Vi per cent per annum, implies that the steady-state net capital stock will be just over Wi times the level of output at constant prices.4 In 1995 the stock of inward foreign direct investment in the UK manufacturing sector was approximately VA times the size of value added output in foreign-owned firms. The discrepancy may reflect the extent to which foreign affiliates have borrowed funds in the UK, or it may reflect differences between the FDI stock at book value and the present market value. Cantwell and Bellak (1998) estimate that the value of the total stock of inward investment in the UK in 1990 at current replacement cost was 28 per cent greater than the book value. In either case there is a strong argument for focusing on the activities undertaken by foreign-owned firms, as well as the level of inward direct investment, in assessing the impact of inward investors on the UK economy and the comparative success of the UK in attracting inward investment. One immediate difficulty stems from the absence of any official data on the activities of foreign-owned firms throughout the UK economy as a whole. Detailed statistics are available on the output, employment and fixed capital formation of foreign firms in the manufacturing sector. These are drawn upon in the studies by Oulton, Driffield, and Hubert and Pain. Outside the manufacturing sector comprehensive data
The growth and impact of inward investment in the UK
11
Table 1.3. Foreign-owned firms in the UK manufacturing sector (annual averages) 1973-79
1981-89
1990-97
Share of foreign-owned firms (%) Gross value added Net capital expenditure Total employment Employment of operatives Employment of non-operatives
17.4 18.1 12.8 12.0 15.2
18.5 22.2 13.8 12.8 16.4
23.9 30.9 17.4 16.0 19.9
Nationality of investor (%) United States Western Europe Japan Rest
72.9 18.7 0.1 8.3
65.2 20.3 1.0 13.5
53.1 29.8 6.6 10.5
Sources: Annual Census of Production, various issues. Note: Nationality shares based on share of gross value added. Employment of operatives and non-operatives for 1990-97 based on data for 1990-95 only. can be obtained only from commercial databases of company accounts. One particular database, the OneSource database, is also used by Oulton, as well as in the paper by Sourafel Girma, David Greenaway, Katharine Wakelin and Nuno Sousa, although they confine their attention to the manufacturing sector. The long-term trends in the share of foreign-owned firms in the UK manufacturing sector are summarised in Table 1.3. It can be seen that their share of total output, employment and investment has risen over time. As with the foreign direct investment data, the dominance of UScontrolled affiliates is clear. They continue to account for over half of the value added output produced in foreign-owned firms, even though their relative importance has declined over time as the share of Western European and, to a lesser extent, Japanese firms has risen. In aggregate it is clear that the labour productivity of foreign-owned firms, measured in terms of value added output per employee, has been continuously around 40 per cent higher than in UK-owned firms, given the respective shares of output and employment. A similar differential between the productivity of foreign and domestic firms can be seen in many other Western European economies. This appears to stem from more than just a concentration of foreign firms in sectors with high value
12
Inward investment, technological change and growth
added. Hubert and Pain show that the labour productivity advantage is also apparent within industrial classes in the manufacturing sector, and Oulton reports a similar figure from a sample of establishments drawn from the longitudinal database of the Annual Census of Production. The sample consists only of establishments that have survived throughout the period 1973-93. After controlling for industrial structure, Oulton finds that labour productivity in establishments owned by US companies in 1993 was 32 per cent higher than in UK-owned establishments. Other foreign-owned establishments had a productivity advantage of 15 per cent. Two potential explanations for the higher level of labour productivity in foreign firms are apparent from Table 1.3: foreign firms are more capital-intensive on average than domestic ones and have a higher ratio of non-operatives in total employment. The latter can be regarded as an (imperfect) proxy for the share of skilled labour. Oulton estimates that the average level of capital per worker in his sample of establishments is around 50 per cent higher in foreign firms than in UK ones. Another potential difference may be scale, with the average size of foreign establishments being larger than domestic ones. All these factors appear to be important, but cannot account fully for the productivity advantages of foreign firms. Oulton reports that greater usage of other inputs can explain some 61 per cent of the higher labour productivity of US-owned establishments and 97 per cent of that of other foreign-owned ones. Girma et al. find that after accounting for scale and industrial structure the labour productivity of foreign-owned companies during 1991-6 was 10 per cent higher than UK-owned ones, and the level of total factor productivity was 5XA per cent higher. Both these differences are statistically significant, and considerably larger for US-owned firms than for other foreign firms, especially Japanese ones. A very similar picture is shown by the results reported by Oulton for the non-manufacturing sector using company-level data for 1995. The labour productivity advantage of foreign firms is estimated to be somewhat higher than in the manufacturing sector, at 49 and 46 per cent respectively for US and other foreign-owned companies. Approximately two-thirds of these differentials can be accounted for by greater capital intensity and labour quality. Thus overall it seems clear that there are important firm-specific advantages over and above the scale of operations and the mix of measurable inputs that account for the comparatively higher labour
The growth and impact of inward investment in the UK
13
productivity of foreign firms in the UK. These may reflect factors such as better organisational efficiency, greater exposure to international competition and the quality of knowledge-based assets. If foreign firms did not possess such firm-specific advantages, it would be difficult to explain why they are able to take advantage of profitable opportunities in the UK whilst UK-owned firms are not. There are often said to be significant asymmetries in the provision of public financial incentives for inward investors and indigenous firms, but there is little hard evidence of this, not least because it is not permitted under EU competition rules. Reports by the Welsh Affairs and Scottish Affairs Committees of the House of Commons indicate that, whilst the present schemes for Regional Selective Assistance (RSA) in the UK are not particularly transparent, there is little evidence of significant discrimination between comparable projects proposed by foreign and domestic firms (House of Commons, 1998, 1999). An alternative explanation might be that domestic firms suffer from capital market constraints that do not impinge on foreign firms, although again there is little evidence in favour of this proposition, as the results in the paper by Oulton make clear.
PRODUCTIVITY SPILLOVERS FROM INWARD INVESTORS Given that foreign firms in the UK possess firm-specific advantages, it is natural to ask whether these spill over and become available to domestic companies. If that was the case, inward investment would be associated with significant positive externalities for the economy as a whole, providing a justification for government intervention in the form of investment incentives and promotional activities designed to attract potential foreign investors. It is important to be clear about how spillovers from inward investors are best evaluated. All too often the focus of investment promotion activities is on the number of jobs associated with new investments, either created directly by the investor or indirectly as a result of associated benefits to indigenous firms. Yet there is little evidence that the long-run level of unemployment in the UK and other industrialised countries is affected by the industrial structure of the economy. Instead, emphasis is placed on supply-side factors and labour market institutions such as the workings of the social security system, trade union cover-
14
Inward investment, technological change and growth
age and minimum wages (Layarde£tf/., 1991). If real (producer) wages are sufficiently flexible in response to changes in labour market pressures then a rise in the demand for labour resulting from inward investment will eventually push up real wages and reduce the demand for labour in other firms. A more appropriate means of judging the benefits from inward investors is to look at the overall level of national income. Even if inward investment does not affect the long-run level of employment, it may still affect the productivity of those employed, their wage levels and the organisational efficiency of the companies in which they work. The paper by Hubert and Pain examines the impact of inward investment on the pace of technical change in the UK. This encompasses two factors, new advances which serve to push out the production possibility frontier and demonstration effects which serve to eliminate technical inefficiencies. Either show up as a measured increase in output per unit of input. Hubert and Pain assume that the benefits from inward investment are labour-augmenting, so that it reduces the level of labour input required to produce any given level of output, given that other factors of production remain unchanged, and employ the modelling framework developed in Barrell and Pain (1997, 1999b). This allows for endogenous technical progress within labour demand relationships derived consistently from an underlying production function. There are two different studies reported in their paper. The first, which uses sector-specific stocks of foreign direct investment as indicators of the extent of inward investment, examines whether inward investment has raised technical progress in manufacturing, public services and three private service sectors in the UK economy since the early 1970s. There is clear evidence that it has done so in manufacturing, as first found by Barrell and Pain (1997), and it also appears to have been important within some, although not all, private sector service activities. The findings do not indicate that inward investment is the only source of technical change in the UK economy, but they do indicate that it is an important and significant one, accounting for up to one-third of output growth in manufacturing and the three private service sectors from 1972 to 1996. Given the limitations of the data available for the non-manufacturing sector it is possible that some of the observed benefits from inward investment simply reflect the higher productivity of foreign-owned firms, rather than spillovers into domestic companies. This does not appear to be the case for the manufacturing sector. In their second study, Hubert
The growth and impact of inward investment in the UK
15
and Pain use a panel data set of the activities of UK-owned firms in eleven manufacturing classes over 1983-92 to test directly whether the presence and scale of foreign-owned firms, as measured by the level of their value added output, have affected the performance of domestically owned firms. The results indicate that the presence of foreign firms has had a significant positive impact on the technical efficiency of domestic firms. A particularly interesting finding is that the magnitude of inter-industry spillovers from inward investment is statistically similar to the magnitude of intra-industry spillovers. This suggests that inward investment has helped to transfer innovative business techniques and management practices that can be applied across a wide range of industries rather than just new processes and products that are specific to a particular industry. Driffield uses a similar data set to consider a different issue, namely whether the benefits of inward investment are confined to the region in which it takes place, or whether there are wider spillovers to other regions. His results indicate that there are significant intra-industry and intraregional spillovers from inward investment on the productivity of domestic firms, but that there are few benefits outside the host region. There is also marginally significant evidence of inter-industry spillovers within the host region. Girma et al. also test for the existence of productivity spillovers, using a sample of 2,342 UK-owned manufacturing companies, and focus on the extent of intra-sector effects from the presence of foreign firms measured in terms of their share of total sectoral employment. Their results indicate that on average the foreign share has a positive impact on the level of productivity in domestic firms, but that it is not statistically significant. However they note evidence of significant positive effects in sectors characterised by a high proportion of skilled employees. Finally, they also fail to find any evidence that the presence of foreign firms has any effect on the rate of productivity growth of domestic firms. This is in line with the relatively constant labour productivity differential for the aggregate manufacturing sector apparent from Table 1.3. Thus there would appear to be a number of unresolved questions, even if the weight of the evidence does suggest that there are significant productivity spillovers from foreign firms in at least some industries and locations. One implication of the different results is that the presence of a single inward investor may sometimes matter as much as the presence of a number of competing firms. As we discuss below, greater competition from foreign firms might reasonably be expected to affect
16
Inward investment, technological change and growth
the speed with which domestic firms seek to adopt new practices, but a single, high-profile investor is all that is required to indicate whether new techniques and business methods are viable in the host economy. The wider economic impact of productivity spillovers from inward investment on host regions is considered in the paper by Gary Gillespie, Peter McGregor, Kim Swales and Ya Ping Yin. They use a specially modified version of their computable general equilibrium model of the Scottish economy to simulate the impact of a 3 per cent rise in the fixed capital stock of foreign-owned affiliates in the manufacturing sector in Scotland. This additional investment is allowed to have a positive impact on the level of labour-augmenting technical progress in the indigenous manufacturing sector, with the long-term effect calibrated using the findings reported in Barrell and Pain (1997). It also raises the level of output exported from Scotland to the rest of the UK and overseas. The long-run outcome of the simulation is that the additional inward investment has a permanent effect on the size of the Scottish economy, with the volume of GDP eventually 0.7 per cent higher than it would otherwise have been. Output in the indigenous manufacturing sector is raised by almost 1 per cent, with approximately three-quarters of this resulting from supply-side spillovers. Other non-manufacturing sectors also benefit, although in this case it reflects the impact of demand-side effects generated by the expansion in export sales. Related findings for the UK economy as a whole are reported by Pain and Young (2000) using a version of the National Institute macroeonometric model of the UK economy in which technical progress in the manufacturing and nonmanufacturing sector is endogenous and determined according to the results reported in the paper by Hubert and Pain in this volume. One interesting difference between these two models is that the initial rise in real wages that results from a higher level of activity has less effect on the demand for labour in the regional model because it also acts to induce additional inward labour migration which subsequently eases labour market constraints. Gillespie et al. also consider the implications of their results for the design of regional investment promotional policies. Supply-side spillovers take some time to emerge in the particular version of the model they use. In the short term demand-side effects dominate, so that, whilst a sole focus on demand-side effects might be inappropriate, it could still be a reasonable guide to the impact of policies over short planning horizons.
The growth and impact of inward investment in the UK
17
O T H E R F O R M S OF SPILLOVERS F R O M I N W A R D INVESTMENT The discussion to date has considered direct spillovers from foreignowned firms to domestic firms which improve productive efficiency. There are also several channels through which inward investment might have an indirect impact on the performance of domestic firms. Three of these are considered in papers in this volume; the impact of greater competition in product markets, the impact on national innovation and R&D, and the impact on the openness of the host economy. All of these are known to be potentially important determinants of productivity growth.
PRODUCT MARKET COMPETITION One impact of greater competition, or contestability, in product markets is likely to be the elimination of productive inefficiencies amongst underperforming firms and the efficient allocation of resources across industries in line with comparative advantages. Evidence of a significant positive effect between competition and total factor productivity growth in UK-resident firms is reported in Nickell (1996). Evidence on the importance of competition has also emerged from studies employing stochastic frontier production functions to estimate the gap between average and best plant productivity in the UK (Hay and Liu, 1997). Blomstrom, Kokko and Globerman argue that competitive markets in host economies facilitate the transfer of new technologies by inward investors and raise the incentives for indigenous firms to incur the costs required in order to adopt those technologies. However it is also possible that at times multinational companies may be reluctant to transfer technologies, or indeed undertake further innovation, if they feel the resulting economic rents are soon going to disappear as a result of competition. Whilst there is less empirical evidence on this issue than might be expected given the importance attached to competitive markets by policymakers (DTI, 1998), the balance of evidence across countries cited by Blomstrom et al. does appear to point to favourable effects of spillovers due to greater competition as a result of additional inward investment. However the evidence reported for the UK in this volume is mixed. Girma et al. report that while there appears to be little general evidence
18
Inward investment, technological change and growth
that the productivity of domestic firms is significantly related to a rising intra-sector share of foreign firms, there is a stronger effect on those firms with a low productivity gap relative to the sector average. Driffield finds that inward investment in a given industry has a negative effect on the productivity of domestic firms in that industry outside the region in which investment has taken place. One potential explanation of this finding is that competition has reduced the scale of operations of domestic firms and raised average costs.
INWARD INVESTMENT AND R & D The creation, application and refinement of knowledge are three of the key factors driving the growth process in many theoretical and empirical models of growth and technological change. There have been many studies that have indicated that R&D expenditures have a significant influence on productivity in developed and developing economies (Griliches, 1998; CoeetaL, 1997). Foreign firms have come to play an increasingly important role in both the financing and performance of research and development in the UK. Table 1.4 summarises the trends over the last decade for business enterprise R&D. By 1998, this represented approximately two-thirds of the total R&D undertaken in the UK. The share of business enterprise R&D expenditure funded by foreign sources and performed by foreign-owned firms has risen sharply since 1989. 5 Of the total level of expenditure by foreign firms in 1998, some 52V^ per cent was undertaken by US-owned affiliates, and 263A per cent by EU-owned affiliates. Japanese affiliates accounted for just 8V2 per cent of expenditure. The importance of foreign firms in total R&D becomes even more apparent if the figures in Table 1.4 are converted into constant prices using the GDP deflator from the UK national accounts. In 1995 prices, the total expenditure on R&D in 1998 was 3Vi per cent lower than in 1989 and 0.7 per cent lower than in 1994. Funding from foreign sources at constant prices rose by 57.3 per cent between 1989 and 1998, and expenditure by foreign-owned firms rose by just over 75 per cent. Funding from, and expenditure by domestic sources declined by 13 and 19 per cent respectively in real terms over the same period. One distinctive feature of the UK is that it is the only major country
The growth and impact of inward investment in the UK
19
Table 1.4. Business enterprise R&D in the UK (£bn, percentage share in parentheses)
Total R&D expenditure Financed by foreign funds Performed by foreign firms Memorandum item GDP deflator (1995=100)
1989
1994
1998
7.60 1.02 (13.5) 1.26 (16.5)
9.20 1.51 (16.4) 2.57 (28.0)
10.23 2.25 (22.0) 3.08 (30.1)
78.2
97.5
109.1
Sources: Business Monitors M014 (1989) and MA14 (1998), and UK National Accounts 1999 (Table 1.4).
in which the share of foreign firms in manufacturing R&D is greater than their share in production (OECD, 1998). Put differently, their average R&D intensity is higher than that of domestic firms. Canada is the sole G7 economy in which the foreign contribution to R&D is similar to that in the UK. But there is an important difference between the UK experience and that of Canada. The R&D expenditure of domestic firms in the UK has fallen in real terms, whereas it has risen rapidly in Canada (OECD, 1998, p. 63). A further dimension of the contribution of foreign-owned firms is provided by their research output in the form of patentable products and processes. Comparative statistics for the UK, Germany and Italy are provided in the paper by John Cantwell, Simona lammarino and Camilla Noonan. Their figures (Table 8.1) show that foreignowned firms received 45 per cent of the patents granted by the United States Patent and Trademark Office (USPTO) to large firms resident in the UK during 1991-5, compared to just over one-third of the patents granted in the 1980s. The penetration of foreign firms in the UK is lower than in Italy, but well above the average for Western Europe as a whole. Of course the proportion of R&D funded and undertaken by foreign-controlled entities does not in itself matter, although it probably raises the direct influence of foreign economic conditions on domestic expenditure. Indeed part of the explanation for the increase in the foreign share may just be the takeover of R&D-intensive indigenous companies by foreign companies. In this sense it might be said
20
Inward investment, technological change and growth
that the attributes of the British science and engineering base have helped to attract inward investment. Mergers and acquisitions of this kind may help to strengthen further the technological position of the acquired companies. While the additional foreign funding of R&D in the UK is welcome, there are many areas where further research is needed before the implications for the UK economy are well understood. Blomstrom, Kokko and Globerman highlight several key issues in their discussion of the potential spillovers from R&D performed in foreign-owned firms. These include the extent to which multinational firms may choose to centralise their R&D operations in the parent company and use foreign facilities just to adapt parent company technologies for local markets, the extent to which domestic firms need to undertake their own R&D to facilitate technology adoption, and the magnitude of any spillovers. In these respects the decline in the volume of R&D expenditure by UK firms over the past decade is somewhat worrying. If foreign firms are to become an ever more important source of R&D, it becomes important to understand the factors that lie behind their choice of location and the range of activities they choose to undertake. These issues are discussed in the paper by Cantwell et al. They argue that the extent to which agglomerations of technological knowledge and expertise serve to attract additional investment varies across industries and regions. Some regions offer only knowledge in specific fields, and attract sector-specific investments. Others, termed higher-order research centres, offer more general technological expertise and locational advantages and can attract investments from a wider range of industries. In the UK it is only the South East which appears to act as a higher-order research centre, attracting investments from foreign firms across a wide range of technologies independently of the relative investment in those technologies by domestic firms. Around 60 per cent of the patents granted by the USPTO to large foreign firms resident in the UK during 1969-95 were granted to firms located in the South East. A comparable figure was observed for foreign firms located in the Piemonte region of Italy. The South East and Piemonte were also the most popular locations for indigenous companies, although to a lesser extent. In contrast there is a much broader dispersion of foreign-owned firms found within Germany, and the most popular location for foreign firms is the third most popular only for domestic firms. The authors argue that this reflects a desire to gain access to core technological
The growth and impact of inward investment in the UK
21
expertise while avoiding direct competition with local, regionally specialised research centres.
I N W A R D I N V E S T M E N T A N D I N T E R N A T I O N A L TRADE Inward investment might also change the level of international trade undertaken by firms located in the host country, since multinational firms dominate merchandise trade and play an increasing role in trade in services. This will almost certainly be associated with some changes in the industrial structure of the host economy, and possibly also the long-run size of that economy if there are beneficial externalities from exporting. There is certainly evidence that greater international openness to trade is associated with faster economic growth, both in the UK and in other European economies, although the evidence is less clearcut from the comparative performance of individual exporters and nonexporters (Bernard and Jensen, 1999). 6 One important means through which inward investors can expand exports is through sales to their parent companies. Unfortunately there are no official data on intra-firm trade by inward investors in the UK. Data on the total exports of foreign-owned firms were collected in the past but the surveys were discontinued in the early 1980s. At that time it was already clear that foreign-owned firms were relatively exportintensive, accounting for 34 per cent of all manufacturing exports in 1981, compared to \%XA percent of gross value added output. Some current evidence on the importance of inward investors in UK export performance can be gleaned from individual company accounts data. Table 1.5 summarises the activities in 1997 of the 85 largest UK exporters, who accounted for around 30 per cent of all exports that year and had a total turnover of £179.9 billion. Of these, it is possible to distinguish the destination of the exports of 67 firms. The exports of this sample of firms were worth £57.9 billion in 1997, around 25 per cent of total UK exports. The destination of their exports is similar to the composition for total UK merchandise exports, with just over half of all exports destined for European markets. The foreign-owned companies account for just under a half of total exports in this sample. There are some interesting differences between the foreign and UK companies. Exports are a higher proportion of turnover for foreign companies at 44 per cent, compared with 37 per cent for domestically owned companies. Europe is a more important market for foreign firms
22
Inward investment, technological change and growth
Table 1.5. The UK's major exporters in 1997 All firms UK firms European firms Number of firms 67 57,882 Total exports (£m) 31,077 Europe exports: (£m) (% of total exports) (53.7) Non-Europe exports: (£m) 26,805 (% of total exports) (46.3) Total exports/turnover 40.4% UK employees 704,738
37 9 29,671 9,935 13,023 4,137 (43.9) (41.6) 16,650 5,798 (56.1) (58.4) 37.3% 39.5% 505,251 102,127
Non-European firms 21 18,276 13,917 (76.1) 4,357 (23.9) 47.6% 97,360
Source: Pain and Young (2000, Table 3). than British ones, accounting for 64 per cent of foreign-owned sales compared to 44 per cent for British firms. More noticeable still is the importance of the European market to non-European firms in the UK. Other European-owned firms are in many ways similar to their British counterparts, exporting a similar proportion of their overall turnover to Europe and the rest of the world. By contrast, the non-European foreign firms export almost half of their turnover with the bulk, over three-quarters of exports, going to Europe. In the sample of 67 companies, there are thirteen whose exports to Europe account for more than 40 per cent of their turnover (and thus almost certainly over half of their value-added output). Nine of these are foreign-owned firms and all are controlled by parent companies from non-EU locations. If the UK were to leave the European Union (and the European Economic Area) such firms might well choose to relocate elsewhere inside the EU to bypass any tariff or non-tariff barriers to trade (Barrell and Pain, 1999a). Other evidence on the impact of inward investors on UK exports can be constructed from the detailed statistics available in the United States on the operations of overseas affiliates of US parent companies. These indicate that US non-bank foreign affiliates in the UK exported $8.36 billion of goods to the United States in 1997, representing approximately one-quarter of total UK merchandise exports to the US that year. 7 Inward investors might also affect import penetration in the UK, particularly as foreign-owned firms tend to employ a higher proportion of intermediate inputs than UK-owned firms, as the papers by Hubert
The growth and impact of inward investment in the UK
23
and Pain, and Oulton both report. Many of these inputs are likely to be sourced from overseas, at least initially. The issue of the impact of both inward and outward foreign direct investment on bilateral, industry-level trade is considered in detail in the paper by Fontagne and Pajot. The extent of the complementarity between international trade and investment has recently been a subject of considerable interest to policymakers at the OECD, the World Trade Organisation and elsewhere. The relationship between trade and investment may well vary across countries, time and the level at which any analysis is undertaken. For vertically integrated firms, foreign production and trade are likely to be complements. For horizontally integrated firms, foreign production of finished goods may substitute for trade at the level of the firm, but may raise trade at the industry level if the size of the foreign market is expanded for other goods produced in the investing country, or if the foreign affiliate purchases intermediate goods from other supplier firms in the investing country. Fontagne and Pajot use a gravity model of bilateral trade to examine the influence of inward and outward investment stocks on the bilateral pattern and level of UK trade in a panel of industries, and draw comparisons with the experiences of France and the United States. Inward investment in a particular industry in the UK is found to have a significant positive effect on the level of exports to the investing country in that industry and a significant negative impact on the level of imports. One possible explanation for this is that foreign firms have chosen to relocate labour-intensive export-orientated operations to the UK to take advantage of the comparatively low labour costs available here compared to other potential European hosts (Barrell and Pain, 1999a, 1999b). In contrast, inward investment in France and the US is found to be associated with an overall deterioration in the balance of trade, although it does act to raise the overall level of trade relative to income. These findings imply that multinational corporations can influence employment prospects through their indirect effects on exports and imports, as well as through direct effects from their own employment. A related question into which comparatively little research has been undertaken is whether they also affect the export performance of domestic firms. For instance, foreign companies can provide knowledge and expertise that can help domestic firms to overcome the fixed costs of entry into export markets. This issue is discussed in greater detail by Girma, Greenaway, Wakelin and Sousa.
24
Inward investment, technological change and growth
They use a firm-level data set for the UK manufacturing sector over the period 1992-6 to test whether foreign-owned firms affect both the probability of domestic firms exporting and their export propensity if they do so. The probability of a firm in a particular industry choosing to export is found to have a significant positive association with the share of foreign-owned firms in employment in that sector and the R&D expenditure carried out by foreign firms in that sector. Both variables are also found to have a significant positive link with export propensity. These findings are consistent with the hypothesis that greater competition from foreign firms raises the potential for beneficial spillovers of knowledge to domestic firms. There is less evidence in favour of the hypothesis that the effects are strongest in those sectors in which foreign-owned firms account for the largest shares of total export activity, possibly suggesting that country-specific information from foreign firms may be more important than industry-specific information.
T H E RESEARCH A G E N D A The papers in this volume provide a comprehensive overview of the current empirical evidence regarding the impact of inward investors on the British economy. Whilst interest in this topic has expanded markedly in recent years, there remain a large number of topics which have yet to be investigated in much detail either in the UK or elsewhere. Most research to date has concentrated on the potential existence of spillovers rather than the factors that determine their magnitude or their distributional impact. The overview of the literature provided by Blomstrom, Kokko and Globerman highlights a number of areas in which further research might be merited. One important issue is whether spillovers are generic to all inward investors, or whether there are particular types of investor or particular sectors which are more important than others. For instance R&D expenditures are dominated by a fairly limited number of large firms and are concentrated in relatively capital-intensive industries. It would be of interest to know whether the effects from inward investment by such firms differ from those in other industries. A related area where further work is necessary lies in the question of whether there are significant inter-industry spillovers from inward investment, as suggested by the econometric results of Hubert and Pain,
The growth and impact of inward investment in the UK
25
and, if so, what they are and how they are diffused. In principle there would seem to be little reason why some innovative managerial techniques and production standards brought in by individual firms could not be applied throughout the host economy. Romer (1993) highlights developments in the automobile industry over the past century, with international investments being the conduit for the transmission of ideas such as mass production and subsequently 'just-in-time' techniques and quality control standards. In both instances such techniques could be utilised across many different industries. This would seem to be an issue in which case studies comparing the organisation of foreign and domestic firms within particular industries would be able usefully to complement the findings from empirical analysis. It would also be of interest to investigate whether there are spillovers between the manufacturing and non-manufacturing sectors. For instance 'just-in-time' inventory control techniques are widely used in the retail and wholesale trade industries, as well as in manufacturing ones. Relatively little is known about the impact of the mode of entry on the transmission of knowledge and technologies to host country firms. Firm-specific knowledge and practices can in theory be introduced as easily through the reorganisation of existing companies as through the establishment of 'greenfield' sites. However there is little empirical evidence on the relative impact of these forms of investment or, indeed, on the impact of other forms of relationship such as joint ventures and strategic alliances. Another issue where limited research has been undertaken is the role of labour mobility in the transmission of knowledge. Blomstrom, Kokko and Globerman highlight the virtual absence of any detailed empirical evidence on the potential existence and magnitude of spillovers resulting from the movement of managers and other skilled employees from foreign-owned firms to domestic ones. Girma et al. find that on average wage levels per employee in foreign-owned firms in the UK manufacturing sector are SVASVI per cent higher than in UK-owned firms, even after controlling for productivity differentials. One explanation might be that some foreign firms pay a premium in order to try and prevent key employees from leaving. A related issue to do with human capital is the extent to which industries employing workers with comparatively high levels of human capital are better able to assimilate new ideas and technologies from inward investors. Borensztein et al. (1998) find that there are significant complementarities between the level of human capital and the
26
Inward investment, technological change and growth
benefits of inward investment in their multi-country study, although their results largely relate to developing economies. The preliminary evidence discussed by Girma et al. would seem to suggest that this might be important in the UK as well. Even if there are positive spillovers from inward investment, as the balance of evidence in this volume suggests, there is no necessary reason why the gains should be felt equally in all parts of the country, or by all factors of production. The increasing importance of regional agglomerations of skilled labour, R&D facilities or large markets in the location decisions of multinational firms could lead to geographic concentration and widen regional disparities. There is clear evidence that such forces could be at work in the UK in the evidence provided in the papers by Driffield, and by Cantwell, lammarino and Noonan. Equally, if there are important complementarities between fixed capital and skilled labour, the factor bias of the new technologies introduced by foreign firms might also help to raise the relative demand for skilled labour in host economies, and potentially widen wage inequality. Little empirical work has been undertaken on this question to date, especially in the UK. Hubert and Pain allow for labour augmenting technical progress, but do not seek to distinguish between the impact of inward investment on different types of labour.
POLICY IMPLICATIONS Productivity spillovers from foreign firms to indigenous ones provide a justification for proactive government policies to try and attract inward investment. The contingent nature of location and the scope for selfreinforcing agglomeration effects have important implications for these policies if other countries that compete for investments in 'location tournaments' provide similar incentives. There is little empirical evidence that investment incentives have been an important determinant of the total level of inward investment in either developed or developing countries, although they may make a difference at the margin for some projects (UNCTAD, 1998). In contrast, there is plenty of evidence which suggests that incentives can affect the distribution of inward investments within host countries, partly because different levels of assistance are offered in different regions. Even if incentives make little difference for the country as whole, it does not necessarily follow that they should be abandoned. Offering incentives may not in itself produce many gains,
The growth and impact of inward investment in the UK
27
but forgoing them could prove costly and difficult to reverse if there are self-reinforcing agglomeration effects (Barrell and Pain, 1999b; Head etal.,1999). Of course it is a matter of judgement whether the externalities from inward investments exceed the value of the resources that would be released if incentives were forgone. The evidence suggests that there are no automatic benefits from inward investment. Investments have to fit in with the capabilities of the host location and the state of development. Carefully formulated, proactive industrial policies aimed at particular sectors can be highly successful, as the experience of Ireland indicates (Ruane and Gorg, 1999). Complementary public policies designed to improve the contestability of national markets and strengthen the local science base and the adaptability of the local workforce may help to raise the magnitude and speed with which new ideas and technologies spill over to local firms. A related question of policy interest is whether the effects of inward investment differ according to the nationality of the investor. The findings in this volume highlight the particularly important role of the foreign affiliates of US companies in the UK. The papers by Oulton and Girma et al. illustrate that the ownership advantages of these affiliates are greater than those of other foreign-owned affiliates. Hubert and Pain obtain evidence that the spillovers from US affiliates in the manufacturing sector are larger than from other foreign affiliates, although the difference is not statistically significant. There is little evidence in any of these papers that inward investments from Asian countries have been the primary source of spillovers. Thus there may be some merit in favour of greater concentration of investment promotion activities on potential investors from the United States. It is worth noting that a simple concentration on the relative levels of foreign direct investment received from the United States provides a misleading guide to the relative success the UK has had in attracting inward investment (Barrell and Pain, 1997, 1999c). In the manufacturing sector, US affiliates produce a greater volume of output and undertake more R&D expenditures in Germany than they do in the UK, although employment is higher in British affiliates. The average labour productivity of US affiliates in the UK is over 15 per cent lower than in all US affiliates in the European Union and 25 per cent lower than in affiliates in Germany. Providing an explanation for these differences may do much to highlight factors which currently act to constrain the magnitude of spillovers from inward investment in the UK. Barrell and Pain
28
Inward investment, technological change and growth
(1999b) illustrate that agglomeration economies arising from the size and strength of the research base in Germany are one important factor. Given the potential for spillovers from inward investment, there would seem to be a strong case for well designed after-care programmes to help ensure that linkages between investors and local suppliers are developed and that any potential supply-side spillovers are actually realised (Young and Hood, 1994). With reinvested profits worth approximately 0.5 per cent of GDP per annum in the UK, there is already considerable capital deepening taking place. However it would be a mistake for investment promotion agencies to think that all transnational affiliates can be left to fend for themselves after entry. Persuading investors to stay matters. Mudambi (1998) presents evidence for the UK which suggests that multinational investment is significantly durationdependent, so that firms with a longer tenure of operations are significantly more likely to undertake additional investment in any given period. The papers in this volume also help to highlight the potential for tension between the preferences of regional and local development agencies and those of central government. Given the importance of agglomeration forces, the optimal strategy for an individual development agency is to seek to attract as much inward investment as possible. Losing out on one project raises the chances of losing out on related future investments if there are important external economies of scale from agglomerations due to vertical linkages of suppliers and customers, or horizontal linkages of specialised local pools of skilled labour. The findings in the papers by Driffield, on the regional scope of spillovers, and by Cantwell et al.^ on the concentration of the research activity of inward investors, emphasise the local incentives to gain investments. However, for the country as a whole, the optimal strategy is to use public funds to seek to ensure that inward investment is located so as to maximise the potential spillovers and agglomeration economies from that investment. Policies that seek to disperse foreign firms to regions with high unemployment may forgo some of the important spillovers that they could generate and forgo some of the agglomeration economies that might serve to attract additional investments. In principle at least, the losses that could result from excessive local competition might outweigh the benefits to the individual regions. This is not to imply that local agencies do not have an important role to play. It is highly unlikely that a single centralised authority can be fully conversant with all the relative strengths of each local economy. But there would appear to be a strong case for ensuring greater central coordination.
The growth and impact of inward investment in the UK
29
SUMMARY There have been many earlier reviews of the impact of inward investment on the British economy. Reviewing developments in the 1970s and early 1980s from an international business perspective, Young et al. (1988, Chapter 7) concluded that the foreign sector had 'made a positive and welcome contribution to the economic well being of the country' but that linkages between multinational companies and indigenous firms were weak and spillovers were limited. The evidence in this book allows a somewhat more optimistic conclusion to be drawn. Foreign-owned firms have helped to raise the level of national income and improve export performance, and have had beneficial supply-side effects on the performance of UK-owned companies. However the externalities available from inward investment do not always appear to have been distributed equally amongst industries or regions, and there appears to be little evidence that the average productivity gap between foreign and UK firms has been closed over time. Whilst some gap might reasonably be expected to persist, since the average foreign-owned company operates at a larger scale than the average British company, the apparent failure to narrow the gap does suggest that there may be policies or institutional reforms which could help to raise the magnitude of spillovers still further. It is to be hoped that the papers in this volume help in the formation of a research agenda to discover exactly what those policies and reforms might be.
NOTES 1 I am grateful to colleagues at the National Institute, and in particular to Florence Hubert, for help in preparing this introduction. I am also grateful to the ESRC for financial support (grant number L138251022). 2 The International Monetary Fund estimates the UK share of world GDP in 1998 to be 3.2 per cent on a PPP basis. See the Spring 2000 World Economic Outlook Database. 3 The comparative size of inward FDI from the United States and Japan appears to be broadly consistent with the data available in these countries on employment in their UK affiliates. Toyoukeizai data indicate that Japanese affiliates in the UK had approximately 120,000 employees in 1995. Employment in the non-bank affiliates
30
4
5
6
7
Inward investment, technological change and growth
of US companies in the UK was 928,800 {Survey of Current Business^ October 1997, p. 62). I am grateful to Kyoji Fukao for assistance with the Toyoukeizai data. If the annual depreciation rate was assumed to be 5 per cent, then the steady-state value of the net capital stock would be 2.05 times the level of output under these assumptions. The increase in foreign funding for R&D appears to continue a trend apparent in earlier years as well. For example foreign source funding covered lllA per cent of total R&D expenditure in 1985. However detailed comparisons are hampered by privatisations, which have moved corporations from the public sector into the business enterprise sector. The most R&D-intensive of these, the Atomic Energy Authority, raised recorded business enterprise R&D expenditure in the UK by over 10 per cent when it was transferred in 1986. A comprehensive survey of the empirical literature on exposure to international markets and corporate performance is provided by Pain and teVelde (2000). See the tables on pages 29 and 113 of the July 1999 issue of the Survey of Current Business.
REFERENCES Arrowsmith, J., Barrell, R., Pain, N., Young, G. and Wlodek, K. (1997), Capital Market Liberalization In Europe, London, Kogan Page. Barrell, R. and Pain, N. (1997), 'Foreign direct investment, technological change and economic growth within Europe', Economic Journal, 107, pp. 1770-86. (1999a), 'Trade restraints and Japanese direct investment flows', European Economic Review, 43, pp. 29-45. (1999b), 'Domestic institutions, agglomerations and foreign direct investment in Europe', European Economic Review, 43, pp. 9 2 5 34. (1999c), 'The growth of foreign direct investment in Europe', in Barrell, R. and Pain, N. (eds), Investment, Innovation and the Diffusion of Technology in Europe, Cambridge, Cambridge University Press. Belderbos, R. and Sleuwaegen, L. (1998), 'Tariff jumping DFI and
The growth and impact of inward investment in the UK
31
export substitution: Japanese electronics firms in Europe', International Journal of Industrial Organisation, 16, pp. 601-38. Bernard, A.B. and Jensen, J.B. (1999), 'Exceptional exporter performance: cause, effect or both?', Journal of International Economics, 47, pp. 1-25. Borensztein, E., De Gregorio, J. and Lee, J.-W. (1998), 'How does foreign direct investment affect economic growth?', Journal of International Economics, 45, pp. 115-35. Cantwell, J. and Bellak, C. (1998), 'How important is foreign direct investment?', Oxford Bulletin of Economics and Statistics, 60, pp. 99-106. Coe, D.T., Helpman, E. and Hoffmaister, A.W. (1997), 'North-South R&D spillovers', The Economic Journal, 107, pp. 134-49. Devereux, M.P. and Griffith, R. (1998), 'Taxes and the location of production: evidence from a panel of US multinationals', Journal of Public Economics, 68, pp. 335-68. DTI (1998), Our Competitive Future: Building The Knowledge Driven Economy, The Competitiveness White Paper: Analysis and Background, London, Department of Trade and Industry. Dunning, J.H. (1988), Explaining Inter national Production, London, Harper Collins. EAG (1998), Foreign Direct Investment, London, Kogan Page. Eltis, W. (1996), 'How low profitability and weak innovativeness undermines UK industrial growth.' Economic Journal, 106, pp. 184-95. Eltis, W. and Higham, D. (1995), 'Closing the UK competitiveness gap', National Institute Economic Review, 154, November, pp. 71-84. Fujita, M., Krugman, P. and Venables, A.J. (1999), The Spatial Economy: Cities, Regions and International Trade, Cambridge MA., The MIT Press. Griliches, Z. (1998), R&D and Productivity: The Econometric Evidence, Chicago, University of Chicago Press. Hay, D.A. and Liu, G.S. (1997), 'The efficiency of firms: what difference does competition make?', Economic Journal, 107, pp. 597-617. Head, C.K., Ries, J.C. and Swenson, D.L. (1995), 'Agglomeration benefits and locational choice: evidence from Japanese manufacturing investments in the United States', Journal of International Economics, 10, pp. 92-116. (1999), 'Attracting foreign manufacturing: investment promotion and
32
Inward investment, technological change and growth
agglomeration', Regional Science and Urban Economics, 29, pp. 197-218. Hoeller, P., Girouard, N. and Colecchia, A. (1998), 'The European Union's trade policies and their economic effects', OECD Economics Department Working Paper No. 194. House of Commons (1998), Investment in Industry in Wales, Welsh Affairs Committee Fourth Report, House of Commons Session 1997-98, HC 821. (1999), Inward/Outward Investment in Scotland, Scottish Affairs Committee First Report, House of Commons Session 1998-99, HC 84-1. Hubert, F. and Pain, N. (1999), 'Innovation and the regional and industrial pattern of German foreign direct investment', in Barrell, R. and Pain, N. (eds), Investment, Innovation and the Diffusion of Technology in Europe, Cambridge, Cambridge University Press. Layard, R., Nickell, S. and Jackman, R. (1991), Unemployment: Macroeconomic Performance and the Labour Market, Oxford, Oxford University Press. Mansfield, E. and Romeo, A. (1980), 'Technology transfer to overseas subsidiaries by US based firms', Quarterly Journal of Economics, 95, pp. 737-50. Mody, A. and Srinivasan, K. (1998), 'Japanese and US firms as foreign investors: do they march to the same tune?', Canadian Journal of Economics, 31, pp. 778-99. Mudambi, R. (1998), 'The role of duration in multinational investment strategies', Journal of International Business Studies, 29, pp. 2 3 9 62. Nickell, S.J. (1996), 'Competition and corporate performance', Journal of Political Economy, 104, pp. 724-46. OECD (1998), Internationalisation of Industrial R&D: Patterns and Trends, Paris, OECD. PACEC (1995), Assessment of the wider effects of foreign direct investment in manufacturing in the UK, Report by PA Cambridge Economic Consultants for Department of Trade and Industry. Pain, N. and te Velde, D.W. (2000), Exposure to International Markets and Corporate Performance, report prepared for UK Department of Trade and Industry. Pain, N. and Young, G. (2000), Continent Cut Off? The Macroeconomic Impact of British Withdrawal from the EU, report prepared for Britain in Europe.
The growth and impact of inward investment in the UK
33
Proudman. J. and Redding, S. (eds) (1998), Openness and Growth, London, Bank of England. Romer, P. (1993), 'Idea gaps and object gaps in economic development', Journal of Monetary Economics, 32, pp. 543-73. Ruane, F. and Gorg, H, (1999), 'Irish FDI policy and investment from the EU', in Barrell, R. and Pain, N. (eds), Investment, Innovation and the Diffusion of Technology in Europe, Cambridge, Cambridge University Press. UNCTAD (1998), 'Investment policy issues', World Investment Report 1998, Chapter 3, Geneva, United Nations. (1999), World Investment Report 1999, Geneva, United Nations. Wheeler, D. and Mody, A. (1992), 'International investment location decisions: the case of U.S. firms', Journal of International Economics, 33, pp. 57-76. Young, S. and Hood, N. (1994), 'Designing developmental after-care programmes for foreign direct investors in the European Union', Transnational Corporations, 3, pp. 45-72. Young, S., Hood, N. and Hamill, J. (1988), Foreign Multinationals and The British Economy, London, Croon Helm.
2
The determinants of host country spillovers from foreign direct investment: a review and synthesis of the literature Magnus Blomstrom, Ari Kokko and Steven Globerman
INTRODUCTION The existence of spillover efficiency benefits to host country economies from inward foreign direct investment (FDI) is well documented in the literature. 1 The determinants of the size and scope of the spillover benefits have also been studied, but they are not as clearly and consistently documented as the existence and magnitude of the relevant externalities. There are many factors that are thought to affect the size and nature of productivity spillovers from inward investment, including host country characteristics such as industrial market structure, technological sophistication and overall economic size. Attributes of the nature of the inward investment have also been considered, such as whether the dominant mode of international business is a wholly owned affiliate, a minority-owned affiliate, or a strategic alliance. Some attention has also been paid to the motives and attributes of the foreign investor. Nevertheless we would argue that theoretical consideration of the determinants of FDI spillovers has been relatively ad hoc and limited. Perhaps as a consequence, there is no well-established theoretical paradigm for the determinants of spillover efficiency benefits to guide empirical research. Indeed a casual reading of the relevant literature conveys a sense of conflicting or inconclusive theoretical, as well as empirical evidence.2 A good understanding of the determinants of the nature and magnitude of FDI efficiency spillovers is of crucial importance to policymakers. Sectoral and related restrictions on inward FDI continue to keep
34
The determinants of host country spillovers
35
capital markets segmented. The ability to leverage spillover efficiency benefits from inward FDI is an important stimulant to a host government's willingness to reduce or eliminate the relevant restrictions. Moreover many government policies not directly related to FDI would be seen as even more (or less) in the public interest if they had substantive positive (or negative) impacts on the benefits that the economy derived from inward FDI. The primary purpose of this chapter is to review and synthesise the available literature, focusing on the determinants of efficiency spillovers from inward FDI. In particular, we attempt to identify points of agreement and to reconcile, if possible, points of disagreement. We also identify issues that merit additional study for the potential light they might shed upon the FDI spillover process. In order to do so, we outline a theoretical framework for understanding the underlying 'supply' and 'demand' forces determining the scope and magnitude of FDI spillovers to host economies. The chapter proceeds as follows. In the next section, we present and discuss a theoretical framework for identifying determinants of FDI spillovers. We also discuss the main empirical hypotheses linking FDI spillovers to their underlying determinants. The third section summarises and assesses the available empirical evidence bearing upon these hypotheses. Finally, we address policy implications and areas for future research.
D E T E R M I N A N T S O F FDI SPILLOVERS In a stylised fashion, one can think of spillovers in a traditional market supply and demand context. In the case of FDI, foreign investors make available (directly or indirectly) appropriable technology to host country businesses. Appropriable technology should be viewed broadly as any tangible or intangible resource that can generate economic rent for host country firms, for example by improving technical efficiency and total factor productivity. However for convenience we shall discuss the concept in its more narrow and traditional meaning, that is, embodied or disembodied knowledge about production and distribution. Foreign investors presumably recognise that host country firms may be able to benefit from spillovers. They also presumably recognise that they can expend resources to mitigate spillovers if it is a profit-maximising strategy to do so. For example, if conveying trade secrets to the
36
Inward investment, technological change and growth
managers of a local affiliate is believed to create unacceptable risks that those managers will take their knowledge to rival local firms (or to other foreign affiliates) in exchange for better pay or improved working conditions, the multinational enterprise (MNE) might consider paying 'efficiency wages' to the managers entrusted with trade secrets.3 Alternatively, the MNE might use expatriate managers from the home country in the foreign affiliate rather than local managers, since the former are less likely than the latter to defect to rival domestic companies. The point here is that the technology (whatever its specific nature) made available to domestic firms, and hence the potential level of spillovers, will be at least partly (if not totally) determined endogenously by the actions of foreign investors.4 Presumably, when it is relatively expensive to constrain the supply of appropriable technology in the host economy, as opposed to internalising the technology within foreign affiliates, greater opportunities for spillovers will willingly be made available by MNEs. When it is relatively cheap to constrain that supply, as compared with the alternative, there will be fewer opportunities for host country firms to profit from FDI spillovers. At the same time, host country firms will need to expend resources in order to adopt the appropriable technology or, equivalently, to use that technology to generate improved productivity. When the value of the resources required for adoption is small relative to the value of the underlying technology, the demand on the part of host country firms to adopt the appropriable technology will be strong. All other things being equal, the economic value of host country spillovers should also be relatively large. On the other hand, if costs of adoption are large relative to the underlying economic value of the technology (to host country firms), there will be relatively little adoption and relatively few realised FDI spillovers. This simple, stylised supply and demand framework for viewing the determinants of FDI spillovers will be developed further below. At this point, it should be noted that a comparable framework was previously suggested (at least implicitly) by Cantwell (1991). Nevertheless there may be value in further developing the framework, since its essential insight is that host government policies may have an important effect on the costs and benefits to the multinational enterprise of allowing technology to be appropriated by host country firms. That is, both supply and demand considerations may be important for leveraging the potential benefits of inward FDI spillovers. Identification and (if pos-
The determinants of host country spillovers
37
sible) quantification of factors on both the supply and demand sides of the 'market for spillovers' might, therefore, offer up valuable additional degrees of (policy) freedom to decision makers.
Determinants of supply As noted above, making (or allowing) technology to be available for appropriation by host country firms has an obvious potential cost to foreign investors. Specifically, it could lead to a reduction in future profits owing to a dissipation of the MNE's underlying firm-specific advantages that are tied to the technology in question. The potential benefits of making technology available for appropriation are less obvious; one example is that it might facilitate the acquisition of even more valuable (to the MNE) technology from the host country. A related benefit is that it might promote the realisation of economies of scale and scope in the innovation activity by indirectly contracting out research work to a larger organisation. For example, the MNE might join a technology consortium with a number of host country firms in which basic and applied research results are exchanged among firms within an integrated research and development (R&D) setting. 5 The paper by Cantwell, lammarino and Noonan in this volume discusses the choice of location for innovation activity in more detail. A second potential benefit is that it may encourage host country governments to provide certain commercial advantages for the multinational enterprises that potentially are worth more than the commercial value of the appropriable technology. When host country governments screen foreign investments for the net benefits they will make to the host economy, direct and indirect technology transfers to host country firms are usually seen as an important source of net benefits. In other cases, government contracts in technologyintensive activities such as communications and information technology require bidding firms to perform a share of value-added activities in the host country. Successful bidders must also usually perform some underlying research and development in the host country. The technological and related expertise that is made available to host country factors of production, such as host country scientists and engineers working for the MNE, is (effectively) potentially appropriable technology. More generally, by directly or indirectly putting proprietary technology at risk for appropriation, the MNE reduces, or avoids, the costs that
38
Inward investment, technological change and growth
would be required to reduce, or eliminate, the risk of appropriation. Equivalently, accepting a certain risk of appropriation may facilitate increased efficiency within the MNE's global network. For example, codifying technology in the form of operating and technical manuals, handbooks and the like should allow easier and quicker transfers of technology from one MNE affiliate to another.6 Alternatively the MNE might rely upon the temporary expatriation of parent company personnel with the requisite expertise to transfer technology to host country affiliates. The latter approach is likely to involve smaller risks of rival host country firms gaining access to the technology in question; however, it is also likely to be more expensive and time consuming than the former approach. In short, the MNE ordinarily faces a set of benefits and costs associated with its decision about whether and how much technology it should put at risk of appropriation. The greater the expected benefits relative to the expected costs, the greater the expected value of the technology the MNE would willingly put at risk. In this context, an understanding of the factors determining the value of FDI spillovers should focus, in part, on the factors conditioning the expected benefits and costs of putting technology at risk of appropriation. In broad terms, the relevant costs will be a positive function of the underlying commercial value of the technology in question, as well as the extent to which host country firms are capable of exploiting this underlying value. The benefits will be a function of the degree to which host governments protect intellectual property through policies such as patent and trade secrets legislation.7 Benefits should also be a positive function of the value of the technology (or other competitive advantage) that is more readily or cheaply accessible to the MNE as a consequence of putting its own technology at risk of appropriation. As noted above, benefits include host country technologies that are directly or indirectly made available to the MNE affiliate that makes its technology available to others, or host government subsidies or purchases that are predicated on the MNE making certain technology (or technological occupations) available to host country nationals. Determinants of demand The demand for the technology put at risk by the MNE should reflect optimisation decisions of host country firms. Commercial exploitation of the technology will impose expected costs on host country firms. For
The determinants of host country spillovers
39
example, it may oblige a host country firm to undertake certain technological activities (and incur associated costs and risks) in order to increase the complementarities between its technological competence and that required to benefit from the MNE's technology. In other cases, it may oblige the host country firm to undertake reverse engineering efforts, hire personnel away from foreign affiliates, pay licensing and management fees to the MNE, or 'patent around' intellectual property restrictions to using the MNE's technology. The expected benefits, in return, are presumably related to the lower costs and/or increased revenues that the host country firm anticipates as a consequence of internalizing foreign-developed technology. The larger the expected benefits relative to the expected costs, the greater should be the demand for internalising foreign-developed technology. Summary Table 2.1 summarises the main proximate determinants of the equilibrium value of technological spillovers. Several of these determinants are, in turn, related to host country characteristics discussed in the literature. For example, technology that is potentially available to the MNE in direct or indirect exchange for putting its own technology at risk of appropriation should be positively related to the technological competence of host country firms. At the same time, the costs associated with absorbing foreign-developed technology should be negatively related to the technological competence of host country firms. Another factor discussed in the literature is the degree of competition in host country markets. The effect of greater competition on the demand for foreign technology is ambiguous. On the one hand, host country firms have less scope to ignore appropriable foreign technology in highly competitive markets since the resulting loss in sales and profits is likely to be substantial. Put differently, managers who indulge in 'X-inefficiency' by ignoring foreign technology can expect to suffer potentially severe (to their organisations) adverse economic consequences in highly competitive markets. To this extent, increased competition should increase demand for appropriable foreign technology. On the other hand, there may be smaller ex ante profits associated with absorbing foreign technology in competitive markets compared with markets where firms enjoy moderate degrees of market power. All other things being equal, this situation would discourage demand for foreign technology.
40
Inward investment, technological change and growth
Table 2.1. The proximate determinants of spillovers Supply (-) Value of underlying technology (-) Intellectual property protection (+) Technology available in exchange (?) Competition in host markets ( + ) Other commercial benefits
Demand (+) Value of underlying technology (-) Costs of absorbing technology (?) Competition in host markets
Host country market structure can also potentially affect the supply of foreign technology. For example, Wang and Blomstrom (1992) discuss a model in which stronger competition in host country markets obliges MNEs to transfer more technology to their host country affiliates to enable the latter to compete better against host country firms. However, if the potential costs of having technology appropriated by host country firms are sufficiently large, MNEs may simply choose to abandon the relevant host country market. Hence, if competition obliges MNEs to accept increased commercial risks associated with the loss of firm-specific advantages, there may, at some point, be reduced inward FDI and, by extension, a reduced supply of appropriable technology. Other host country attributes might influence the commercial benefits available to MNEs in compensation for putting firm-specific technology at risk of appropriation. In particular, the size and average real income level of the host country have been found to attract inward FDI, presumably because they are positively related to the demand for products embodying relatively high levels of intangible capital such as brand names.8 All other things being equal, this greater demand should make the host market a more profitable place to do business. The presence of technological centres of excellence in host countries creates the potential for the MNE to benefit from reverse technology flows by participating in technological activities in those centres. Along the same line, other strong locational advantages, such as the presence of relatively cheap factors of production that are intensively used by the MNE, should also increase the profitability of specific locations to foreign investors.9 The potential influence of government policy on the benefits to investing and doing business in a specific host country is potentially important but also complex. Most relevant, perhaps, are host govern-
The determinants of host country spillovers
41
ment restrictions on the extent and nature of foreign ownership. The latter are largely sector-specific; however, they tend to cover a fairly wide range of sectors in many countries. On the one hand, such restrictions should reduce inflows of inward FDI and, therefore, the technology that would accompany those inflows. On the other hand, foreign ownership restrictions might encourage a substitution, at the margin, of other forms of investment, such as joint ventures. If strategic alliances, such as joint ventures, facilitate increased appropriation and usage of foreign technology, their substitution for wholly owned investments might more than offset the income effect of a reduced inflow of inward investment and/or a reduced transfer of technology from the parent affiliate to its host country affiliate.10 Finally, the technology available to the MNE in the host country should also be a positive function of the complementarity between the MNE's technological competence and the technological competence of host country firms. This variable is conceptually distinct from the overall technological competence of host country firms. Obviously the value of appropriable MNE technology to host country firms will also be a positive function of the degree of complementarity. These various hypotheses are summarised in Table 2.2. Specifically the table shows variables that are plausibly theoretically related to the magnitude of spillovers from FDI. In the next section, we review available evidence on the impact of a number of these variables.
EMPIRICAL EVIDENCE O N SPILLOVER DETERMINANTS Attempts to identify the magnitude and nature of FDI spillovers have employed various direct and indirect approaches. The direct approach has been to relate productivity measures of host country firms or industries to, among other things, the extent of foreign ownership in the host country. Indirect approaches examine different aspects of the interaction between MNEs and host country residents that are plausibly related to FDI spillovers. These include: • • • •
technology licences, vertical linkages, copying of technology introduced by foreign investors, the impact of FDI on host country market structure, especially
42
Inward investment, technological change and growth
Table 2.2. Variables related to FDI spillovers Variable 1. 2 3. 4. 5. 6. 7.
• •
Technological complementarities between MNE and host country firms Strength of intellectual property protection in host country Competition in host country markets Size and wealth of host country Technical centres of excellence in host country Technical competence of host country firms Government policies
Sign + ? + + + ?
competitiveness, labour training, the performance of research and development ( R & D ) by M N E affiliates in the host country.
Direct productivity measures The relevant studies typically focus on a partial productivity variable, such as value-added per employee, and relate differences across industries to differences in foreign presence, holding other factors constant. For developed countries, the limited evidence is fairly consistent in showing t h a t the productivity of domestically owned firms is positively related to foreign presence. For example, Caves (1974) and Globerman (1979) identify a positive relationship for Australia and Canada, respectively, using indices of labour productivity. Nadiri (1991) concludes that increases in the capital stock owned by US M N E s appear to have a positive impact on the g r o w t h of total factor productivity in the manufacturing sectors of France, Germany, the UK and Japan. Barrell and Pain (1999) provide evidence that inward FDI has had a positive impact on the aggregate level of labour-augmenting technical progress in four European economies. Several papers in this volume provide related evidence for the UK. The available studies do not, however, provide substantial insight into the determinants of the productivity spillovers identified. A notable exception is the study by Imbriani and Reganati (1997) focusing on the effects of FDI on the technical efficiency of Italian firms. The authors conclude that productivity levels are higher in domestic manufacturing
The determinants of host country spillovers
43
sectors where MNEs account for larger shares. Moreover efficiency spillovers are higher when the size of the initial technology gap between domestic and foreign firms is small. There are a greater number of studies estimating direct productivity spillovers for developing countries than for developed countries. The former tend to produce more mixed results than the latter. In particular, a number of studies for developing countries document that a foreign presence promotes higher productivity in host country sectors, while other studies point to limited or no significant efficiency spillovers.11 The diverse experiences of developing countries in gaining FDI efficiency spillovers provide potential insights into the determinants of those spillovers. Perhaps the most consistent finding is that the capability of host country firms to absorb foreign technology is an important conditioner of the magnitude of any realised spillovers. For example, Blomstrom (1986) finds that foreign presence lowers the average dispersion of a sector's productivity, but that this effect is more significant in sectors with simpler technology. One interpretation of this is that the presence of foreign firms forces local firms to become more productive in sectors where best practice technology lies within their grasp. In fact, the ability of host country firms to absorb best practice technology will depend largely upon the firm-specific attributes required to absorb the technology relative to the attributes possessed by home country firms. Where there is a high degree of congruence between the two, spillovers should be more completely captured, all other things constant. That is, in terms of the terminology in Table 2.2, the empirical identification of spillovers should be easier when there is greater complementarity between the technological attributes of the host country firms and MNE investors. Blomstrom's results can be interpreted along these lines. More direct evidence bearing upon this hypothesis is provided by Kokko (1994) and Kokko et al. (1996) who find, for Mexico and Uruguay respectively, that spillovers are difficult to identify in industries where foreign affiliates have much higher productivity levels than local firms. Conversely spillovers can be identified when foreign investors are not a self-contained enclave of firms. Since technology gaps are more likely to be small when MNEs invest in developed countries than in developing countries, this consideration might help explain the more consistent identification of FDI spillovers in developed host economies.
44
Inward investment, technological change and growth
Licensing Host country firms gain the equivalent of spillover efficiency benefits when they are able to purchase or license technology from foreign firms at a cost that is less than the value of the foreign technology to them. By doing so, they presumably gain improvements in efficiency (directly or indirectly) above and beyond the costs associated with acquiring the necessary tangible or intangible assets.12 In this context, the relevant issue is whether any factors have been identified in the literature that are systematically related to the surpluses realised by host country firms on technology licensed from foreign investors. In fact, foreign investors appear to regard foreign investment and licensing as direct alternatives (Caves, 1996). In this regard, MNEs are seen as resorting to licensing when there are restrictions on direct investment, either from government policies or from some other constraint. This is because there is usually a greater perceived risk of losing proprietary technology to host country firms when technology is brought to the host country via the licensing option. As a result, it is unsurprising to observe that technology tends to be introduced more quickly into host countries when MNEs have the option of introducing the technology through their affiliates rather than through joint ventures or arm's-length licensing agreements.13 Given that 'forced' licensing might be a second-best way to promote spillovers, the relevant issue is whether technology leakages to host country firms under licensing arrangements are at all conditioned by the factors identified in Table 2.2 or other factors. Unfortunately we have been unable to identify any studies that focus on this issue.
Vertical linkages Closer linkages between MNEs and domestically owned firms are presumed to enhance spillovers in the host country. For example, it is (often implicitly) presumed that closer commercial ties between MNE affiliates and upstream suppliers and downstream customers lead to a greater (uncompensated) transfer of technical and commercial information to suppliers and customers. They may also lead to the defection of key personnel from the MNE affiliate to domestically owned firms. To the extent that closer vertical linkages are associated with larger spillovers to host country firms, the empirical issue is to identify the factors promoting or discouraging closer vertical linkages. With respect to
The determinants of host country spillovers
45
the first point, the available evidence, on balance, points to the existence of efficiency benefits associated with closer vertical linkages between MNE affiliates and host country firms. In particular, there is evidence that MNEs place pressures on upstream suppliers to meet higher standards of quality, reliability and speed of delivery (see, for example, Brash, 1966, for Australia; Katz, 1969, for Argentina; and Watanabe, 1983, for the Philippines). It is often unclear whether the benefits have a significant spillover component or whether host country firms must fully compensate foreign-owned affiliates for the tangible and intangible resource transfers involved. Nevertheless, it seems unlikely that MNEs will ordinarily be able to fully appropriate all of the value of the explicit and implicit transfers with their upstream and downstream host country business partners. There is some evidence bearing upon the factors that promote vertical linkages with foreign-owned firms. Most of the available evidence is from the experience of developing countries. One robust finding is that the technical capability of potential local suppliers positively influences the extent of backward linkages. A second is that linkages are more pronounced, the larger the size of the host country market. A less robust finding is that local content requirements can promote increased local purchasing by MNE affiliates. This is because the potential for such requirements to discourage inward FDI is usually not factored into the analysis (Blomstrom and Kokko, 1998). A study of vertical linkages in the Canadian aircraft industries sheds some light on the relevance of vertical linkages to productivity spillovers, as well as the determinants of the nature of those linkages (De Bresson etal., 1991). The authors conclude that foreign ownership has only an indirect impact on the build-up of technological capability by domestically owned suppliers. Specifically the opportunity to design and manufacture whole systems in the commercial field over a long period of time was found to be the key to building up domestic technological capability, and this opportunity was not automatically associated with inward investment. However the opportunity to design and manufacture whole commercial systems could be enhanced by the ability of domestically owned firms to engage in two-way technology transfer at the early stage of the technology's life cycle.
Demonstration effects A related issue is the diffusion of technology following an early and sue-
46
Inward investment, technological change and growth
cessful introduction by MNE affiliates. The successful introduction of new production techniques and new products reduces the subjective risk surrounding the adoption of the innovation and should, therefore, promote its adoption more widely throughout the population of potential adopters in the host country. Successful demonstration effects would be reinforced by the increased competition supplied by the entry of successful and innovative foreign-owned firms within host country markets. In turn, increased competition should stimulate a faster rate of adoption of new technology.14 Two related empirical issues that are relevant in this context are: • •
do foreign-owned firms adopt innovations sooner than domestically owned firms? what factors condition any adoption lags between foreign and domestically owned firms?
There are numerous studies that seek to identify and assess differences in production conditions (including technology) across foreign and domestically owned firms. However very few consider the importance of the MNE in the introduction and spread of the technology in the host country when the new technology is appropriate for domestically owned firms to use. Relevant in this regard is the study by Mansfield and Romeo (1980) which obtained from domestic firms in the United Kingdom estimates of how often their innovative efforts had been hastened in response to technology transfers from US MNEs to their competing subsidiaries in the United Kingdom. The majority expressed a belief that at least some of their products and processes had been introduced (or introduced sooner) to meet the competitive effects of the transfers. Note that this finding suggests that it is increased competition rather than demonstration that promotes technology adoption by domestic firms. Several other studies suggest that the presence of foreign-owned affiliates might accelerate the introduction and adoption of new technology by host country firms. For example, a study of the Canadian paper industry found that foreign-owned affiliates were quicker to adopt a processing innovation than their Canadian-owned counterparts, although it was unclear how important demonstration effects were in encouraging the latter also to adopt the innovation.15 A positive association between the speed of technological diffusion and the share of foreign ownership was shown in four Hong Kong industries by Chen (1983).
The determinants of host country spillovers
47
The results of studies that compare new technology adoption by foreign-owned and domestically owned firms therefore tend to conclude that new technology is frequently introduced sooner by foreign-owned affiliates, and that competition spurs quicker adoption of innovations by both domestically owned and foreign-owned firms. Moreover MNEs are quicker to introduce new technology into host markets characterised by relatively high per capita incomes and high literacy rates (McFetridge, 1987). The latter can be taken as a proxy for the technical competence of the host economy, while the former might capture the attractiveness of doing business in a host market.
FDI and market structures Inward FDI could have an important indirect impact on spillover efficiency benefits if it alters host country market structures in ways that affect those benefits. The preceding section suggests one such indirect channel of influence. If inward investment contributes to host country markets being more competitive (or more contestable), it would contribute to the faster adoption of new technology by domestic firms. It might also encourage MNEs to introduce new technology sooner and more extensively into their foreign-owned affiliates in order to enable the latter to compete more effectively.16 More simply, increased competition might encourage a more efficient allocation of resources across industrial sectors and production establishments which, in turn, is manifested by increases in sectoral and economy-wide measures of productivity. There are substantial problems associated with linking inward FDI to changes in market structure in host economies, especially if one focuses on simple measures of market structure such as industrial concentration ratios. 17 Nevertheless data restrictions frequently constrain the research focus to simple summary measures of this kind. The studies undertaken for developed countries tend to conclude that MNEs are more likely than domestically-owned firms to enter concentrated industries with relatively high barriers to entry. At the same time, actual inward FDI tends to have no significant long-run relationship to industrial concentration (Dunning, 1993). This suggests that the potential for inward investment contributes to domestic markets being more contestable, but that actual long-run changes in market structure are influenced by other, more fundamental, factors such as minimum efficient scale relative to market size. While there is some evidence from developing
48
Inward investment, technological change and growth
countries that inward FDI is associated with increased seller concentration, it cannot be assumed that foreign presence leads to reduced competition in developing countries. This is because foreign-owned affiliates are often more aggressive domestic competitors than locally owned firms. The potential linkage between competition and FDI spillover benefits highlights the general equilibrium nature of the structural determinants of spillover benefits. For example, domestic market structure might itself be an important indirect determinant of vertical linkages, the technical competence of host country firms and even government policy. In principle, all these factors should be controlled for in any empirical study. Nevertheless, given the apparently weak long-run relationship between inward FDI and domestic market structures, a failure to account explicitly for these indirect structural linkages may not be a serious liability in the existing literature. Investments in human capital Technology is embodied not only in machinery, equipment, patent rights and expatriate managers and technicians, but also in the human capital of the affiliates' local employees. In turn, the latter may acquire much of their human capital through direct and indirect training received while working for foreign affiliates. In theory, employees should pay indirectly for any general human capital, that is, fungible skills acquired from employers, typically in the form of lower wages during the training period and so forth. In fact there is some evidence that, in practice, MNEs may pay efficiency wages to productive employees in their foreign affiliates in order to keep them from defecting to domestically owned competitors (Globerman etal., 1994). The training given to host country employees working for foreignowned firms could affect most levels of employees - from manufacturing operatives, through supervisors, to advanced professionals and top-level managers. The various fungible skills gained while working for foreignowned affiliates may, in turn, generate spillover benefits for the host economy. Employees can migrate to domestically owned firms or start their own businesses using the knowledge and skills gained through training to enhance their productivity in other organizations. This perspective on FDI spillovers generates two interrelated empirical questions: •
what evidence is there that a significant amount of 'spillover-gener-
The determinants of host country spillovers
•
49
ating' general human capital is accumulated in MNE affiliates for which employees do not directly or indirectly pay full value; what factors condition the accumulation of host country human capital via inward investments?
There is only scattered evidence on these two issues, the majority of which focuses on the experiences of developing host economies. The evidence indicates that MNEs offer more training to managers and other types of employees than do privately owned local firms, and that the movement of employees from foreign-owned affiliates to other firms contributes to the diffusion of know-how.18 The limited evidence that exists for developed countries suggests that manager mobility has contributed to the international diffusion of specific management practices. Moreover the mobility of employees from MNEs in the computer and software industries contributes to spillovers, both within the industry and elsewhere. Unfortunately, we have been unable to identify studies that attempt to quantify the magnitude of the host country spillovers associated with the training and subsequent mobility of MNE affiliate employees. Nor have we found studies identifying the market-related factors that condition such spillovers.19 There is clearly considerable research that could be done in this area.
Research and development Another activity indirectly related to the human capital of foreign affiliate employees is the performance of research and development (R&D) and other technology-promoting efforts by MNE affiliates. A longstanding policy concern about foreign ownership is that MNEs will centralise R&D in the parent company and leave their foreign affiliates to perform a very modest amount of R&D that focuses primarily upon modifying parent company technology for the foreign market.20 Domestically performed R&D expenditures have been found to generate significant spillover efficiency gains, both within and across industries in the R&D performing country (Bernstein, 1988, 1989). While the precise results vary from industry to industry, the available evidence suggests that the cost reduction impact of R&D spillovers is larger for inter-industry than for intra-industry spillovers. Most important, perhaps, is the finding that social rates of return to R&D capital are higher in domestic industries with relatively large R&D spending propensities (Bernstein, 1988). Hence the relatively limited performance of R&D in host countries
50
Inward investment, technological change and growth
by some MNEs may deprive host countries of the productivity spillovers associated with domestic R&D performance. Furthermore, domestic R&D performance is also positively related to the adoption of new technology, particularly in many models with endogenous growth. That is, some amount of R&D performance is apparently required to facilitate new technology adoption. A truncated R&D function in MNE affiliates has therefore also been identified as a factor slowing technology transfers from the MNE's home country to host country firms. Concerns expressed about the truncation of innovation in host countries, as a result of the centralisation of R&D in the head offices of MNEs, implicitly draw upon other premises. One is that the successful appropriation of R&D benefits requires companies to be in relatively close physical proximity to the R&D performing organisations. But if host country firms are readily able to exploit foreign technology developed abroad, any decentralisation of R&D from home to host countries might have little overall impact on the spillover efficiency benefits that host country firms derive from the MNE's R&D performance. For example, technology generated abroad might be indirectly captured by host country firms through imports, or directly captured through reverse engineering activities. There is relatively little direct evidence on the magnitude of the spillovers from host versus home country located MNE R&D activities. Fors (1996) estimates that around 80 per cent of the gain in value-added attributable to the home country R&D performed by Swedish MNEs was realised in those companies' home plants. Thus around 20 per cent of the gain was realised by foreign plants. Related evidence is obtained by Bernstein and Mohnen (1998) who identify substantial R&D spillovers from the United States to Japan, notwithstanding the relatively small amount of American foreign ownership of Japanese companies. However they are not able to identify R&D spillovers from Japan to the United States, even allowing for the relatively large amount of Japanese foreign direct investment in the United States. In short, while the evidence is cursory, it suggests that robust channels for the international transmission of R&D spillovers exist independently of the R&D activities that MNEs conduct (or do not conduct) in host countries. A second premise is that the R&D expenditures of domestically owned firms are independent of the R&D expenditures of foreignowned firms in the host economy. Hence, less R&D performed by foreign-owned affiliates implies less R&D in total performed in the host
The determinants of host country spillovers
51
country. In contrast, one might argue that the R&D performed by foreign affiliates is either a substitute for, or a complement to, the R&D performed by domestically owned firms. In this case, the impact on host country spillovers from centralising R&D in the home country might be either mitigated or magnified. This brief discussion of the relationship between the R&D performed by MNEs and FDI spillover benefits to host country economies highlights a number of complex empirical issues including: •
• •
How much difference does the actual physical location of MNE R&D performance make to the magnitude of host country spillover benefits? If location matters, what factors condition the R&D location decisions made by MNEs? What factors condition the magnitude and the nature of the spillover efficiency benefits that host countries derive from the R&D performed by foreign-owned affiliates?
There is some evidence available on all these issues, although it is far from conclusive. One finding (as noted above) seems to be that host country firms can appropriate productivity benefits from the R&D performed by foreign-owned firms whether that R&D is performed in the host country or the home country (Bernstein and Mohnen, 1998; Braconier and Sjoholm, 1998). Moreover, there are several possible channels through which the benefits of the R&D performed by foreignowned firms can be accessed by host country firms, including imports of intermediate goods produced by foreign-owned firms (Coe and Helpman, 1995). A second broad finding is that the R&D performed by foreign-owned firms affects the rates of return to R&D (and related innovation-generating activities) in domestically owned firms. The relationships appear to be quite complex, however, and seem to vary across countries and across industries within countries. Thus in some cases the R&D performed by foreign-owned firms is complementary to the R&D performed by domestically owned firms, but it is a substitute for the licensing of foreign technology (Basant and Fikkert, 1996). In other cases, the R&D performed by foreign-owned firms is a complement to the R&D performed by domestically owned firms in certain industries, while it is a substitute for the latter in other industries (Bernstein and Mohnen, 1998).
52
Inward investment, technological change and growth
A third broad finding is that the rates of return to domestically performed R&D are not uniformly higher or lower than rates of return to foreign-performed R&D. That is, there is no consistent evidence that host country firms benefit more from R&D performed domestically than from R&D performed elsewhere. This is not to say that firms performing R&D are indifferent as to where they locate their R&D activities. There is abundant evidence of locational advantages to performing certain types of R&D in specific locations, and that at least one major source of locational advantage is agglomeration economies. These indicate that it may be efficient to cluster specific R&D expertise in a given geographical location, at least up to some critical scale.21 Also the locational advantages are fairly durable over time, such that MNEs tend to keep most of their R&D centralised at home headquarters (Globerman, 1979). On balance, these findings suggest that the spillovers from MNE R&D may be available to host country firms regardless of where that R&D is undertaken, although the underlying productivity of the R&D will be sensitive to where the R&D is performed. Also the spillover benefits to host country firms from the R&D performed by foreign-owned firms are often quite significant. The key issue is then to establish which factors affect the magnitude of the spillover benefits from foreign R&D that are captured by host country firms. One factor that has been identified as conditioning the 'capture' of international knowledge is human capital. For example, Engelbrecht (1997) identifies general human capital as a vehicle for international knowledge transfer associated with productivity catch-up among OECD countries, although the technology transferred was not necessarily generated in MNEs. Openness to imports also promotes the capture of international technology transfers, in part because a substantial share of technology is embodied in intermediate inputs, and in part because openness to imports is a stimulant to domestic competition (Bayoumi et al., 1999).22 Interestingly it is openness to US exports that is an especially important determinant of international technology transfers, since the United States is thought to be a disproportionate generator of appropriable commercial technologies (Park, 1995).
Summary The preceding evidence can be summed up as follows. 1. FDI spillovers have been investigated both through statistical
The determinants of host country spillovers
53
studies directly linking host country spillovers to foreign presence, as well as through more structurally oriented studies that identify channels through which FDI spillovers might be realised and then evaluate the robustness of those channels. The evidence is convincing in showing the existence of FDI efficiency spillovers in host countries, although there is no strong consensus on the associated magnitudes. 2. Studies focusing on potential channels through which FDI spillovers are realised also tend to support the empirical relevance of those channels, although there is a fair amount of heterogeneity across the relevant studies. The implication is that FDI spillovers are likely to be generated through a variety of activities in the host economy, including labour and management training, technological copying, direct licensing of technology, vertical linkages in production and distribution value chains and so forth. No consensus, however, can be inferred about the relative importance of the different channels. 3. Certain factors appear to have a reasonably consistent and significant influence on the magnitude of the efficiency spillovers captured by host country firms, including spillovers that are generated outside the host country from sources such as the R&D activities of firms based in foreign countries. One such factor seems to be the openness of the host country to trade and investment. As noted above, such openness, particularly to imports, is a proxy for the competitiveness of domestic markets. It is also a proxy for investment in technology embodied in intermediate inputs and capital equipment. There is also evidence that direct competition between host country firms stimulates the capture of appropriable spillover benefits. Another broad factor related to the magnitude of host country spillovers is the technical capability of local firms. Specifically inward FDI seems to have more beneficial effects on host country productivity when host country firms enjoy traditional technological strengths.23 4. Obviously there will be few spillover efficiency benefits to capture if foreign-owned firms are less efficient and less technologically advanced than host country firms. Nevertheless there is some evidence that too large a gap between the capabilities of MNEs and their domestically owned counterparts may attenuate the capture of spillovers by host country factors of production. A closely related notion is that potential spillovers will be more extensively captured if foreign and host country technologies (broadly defined) are com-
54
Inward investment, technological change and growth
plementary. Unfortunately, there is less evidence on this latter point than one might have expected. 5. The degree to which other modes of international business (besides traditional inward FDI) generate appropriable spillover benefits for the host country is an exceedingly important policy issue for which there is a disappointing amount of evidence. Moreover the available evidence provides no clear insight for policymakers. Some studies suggest that spillover benefits are not monotonically related to foreign presence. For example, in some studies, spillover benefits are as significant in environments where foreign investors are primarily involved in joint ventures as when foreign investors own controlling interests in the host affiliates. Other studies show, however, that the vintages of technologies transferred through the MNE vary according to the degree of foreign ownership of the affiliate. Moreover some forms of foreign presence may be discouraged altogether if the foreign investor is denied a controlling interest. In this case, potential spillovers may be extinguished. 6. Many of the channels through which FDI spillovers can be captured exhibit interdependence. For example, training offered by MNEs might not only increase the technical competence of host country firms directly, as former MNE employees migrate to domestically owned firms, but also encourage increased domestic competition by promoting new firm start-ups by former MNE managers and technical employees. This and other manifestations of interdependence make it more difficult to synthesise the available literature, especially since most studies do not try to identify, or sort out, important interdependencies of this kind.
POLICY IMPLICATIONS A N D T H E RESEARCH AGENDA Some areas of government policy are obvious determinants of FDI spillovers. The most obvious example is the host government's policy towards inward FDI. In general terms, there are two dimensions in which FDI policy can be characterised: • •
the degree to which foreign ownership is constrained, either in specific sectors or in the economy as a whole; the degree to which business decisions of foreign investors are con-
The determinants of host country spillovers 55
strained or regulated via formal or informal requirements to carry out certain activities in the host country. At one extreme, policies that discourage inward FDI in any form will close off channels for spillover benefits from foreign technology that can be captured only if foreign investors are present in the (potential) host economy. Moreover such policies are likely to reduce the contestability of host country industries, especially those that are characterised by relatively high levels of industrial concentration. The available evidence suggests that reductions in contestability will reduce the incentives for domestically owned firms to exploit foreign technology that is available through channels other than inward investment. Hence virtually all countries can count on paying a significant price, in the form of forgone spillover efficiency benefits from inward FDI, as the quid pro quo for indulging their preferences for economic or political sovereignty.24 At the same time, it is more difficult to be unequivocal about the implications of policies designed to constrain or regulate the behaviour of foreign investors. For example, policies that require (or encourage) MNEs to transfer technology more quickly to the host country could enhance potential spillover benefits by enriching the pool of appropriable technology in the host country. On the other hand, if the relevant policies reduce the ex ante profitability of foreign investment in the host country, the overall pool of appropriable technology might actually decline as a result of substantial reductions in inward FDI. The potential trade-off between these two effects is likely to vary across host countries and even across industries within host countries. For example, in countries with high capabilities of absorbing foreign technology and attractive markets for foreign investors, appropriate regulations might bring about a net expansion of the pool of appropriable foreign technology.25 Countries with the opposite set of characteristics are more likely to suffer net decreases in the pool. Similar contradictory effects are related to host government policies that oblige MNEs to form joint ventures or other strategic alliances with domestically owned companies. Although the diffusion of technology to local users is likely to be faster when there is a technically competent local equity partner, it is also likely that the commercial value of foreign technology invested in the venture by the foreign partner will be lower. In this regard, Blomstrom and Zejan (1991) argue that the MNEs most likely to accept joint ventures are typically not those with the most valuable technological assets. On the other hand, Blomstrom
56
Inward investment, technological change and growth
and Sjoholm (1999) find no significant differences in the spillover effects from wholly-owned MNE affiliates and joint ventures in Indonesia. Trade policy has the indirect potential to influence spillovers. In particular, it has been found that an open regime towards imports can stimulate the capture of foreign technology spillovers in at least two ways. One is by facilitating the import of technology embodied in capital goods and intermediate inputs. A second is by stimulating increased competition in domestic industries, thereby encouraging foreign-owned firms to transfer technology more quickly to their host country affiliates, while encouraging domestically owned firms to capitalise on the available appropriable foreign technology. One potential source of ambiguity with respect to the claim that freer trade encourages host country spillovers is the recognition that tariff barriers can encourage inward FDI as a substitute for exporting. As a result, spillovers associated with the presence of foreign-owned affiliates might actually increase. However protectionism has a variety of other adverse consequences for host economies that might reasonably be expected to discourage FDI spillovers in the long run, including slower economic growth and the slower accumulation of technical competence and other manifestations of human capital. Technology policy is a third area that has potentially important impacts upon FDI spillover benefits. For example, government policies that encourage the performance of R&D in the host economy should enhance the technical capability of local firms. Other things being equal, this should enhance the capability of host country firms to exploit appropriable foreign technology. Yet even this seemingly obvious conclusion must be hedged. For instance, it can be argued that complementarities between the technical competencies of foreign and domestically owned firms strongly condition the magnitude of actual spillover benefits. Hence government policies might increase the technical competence of local firms in meaningful ways, yet still reduce the 'fit' between local technical competencies and those enjoyed by likely foreign investors, thereby actually discouraging net R&D capital accumulation. It would be useful to have more direct evidence that would permit this hypothesis to be evaluated. A possible example in this context is offered by Canada's technology policies. The Canadian government has tended to encourage R&D in areas such as telecommunications and commercial aircraft. These are traditional areas of technological strength in the United States, which is the single largest source of inward FDI for Canada. Much of the
The determinants of host country spillovers
57
technology developed in Canada is in competition with that in the US. Indeed an intention of the Canadian government is to protect Canadian economic sovereignty in these technically advanced industries. As a result, the technology developed by Canadian companies in the relevant industries is often a substitute for technologies developed by counterpart US companies. The latter will usually find it cheaper to conduct similar types of innovation in their home country affiliates. Another potentially important aspect of government technology policy is intellectual property protection. Industry representatives from a number of technology industries, most notably the pharmaceutical industry, argue that secure intellectual property rights in host countries are an important precondition for inflows of valuable technologies to host countries, including those associated with the performance of domestic R&D by foreign affiliates. In this regard, Mansfield (1994) has shown that the character of a country's system of intellectual property protection has a significant effect on FDI flows, as well as on the quality and quantity of technology transferred by US firms to their foreign affiliates. At the same time, there is ample research showing that formal intellectual property protection is not an important vehicle for protecting proprietary technology in most industries. In fact, there has been almost no attention paid in the literature to intellectual property regimes as a determinant of the size and scope of FDI spillovers. The existence of spillover efficiency benefits would seem to call for some sort of government policy reaction, as it seems a clear manifestation of an externality. In particular, it would seem to argue for aggressive actions by government to encourage inward FDI. To do so effectively, of course, implies the need for knowledge about cost effective levers to encourage inward FDI and to capture FDI spillovers. In fact, the existing literature offers only limited and fairly circumscribed insight into the determinants of FDI spillovers. Nevertheless the insights offered are gratifying inasmuch as they reinforce conventional verities of public policy, including the promotion of competitive markets, scientific training and education and the performance of research and development. The existence of spillover efficiency benefits would also seem to call more broadly for technology transfer policies on the part of host governments. Unfortunately, such policies are often counterproductive in that they may discourage inward FDI, and associated inflows of technology, especially when they impose additional costs on foreign investors. There is a natural temptation for host governments to attempt to extract technology transfer and other operating commitments from
58
Inward investment, technological change and growth
foreign investors with a view to those investors earning economic rent by operating in the host country. Hence additional costs imposed on foreign investors, up to some maximum, may not discourage or even modify investment plans. While economic rent is undoubtedly anticipated across a range of foreign investments, host country bureaucrats are unlikely to be able to identify their existence beforehand, let alone calibrate the limit to the costs they can impose upon the foreign investor without discouraging investments. To be sure, governments will continue to implement policies designed to lever greater spillover benefits from foreign investment. Hence setting an agenda for future research seems appropriate. Without worrying about the costs, or even the practicality of the research projects implied, it seems to us that one area for research with significant potential policy payoffs is to identify whether spillovers are generic to most types of inward FDI, or whether specific MNEs account for the bulk of host country spillovers. R&D and patent holdings tend to be concentrated among a relatively small number of leading MNEs, thereby suggesting that the source of spillovers is concentrated among a relatively small number of foreign investors. Moreover government policies are frequently aimed at attracting specific MNE investors, or types of investment. These considerations point to the relevance of a more 'micro' focused analysis of FDI spillovers. The effectiveness and efficiency of firm or sector-specific FDI policies could obviously be improved if identification of the major MNEs generating spillovers was possible. On the other hand, if identification were either impossible or irrelevant, additional arguments would be generated for focusing public policy exclusively on getting the broad economic environment right for generating and exploiting FDI spillovers. A second area for research might focus on the distributional implications of FDI spillovers. Whilst neoclassical theory broadly suggests that host country workers gain, and host country capitalists lose, from inward FDI, modern theories of the FDI process indicate that this perception is much too simple and stylised to help predict the distributional effects of liberalised capital markets. Yet it may also be too simplistic to assume that the spillover benefits from inward FDI are broadly distributed in the host country, either geographically or demographically. While adverse distributional consequences of inward FDI may not obviate the desirability of foreign investment based upon efficiency considerations, they may highlight the need for, and the nature of, policies to ensure that the benefits of open capital markets are broadly and
The determinants of host country spillovers
59
equitably distributed. Finally the research agenda should accommodate studies of the linkages between intellectual property regimes and spillover benefits. In smaller developed and developing countries, there has been a temptation on the part of governments to implement relatively weak intellectual property protection given the (valid) perception that the bulk of the property involved is owned by large, foreign-owned companies. At the same time, the foreign-owned companies argue that host countries are hurting themselves with weak intellectual property laws by discouraging transfers of technology from abroad. The emergence and growth of new high-technology areas such as biotechnology and computer communications has heightened the policy related importance of understanding the interactions between intellectual property protection and FDI spillover benefits.
NOTES 1 For reviews of the relevant literature, see Blomstrom (1991) and Blomstrom and Kokko (1998). 2 For some studies that explicitly address the issue of the determinants of FDI spillovers, see Cantwell (1991), Kokko (1994, 1996) and Sjoholm(1999). 3 Efficiency wages are payments above the value of the marginal product produced by the employee or, equivalently, a side payment for organisational loyalty. 4 It is more typical in the literature to assume that this supply is determined exogenously. 5 An example of the latter is Forintek, a Canadian research cooperative that includes the leading forest products companies in Canada, including affiliates of foreign-owned companies. 6 The presumption, of course, is that some technology can be relatively easily codified. 7 The more extensive and effective the protection offered, the lower the costs that MNEs would presumably face associated with the requirement to make proprietary technology inaccessible to host country firms through other means, for example, by minimising the codification of the technology in inter-affiliate transfers. 8 For a review of studies linking demand to inward FDI, see Dunning (1993) and Globerman and Shapiro (1999).
60
Inward investment, technological change and growth
9 For instance, Barrell and Pain (1999) find that the location of US foreign direct investment in Europe is affected in part by the relative unit labour costs of different locations. 10 There is ample evidence that MNEs will transfer more up-to-date technology to wholly-owned affiliates than to partly-owned affiliates. See, for example, the discussion of technology transfer in Caves (1996, pp. 166-74). However, there is very little evidence on how the form of foreign investment affects host country spillovers. For one such study, see Blomstrom and Sjoholm (1999). 11 Blomstrom and Persson (1983), Blomstrom (1986) and Kokko (1994) find evidence of FDI spillovers for Mexico. Blomstrom etal. (1994) find evidence for Uruguay, and Sjoholm (1999) identifies spillovers for Indonesia. On the other hand, Haddad and Harrison (1993) and Aitken and Harrison (1999) find limited or no evidence for Morocco and Venezuela, respectively. 12 Evidence documenting the capturing of economic surplus by licensees is discussed in Caves (1996). 13 Evidence on this phenomenon is provided in Mansfield and Romeo (1980) and McFetridge (1987). An exception to these findings is Blomstrom and Sjoholm's (1999) study of Indonesian establishments. The latter find that spillovers from the presence of foreign minority-owned establishments on domestic productivity are no different from those related to the presence of wholly foreign-owned establishments. They speculate that this result may be due to compensating benefits that foreign investors receive, such as government subsidies, for giving up some control over proprietary technology. 14 We shall discuss the potential linkages between FDI and market structure below. 15 See Globerman (1976). However, in a study of the Canadian tool and die industry, Globerman (1975) found that foreign-owned companies were no quicker than their domestically owned counterparts to adopt a processing innovation. Increased competition in domestic markets encouraged both foreign and domestically owned tool and die firms to adopt the innovation sooner. 16 Conversely, as mentioned earlier, vigorous competition might truncate the incentive of MNEs to introduce new technology because the opportunities to earn 'economic rent' are attenuated by competition. 17 For an extensive discussion of these problems, see Dunning (1993, Chapter 15).
The determinants of host country spillovers
61
18 A review of a number of relevant studies can be found in Blomstrom etal. (2000). 19 Zucker et al. (1998) identify investment-related ties between universities and biotechnology firms as critical to the generation of localised commercial spillovers. 20 A review of this policy issue is provided in Globerman (1985). 21 For evidence on these points, see Cantwell (1992), Jaffee et al. (1993), Zucker etal. (1998), Barrell and Pain (1999) and Chapter 8 by Cantwell, lammarino and Noonan in this volume. 22 Note that not all studies conclude that openness to imports is a strong stimulant to the capture of international technology transfers. For example, Sjoholm (1999) concludes that it is primarily domestic competition rather than competition from imports that promotes spillovers from FDI. 23 Cantwell (1989) argues that the technological capacity of indigenous firms was the major factor determining the success of the European corporate response to inward FDI by US MNEs. Other local firms were forced out of business, especially in countries with small domestic markets. The latter presumably did not provide the same 'niche' opportunities as larger markets for the less technologically sophisticated firms to escape from competing with foreign-owned affiliates. 24 In many cases, policies restricting inward FDI will reflect nothing more than the successful political lobbying of local owners of capital who want protection against competition from foreign investors. 25 Until recently, Japan would appear to offer the quintessential example of the successful exploitation of regulation and restrictions to encourage technology transfer with the minimum associated amount of inward FDI.
REFERENCES Aitken, B.J. and Harrison, A.E. (1999), 'Do domestic firms benefit from direct foreign investment? Evidence from Venezuela', American Economic Review, 89, 3, pp. 605-18. Barrell, R.J. and Pain, N. (1999), 'Domestic institutions, agglomerations and foreign direct investment in Europe', European Economic Review, 43, pp. 925-34. Basant, R. and Fikkert, B. (1996), 'The effects of R&D, foreign tech-
62
Inward investment, technological change and growth
nology purchase and domestic and international spillovers on productivity in Indian firms', The Review of Economics and Statistics, Vol.LVXII, pp. 187-98. Bayoumi, T., Coe, D.T. and Helpman, E. (1999), 'R&D spillovers and global growth', Journal of International Economics, 47, pp. 3 9 9 428. Bernstein, J.I. (1988), 'Cost of production, intra- and interindustry R&D spillovers: Canadian evidence', Canadian)rout-nal of Economics, 2 1 , pp. 324-47. (1989), 'The structure of Canadian interindustry R&D spillovers, and the rates of return to R&D', Journal of Industrial Economics, 37, pp. 315-28. Bernstein, J.I. and Mohnen, P. (1998), 'International R&D spillovers between US and Japanese R&D intensive sectors', Journal of International Economics, 44, pp. 315-38. Blomstrom, M. (1986), 'Foreign investment and productive efficiency: the case of Mexico', Journal of Industrial Economics, 15, pp. 97110. (1991), 'Host country benefits of foreign investment', in McFetridge, D.G. (ed.), Foreign Investment, Technology and Economic Growth, Toronto and London, Toronto University Press. Blomstrom, M. and Kokko, A. (1998), 'Multinational corporations and spillovers', Journal of Economic Surveys, 12, pp. 247-8. Blomstrom, M. and Persson, H. (1983), 'Foreign investment and spillover efficiency in an underdeveloped economy: evidence from the Mexican manufacturing industry', World Development, 11, pp.493-501. Blomstrom, M. and Sjoholm, F. (1999), 'Technology transfer and spillovers: does local participation with multinationals matter?', European Economic Review, 43, pp. 915-23. Blomstrom, M. and Zejan, M. (1991), 'Why do multinational firms seek out joint ventures?', Journal of International Development, 3, pp. 53-63. Blomstrom, M., Kokko, A. and Zejan, M. (1994), 'Host country competition and technology transfer by multinationals', Weltwirtschaftliches Archiv, 130, pp. 521-33. (2000), Foreign Direct Investment: Firm and Host Country Strategies, London, Macmillan. Braconier, H. and Sjoholm, F. (1998), 'National and international spillovers from R&D: comparing a neoclassical and an endogenous
The determinants of host country spillovers
63
growth approach', Weltwirtschaftliches Archiv, 134, pp. 638-65. Brash, D.T. (1966), American Investment in Australian Industry, Cambridge, Mass., Harvard University Press. Cantwell, J. (1989), Technological Innovation and Multinational Corporations, Oxford, Basil Blackwell. (1991), 'A theory of technological competence and its application to international production', in McFetridge, D. (ed.), Investment, Technology and Economic Growth, Calgary, The University of Calgary Press, pp. 93-109. (1992), 'Innovation and technological competitiveness', in Buckley, P.J. and Casson, M. (eds.), Multinational Enterprises in the World Economy: Essays in Honour of John Dunning, Aldershot, Edward Elgar Publishing, pp. 20-40. Caves, R.E. (1974), 'Multinational firms, competition and productivity in host-country markets', Economica, 41, pp. 176-93. (1996), Multinational Enterprise and Economic Analysis, 2nd edn, Cambridge, Cambridge University Press. Chen, E.K.Y. (1983), Multinational Corporations, Technology and Employment, London, Macmillan. Coe, D.T. and Helpman, E. (1995), 'International R&D spillovers', European Economic Review, 39, pp. 859-86. De Bresson, J. N., Dalpe, R. and Winer, D. (1991), 'Technological linkages and foreign ownership in the Canadian aircraft industry', in McFetridge, D. (ed.), Foreign Investment, Technology and Economic Growth, Calgary, The University of Calgary Press, pp. 317-58. Dunning, J. (1993), Multinational Enterprises and the Global Economy, Reading, Addison-Wesley Publishing Co. Engelbrecht, H.J. (1997), 'International R&D spillovers, human capital and productivity in OECD economies: an empirical investigation', European Economic Review, 4 1 , pp. 1479-88. Fors, G. (1996), R&D and Technology Transfer By Multinational Enterprises, Stockholm, Almqvist and Wiksell International. Globerman, S. (1975), 'Technological diffusion in the Canadian tool and die industry', Review of Economics and Statistics, 57, pp. 428-34. (1976), 'New technology adoption in the Canadian paper industry', Industrial Organization Review, 4, pp. 5-12. (1979), 'Foreign direct investment and spillover efficiency benefits in Canadian manufacturing industries', Canadian Journal of Economics, 12, pp. 42-56.
64
Inward investment, technological change and growth
(1985), 'Canada', in Dunning, J.H. (ed.), Multinational Enterprises, Economic Structure and International Competitiveness, Chichester, John Wiley and Sons. Globerman, S. and Shapiro, D.M. (1999), 'The impacts of government policies on foreign direct investment: the Canadian experience', Journal of International Business Studies, 30, pp. 513-32. Globerman, S., Ries, J.C. and Vertinsky, I. (1994), 'The economic performance of foreign affiliates in Canada', Canadian Journal of Economics, 27, pp. 143-56. Haddad, M. and Harrison, A. (1993), 'Are there positive spillovers from direct foreign investment? Evidence from panel data for Morocco', Journal of Development Economics, 42, pp. 51-74. Imbriani, C. and Reganati, F. (1997), 'International efficiency spillovers into the Italian manufacturing sector - English summary', Economia Internazionale, 50, pp. 583-95. Jaffee, A.B., Trajtenberg and Henderson, R. (1993), 'Geographic localization of knowledge spillovers as evidenced by patent citations', Quarterly Journal of Economics, August, pp. 577-98 Katz, J.M. (1969), Production Functions, Foreign Investment and Growth, Amsterdam, North Holland. Kokko, A. (1994), 'Technology, market characteristics, and spillovers', Journal of Development Economics, 43, pp. 279-93. (1996), 'Productivity spillovers from competition between local firms and foreign affiliates', Journal of International Development, 8, pp. 517-30. Kokko, A., Tansini, R. and Zejan, M. (1996), 'Local technological capability and spillovers from FDI in the Uruguayan manufacturing sector', Journal of Development Studies, 34, pp. 602-11. Mansfield, E. and Romeo, A. (1980), 'Technology transfer to overseas subsidiaries by US-based firms', Quarterly Journal of Economics, 95, pp. 737-50. Mansfield, W. (1994), 'Intellectual property protection, foreign direct investment and technology transfer', IFC Discussion Paper no. 19, International Finance Corporation. McFetridge, D.G. (1987), 'The timing, mode and terms of technology transfer: some recent findings', in Safarian, A. and Bertin, G.Y. (eds), Governments, Multinationals and International Technology Transfer, New York, St. Martin's Press, pp. 135-50. Nadiri, M.I. (1991), 'Innovations and technological spillovers', mimeo, New York University and NBER, September.
The determinants of host country spillovers
65
Park, W.G. (1995), 'International R&D spillovers and OECD economic growth', Economic Inquiry, 33, pp. 571-91. Sjoholm, F. (1999), 'Productivity growth in Indonesia: the role of regional characteristics and direct foreign investment', Economic Development and Cultural Change, 47, pp. 559-84. Wang, Y. and Blomstrom, M. (1992), 'Foreign investment and technology transfer: a simple model', European Economic Review, 36, pp. 137-55. Watanabe, S. (1983), 'Technical co-operation between large and small firms in the Filipino automobile industry', in Watanabe, S. (ed.), Technology Marketing and Industrialization: Linkages between Small and Large Enterprises, New Delhi, Macmillan. Zucker, L.G., Darby, M.R. and Armstrong, J. (1998), 'Geographically localised knowledge: spillovers or markets', Economic Inquiry, 36, pp. 65-86.
3
Inward investment and technical progress in the UK Florence Hubert and Nigel Pain1
INTRODUCTION Theories of the multinational firm (Dunning, 1988a) and econometric evidence on the determinants of foreign direct investment (FDI) (Barrell and Pain, 1997,1999b) both highlight the extent to which the decision to establish foreign subsidiaries is influenced by the need to appropriate the rents accruing from investment in firm-specific knowledge-based assets and practices. Many new theoretical models also view the creation and exploitation of knowledge as the key factors driving the process of economic growth (Aghion and Howitt, 1998; Grossman and Helpman, 1991). Hence foreign-owned firms may help to transmit new ideas and technologies across national borders, particularly to industrialised economies and regional markets with location-specific advantages. This suggests that the location of economic activity could be an important endogenous influence on national growth prospects. In this chapter we present new evidence on the impact of inward investment on technical progress, and hence labour productivity and growth, in a number of different UK industries. We utilise two different data sets and a number of different estimation techniques in an attempt to assess the robustness of our findings. For many years the popular discussion of inward investment and the focus of investment promotion agencies have tended to centre around the gross number of jobs believed to be created or safeguarded by such investments. Whilst this can be of obvious importance in particular regions with high unemployment rates, it is not the best way of assessing the impact of inward investment on national income and living standards. In an economy such as the UK with flexible real wages and limited income support for those out of work, inward investment should affect
66
Inward investment and technical progress in the UK 67
only the types of jobs available, rather than the quantity, unless it has wider effects on the growth process. There would obviously be significant adjustment costs if all foreign firms suddenly left the economy, but ultimately employment could be expected to recover in the economy as a whole if the real cost of labour fell and new jobs were eventually created. A more appropriate question is whether inward investments raise the incomes of UK residents above the levels they would otherwise be at in the absence of any such investments. The main factor behind the growth of national income in most of the advanced western economies over the past fifty years has been technical and organisational change, rather than the accumulation of inputs into production (Crafts and Toniolo, 1996). Hence it is natural to focus on whether inward investment affects technical progress. The research in this paper builds on the approach of Barrell and Pain (1997), who use time-series models of labour demand in the UK to investigate the impact of the stock of inward foreign direct investment on technical progress in the manufacturing and non-manufacturing sectors. Their results suggest that inward investment has been a major source of technological progress in the manufacturing sector, but has not brought any wider benefits in non-manufacturing. There are a number of important extensions that can be made to this work. First, there is the question of whether the use of a single aggregate non-manufacturing sector masks the existence of significant effects from inward investment in particular industries within this sector. Second, there is a possibility that all that is being observed is 'batting average' type effects rather than genuine spillovers; foreign firms are more productive on average than domestic ones, so a rising share of them within the total population of firms will raise the average level of productivity. This requires the use of a sample that omits foreign-owned firms. Finally there is also the issue of whether the impact of inward investment varies according to the nationality of the investor. All these factors are important if investment promotion schemes are to be designed effectively. We undertake two different studies. The first comprises an investigation of the influence of the sector-specific stock of inward direct investment on technical progress in five broad economic sectors - manufacturing, public services and three private service sectors over a period of 25 years. Our results suggest that there is considerable heterogeneity in the structure of production across the various sectors, illustrating the difficulties in drawing conclusions from single studies of the source
68
Inward investment, technological change and growth
of economy-wide technical progress. However, using appropriate econometric techniques, we find evidence of a significant impact from inward investment. It does not indicate that inward investment is the only source of technical change in the UK economy, but it does indicate that it is an important and significant one, accounting for around one-third of output growth in manufacturing and the three private service sectors from 1972 to 1996. For a single sector, such as manufacturing, an aggregate time-series analysis allows indirect across-industry spillovers to be captured as well as direct spillovers within the industry in which the inward investor is located. However a potential disadvantage of an aggregate study is that the econometric estimates may be picking up the overall rise in the productivity level from the increasing numbers of foreign firms with above-average productivity levels. We explore these issues with a second panel data set, drawn from the annual Census of Production. The Census identifies foreign-owned firms within manufacturing industries, and thus permits a direct test of whether the presence and scale of foreign-owned firms affects the performance of domestically owned firms both within and across the manufacturing sector. Results from a sample of two-digit industries from 1984-92 indicate that there appear to be important intra-industry and inter-industry spillovers from the total scale of output of foreign firms in the manufacturing sector. This is consistent with the transfer of innovative business techniques and management practices, rather than just the spillover of new processes and products that are specific to a particular industry. One important implication of these two different studies is that use of the stock of FDI does provide a reasonable proxy for the impact of inward investment, even though it is technically just one means of financing the activities of inward investors. This matters, as in many cases information on foreign direct investment is all that is readily available to researchers. A subsidiary aim of this chapter is to draw attention to some of the problems associated with the approaches frequently used to model the determinants of either total factor productivity or labour productivity. In particular, many studies employ models which are derived from an underlying assumption that the production process can be described by a Cobb-Douglas function. This has strong implications for the extent of substitution between different factors of production, and assumes that the (long-run) share of each factor of production in output is constant. If these assumptions are not valid, then the associated econometric work
Inward investment and technical progress in the UK 69
may yield misleading, and potentially biased, results. The plan of the remainder of the chapter is as follows. In the following section, we set out a modelling framework for relating inward investment to technical progress and compare the specifications that arise from the use of alternative forms of production function. The third section gives a brief description of the data used in the empirical analysis. The first set of econometric results using a long time-series data set for manufacturing and services is reported in the fourth section, and the second set of results using industry-level data on domestically owned firms is reported in the fifth section. Some concluding comments are given in the final section.
M O D E L L I N G T E C H N I C A L PROGRESS A N D SPILLOVERS Snapshots based on detailed company sector survey data provide a picture of the comparative advantages of foreign firms at particular points in time. A key question left unanswered in any cross-sectional analysis is whether the build-up of foreign investment over time consistently raises the level of labour productivity across the economy as a whole. If foreign firms help to close both idea and object gaps (Romer, 1993), then technical progress is endogenous, providing an explanation and a motive for the efforts made to attract new inward investment. Qualitative evidence also leaves open the wider question of whether the identified effects are in fact quantitatively significant. One common way of investigating the impact of absorbed technology on growth is to look at the determinants of total factor productivity (TFP), using an assumed Cobb-Douglas production function: Q = ALaK^ with a + (3 = 1
(3.1)
where L, K and A represent labour, capital and an indicator of technology respectively. Equation (3.1) can be rearranged to 'back-out' an estimate of TFP growth: Aln(A) = Aln(Q)-aAln(L)-(l-a)Aln(K)
(3.2a)
with a given by the share of labour in national income. The constructed measure of TFP, which will reflect factors such as learning-by-doing and
70
Inward investment, technological change and growth
organisational change as well as technological advances, is then regressed on a number of factors which are thought to determine it. Coe and Helpman (1995) provide one recent example of this approach, with TFP related both to domestic R&D and foreign R&D embodied in trade. There are a number of difficulties with this method. One general problem is that the Cobb-Douglas function imposes an elasticity of substitution of unity. If this is invalid, then the constructed measures of TFP will be biased. For instance with a two-factor CES production function it can be shown (Nelson, 1966) that: A ln( A) = A ln(Q) - aA ln(L) - (3A ln(K) -^-a(l-a)[(o-l)/o]fAln(K)-Aln(L)]2
(3-2b)
Rodrik (1997) provides a good illustration of the biases that can arise from the inappropriate use of Cobb-Douglas functions by comparing estimates of the sources of economic growth in East Asia. Use of a CobbDouglas function also forces technical progress to be neutral, with the possibility of factor biased technical change ruled out. An additional problem with the growth accounting approach is that the TFP calculation makes the assumption that firms are always operating efficiently. In practice firms face adjustment costs, such as hiring and firing impediments and delays in ordering investment goods. Shortterm demand fluctuations may also lead to changes in factor utilisation rates, and potentially productivity, at different points in time for reasons that have nothing to do with technological or organisational advances. A weak competitive environment might also mean that other firms are slow to adopt best practice techniques, and operate inside the production frontier. An alternative approach is to allow for endogenous technical progress within estimated dynamic factor demand equations consistent with a particular underlying production structure. In this chapter we adopt the methodology used by Barrell and Pain (1997) and examine the impact of international linkages on technical change using an estimated labour demand model consistent with an underlying CES production function of the form:
Q= y[s(Kr + (1- s)(Lextrp\ivl(>)
(3.3)
Inward investment and technical progress in the UK 71
Here v denotes returns to scale, Y and s are production function scale parameters, and the elasticity of substitution (a) is given by 1 / (1 + p). Technical progress is assumed to be labour-augmenting. We wish to investigate the extent to which technology transfers and other spillovers from foreign-owned firms affect the pace of technical change and hence economic growth. It is convenient to investigate this by estimating a labour demand relationship consistent with (3.3). The first-order condition that the marginal product of each input should equal its (mark-up adjusted) real price can be used to derive a log-linear labour demand equation. 2 This has the form: $(w/p) = 8Q/5L
= v(yyP/»(l-s)Q{UPh)(Leltr{1+v)eXt
{3A)
Here w and p denote labour costs per person hour and the price of value added (at factor cost) and (3 denotes the mark-up. Taking logs and rearranging yields the long-run labour demand equation: \n(L')=Ua{v~1hn{Q)-o\n(w/p)-(l-o)Xt +k (3.5) v where k denotes a constant. The coefficient on the real producer wage provides a direct point estimate of the elasticity of substitution, allowing the technical progress parameter(s) to be identified. Ideally it would be efficient to estimate (3.5) jointly with the equivalent capital demand equation, with appropriate cross-equation restrictions being imposed (Barrell and Pain, 1999c). However this would require good quality data on both the size of the capital stock and the user cost of capital. 3 Thus, whilst we refer to any observed degree of technical progress in (3.5) as labour-augmenting, we cannot rule out the possibility that it is in fact neutral. One additional feature of interest apparent from (3.5) is that a finding of a unit output elasticity could stem from either a unit elasticity of substitution (a = 1) or from constant returns to scale (v = 1). Three restrictions are required on (3.5) to yield a long-run relationship consistent with a Cobb-Douglas production function, a unit elasticity on output and real wages and a zero coefficient on the proxy for technical progress. Imposing these gives a constant factor share (wL/pQ = k). These can be tested jointly in any empirical exercise using a factor demand relationship of this kind.
72
Inward investment, technological change and growth
Many studies of labour demand typically capture technical progress (Xt) by either a deterministic time trend (Barrell et al., 1996) or a stochastic one (Harvey etal., 1986). The former implies that technical progress is exogenous, rising at a constant rate over time. The latter provides a means of capturing any underlying change in the rate of technical progress, but does not provide an explanation of why it has occurred. However our main interest lies in explicitly testing whether inward investment affects technical change and hence output. Endogenous technical change is typically investigated either by introducing specific variables in the production function, or by endogenising technical progress (Keller, 1989; Griliches, 1998). We follow Barrell and Pain (1997) by allowing technical progress in any given sector at time t to be dependent on the aggregate stock of foreign direct investment in that sector in real terms (FDI), plus an exogenous deterministic trend (TIME):4 Xt =
XJJMETIME+X^
]n(FDI)t_i
(3.6)
The specification used here implies that technical progress will grow at a constant rate if the inward investment stock does so. As we use annual data, we arbitrarily assume that the endogenous influences enter with a one-year lag (/ = 1). Substituting (3.6) into (3.5) gives a long-run desired labour demand (L*) relationship of the form:
HQ--
l-a(l-zz)
HQt)-c\n(Wt/Pt) +k v ( 1 - o)[X11METIME+XFDI ln(FDIt_,)]
(3.7)
The potential existence of adjustment costs arising from hiring and firing costs is allowed for by estimating a data-based dynamic equilibrium-correction model for employment in which the factor demand expression implied by the marginal productivity condition (3.7) is embedded as the long-run steady-state solution. Failure to allow for any cyclical effects would imply the strong assumption that companies always use the minimum inputs necessary to produce a given level of output. We estimate a relationship of the form: A \n(Lt) = a + px A \n(Qt) + p2A ln(wt I pt) +
p 3 ln(L,_ 1 /L;_ 1 ) + 8f
(3 8)
*
Inward investment and technical progress in the UK 73
Models of labour productivity An alternative, but related, approach often employed in the industrial economics literature is to seek to model labour productivity directly. Examples include the models employed by Caves (1974), Globerman (1979), Blomstrom (1986) and Blomstrom and Sjoholm (1999) which all suggest that the presence of inward investors can affect the productivity of local firms. Such studies typically employ a detailed cross-sectional data set and model labour productivity in domestically owned firms using a mixture of firm and industry characteristics, such as the capital-labour ratio, the quality of the labour force (typically proxied by the share of skilled labour in the workforce), capacity utilisation, indicators of the presence of foreign firms (dummy variables or their output share) and proxies for factors such as scale economies and concentration ratios. A similar type of approach can be found in the separate literature on externalities from R&D (Griliches, 1998). The studies on inward investment cited above say little about the precise form of the production function implied by the empirical analysis. However, it is straightforward to derive models of this kind from a Cobb-Douglas production function for industry i of the form: Q = ®tFDlfHmKf
(3.9)
This augments (3.1) with measures of FDI and human capital (H). The stock of inward investment is assumed to be the main driving force behind the growth of disembodied technical change. The production function can be expressed in log-linear form as: \n(Q) = ln(Ot) + X\n(FDIl) + yln(Ht) + aln(Lt) + ^\n(Kl) + El or equivalently for labour productivity: \n(Qi/Li)
= ln(®i) + 'k\n(FDIi) +
yln(Hi/Li)
+ kln(Li) + $ln(Ki/Li)
+ ei
where k = a + $ +
(3 10)
'
y-l.
We assume for simplicity that human capital is an ordinary input into the production process.5 The restriction of constant returns to scale from
74
Inward investment, technological change and growth
skills (human capital), labour and physical capital can be tested. If accepted, then the separate levels term in labour can be discarded from (3.10).6 Other industry or firm-specific characteristics can be introduced through the O, term. This formulation obviously forces technical change to be factor-neutral and imposes a unit elasticity of substitution, both of which may bias the results if they are not valid restrictions.
DATA ISSUES In this section we describe the data set we use to investigate the determinants of technical progress in five separate sectors of the UK manufacturing, transport and communications, distribution, business services and public services. Barrell and Pain (1997) did not find any evidence of significant effects from inward investment in the non-manufacturing sector as a whole. The inclusion of four separate nonmanufacturing sectors in the present chapter allows us to investigate whether this is true for all non-manufacturing sectors, or whether there is some evidence of heterogeneity. This is an important issue, given that well over half of all inward investment is now outside manufacturing. Employment and productivity Tables 3.1 and 3.2, and Figure 3.1 provide some indication of the relative scale of each of the sectors and the trends in labour productivity per employee hour. The sectoral employment data comprise employees in employment plus the numbers of self-employed whose primary occupation is in the relevant sector. This adjustment helps to limit the scale of potential inconsistencies that might otherwise arise as a result of the contracting out of activities previously performed 'in-house' to selfemployed businesses run by former employees of the firm.7 The broad picture shown in Table 3.1 is the sharp decline in manufacturing employment, both in absolute terms and as a share of total employment. In contrast, service sector employment has risen steadily over time, especially in financial and public services.8 The five sectors shown cover just under 90 per cent of the total workforce, with the rest largely employed in oil production, construction and power supply. The growth of labour productivity differs markedly across sectors. It has been fastest in the manufacturing and transport/communications sectors. In the latter this has been accompanied by broadly stable em-
Inward investment and technical progress in the UK
75
Table 3.1. Sectoral composition of employment (thousands and per cent share of whole economy)
1974 1984 1994
Manufacturing
Distribution
7,353 (30.1) 5,172 (22.1) 4,193 (17.2)
4,623 (19.0) 5,052 (21.6) 5,469 (22.5)
Financial Transport & Public services communiservices(a) cations 2,160 (8.9) 2,889 (12.4) 3,889 (16.0)
1,555 (6.4) 1,450 (6.2) 1,474 (6.1)
5,271 (21.6) 6,005 (25.7) 6,954 (28.6)
Source: Authors' calculations from Labour Market Trends and Employment Gazette Historical Supplement. Note: (a) This covers education, health, public administration plus miscellaneous market services. Table 3.2. Labour productivity per employee hour (1972 = 100)
1972 1984 1996
Manufacturing
Distribution
100.0 140.8 212.5
100.0 104.1 137.1
100.0 131.5 159.6
100.0 144.0 235.1
100.0 109.2 113.8
0.3 2.3
2.3 1.6
3.1 4.2
0.7 0.3
Growth rates per annum 1972-84 2.9 1984-96 3.5
Financial Transport & services communications
Public services
Source: Authors' calculations using gross domestic product at current factor cost (United Kingdom National Accounts 1998, Table 2.2), hours data from the National Institute Domestic Econometric Model (NiDEM) database and employment from Table 3.1.
ployment levels over time and sustained output growth, whereas total employment has fallen sharply in manufacturing with total output rising only slowly over time. As Figure 3.1 illustrates, there have been some periods, especially in the second half of the 1970s and the mid-1990s, when manufacturing productivity has been stagnant for a number of years. To the extent that this is attributable to technical change, it might suggest that more is needed than just a simple deterministic trend. The
76
Inward investment, technological change and growth
5.6
a 5.4 rt < /D bJ>
11
5.2
Transport & communications
Manufacturing
4.8
Financial services
ON
Distribution i
i
i
i
O f N ^ J - ^ O O O O C ^ ^ I - ^ O O O O C N l ^ - ^ s O
Figure 3 . 1 . Labour productivity
per employee
hour
distribution sector has seen a marked rise in the rate of labour productivity in the latter half of our sample period, whilst productivity in financial services has risen at a relatively constant rate over time. Finally there is little measured productivity growth in public services over time. The diverse patterns of productivity growth across these five sectors suggest that a significant degree of heterogeneity might be expected to emerge in any empirical study and t h a t an a p p r o p r i a t e econometric technique that is robust to such effects should be used. It also suggests that the implicit assumption of c o m m o n sectoral effects made in many aggregate cross-country studies may not be especially realistic. Foreign direct i n v e s t m e n t The composition of the stock of inward direct investment in the UK is shown in Table 3.3. Data on the stock of FDI by industry are available on a triennial basis up to 1987 and annually thereafter. However flow data are available on an annual basis, allowing the missing values for the stock of investment to be derived through interpolation. Historically inward investment had largely taken place in manufacturing and the energy sectors, the latter primarily reflecting oil-related investments. M o r e recently the stock of investment in the financial services sector
Inward investment and technical progress in the UK 77
Table 3.3. Industrial composition of inward FDI stock in the UK (£bn) Manufacturing 1974 1984 1994
4.77 15.69 39.58
Distri- Financial Transport bution'^ services 0.95 3.50 13.53
0.75 4.53 24.02
0.11 0.22 1.59
Energy
Other(b)
3.05 13.02 30.90
0.71 3.26 12.96
Source: Authors' calculations from Business Monitor MA4, various issues. Notes: (a) Includes imputed allowance for investment in hotels and restaurants in 1974 and 1984. (b) Includes direct investment in property by foreign individuals.
has risen rapidly, reflecting the large number of investments attracted by the potential agglomeration economies available in the City of London. Inward investment in the transport and communications industry has grown rapidly in recent years, especially after privatisation of the railway network, but remains a small proportion of total investment. Privatisation has also led to substantial new investments in the power generation and water supply industries.
T H E I M P A C T O F I N W A R D I N V E S T M E N T IN M A N U F A C T U R I N G A N D SERVICES We begin by using annual data from 1972 to 1996 to estimate a panel data model of labour demand in five separate sectors of the UK - manufacturing, transport and communications, distribution, business services and public services. Labour input is measured in terms of employee hours, output is measured on a value added basis at 1990 prices, while total labour compensation per employee is used for wages, thus including social security costs as well as basic wages. Sector-specific price deflators are constructed from the value and volume output estimates in the national accounts. The sector stock of inward FDI was converted to 1990 prices using sector-specific price deflators. Current dated terms in output and real wages are treated as endogenous in estimation. We initially estimated separate relationships for each of the sectors using equation (3.8) and tested the statistical adequacy of the implied long-run equilibrium solution using the tECM statistic proposed by Kremers et al. (1992). This provided asymptotic evidence in favour of
78
Inward investment, technological change and growth
cointegration and thus the existence of a meaningful long-run relationship in all the sectors.9 Inward investment was found to have a significant effect on technical progress within the manufacturing sector, in line with the results reported by Barrell and Pain (1997), and also in two service sectors distribution and financial services. However there was no evidence of any significant effect in the transport and communications sector, and there has not been any inward investment in public services. There was also found to be a significant exogenous element to technical progress in both the manufacturing sector and the transport and communications sector, but not in the other sectors. While these preliminary findings suggest that there is likely to be a considerable degree of heterogeneity present in any standard panel model including all these sectors, it may still be worthwhile pooling across the sectors since some of the long-run parameters of interest were not particularly well defined in the individual equations. It is known that conventional panel data models with fixed effects and common slope parameters and variances will yield biased and inconsistent parameter estimates if significant heterogeneity is present in the panel (Pesaran and Smith, 1995; Pesaran etal., 1999). This is particularly likely if non-stationary regressors are being used and the long-run relationship differs between panel members, since the pooled regression will introduce a non-stationary component into the error term. A more appropriate procedure when sufficient observations are available is to use the mean-group estimator, with the model being estimated separately for each panel member and the 'panel' coefficients being obtained by averaging the individual group estimates. Such an estimator provides consistent slope estimates even if significant heterogeneity is present. An additional issue is whether the public services sector should be included in the panel, as it appears quite distinct from the other sectors. If output in this sector is simply estimated on the basis of inputs, the econometric results will tell us more about the measurement of output in the national accounts than about the true underlying production relationships. We therefore report two sets of estimates, with and without public services. Tests of a fixed effects panel against the mean group model provide evidence of significant heterogeneity. Table 3.4 summarises the maximised log-likelihood for the mean group model, a panel model with slope heterogeneity but common variances, a fixed effects panel with com-
Inward investment and technical progress in the UK
79
Table 3.4. Choosing between different panel estimators Manufacturing, transport, Manufacturing, transport, financial services & financial services, public distribution services & distribution LogL Mean group Common variances Common long-run parameters Standard fixed effects panel
Test statistic
LogL
Test statistic
330.55 328.26
LR(3) = 4.58
418.06 414.90
LR(4) = 6.32
296.29
LR(12)= 63.95* 377.76
LR(15)= 74.28*
282.75
LR(21)= 91.02* 351.27
LR(27)= 127.3*
Note: An asterisk denotes a significant test statistic. m o n long-run parameters only, and finally the standard fixed effects model with c o m m o n long and short-run parameters. 1 0 It is clear that the joint imposition of c o m m o n variances and parameters is strongly rejected against the mean group estimates. This largely appears to arise from the imposition of c o m m o n parameters, since the imposition of a c o m m o n error variance for each sector cannot be rejected. These tests suggest that the mean group estimates are more likely to provide consistent parameter estimates. The m e a n g r o u p estimates are s h o w n in columns [1] and [2] of Table 3.5. There are three sets of parameters reported. The long-run parameters correspond to those in equation (3.7). The dynamic parameters correspond to p l 5 p 2 and p 3 in equation (3.8) and the implied technical progress parameters are those corresponding to equation (3.6). The long-run coefficient on the stock of inward investment at constant prices is significantly different from zero, and implies that a 1 per cent rise in the stock will raise the level of technical progress by 0.18 per cent. There is also a significant effect from the deterministic time trend, implying exogenous technical progress of 2.18 per cent per a n n u m in column [1] and 1.72 per cent per a n n u m if the public services sector is included. This is quite close to the trend rate of growth of whole economy labour productivity in the UK. The other notable features in Table 3.5 are that the coefficient on the equilibrium-correction term (P 3 in equation (3.8)) is well determined, providing evidence of a valid long-run relationship, and that the elasticity of substitution is significantly less than unity. There is strong evidence against the restrictions that would be required to return to a long-run Cobb-Douglas production function.
80
Inward investment, technological change and growth
Table 3.5. Mean group estimates with endogenous technical change Sample: 1972-1996 Dependent variable:
Mean-group estimates [1] 4 Sectors
Long-run parameters Output 1.1098 (5.0) -0.2993 (4.9) Real wage -0.0145 (2.2) Time -0.1165 (3.4) Inward FDI Dynamic parameters Equilibrium-0.5774 (3.6) correction 0.2963 (2.4) Aln(Qf.), -0.1339(1.7) Aln(w7P7), Technical progress parameters 0.0218 (2.5) A- TIMF. 0.1814(2.8) A- FDI
Fixed effects panel
[2] 5 Sectors
[3] 4 sectors
1.0892(6.2) -0.3271 (5.9) -0.0114(2.0) -0.1165(3.4)
1.4601(14.5) -0.4720 (4.2) -0.0191 (5.5) -0.0862 (2.6)
-0.5102(3.6) 0.5106(2.2) -0.1947(2.3)
-0.2161 (5.8) 0.5534 (9.6) -0.2161 (3.8)
0.0172(2.1) 0.1814(2.8)
0.0361 (9.3) 0.1633 (2.2)
Notes: ^-statistics in parentheses. Standard error of fixed-effects model 1.57 per cent. The strength of the inward investment effect from the mean group estimates partially reflects the strong effect in the manufacturing sector. However the mean group estimates for the services sectors alone still reveal a significant effect from inward investment, with the longrun FDI parameter estimated to be - 0 . 0 9 6 5 (standard error 0.040) and the technical progress parameter (XFDI) estimated to be 0.1349 (standard error 0.064). The corresponding p a r a m e t e r estimates for manufacturing are - 0 . 1 7 6 3 (standard error 0.050) and 0.321 (standard error 0.084) respectively. These results suggest t h a t there are significant benefits from disaggregating the private non-manufacturing sector into different industries. Barrell and Pain (1997) were unable to find a significant effect from inward investment in the private non-manufacturing sector as a whole. O u r results suggest that this is largely because the impact of inward investment differs considerably across the different service sectors. However it is still the case that the weaker effects found in these sectors might suggest that some caution be applied to the extent to which the benefits of inward investment are extolled. The profile of labour-augmenting technical progress implied by the
Inward investment and technical progress in the UK
5.7
r
u 5.5
Distribution
n
/
Financial services
en
bJD 5 . 3
o
81
N*
5.1
II
r^ |\ ON
Manufacturing
f
4.7
Transport & communications I
I
I
I
I
( N ^ I - V O O O O f N ^ I - ^ O O O O C N l ^ r ' ^ D
| \ | \ | \ t \ o o o o o o o o o o a N a > a N a \
Figure 3.2. Implied technical progress
functions
mean group estimates is shown in Figure 3.2 for the four private market sectors. Comparing this figure with that for labour productivity in Figure 3.1 indicates t h a t a close relationship is obtained for b o t h manufacturing and financial services. For distribution, the technical progress function does point to faster improvements in technical progress in the second half of the sample period, but cannot account fully for the observed pick-up in labour productivity. The opposite is true for transport and communications. Technical progress is estimated to have risen rapidly in the latter half of the sample period, but much more slowly in the 1970s. In contrast, the labour productivity estimates show a fairly constant rate of growth. This may indicate that other factors are also driving technical efficiency, although it should also be remembered that labour productivity and labour-augmenting technical progress are not the same thing with a CES production function. The parameter estimates from the standard fixed effects panel model are reported in column [3] of Table 3.5. The effect of enforcing slope homogeneity in the standard panel estimates is much as expected given the analysis of Pesaran et al. (1999), with the coefficient on the lagged dependent variable (the equilibrium-correction parameter) pushed towards zero. There are also sizeable differences in the long-run output
82
Inward investment, technological change and growth
and real wage elasticities, with the latter (the elasticity of substitution) now close to one half. However it is notable that the inward investment term remains significant in the fixed effects model, with the implied impact on technical progress little different from that obtained from the mean group estimates.
Inward investment and output growth The regressions can be used to estimate the contribution of inward investment to the growth of output at constant prices by differentiating the implied production function with respect to inward investment, following the approach of Barrell and Pain (1997). The implied production function for each sector can be written as:
Q = y s(K)-?+(l-s)(Le
.(XTIMET+kFDI\n(FDI))\
p
-d/p)
(3.11)
The derivative of (3.11) with respect to inward investment is given by: Sln(Q) _ 81n(FD7)
fD
' 5L
Q
= XFDI WL PQ
(3.12)
Thus the impact of a given level of growth in the stock of inward FDI on the growth of output is given by the estimated technical progress coefficient Xpj^j multiplied by the labour share. 11 We use the coefficient from the combined panel estimate (XFDj = 0.1814) and weight together the results for the impact on output in manufacturing and the three private service sectors using fixed weights based on their relative sizes at the start of the period considered. Together these four sectors account for around two-thirds of the whole economy. The results are summarised in Table 3.6. The results indicate that the growth of inward FDI accounts for about one-third of average annual output growth in these sectors of the UK economy, whether based on the full sample period of 1972-96 or on the sub-sample of 1984-96. This should probably be regarded as an estimate of the upper limit of the contribution of inward investment to output growth, given the number of other potential determinants of growth which have not been included in our analysis.
Inward investment and technical progress in the UK
83
Table 3.6. The contribution of inward investment to UK output growth Output growth (% pa, 1990 prices) 1972-96 1984-96
Contribution of growth in FDI (% pa)
2.06 3.04
0.73 1.22
Note: 1984-96 using 1972 weights - FDI contribution 1.12 per cent per annum.
CENSUS OF PRODUCTION DATA FOR MANUFACTURING Any attempt to quantify the impact of inward investment needs to allow for the possibility of indirect effects across sectors, as well as direct effects within the sectors where investment occurs. One advantage of the aggregate approach used thus far is that it captures inter-industry spillovers as well as intra-industry effects in the industries in which the inward investment takes place. However it raises a number of important issues which can only be addressed with a more disaggregated data set. In particular, it is important to investigate whether the reported sector-level effects from inward investment represent genuine spillovers into domestically owned companies. It is possible that they simply reflect the fact that on average foreign-owned firms have a higher labour productivity than domestic ones and are disproportionately represented in higher-productivity industries (Solomon and Ingham, 1977; Davies and Lyons, 1991). This can be investigated by utilising a data set which allows separate factor demand functions to be estimated for domestic firms by industry. This permits a more detailed investigation into the channels through which foreign firms affect the productivity of domestic firms. It is also necessary to consider alternative measures of the scale of inward investment. Foreign direct investment is a potentially imperfect measure of the scale of the operations of foreign-owned firms, as it is a financial flow and excludes domestic assets purchased using finance raised from UK sources. In some cases it is also possible to use more appropriate indicators, such as the total production or the total fixed capital investments of foreign-owned firms in the UK. These issues can be investigated using data on the activities of foreign-owned firms in manufacturing industries from the Census of
84
Inward investment, technological change and growth
Production. Data on the output, employment, wage bill and fixed capital investments of foreign-owned firms are available in current prices from the annual Census of Production and can be subtracted from data for the total population of firms to obtain data for UK-owned firms alone. Figures were published biannually from 1971 to 1983, but have been issued on an annual basis since 1984. 12 One immediate problem with these data is that three different industrial classifications [SIC(68), SIQ80) and SIC(92)] have been used over the past twenty-five years, making it difficult to obtain a long span of data on a consistent basis. We converted SIC(68) data for 1971-9 into an SIQ80) basis, using information from CSO (1980). The resulting data set for 1971-92 was then linked onto SIC(92) data for 1993-6 using an unpublished concordance supplied to us by the ONS. We found that this concordance gave very different, but considerably more plausible, SIC(92) based estimates for 1987-91 than the figures originally published in CSO(1994), particularly for metals and machinery manufacturing (SIC(92) sub-sections DJ and DK). This may be of interest to other researchers who wish to link data derived from the different industrial classifications. The table in Appendix 3 A reports some simple approximations that can be used to link data at the two-digit industry level across the different industrial classifications. These provide a summary of the underlying conversion factors at the three and four digit levels.13 Time series data on annual hours worked by industry were constructed using information from the New Earnings Survey and the benchmark estimates available in O'Mahony and Wagner (1994). Output prices were measured using two-digit producer price indices. In the absence of any information to the contrary we assumed that prices and hours were similar for domestic and foreign-owned firms in each industry. Initial data analysis sought to establish the period over which a more detailed analysis could take place. The longest span of annual data we were able to construct was for 1974-96 for all firms (foreign and domestic) for eleven of the SIC(92) sub-sections reported in the Table in Appendix 3A. 14 We used this panel data set to estimate a conventional labour demand relationship of the form of (3.8), but with exogenous technical progess only, over different samples and tested for parameter stability across potential break points by allowing the slope parameters to vary from a particular year. Some of the key production function parameters and diagnostic statistics are summarised in Table 3.7. In each sample period we were able to find a significant negative re-
Inward investment and technical progress in the UK
85
Table 3.7. Summary statistics from all firms' panel estimates Elasticity of substitution (o) Technical progress (X) Returns to scale (v)
0.259 0.034 0.845
(0.047) (0.003) (0.126)
Sample period 1974-96 No. of observations 253 Standard error 4.09% Structural break date 1993 Parameter stability [p value] [0.00]
0.254 (0.031) 0.035 (0.002) 1.027 (0.127) 1974-92 209 3.60% 1980 [0.00]
0.247 (0.074) 0.020 (0.002) 1.442 (0.425) 1982-92 121 2.03% 1989 [0.61]
Note: Standard errors of coefficients in parentheses.
lationship between labour demand and the real wage and evidence of significant positive technical progress. However the parameters varied considerably, and there was clear evidence of a structural break in both 1980 and 1993, years in which one industrial classification is linked onto another. To our knowledge this has not been explicitly tested in other studies that have used data linked in this way, although some have found evidence of parameter stability at the end of the 1970s (Cameron and Muellbauer, 1996). In contrast, there was n o evidence of a significant break in the 1 9 8 2 - 9 2 sub-sample using data measured on a single industrial classification. We therefore chose to undertake the rest of our empirical w o r k using data from within this sub-sample. The results in the remainder of this chapter are obtained using a panel data set with the eleven separate industry groups shown in Table 3.8, with data based on the SIC(80). These eleven groups were chosen as they could be readily matched with the available data on hours worked by industry. As can be seen in all industries, the average labour productivity of foreign-owned firms is above that of domestic-owned firms, since foreign firms account for a higher proportion of (value added) outp u t t h a n e m p l o y m e n t . Foreign firms account for a relatively large proportion of the electrical engineering, transport equipment and chemicals industries, but account for only a small p r o p o r t i o n of firms in resource-based industries such as textiles, w o o d products and non-metallic mineral products. 1 5 Tables 3.9A and 3.9B summarise the relative characteristics of foreign and domestic firms over the whole sample period. It can be seen that labour productivity in foreign firms has been consistently higher than in domestic ones, with an average differential of 3 0 - 4 0 per cent,
86
Inward investment, technological change and growth
Table 3.8. Share of foreign firms in manufacturing in 1992 SIQ80) Share of total Share of foreign-owned order manufacturing firms (%) output (%) Value added Employment Basic metals and metal manufacturing Non-metallic mineral products Chemicals and manmade fibres Mechanical engineering Electrical engineering and office machinery Transport equipment Food, drink and tobacco Textiles, footwear and clothing Timber and wooden furniture Paper and publishing Plastics and rubber Total manufacturing
22+31
8.0
17.3
13.0
24
3.9
11.5
11.0
25+26
11.4
36.1
34.0
32
11.6
23.0
19.0
33+34 35+36
10.5 9.9
35.1 28.8
30.5 26.6
41/42
14.4
21.5
13.2
43+45
5.6
7.2
5.2
46 47 48
2.9 12.0 5.1
5.2 19.2 25.9
3.2 14.7 21.1
2-4
100.0
23.4
17.9
Note: Output is measured as gross value added at factor cost.
other than in the food, drink and tobacco sector, where it is considerably higher, and in chemicals and non-metallic mineral products, where it is much smaller. These figures are similar to those found by Nick Oulton in his paper in this volume using a sample of establishments from the Census of Production. Some of the possible explanations for the productivity differential are apparent from the data on fixed investment intensity, the share of skilled labour in the workforce (defined as the ratio of non-operatives to total employment) and the ratio of gross output to value added. Foreign firms tend to be more capital-intensive than domestic firms, with investment being a larger proportion of value added in eight out of the eleven industries. They also have a higher share of non-operatives in total employment in ten out of the eleven industries, and a higher proportion
Inward investment and technical progress in the UK
87
Table 3.9A. Characteristics of foreign and domestic firms, 1983-92 average Foreign labour productivity (Domestic^TOO)
1983-92
Basic metals and metal manufacturing Non-metallic mineral products Chemicals and man-made fibres Mechanical engineering Electrical engineering and office machinery Transport equipment Food, drink and tobacco Textiles, footwear and clothing Timber and wooden furniture Paper and publishing Plastics and rubber
Fixed inv estment/ Value added (%) Domestic Foreign
131.5 107.8 111.1 128.1
10.6 13.7 16.4 7.2
12.4 12.6 16.5 8.0
129.7 130.2 173.0 140.6 142.6 132.6 133.0
9.0 9.7 13.3 7.3 8.3 11.0 14.2
13.6 23.6 12.8 9.1 25.2 13.5 11.3
Table 3.9B. Characteristics of foreign and domestic firms, 1983-92 average 1983-92
Basic metals and metal manufacturing Non-metallic mineral products Chemicals and man-made fibres Mechanical engineering Electrical engineering and office machinery Transport equipment Food, drink and tobacco Textiles, footwear and clothing Timber and wooden furniture Paper and publishing Plastics and rubber
Skilled/total labour (%)
Gross output/ value added
Domestic
Foreign
26.0
30.0
2.97
3.65
26.4
29.4
2.38
2.42
44.3 36.2
50.5 43.2
3.00 2.38
3.07 2.71
42.7 34.7 21.3
45.0 27.3 27.2
2.41 2.60 4.34
3.25 4.20 4.22
17.8
24.3
2.56
2.75
24.4 41.2 27.0
34.0 42.2 31.6
2.98 2.31 2.62
3.19 2.50 2.53
Domestic
Foreign
88
Inward investment, technological change and growth
of intermediate inputs. The ratio of gross to net output is higher for foreign-owned firms in nine of the industries, implying greater use of intermediate inputs. It is of interest to consider the differences in capital intensity, the share of skilled labour and the use of intermediate inputs between industries. There is some evidence that domestic and foreign firms employ a high proportion of skilled labour and use more intermediate inputs in the same industries, with statistically significant Spearman rank correlation coefficients of 0.836 and 0.659, respectively. However there is an insignificant positive correlation between industries ranked by capital intensity in foreign and domestic firms (0.173). A simple cross-sectional regression provides some support for the view that the labour productivity differentials between foreign and domestic firms are related to differences in factors such as capital intensity and usage of skilled labour. Letting PDIFl denote the mean ratio of labour productivity in foreign firms to that in domestic firms over 1983-92 in industry /', IDITt denote the mean ratio of investment-output in foreign firms to that in domestic firms over 1983-92, and SDIFj denote the mean ratio of the share of skilled labour in foreign firms to that in domestic firms, the following coefficients were obtained: FDIFl = 87.07 + 0.082*IDIF1 + 0.256* SDIFt + 45.2*Dum (5.4) (2.1) (2.2) (8.9) R2= 0.67; Standard error = 10.0 (Mean of dependent var. = 132.1). Heteroscedastic-consistent t-statistics are reported in parentheses. The dummy is equal to 1 for the food, drink and tobacco group, reflecting the unusually high relative level of labour productivity there, as shown in Table 3.9A. The results imply that there is a positive relationship between the labour productivity differential and the capital intensity and skill proportions differentials, with both coefficients significant at the 10 per cent level. A 10 per cent differential in capital intensity is associated with a 0.8 per cent labour productivity differential and a 10 per cent differential in the skilled labour share is associated with a 2Vi per cent productivity differential. To investigate whether the presence of foreign firms has had an effect on the behaviour of domestic firms, we estimate a model of labour demand by indigenous companies analogous to (3.8) and that used in Table 3.5, with technical progress modelled using a deterministic trend
Inward investment and technical progress in the UK 89
and alternative measures of the activities of foreign-owned firms. Industry-specific fixed effects are included to control for other industry-specific factors that have remained constant over time. With eleven separate industry groupings, there is a total sample size of 99 annual observations. An instrumental variable estimator is employed to overcome the small sample bias arising from the inclusion of a lagged dependent variable in the model (Nickell, 1981). Higher-order lags of employment are used as instruments. 16 The initial results are summarised in Table 3.10. In the first column we use a deterministic trend as an indicator of technical progress. We obtain significant negative effects from both the time trend and from real wages. The implied elasticity of substitution is 0.45, similar to that obtained in the longer time-series results for the entire manufacturing sector. The coefficients are consistent with increasing returns to scale, with the long-run scale parameter estimated to be 1.53 per cent, although constant returns to scale cannot be rejected. In the second column we endogenise technical progress by assuming that the level is related to the (lagged) total value added of foreign-owned firms in the manufacturing sector at constant prices (denoted Q^MF)- This allows for across-industry spillovers as well as intra-industry ones. The output of foreign-owned firms is found to be significant, and the size and significance of the coefficient on the deterministic time trend drops noticeably. It continues to be possible to accept constant returns to scale, although this restriction has not been imposed in the reported regression. The implied technical progress parameter is much larger than that obtained in the earlier regressions, with a 1 per cent rise in the total output of foreign-owned firms estimated to raise the long-run level of technical progress by 0.92 per cent (standard error 0.44). The exogenous rate of technical change is estimated to be 2.17 per cent per annum (standard error 1 per cent). In the third column of Table 3.10 we employ a log transformation 17 to test whether the direct intra-industry impact of inward investment in a particular industry [denoted ln(QF•-)M] is significantly different from the inter-industry effects arising from investment in other industries. If the own-industry variable does not have a significant coefficient in a regression in which the total manufacturing variable is also included, it is valid to return to the regression in column 2. This is in fact the case. The coefficient on the intra-industry foreign output term is not significant, but the one on the total manufacturing foreign output term is, implying that we cannot reject the hypothesis that intra-industry and
90
Inward investment, technological change and growth
Table 3.10. Inward FDI and technical progress in domestic firms Dependent variable: Mn(LJt Sample: 1984-92 No. of obs. = 99 [1] A ln(Q,),
M-L),-, MQ,)M
ln( W/P.),, TIME '
inisa, ln(Q ,),-,
0.3170 -0.3782 0.3057 -0.1710 -0.0091
(4.0) 0.2587 (3.0) (4.0) -0.2854 (4.1) (5.6) 0.2691 (5.7) (5.5) -0.1386 (4.9) (5.1) -0.0032 (1.7) -0.1350 (2.8)
f
R2
[3]
[2]
0.766 0.783 Standard error 2.11% 2.02% Serial correlation Chi(l)=0.96 Chi(l)=0.003 Production function parameters Returns to 1.539 (0.32) 1.125 (0.27) scale (v) Elasticity of substitution 0.452 (0.02) 0.486 (0.03) (a)
0.2586 -0.2943 0.2737 -0.1420 -0.0034 -0.1398 0.0055
(3.0) (4.2) (5.9) (5.0) (1.8) (2.6) (0.4)
[4] 0.2483 -0.2652 0.2442 -0.1274 -0.0010 -0.1489 -0.0009
(2.8) (3.0) (4.6) (4.3) (0.3) (2.7) (0.0)
0.781
0.785
2.03%
2.01%
Chi(l)=0.01
Chi(l)=0.01
1.157
(0.28) 1.180
(0.30)
0.482
(0.03) 0.480
(0.03)
Notes: Industry fixed effects also included. Current output and lagged employment are endogenous. Heteroscedastic-consistent t-statistics are reported in parentheses, except for derived production function parameters, where standard errors are reported.
inter-industry effects from foreign investment are of a similar magnitude. These results confirm that inward investment has had a significant impact on technical progress, not just in the industry in which the investment has taken place, but also through wider spillover effects to other manufacturing industries. In the final column of Table 3.10 we check whether this finding is sensitive to the measure of inward investment used by re-estimating the model in the third column using the gross output of foreign-owned firms rather than value added. This makes little difference to the findings from the previous regression, suggesting that they are not particularly sensitive to the use of net or gross output. The results also suggest that use of the stock of inward FDI as the proxy for the activities of foreign firms, as in Barrell and Pain (1997) and in Table 3.5 above, does not provide
Inward investment and technical progress in the UK 91
a completely misleading guide to the impact of inward investment, since it also points to the existence of significant effects from inward investment on technical progress. We find larger effects using the output-based variables, but this appears to owe more to the use of a disaggregated panel estimator rather than to bias from using the inward FDI stock. The statistical adequacy of the panel models in Table 3.10 was investigated using a test of first-order serial correlation (as annual data are being used). An auxiliary regression procedure was adopted, with the lagged own-country residuals being added in as additional regressors. 18 In each specification the coefficient on the lagged residuals was insignificant, suggesting that there is no evidence of significant serial correlation and that the model can be regarded as a statistically adequate model of employment demand. An additional finding for all the models reported in Table 3.10 was that the (joint) restrictions required to yield a model consistent with an underlying Cobb-Douglas production function were clearly rejected by the data. For example, the test statistic for the four restrictions required in the equation in column 2 of Table 3.10 was [Wald(4) = 376.6]. The model used so far allows for industry-specific fixed effects, but imposes homogeneity on the slope parameters. Restrictions of this type will yield biased and inconsistent parameter estimates if significant heterogeneity is present in the panel. It is not possible to compute mean group estimates as there are too few observations available to permit efficient independent estimation of the model for each panel member. An alternative procedure, given the relatively small time dimension in our panel, is to use a modified version of the Pooled Mean Group (PMG) estimator proposed by Pesaran et al. (1999), with common longrun parameters but separate dynamic effects for each panel member.19 The panel dynamic effects are computed as the mean of the estimates for the individual panel members. This combines the efficiency of pooled estimation whilst avoiding some of the potential inconsistencies that can arise from the pooling of heterogeneous dynamic relationships. We reestimate the equation in the third column of Table 3.10. This can be expressed as: Aln(Lijt) = at + XltA\n(Qht) + -X3ln(Qi9t-i)-^MW
X^HL^) I P^-XsTIME
(3A3)
92
Inward investment, technological change and growth
Table 3.11. Fooled mean group estimates for eleven manufacturing industries Long-run parameters:
Dynamic parameters:
1.1072
(11.8)
Real wages (A4)
-0.5593
(11.7)
Time trend (A5)
-0.0114
(2.0)
Foreign output: 1(h) -0.6131
(3.2)
(X7)
0.0089
(0.1)
Output growth (A,i) Equilibrium
0.2926
(5.0)
-0.2153
(5.1)
Output (A3)
correction (A^) R2 Standard error Memorandum: individual equilibrium-correction Basic metals and metal manufacturing Non-metallic mineral products Chemicals and man-made fibres Mechanical engineering Electrical engineering and office machinery Transport equipment Food, drink and tobacco Textiles, footwear and clo thing Timber and wooden furniture Paper and publishing Plastics and rubber
0.802 1.92% \Parameters -0.231 -0.187 -0.112 0.036 -0.244 -0.473 -0.183 -0.289 -0.107 -0.185 -0.394
(3.3) (2.6) (2.2) (0.5) (2.7) (4.2) (4.5) (3.5) (1.4) (2.5) (5.3)
The results are reported in Table 3.11. The long-run coefficients are similar to those found in Table 3.10. In particular we still obtain significant effects from the lagged output of foreign-owned firms and the elasticity of substitution remains significantly different from unity. The standard error of the equation is lower than those reported in Table 3.10, and the imposition of common dynamic parameters is rejected by the data [Wald(20) = 89.73], suggesting that there is some heterogeneity present in the panel. The individual equilibrium-correction parameters are reported at the foot of Table 3.11. If the reported parameters are consistent with a valid long-run relationship in each of these industries, then we would expect to obtain a significant negative coefficient. It can be seen that in nine
Inward investment and technical progress in the UK
93
out of the eleven industries this is the case, with particularly well-determined coefficients for transport equipment, food, drink and tobacco, and plastics and rubber. The only industries which do not appear to have a long-run relationship consistent with that reported in Table 3.11 are mechanical engineering and, to a lesser extent, timber and furniture.
Models of labour productivity As we discussed in the second section, an alternative modelling strategy would be to seek to model labour productivity directly and relate it to an indicator of inward investment, the industry capital-labour ratio and the ratio of human capital to the labour force. To obtain an industry-specific estimate of the capital stock, we followed a procedure similar to that described in greater detail in Chapter 5 by Nick Oulton, with the flow of investment being cumulated from 1974 onwards, subject to a suitable depreciation rate. To obtain an industry-specific measure of human capital we create an index based on the average skill level per worker. We make the strong assumption, common in the growth accounting literature, that the internal benefits of human capital are reflected in wages, so that wage differentials reflect the returns to labour skills. We construct an index for domestic firms in each twodigit industry using the shares of operative and non-operative employees: H, = (WN0/W0)*[LN0/(LN0
+
L0)\ + | L N 0 / ( L N 0 + L 0 )]
(3.14)
Wages are normalised on those of operatives. One additional variable is added to the basic specification, with capacity utilisation in the overall manufacturing sector included to allow for common industry effects from the general business cycle. Again we use data on employment, output and wages in domestically owned firms within the same eleven two-digit industries. The results are summarised in Table 3.12, with industry fixed effects included but not reported in the table. We found it possible to accept the restriction of constant returns to scale from labour, physical capital and human capital (k = 0 in (3.10)) and so results are reported with this imposed. The first column includes the two separate terms in the scale of value added output by foreign-owned firms. Again the manufacturing-wide measure is found to be significant, but the additional intra-industry term is not, implying that intra-industry and inter-industry effects are of a similar
94
Inward investment, technological change and growth
Table 3.12. Inward FDI and labour productivity Dependent variable: \n(Y./L.)t Sample: 1984-92 [1] OLS ln(K/L.) ; ln(H./L.), ln(Q fl .),'
ln(Q,J,
In (CUJ R2 Standard error Implied labour coefficient Serial correlation
No. of obs. = 99
[2] OLS
0.2240 (3.8) 0.2636 (2.9) 0.0211 (0.5) 0.2295 (3.8) 0.7167 (5.7) 0.962 5.01%
0.2238 0.2692
0.5123 -
(6.9)
[3] IV
(3.8) (3.0)
0.1691 (2.6) 0.3350 (3.0)
0.2531 (6.5) 0.7228 (5.6) 0.963 4.98%
0.3105 (4.5) 0.8118 (5.4) 0.961 5.07%
0.5070 -
0.4959 (4.4) Chi(l) = 10.35
(6.9)
Notes: Industry fixed effects also included, ^-statistics are reported in parentheses. A dummy is included to deal with outliers in the metals group.
magnitude. Labour productivity appears to be procyclical, with a significant positive effect being obtained from manufacturing capacity utilisation. The first two equations are estimated by OLS, as is commonly done in many empirical applications. In practice the capital-labour ratio, human capital, capacity utilisation and the output of foreign firms are all potentially endogenous. In the final column we re-estimate the model using instrumental variables, with lagged values being used as instruments. This makes little difference to the estimated effects from inward investment, but raises the exponent on human capital and reduces that on physical capital. The coefficients imply that a 1 per cent rise in the volume of output in foreign-owned firms in the manufacturing sector will raise labour productivity by 0.31 per cent. This is in line with the impact on technical progress derived from the long-run aggregate time series evidence but is considerably smaller than that found in the other panel results for these manufacturing industries. One explanation for this might be that, while seemingly plausible, the labour productivity equations are strictly invalid, since they impose a restriction - an elasticity of substitution of unity - which has already been shown to be rejected by the data. The implicit assumption that firms are always operating on their production possibility frontier, apart from any common effects arising from the general business cycle, and
Inward investment and technical progress in the UK 95
Table 3.13. Do spillovers vary according to nationality of investors? Dependent Variable: Mn(L)t Sample: 1984-92 No. of obs. = 99 [2]
[1] A ln(Q,.),
ML,),., ln(Q,),-, ln(W/P.),,
TIME MQ>.JM MQKJM
ln(Q,„),-, MQ,„JM P2 Standard error Serial correlation
0.3548 -0.3168 0.3134 -0.1575 -0.0067 -0.0114 -0.0849
(3.7) (4.7) (5.7) (5.4) (1.9) (0.1) (1.0)
0.2951(3.8) -0.3225(4.8) 0.3024(6.4) -0.1562(5.6) -0.0098(3.3) -0.2205 (3.7)
[3] 0.4022 (4.4) -0.3099 (4.6) 0.2913 (5.8) -0.1503 (5.3) -0.0049 (2.1) -0.0815 (1.9)
0.1108(2.8) 0.798 1.94% Chi(l)=0.19
0.801 1.93% Chi(l)=0.13
0.0091 (0.3) 0.794 1.96% Chi(l)=0.01
Combined inward investment parameter US: -0.0963 (2.0) EU: -0.1096 (2.4)ROW: -0.0724(1.8) Production function parameters Returns to scale (v) 1.022 (0.24) Elasticity of substitution ( a ) 0.497 (0.03)
1.138 (0.22)
1.132 (0.27)
0.484
0.485 (0.03)
(0.02)
Notes: See Table 3.10. The combined inward investment parameter shows the overall impact from the particular source locations - the United States (US), or the EU, or the rest of the world (ROW). Heteroscedastic-consistent ^-statistics in parentheses, apart from derived production function parameters, where standard errors are given.
can instantaneously adjust their inputs, is also something which is rejected by the data. The serial correlation test indicates the presence of significant first-order serial correlation. A similar finding is apparent in the results using establishment level data in Griffith (1999). This suggests greater weight should be placed on the results and findings from the alternative panel model of labour demand. However it is reassuring that, even allowing for the above factors, we continue to find a significant effect from inward investment. Does the nationality of investors matter? In designing investment promotion schemes it is of interest to know
96
Inward investment, technological change and growth
whether the impact of inward investment varies according to the nationality of the investor. There are distinct differences in the pattern of inward investment by nationality, with the United States having been the most important source of inward investment for many years. In contrast, investment from other EU economies has risen especially rapidly since the advent of the Single Market Programme in Europe in the late 1980s. In a comparison of inward investment by US companies in the 1950s and Japanese companies in the 1980s, Dunning (1988b) argues that the former largely sought to transfer product and marketing innovations to the UK, whereas the latter were more concerned to transfer the managerial practices and quality standard used by the parent company. We constructed data at the aggregate manufacturing level for the level of output attributed to firms from the United States, the EU and the rest of the world (primarily the Asia-Pacific countries) and then tested to see whether the effects from any single set of investors differed significantly from the others. This was done by re-estimating the second equation in Table 3.10, with an additional term in (the log of) the level of value added output produced by firms from each region in turn. If the additional term is significant, we can conclude that the impact of inward investment from the particular region differs significantly from the impact of investments from elsewhere. The overall impact of investment from the individual region is given by the sum of the coefficients on the two inward investment terms. The results are shown in Table 3.13. These indicate that inward investments from firms with US parent companies have a larger point impact than those from other investors, although it is not possible to reject the hypothesis that the two have a common coefficient. For EUbased firms, we find that, while they have a significant positive effect on technical progress, it is significantly smaller than that of other foreign firms. For firms from other locations the results are inconclusive. While this may reflect some heterogeneity within this relatively broad category, there is certainly no evidence that inward investment from Asian countries has been the dominant source of the beneficial spillovers from inward investors. Taken together, these results provide some evidence in favour of greater targeting of investment promotion activities towards potential investors from the US compared to other regions, particularly Europe.
Inward investment and technical progress in the UK 97
CONCLUDING COMMENTS For many years successive UK governments have emphasised the potential beneficial effects of inward investment on the UK economy. However, to date, there has been relatively little quantitative econometric evidence as to the extent to which such investment has affected the growth process. In this chapter we have undertaken a systematic exploration of the impact of inward investment on (labour-augmenting) technical progress throughout the UK economy using a number of different models and data sets. Our general findings indicate that the benefits of inward investment have been particularly apparent in the manufacturing sector, but have also been significant in some private service industries. These findings do not indicate that inward investment is the only source of technical change in the UK economy, but they do indicate that it is an important and significant one. The evidence from a number of manufacturing industries suggests that the presence of foreign firms has affected the behaviour of domestic firms. We find that intra-industry and inter-industry spillovers are of similar magnitude. These findings show up in models in which we treat inward investment as an indicator of technical progress within factor demand equations, as well as in the more commonly adopted framework of estimating labour productivity relationships directly through log-linearisation of a production function. The evidence of significant across-industry spillovers suggests that policies designed to improve the attractiveness of the UK as a business location and improve the dissemination of new business practices across a wide range of industries are of equal benefit to policies aimed at solely promoting investment in selected industries. Our analysis also clarifies a number of unresolved issues in the literature on international spillovers. We find considerable heterogeneity in the structure of production across sectors; inevitably this must call into question the extent to which satisfactory estimates of the determinants of technical progress, or in some cases total factor productivity, can be obtained from a single economy-wide relationship. We also find that significant spillovers can be detected from both the stock of foreign direct investment and from measures of the scale of production by foreign-owned firms in the UK. This matters, as in many instances applied researchers have only FDI data with which to work. Our results suggest that it may still be possible to obtain a reasonable guide as to the impact of foreign firms.
98
Inward investment, technological change and growth
APPENDIX 3A Linkages between different Standard Industrial Classifications SIQ92) Industry subsection DA
SIC 68
SIC 80
SIC 92
Weight "Order Weight "Sector Weight * Division
Food, drink and tobacco
DB+DC Textile, clothing and leather products
III
41-42
15-16
0.9*XIII XIV-XV
43-45 0.2*49
17-19
XVII XIX
46
20 361
V
47
21-22
0.9*V 0.1*XIII 0.7*XIX 0.9*XVI
25-26
24
48 0.9*24
25 26
35 0.9*36 22 0.9*31 0.3*32
34-35
DD +361
Wood and wood products Furniture manufacturing
DE
Paper 6c publishing
DG
Chemicals and man-made fibres
DH DI
Plastics & rubber Other non-metallic mineral products
DM
Transport equipment
DJ
Basic metals and metal products
VI 0.8 * XII
DK
Manufacturing of machinery and equipment not elsewhere specified
0.7*VII
o.rxii
DL
Electrical and optical equipment
DN -361
Other manufacturing
0.9*X-XI
27-28
0.7*32 0.1*34 0.1*36
29
VIII 0.9*IX
33 0.9*34 37
30-33
o.rxn
0.1* 31 0.7*49
362-366 37
0.2 * XIX
Inward investment and technical progress in the UK 99
NOTES 1 We are grateful to participants at the NIESR FDI Conference, the Macroeconomic Modelling Bureau Conference at the University of Warwick, the Econometric Society European Meeting in Santiago de Compostela, the EARIE conference in Torino, the Money, Macro and Finance Annual Conference at the University of Oxford, participants in seminars at the Austrian Institute of Economic Research (WIFO), the University of Reading and the University of East London, Peter Hart and colleagues at NIESR for helpful comments and suggestions on much of this work. We are grateful to the ESRC for financial support for this research under grants LI 16251012 and R000237815. 2 This assumption may be a strong one in some industries, notably public services. However in the empirical work described below we did find a significant (dynamic) relationship between employment and real wages in this sector, suggesting that relative prices have at least some influence on real economic decisions. 3 The need for a measure of the stock of fixed capital poses equal difficulties for the alternative empirical approaches of either investigating the determinants of a constructed measure of total factor productivity, or of direct estimation of the production function parameters. 4 Extensions to allow for the incorporation of additional factors that may affect technical change are contained in Barrell and Pain (1999b) and Hubert and Pain (1999a, 1999b). 5 It would be equally valid to treat it either as a source of innovation and, in this framework, disembodied technical change, or to combine it with employment to obtain a quality adjusted labour input. 6 A separate employment term is sometimes also used as an indicator of firm size (Oulton, 1998). 7 These adjustments have a significant effect on estimated productivity growth rates. Self employment in the manufacturing sector was equivalent to just 2 per cent of total employees in employment in 1970/71, but had risen to 9Vi per cent by the mid-1990s. Self employment is especially important in the distribution sector, being currently equivalent to around 15 per cent of employees in employment. 8 This is more accurately described as 'social and personal services' as the category comprises the whole of what was division 9 in the 1980 Standard Industrial Classification, Appendix 5A in the paper
100
Inward investment, technological change and growth
by Nick Oulton in this volume provides further details. 9 This statistic is strictly valid only under the assumption that there is an unique long-run equilibrium relationship. Johansen tests and the panel cointegration tests proposed by Larsson et al. (1998) confirmed that this is the case. 10 Some care is needed in interpreting the statistics in Table 3.4 as they are from estimates which do not instrument the contemporaneous dynamic terms in output and real wages. 11 The labour share was assumed to be the average value over the particular period for which results are calculated. Returns to labour in a given sector were approximated by income from employment in the relevant sector plus a proportion of aggregate self-employment income, based on the average ratios of self-employment in that sector to total self-employment over the sample period. 12 Annual data series can be constructed for individual industries using the establishment level data on the Annual Business Inquiry Respondents Database (ARD), see Griffith (1999) and Chapter 5 by Nick Oulton in this volume. 13 We omit SIC(92) sub-division DF as the manufacturing of coke, refined petroleum products and nuclear fuel had been included in the energy sector rather than the manufacturing sector in the SIC(80). 14 The other manufacturing sector (sub-section DN) was excluded because of the relative heterogeneity of the sector and the difficulties in constructing consistent series for hours worked and output prices. 15 Dunning (1988b) argues that US and Japanese investors have both tended to favour the more technologically advanced and faster growing industries. 16 We also instrument the contemporaneous dynamic term in output, using lagged output and lagged capacity utilisation for the manufacturing sector as a whole as additional instruments. 17 There are two alternative ways of testing this hypothesis. Suppose A is own-industry output and B is output in all other industries. Then in moving from a regression with one term for total output (T) to a regression with separate terms, we can write: y ln(A + B) = Yi \n(A) + y 2 ln(l + BI A) and test whether Yi = Y2Alternatively, noting that B - (T-A), we can write: Y ln(A + B) = YI HA) + y2 M T / A ) =
ln(A)+ 72 M T )
Inward investment and technical progress in the UK
101
w h e r e cf> = y^ — Y2> a n ( ^ t e s t whether 0 = 0 . We follow the second course here. 18 The use of lagged residuals implies that either the estimation period should be reduced by one year or values should be supplied for 1983, since the estimation period begins in 1984. As the former course would result in the loss of eleven observations, we use the asymptotically valid procedure of setting the pre-sample residuals to zero, their value under the null of no serial correlation. Barrell and Pain (1999a) provide more details of this and other panel diagnostic tests. 19 In contrast to Pesaran et al. (1999) we do not seek to allow for heterogenous variances across industries.
REFERENCES Aghion, P. and Howitt, P. (1998), Endogenous Growth Theory, Cambridge, Mass., MIT Press. Barrell, R. and Pain, N. (1997), 'Foreign direct investment, technological change and economic growth within Europe', Economic Journal, 107, pp. 1770-86. (1999a), 'Trade restraints and Japanese direct investment flows', European Economic Review, 43, pp. 29-45. (1999b), 'Domestic institutions, agglomerations and foreign direct investment in Europe', European Economic Review, 43, pp. 925-34. (1999c), 'Foreign direct investment, innovation and economic growth within Europe', in Driver, C. and Temple, P. (eds.) Investment, Growth And Employment: Perspectives For Policy, Routledge Press. Barrell, R., Pain, N. and Young, G. (1996), 'The cross-country demand for labour in Europe', Weltwirtschaftliches Archiv, 132, pp. 6 3 8 50. Blomstrom, M. (1986), 'Foreign investment and productive efficiency: the case of Mexico', Journal of Industrial Economics, 35, pp. 97110. Blomstrom, M. and Sjoholm, F. (1999), 'Technology transfer and spillovers: does local participation with multinationals matter?', European Economic Review, 43, pp. 915-23. Cameron, G., and Muellbauer, J. (1996), 'Knowledge, increasing returns and the UK production function', in Mayes, D. (ed.), Sources of Productivity Growth, Cambridge, Cambridge University Press. Caves, R.E. (1974), 'Multinational firms, competition and productiv-
102
Inward investment, technological change and growth
ity in host-country markets', Economica, 41, pp. 176-93. Coe, D. and Helpman, E. (1995), 'International R&D spillovers', European Economic Review, 39, pp. 859-87. Crafts, N. (1996), 'Deindustrialisation and economic growth', Economic Journal, 106, pp. 172-83. Crafts, N. and Toniolo, G. (1996), Economic Growth In Europe Since 1945, Cambridge, Cambridge University Press. CSO (1980), Standard Industrial Classification Revised 1980. Reconciliation with Standard Industrial Classification 1968, London, Cabinet Office. (1994), Report on the Census of Production: Analysis of Production Industries by Standard Industrial Classification Revised 1992, Business Monitor PA 1002.1. Davies, S.W. and Lyons, B.R. (1991), 'Characterising relative performance: the productivity advantage of foreign owned firms in the UK', Oxford Economic Papers, 43, pp. 584-95. Dunning, J.H. (1988a), 'The eclectic paradigm of international production: a restatement and some possible extensions', Journal of International Business Studies, 19, pp. 1-31. (1988b), 'US and Japanese manufacturing affiliates in the UK: comparisons and contrasts', in Dunning J.H., Explaining International Production, Harper Collins. Globerman, S. (1979), 'Foreign direct investment and 'spillover' efficiency benefits in Canadian manufacturing industries', Canadian Journal of Economics, XII, pp. 42-56. Griffith, R. (1999), 'Using the ARD establishment level data to look at foreign ownership and productivity in the United Kingdom', Economic Journal, 109, pp. F416-42. Griliches, Z. (1998), R&D and Productivity: The Econometric Evidence, Chicago, University of Chicago Press. Grossman, G.M. and Helpman, E. (1991), Innovation and Growth in the Global Economy, Cambridge, Mass., MIT Press. Harvey, A.C., Henry, S.G.B., Peters, S. and Wren-Lewis, S. (1986), 'Stochastic trends in dynamic regression models: an application to the employment-output equation', The Economic Journal, 96, pp.975-H5. Hubert, F. and Pain, N. (1999a), 'Inward investment and technical progress in the United Kingdom', paper presented at European Association for Research in Industrial Economics Annual Conference, Turin, September.
Inward investment and technical progress in the UK 103
(1999b), 'Foreign direct investment, spillovers and technical progress in the UK', paper presented at the Money, Macro and Finance Research Group Annual Conference, Oxford, September. Keller, A. (1989), 'Econometrics of technical change: techniques and problems', in Hackl, P. (ed.), Statistical Analysis and Forecasting of Economic Structural Change, Hamburg, Springer-Verlag. Kremers, J.J.M., Ericsson, N.R. and Dolado, J.J. (1992), 'The power of cointegration tests', Oxford Bulletin of Economics and Statistics, 54, pp. 325-48. Larsson, R., Lyhagen, J. and Lothgren, M. (1998), 'Likelihood-based cointegration tests in heterogeous panels', Stockholm School of Economics Working Paper Series in Economics and Finance No. 250. Nelson, R.R. (1966), 'The CES production function and economic growth projections', Review of Economics and Statistics, XLVIII, pp. 326-28. Nickell, S. (1981), 'Biases in dynamic models with fixed effects', Econometrica, 49, pp. 1399-416. O'Mahony, M. and Wagner, K. (1994), 'Changing fortunes: an industry study of British and German productivity growth over three decades', National Institute of Economic and Social Research, Report Series No. 7. Oulton, N. (1998), 'Labour productivity and foreign ownership in the UK', National Institute of Economic and Social Research Discussion Paper No. 143. Pesaran, M.H. and Smith, R. (1995), 'Estimation of long-run relationships from dynamic heterogenous panels', Journal of Econometrics, 68, pp. 79-114. Pesaran, M.H., Shin, Y. and Smith, R.P. (1999), 'Pooled mean group estimation of dynamic heterogenous panels', Journal of the American Statistical Association, 94, pp. 621-34. Rodrik, D. (1997), 'TFPG controversies, institutions and economic performance in East Asia', NBER Working Paper No. 5914. Romer, P. (1993), 'Idea gaps and object gaps in economic development', Journal of Monetary Economics, 32, pp. 543-73. Solomon, R.F. and Ingham, K.P.D. (1977), 'Discriminating between MNC subsidiaries and indigenous companies: a comparative analysis of the British mechanical engineering industry', Oxford Bulletin of Economics and Statistics, 39, pp. 127-38.
4
Host country effects of FDI in the UK: recent evidence from firm data Sourafel Girma, David Greenaway, Katharine Wakelin and Nuno Sousa1
I N T R O D U C T I O N A N D OUTLINE There has been a general increase in the share of trade in GDP over the last decade or so. In some developing countries this increase is fairly marked as a consequence of rapid liberalisation of areas of tradeable activity. With the exception of the US, the changes in industrialised countries have been more gradual and in many cases almost imperceptible. Yet over this same period there has been a major increase in public awareness of, and sensitivity to, 'globalisation'. This is of course partly fashioned by the prominence given to a succession of public policy issues which are essentially open economy issues - ERM, EMU, Asian currency crises, transatlantic trade spats and so on. In addition, however, it is fashioned by the fact that trade ratios generally only focus on merchandise trade. They do not fully take into account services trade, or, more importantly, inward investment and international production. The growth in the latter has been especially important as a driver of globalisation, but as yet its effects are not well understood and even less well measured. There seems to be a widespread and strong presumption on the part of host governments that inward investment has beneficial economic effects on the host economy, perhaps in the form of employment creation, productivity gains or productivity spillovers. In the UK, as in other OECD economies, significant public funds have been invested in attracting inward investment, presumably on the assumption that externalities accrue. In this chapter we report on two recent studies which have searched for the existence of spillovers. The studies have been completed under a programme of work orchestrated by the Centre for Research
104
Host country effects of FDI in the UK
105
in Globalisation and Labour Markets. This paper synthesises, for the first time, that work. The remainder of the chapter is organised as follows: the following section focuses on wages and productivity spillovers; the third section addresses export spillovers; the final section concludes.
M U L T I N A T I O N A L S , WAGES A N D PRODUCTIVITY SPILLOVERS It is widely assumed that multinational firms (MNCs) have firm-specific assets such as superior technological knowledge, access to international networks and management skills, which enable them to have higher productivity than domestic firms. Inward investment brings two main benefits to the host country. First, it directly introduces new production facilities potentially raising overall output, employment and exports. Second, domestic governments hope that foreign firms will be unable to internalise their advantages fully, so that local firms benefit through spillovers. In addition, MNCs may also pay higher wages, in part as a reflection of higher productivity. However it is also often assumed that foreign firms pay a premium in order to attract high quality workers. There is quite a lot of evidence relating to this issue for industrialised countries. Starting with the UK, most studies seem to find a positive wage and productivity differential, but highlight the potential difficulties in making comparisons and the need to control for some important differences. Aggregate analysis of differences in productivity can fail to control for differences in the industrial distribution of foreign and domestic firms. In addition, foreign firms tend to be larger on average than domestic firms introducing a scale effect that should be controlled for. Davies and Lyons (1991) decompose industry-level data and find that no more than half the productivity advantage of foreign firms can be attributed to structural factors. However, they consider only labour productivity and neglect skills and capital intensity. It is difficult therefore to attribute the remaining differential directly to an ownership effect. Driffield (1996) also finds a significant ownership effect when estimating a wage equation on UK plant data. The wage differential was robust to industrial distribution and to productivity differences; foreign firms paid more even when controlling for their sector and productivity. However there was no differential in the productivity equation when firm size was accounted for. A more recent study (Driffield, 1999) also
106
Inward investment, technological change and growth
showed these patterns for wages and productivity. Foreign firms were found to raise wages in the sector in which they were located, but there were no significant effects on productivity. Griffith (1999) fails to confirm these results for the car industry over the period 1980-92 using establishment-level data. While she found labour productivity to be higher in foreign firms, higher levels of capital intensity and higher use of intermediate inputs could explain the differences. Total factor productivity did not exhibit the same differential, with the exception of US-owned affiliates. But her results do confirm the existence of a wage differential between foreign and domestic firms. Aitken et al. (1996) report similar results for the US. They find only a small wage differential between foreign and domestic firms and much of the apparent difference is accounted for by the greater capital intensity of foreign firms. Evidence for Canada (Globerman etal., 1994) also points to higher labour productivity in foreign establishments but again this can be attributed largely to differences in capital intensity and firm size and foreign firms are found to pay higher wages than domestic firms. To summarise, there appears to be considerable evidence to support a wage differential in favour of foreign-owned firms. However evidence of a productivity differential is more mixed. Moreover the apparent productivity differential noted in aggregate and industry-level data can often be attributed to the larger size of foreign-owned firms, and their greater capital intensity. Productivity and wage differentials Girma etal. (1999) investigate these issues using a large firm-level panel data set of almost 4,000 domestic and foreign firms in UK manufacturing for the period 1991-6. In order to compare similar firms, they chose domestic and foreign subsidiaries that had not experienced a change of ownership between 1991 and 1996. Subsidiaries were chosen so that both types of firms had parent companies. The resulting firms were screened for data availability on wages, employment, value added and fixed assets; firms were included if they had at least three consecutive years of data. To mitigate the impact of outliers, the top and bottom five percentile firms in terms of value added and wages were excluded, as were firms with annual wages or value-added growth exceeding 100 per cent. This left a panel of 2,342 domestic and 1,408 foreign affiliates.
Host country effects of FDI in the UK
107
Table 4.1. Sample means (and standard deviations) Variables Levels Employment Wage rate £000 Output £000 Productivity £000 Capital intensity £000 Growth rates Wages Productivity
Domestic 255.24 (400.18) 15.33 (4.07) 19,618.56 (48,064) 23.91 (10.06) 15.55 (16.15) 2.66% 2.98%
(0.09) (0.18)
Foreign 395.98 (525.64) 17.44 (4.34) 44,990.7 (80,336) 28.59 (11.96) 24.36 (21.50) 3.10% 4.57%
(0.96) (0.20)
Summary statistics for the sample are reported in Table 4.1, which shows that overall foreign-owned manufacturing firms paid higher wages and enjoyed higher productivity than their domestic counterparts. The average annual remuneration and value added per employee in foreign firms was £17,400 and £28,590 respectively, an extra 13.7 per cent and 19.6 per cent relative to the sample means for domestic subsidiaries. In general, foreign firms paid around 6 per cent and produced 11 per cent more than their (four-digit) industry average. In the sample, foreign firms also had a higher scale and capital intensity (defined as fixed assets per employee) than domestic firms. This is also the case for growth rates; the growth rates of productivity and wages for foreign firms are 4.6 per cent and 3.1 per cent respectively, while domestic establishments experienced more modest growth of 3.0 per cent and 2.7 per cent. The basic equation used to analyse productivity and wage differentials was: Ylt = bForft + VXz + DSIC + Dt+f, + elt
(4.1)
where i and t index firms and years respectively, For is a foreign ownership dummy, Dsic is a five-digit SIC92 dummy for fixed industry effects. Dt are time dummies that account for aggregate shocks, f is a timeinvariant firm-specific random effect and e denotes a possible heteroscedastic random noise term with unrestricted (within-firm) serial correlation structure. The dependent variable Y is the log of labour productivity, value added or wages and X is a vector of control vari-
108
Inward investment, technological change and growth
Table 4.2. Differentials between domestic and foreign firms(a) Levels Labour productivity Scale Foreign Total factor productivity Labour Capital Scale Foreign Wages Scale Foreign Wages (with productivity) Scale Productivity Foreign
Growth rates
0.042 9.97%
(13.47) (9.24)
0.050 1.48%
(9.10) (5.42)
0.794 0.145 0.057 5.29%
(89.16) (24.70) (11.35) (5.22)
0.844 0.021 0.072 1.44%
(60.81) (3.38) (11.22) (5.32)
0.030 9.51%
(13.57) (13.29)
0.021 0.40%
(7.80) (2.58)
0.012 0.418 5.34%
(8.72) (54.64) (10.06)
0.010 0.217 0.00%
(4.59) (39.23) (0.00)
Note: (a) In this and all subsequent tables, figures in parentheses indicate asymptotic t statistics. ables. In the labour productivity equation, X consists of Scale, which is defined as the firm's output divided by average (four-digit level) industry output. Estimation of total factor productivity (TFP) differentials was conducted via a standard Cobb-Douglas production function where scale is also used as an additional control variable. 2 In the wage equation, scale and labour productivity were used as firm-level controls. The coefficient on the foreign ownership d u m m y provides a point estimate of the average wage or productivity differential between foreign and domestic firms. Table 4.2 shows that there was a 9.97 per cent labour productivity, 5.29 per cent TFP and 9.51 per cent wages differential in favour of foreign firms. All these differentials are statistically significant. The difference between labour productivity and TFP suggests that foreign firms are more capital intensive than domestic firms. (A similar finding is app a r e n t in the p a p e r by Nick O u l t o n in this volume.) Nevertheless, although higher wages and productivity differentials can be explained partly by scale and capital intensity, there are still substantial differences that can be ascribed to foreign ownership. As well as accounting for scale, the estimations control for the sector in which the firm is located.
Host country effects of FDI in the UK
109
Table 4.3. Differentials by nationality of ownership Levels Labour productivity USA Japan Others Total factor USA Japan Others
Growth rates
13.2 6.14 8.45
(8.43) (1.55) (6.81)
1.78 1.73 1.27
(4.25) (1.61) (3.97)
8.93 -2.99 3.87
(6.13) (0.85) (3.36)
1.71 1.87 1.22
(4.19) (1.76) (3.84)
11.39 5.24 8.79
(10.99) (2.08) (10.75)
0.51 1.04 0.22
(2.39) (2.21) (1.30)
5.87 2.67 5.26
(7.79) (1.56) (8.60)
0.11 0.62 0.00
(0.61) (1.49) (0.35)
productivity
Wages USA Japan Others with productivity) Wages (\ USA Japan Others
The last lines of Table 4.2 reveal that, after controlling for labour productivity in the wage equation, wages in foreign firms are still over 5 per cent higher than in domestic firms. G r o w t h differentials between foreign and domestic firms were also analysed by estimating Equation 4.1 in first differences. It can be seen from Table 4.2 (last column) that labour and total factor productivity growth in foreign firms were higher by about one and a half percentage points compared with domestic firms. However the wage growth rates are not significantly different once productivity is accounted for. It appears that foreign ownership in the UK leads to a level effect. There is no evidence of convergence in productivity levels between domestic and foreign firms. Domestic firms would need a higher growth rate than foreign firms to catch up, but instead, foreign firms are experiencing higher growth rates of productivity, indicating that the gap between foreign and domestic productivity is widening. While this is not the case for the wage differential, there is also no evidence of convergence - domestic firms are not experiencing higher wage growth than foreign firms. Finally, in order to investigate whether productivity and wage differentials were related to the home country, foreign firms were divided into three groups - US-owned, J a p a n e s e - o w n e d a n d others. Table 4.3
110
Inward investment, technological change and growth
presents the results. American firms clearly pay more and are more productive, regardless of the measure used. In contrast, the wage differential of non-American firms is smaller, and is not significant when productivity is controlled for in the wage equation. However wages are growing faster in these firms, though this difference also becomes insignificant once productivity is controlled for. These results contrast with those found for Canada; Globerman etal. (1994) found no significant differences in productivity or wage differentials based on nationality. However they confirm other results for the UK. Griffith (1999) found that US-owned establishments in the UK car industry have higher total factor productivity than domestic firms but no difference was found for firms of other nationalities. Oulton (Chapter 5) also finds that productivity is highest in US-owned affiliates in the UK.
Productivity and wages spillovers In the previous sub-section we reported evidence suggesting that foreign firms pay higher wages and enjoy higher productivity than their domestic counterparts. This implies that there is a composition effect from FDI - a higher proportion of foreign firms in a sector is likely to raise productivity in that sector. Girma etal. (1999) also investigate the dynamic implications of FDI and test whether the presence of foreign firms raises the productivity of domestic firms in the same sector. They estimate possible wage and productivity spillovers by replacing the foreign ownership dummy in equation (4.1) with a measure of foreign presence (FDI): Ylt = SFDIlt + PX„ + Dslc + D, + f,+ elt
(4-2)
The regressions were run on data for domestic firms alone, with industry fixed effects to exploit within-sector variations. The estimation results reported in Table 4.4 show that, on average, there were no wage and productivity spillovers to domestic firms as a result of foreign presence, measured as the sector share of foreign employment. 3 Thus, in aggregate, domestic firms do not gain as a result of a rising share of foreign firms within the same sector. Some experience positive effects, others negative. However additional results show that firms with a low productivity gap relative to the sector average benefit from the presence of foreign firms, as do firms in sectors with high skills and high levels of import competition (see Girma etal., 1999, for details). Financial in-
Host country effects of FDI in the UK
111
Table 4.4. The impact of FDI on the productivity and wages of domestic firms Employment Level Labour Scale FDI
Growth
productivity 0.051 0.018
(8.43) (0.31)
0.057 -0.021
(5.94) (0.47)
Total factor productivity Labour Capital Scale FDI
0.796 0.151 0.063 0.003
(66.11) (21.06) (6.51) (0.05)
0.818 0.034 0.081 -0.048
(48.07) (4.21) (6.54) (1.06)
Wages Scale FDI
0.033 0.028
(8.68) (0.77)
0.021 -0.048
(4.78) (2.16)
Wages (with productivity) Scale Productivity FDI
0.011 0.449 0.019
(5.47) (45.17) (0.65)
0.010 0.247 -0.043
(2.24) (32.50) (2.17)
tervention to foreign firms based on the assumption of universal spillover effects may therefore be misguided. Girma et al. (1999) also test the impact of foreign presence on productivity and wage growth. As can be seen from Table 4.4, there is no substantial evidence linking productivity and wages growth with the growth of FDI, after accounting for domestic firms' scale of production growth.
M U L T I N A T I O N A L S A N D E X P O R T SPILLOVERS Sousa et al. (2000) focus on one potential spillover which has been underexplored, namely the impact of multinational enterprises (MNEs) on the export behaviour of domestic firms. The importance of the export-enhancing role of FDI in host countries has been recognised by, for example, Blake and Pain (1994) for the United Kingdom, O'Sullivan (1993) and Barry and Bradley (1997) for Ireland, and Cabral (1995) for Portugal. This literature typically focuses on the export performance of foreign affiliates themselves. However, in addition, it is possible that
112
Inward investment, technological change and growth
domestic firms become more export-oriented in response to the presence of multinationals. The mechanisms for this can include information spillovers, demonstration effects and stronger product market competition. With regard to the first, exporting implies fixed costs, such as the establishment of distribution networks, the creation of transport infrastructures, investment in advertising to gain public exposure, research about the foreign market to gain intelligence on consumers' tastes, market structure, competitors, regulations and so on. These may be lower for MNEs as they already have knowledge and experience of operating in foreign markets. Such knowledge might spill over to domestic firms. MNEs can also be a source of another sort of information not directly related to exporting, namely new technologies and management techniques. Given this, domestic firms could benefit through processes of demonstration and imitation, due for instance to contact with local clients and suppliers and training of personnel and management staff. The presence of MNEs would thus complement indigenous firms' innovation activities, and contribute to the emergence of a more competitive pool of local firms geared to exporting. 4 Finally entry of foreign companies will typically lead to increased competition. Their presence leaves fewer opportunities for local firms to expand business in the domestic market. Entering into foreign markets is likely to be considered in the expansion plans of local firms. Increased competition in the domestic market may also be responsible for reinforcing the imitation effect, as it constitutes an incentive to engage in more efficient, leaner production techniques. This in turn can also facilitate entry into foreign markets. Empirical evidence on export spillovers is very limited. There are case studies which provide some support for export externalities in developing countries; see for instance Rhee and Belot (1990). Aitken et al. (1997) is the only study to date to test specifically the role of MNEs as export catalysts in the host country. Using panel data on 4,104 Mexican manufacturing plants for the period 1986-90, the authors start by analysing a firm's decision to serve the domestic market or to export, taking into account fixed costs of supplying foreign markets, such as advertising, adaptation to local consumers' tastes and transportation. They argue that these market-specific costs decrease due to information externalities resulting from the local concentration of export activity in general and MNEs' export performance in particular. The impact of
Host country effects of FDI in the UK
113
MNEs on the domestic firm's decision to export is tested using a probit model. Spillovers from both MNE export activity and export activity in general are found to be important. However, there is evidence of heterogeneity. If industries intensive in the use of natural resources and those which face high transport costs are excluded, the effects from the local concentration of export activity become insignificant.
Modelling export spillovers Sousa et al. (2000) construct a theoretical model which extends the work of Aitken et al. (1997). As noted above, that particular model focuses on the decision of whether to export or to supply the domestic market, rather than testing for export spillovers channelled through information spillovers. The model also includes information spillovers but, in addition, explicitly introduces imitation (or demonstration) effects and competition effects as additional possible conduits. The model is tested using a sample of UK domestically-owned firms over a five-year period. The export behaviour of a firm involves both the decision to export and the proportion of production to be exported. The econometric representation chosen takes into account these two decisions and the fact that they are interdependent. Moreover, the second decision takes place only when the firm becomes an exporter.5 Thus the export decision equation for firm i in sector s at time t is written as: EXPORTSlst
=a + $nFORR&c Dst + $2MNEMKTSst +$3MNEEXPORTSst
+$4DOMRd>cDst
+V5SEIst+V6IIst+$7DPRICEst +(38PCS, + $9AVREMit + $10AVASSit
(4 3)
*
+$nTURNOVERit+$nFUNDSit+vit and the export propensity equation as:
EXPROPENSlst =a + XtlFORR&cDst + X2MNEMKTSst +X3MNEEXPORTSst+X4DOMR&cDst +X5SEIst + X6IIst + X7DPRICEst +XsPCst+X9AVREMt +XwAVASSlt + KnTURNOVERit + ult
(4A)
114
Inward investment, technological change and growth
v~N(0, 8) u~N(0,l) corr (vp u) = p
where the subscript / refers to the firms.
The first equation was estimated for the full sample of firms, using a dichotomous dependent variable; EXPORTS takes the value 1 or 0 depending on whether the domestic firm exports or not. This equation also performs the sample selection for the second model that focuses exclusively on export propensity. The dependent variable for this equation is the proportion of the firms' output that is exported. To test for the three channels through which export spillovers may arise, three regressors associated with the presence of multinational enterprises (MNEs), namely MNEMKTS, MNEEXPORTS, and FORR&cD are included. The variable FORR&cD, the expenditure on R&D carried out by foreign MNEs in the UK, is included to control for the demonstration effect.6 We expect that the more innovation activities are carried out by MNEs, the larger is the potential for beneficial imitation by domestically-owned firms. The second variable, MNEMKTS, the share of foreign firms in total employment in a given sector, is included to account for the relative importance of MNEs at the sector level in the domestic market and thus to test for the existence of competition effects. The greater the relative importance of MNEs, the stronger will be the competitive pressure on domestic firms. Finally, the variable MNEEXPORTS, the relative importance of MNEs' export activities in a given sector scaled by the relative importance of MNEs' exports in total export activity, is included to test for spillovers from MNEs' export activities. Positive coefficients were expected for each of these three variables. Because of colinearity the MNEMKTS and MNEEXPORTS variables are not included together in the regressions reported below. The existence of spillovers from export activity in general was also controlled for by including a Sectoral Exports Index (SEI), denoting the relative importance of sector i in domestic exports. This complements the analysis, since it controls for the export structure of the host country and allows the effect of MNEs' export activities to be isolated. Finally the variable II, showing the size of industries at national level in terms of employment, is used to control for possible general spillovers not directly associated with export activity. Other firm- and sector-specific variables were also included, such as domestic producer price (DPRICE), average production costs (PC),
Host country effects of FDI in the UK
115
average wages (AVREM) and fixed assets per employee (AVASS) to control for labour skills and physical capital. The variable TURNOVER was included as a proxy for the size of the firm. Finally, the variable FUNDS, defined as shareholders' funds per unit of output available to the domestic firm, was included to capture domestic firms' financial capacity to meet the extra costs associated with the setting up of export operations. This variable was expected to be important only for the domestic firm's decision of whether to export or not, and is thus not included in equation (4.4). Estimation was carried out on a large sample of highly disaggregated firm-level data comprising 3,662 UK-owned manufacturing firms covering a five-year period from 1992 to 1996. Most of the data used were taken from the OneSource database . Time dummies were also included. Use of firm level data is crucial, since it allows the investigation of the potential export promotion effect of the presence of foreign MNEs to be complemented by an investigation of the specific role of export spillovers to domestic firms.
Results Results for the two equations are reported separately in Tables 4.5 and 4.6. Table 4.5 reports the coefficients for the export decision.7 This model is able to correctly predict 71 per cent of firms' decisions of whether to export or not. 8 There is evidence to support the hypothesis of the existence of export spillovers. It appears that the presence of inward investors increases the probability of an indigenous firm becoming an exporter. The positive and significant coefficient for the variable FORR&cD confirms the presence of a demonstration effect. There is also marginally significant evidence of information spillovers from the export activities of MNEs (MNEEXPORTS) affecting the domestic firms' decision to sell in the foreign markets. This confirms the hypothesis that local firms benefit from contact with the MNEs' exporting strategies and techniques. With respect to the competition effect the relative importance of MNEs in the domestic market, MNEMKTS, is positively and significantly associated with a higher probability that the domestic firm is exporting. The coefficients on the Sectoral Export Index (SEI), which controls for the comparative advantage of the host location, confirm that belonging to an export-oriented sector helps domestic firms to establish export activities. The variable II (Industry Index), which controls for general
116
Inward investment, technological change and growth
Table 4.5. Dependent variable: EXPORTS - export decision Regressor FORR&cD DOMR&cD MNEEXPORTS MNEMKTS SEI II DPRICES PC AVREM AVASS TURNOVER FUNDS Year93 Year94 Year95 Year96 Const. Number of obs. Log likelihood Waldchi 2 (14)
Coeff.
t-stats
0.0001*** 5.787 0.1055 0.695 1.832 0.0987* 8.9422*** 4.65 -32.2887*** -11.432 -0.0055 -1.23 -0.0001 -0.031 1.8918*** 4.185 -0.2568*** -3.695 0.0001 0.903 1.909 0.0679** 0.0039 0.131 -0.1241*** -3.091 -0.628 -0.0363 0.0312 0.45 0.8152* 1.889 11372 -5875.654 309.93
Coeff. 0.0001*** 0.229 1.4151*** 8.741*** -29.8267*** -0.0038 -0.0002 1.6477*** -0.2596*** 0.0001 0.0671** -0.0015 -0.1236*** -0.0214 0.0281 0.4179
t-stats 3.527 1.508 6.158 4.867 --10.463 -0.839 -1.122 3.638 -3.831 0.956 1.835 0.960 -3.066 -0.367 0.402 0.962
11372 -5780.149 312.06
Notes: ** "'significant at 1 per cent; ** significant at 5 per cent; * significant at 10 per cent. spillover effects in the host country not directly associated with exporting, has a significant and negative relationship with the probability of a firm exporting. 9 The coefficients on the variables, TURNOVER, DPRICE (domestic price), and PC (production costs) all have the expected sign but are insignificant. However a positive and significant relationship between labour remuneration costs, AVREM, and the probability of a firm being an exporter are reported, perhaps capturing the importance of labour skills in production for world markets. Thus there is evidence to support the hypothesis that the presence of foreign affiliates in the United Kingdom has a positive influence on the export orientation of domestic firms, although it is not possible to distinguish empirically the competition effect from the information effect. Table 4.6 reports the estimated coefficients for the export propensity equation. Again the model was estimated taking into consideration separately the export information externality effect, MNEEXPORTS, and
Host country effects of FDI in the UK
Table 4.6. Dependent variable: EXPROPEN Regressor FORR&cD DOMR&cD MNEEXPORTS MNEMKTS SEI II DPRICES PC AVREM AVASS TURNOVER Year93 Year94 Year95 Year96 Const. Number of obs. Censored obs. Uncensor. obs Log likelihood Waldchi 2 (14) Rho Wald test of independence of equations, (rho = 0)
t-stats
Coeff. 0.0001* 0.0444 0.0145 0.8533* -2.7719* ** 0.0032* 0.0002* 0.6325* 0.0182 0.0001* -0.0135* ** 0.0062 0.0031 0.0092 -0.2044*
3.034 1.492 1.409 4.612 -4.376 3.795 9.41 6.375 1.078 2.21 -2.763 0.848 0.3 0.717 -2.45
11372 7948 3424 -.5^75.654 309.93 -0.2236 Ch i 2 (l)=32.81 Prob>chi2= 0.0
117
- Propensity to export Coeff. 0.0001* 0.0607** 0.1635*** 0.8561*** -2.5096*** 0.0033*** 0.0002*** 0.6158*** 0.0176 0.0001** -0.013*** 0.0074 0.0069 0.0109 -0.2461**
t-stats 1.623 2.018 3.820 4.692 -3.951 3.920 8.032 6.208 1.041 2.195 -2.675 1.001 0.655 0.851 -2.929
11372 7948 3424 -5780.149 312.06 -0.2281 Chi 2 (l)=34.14 Prob>chi2=:0.00
Notes: **"significant at 1 per cent: ** significant at 5 per cent: * significant at 10 per cent.
the competition effect, MNEMKTS. There is no significant evidence of an export information spillover. The exporting experience of foreign firms seems to contribute little to the export propensity of domestic firms'. There is, nonetheless, evidence that information externalities for export activities in general play an important role, as reflected by the significant positive coefficients on the variable SEE With respect to other spillover channels, there is evidence of demonstration/imitation effects as well as competition effects, given the positive and significant coefficients associated with variables FORR&cD and
118
Inward investment, technological change and growth
MNEMKTS. Thus the presence of foreign MNEs not only affects the decision of domestic firms of whether to enter foreign markets, but also results in an increase in export propensity. A positive and significant coefficient for the variable DOMR&cD suggests that the United Kingdom's expertise in technological activities plays a significant role in the promotion of the export performance of domestic firms. The demonstration effect, associated with the MNEs' own innovation activities, seems to be complementing domestic efforts in terms of technological innovation. As expected, we find a much higher marginal effect associated with domestic R&D efforts than with the innovation activities carried out by MNEs. As far as the other variables are concerned, size, cost and average wage positively and significantly influence export propensity; production costs seem to have a more important influence on export propensity than the decision to export. The results obtained for the export propensity equation confirm the existence of export spillovers associated with the presence of MNEs. The size of the estimated coefficients indicates that the competition effect is again clearly the most important factor. The imitation effect does not seem to be a major factor, owing perhaps to the difficulty in absorbing the technology inflow or to the MNEs' efforts to internalise and protect their technological advantages.
CONCLUSIONS As stated in the introduction to this chapter, policies to attract inward investment feature strongly in the industrial policies of many countries, both developed and developing. This extends not only to attempting to engender an economic environment conducive to investment, but also to the provision of subventions. The latter can be interpreted as reflecting a belief in the presence of positive externalities for MNEs. This chapter has reported work aimed at establishing the presence of spillovers in productivity, wages and exports in the UK. The analysis for productivity spillovers suggests that in the first half of the 1990s, labour productivity in MNEs was about 10 per cent higher than in domestic firms. There is, as one would expect, an associated wage differential, but even after we allow for productivity differences foreign firms appear to pay, on average, about 5 per cent more than domestic firms. However, the results suggest that there is little evidence
Host country effects of FDI in the UK
119
of wage and productivity spillovers to domestic firms, on average. The results for export spillovers are more positive, with foreign presence evidently affecting both the entry of domestic firms into foreign markets and their export propensity.
NOTES 1 The authors gratefully acknowledge financial support from the Leverhulme Trust under Programme Grant No. F114/BF. 2 The results do not rely on the choice of a Cobb-Douglas production function. Alternative estimations using a Translog function give remarkably similar results. We present the Cobb-Douglas results for ease of comparison with other analyses. Wages are defined as average renumeration paid to employees in a year excluding tax, social security and pension payments, and value added is defined as the difference between turnover and the cost of bought-in materials and sources. Employment is given by the average number of employees during the year including full-time and part-time workers. 3 Defining FDI in terms of output yields qualitatively similar results. The variable is measured as the four-digit SIC92 sector's foreign share of manufacturing employment or output, calculated by considering the population of subsidiaries in OneSource. 4 It has been argued elsewhere in the literature that technological innovation plays an important role in promoting the export performance of firms. Empirical evidence supports this view, particularly for the case of developed economies, see Hirsch and Bijaoui (1985) and Wakelin (1998). 5 This specification avoids selectivity biases associated with the option of focusing exclusively on the export propensity of the exporting firms which would cast doubts on the econometric results, see Heckman (1979), Greene (1997). 6 It is important to take into account that the United Kingdom has a clear comparative advantage in technology factors. Technology related activities developed by UK-owned firms are expected to play an important role in determining the country's exports (Wakelin, 1998). We think, therefore, it is important to control for the domestic production of technological know-how when analysing the role played by foreign-owned technology in affecting UK-owned firms' export behaviour. We control for this by including variable
120
Inward investment, technological change and growth
DOMR&D, that captures the domestic contribution to the total innovation activities carried out in the country. 7 The reported Wald test for overall significance indicates that, taken jointly, the coefficients of the regressors are significant. Also the estimated correlation coefficient (rho), between the error terms of the export decision and export propensity equations, is significantly different from zero. This is also confirmed by the likelihood ratio test. Thus the evidence validates our choice of the Heckman selection model for these data. 8 The proportion of correct predictions of the export decision using the probit model. 9 In Aitken etal. (1997) a similar variable (computed at a state level) was introduced to capture site-specific characteristics leading to a concentration of exporting firms in certain regions/states where natural endowments make exporting more feasible. The estimated coefficients associated with this variable were not statistically significant.
REFERENCES Aitken, B.J., Hanson, G.H. and Harrison, A.E. (1997), 'Spillovers, foreign investment and export behavior', Journal of International Economics, 43, pp. 103-32. Aitken, B.J., Harrison, A.E. and Lipsey, R.E. (1996), 'Wages and foreign ownership. A comparative study of Mexico, Venezuela and the United States', Journal of International Economics, 40, pp. 3 4 5 71. Barry, F. and Bradley, J. (1997), 'FDI and the trade: the Irish host-country experience', Economic Journal, 107, pp. 1798-811. Blake, A.P. and Pain, N. (1994), 'Investigating structural change in UK export performance: the role of innovation and direct investment', National Institute of Economic and Social Research, Discussion Paper no. 71. Cabral, S. (1995), 'Comparative export behaviour of foreign and domestic firms in Portugal', Banco de Portugal Economic Bulletin, March, pp. 69-78. Davies, S.W. and Lyons, B.R. (1991), 'Characterising relative performance: the productivity advantage of foreign owned firms in the UK', Oxford Economic Papers, 43, pp. 584-95. Driffield, N.L. (1996), Global Competition and the Labour Market,
Host country effects of FDI in the UK
121
Amsterdam, Harwood Academic Publishers. (1999), 'Indirect employment effects of foreign direct investment into the UK', Bulletin of Economic Research, 51, pp. 207-21. Girma, S. Greenaway, D. and Wakelin, K. (1999), 'Wages, productivity and foreign ownership in UK manufacturing', University of Nottingham GLM Research Paper 99/14. Globerman, S.,Ries,J.C. and Vertinsky, I. (1994), 'The economic performance of foreign affiliates in Canada', Canadian Journal of Economics, 27, pp. 143-56. Greene, W.H. (1997), Econometric Analysis, 3rd edn, Englewood Cliffs, Prentice-Hall. Griffith, R. (1999), 'Productivity and foreign ownership in the UK car industry', The Institute for Fiscal Studies Working Paper Series, No. W99/11. Heckman, J.J. (1979), 'Sample selection bias as a specification error', Econometrica, 47, pp. 153-61. Hirsch, S. and Bijaoui, I. (1985), 'R&D intensity and export performance: a micro view', Weltwirtschaftliches Archiv, 121, pp. 138-251. O'Sullivan, P.J. (1993), 'An assessment of Ireland's export-led growth strategy via foreign direct investment: 1960-1980', Weltwirtschaftliches Archiv, 129, pp. 139-58. Rhee, Y.W. and Belot, T. (1990), 'Export catalysts in low-income countries', World Bank Discussion Paper no. 72. Sousa, N., Greenaway, D. and Wakelin, K. (2000), 'Multinationals and export spillovers in the UK', University of Nottingham GLM Research Paper, forthcoming. Wakelin, K. (1998), 'Innovation and export behaviour at the firm level', Research Policy, 26, pp. 829-41.
5
Why do foreign-owned firms in the UK have higher labour productivity? Nicholas Oulton
INTRODUCTION1 Foreign-owned firms in manufacturing have substantially higher labour productivity than domestically owned ones. This basic feature of the UK economy has been known for some time (Davies and Lyons, 1991). The UK is not unique in this respect: the same is true of countries such as the United States (Doms and Jensen, 1998) and Canada (Globerman etal., 1994). For the UK, the productivity gap at the firm level has been documented by Oulton (1998b); similar findings are reported by Griffith (1999a, 1999b) and in the paper by Girma et al. in this volume. The first aim of this chapter is to analyse the extent to which labour productivity differences are correlated with differences in input intensities. This permits an assessment of the extent to which the higher labour productivity in foreign-owned firms might be explained by the fact that they employ more physical or more human capital per unit of labour. The results indicate that higher human and physical capital intensity can indeed account for most of the observed labour productivity gap in manufacturing. The manufacturing sector accounts for only one-fifth of the UK economy, and there is considerable inward investment in many nonmanufacturing sectors. So the second aim of this chapter is to see whether the foreign-domestic productivity gap is as large in the nonmanufacturing part of the economy. The results suggest that it is, and also that it can be explained largely by differences in capital intensity. These findings prompt the question: if foreign-owned companies, located in Britain and employing British workers, use high human and physical capital intensity to achieve high productivity, why do not British-owned companies do the same? Below we discuss two alterna-
122
Why do foreign-owned firms have higher labour productivity*
123
tive, but not mutually exclusive, hypotheses capable of explaining these findings; one is that the cost of capital is lower for foreign-owned companies and the second is that foreign-owned companies use superior technology.
The two samples The present study will employ two different data sets, one confined to manufacturing and the other covering the whole economy. Our first data set is of establishments in manufacturing drawn from the longitudinal database of the Annual Census of Production (now known as the ARD). We are particularly interested in capital intensity but the ARD contains no estimates of the capital stock, only of investment. So we use the investment series to construct our own estimates of the capital stock by a variant of the perpetual inventory method. Our estimates of human capital are derived from information in the ARD about the number of white collar and blue collar workers and the average wage of each of these groups. Because we need a long series of investment, our sample excludes establishments which are only temporarily present in the ARD, that is establishments which are 'born' or which 'die' during the analysis period 1973-93. Those establishments present continuously in the ARD from 1973 to 1993 are referred to as survivors. There were 1,752 survivors in 1993, of which 411 (24 per cent) were foreign-owned. These survivors employed 1.8 million people in 1973, 27 per cent of all employment recorded in the ARD, and 1.0 million in 1993, or 34.5 per cent of ARD employment. The foreign-owned establishments employed 30.7 per cent of the total in 1993, compared with 17.9 per cent in the total manufacturing sector. The country of ownership of each survivor is known for every year. In practice, we use a threefold breakdown: US-owned, other foreign-owned and UK-owned. The findings cited above on productivity gaps in manufacturing for the UK, the US and Canada have been derived from studying longitudinal databases of each country's production census. Outside manufacturing, no such source exists. Hence to study the non-manufacturing sector we have to utilise data drawn from company accounts. Our second data set is of companies and is drawn from the OneSource database, a large electronic database of company accounts. This database, which has been employed in earlier work on employment growth (Hart and Oulton, 1996,1999) and productivity (Oulton, 1998a) and
124
Inward investment, technological change and growth
is also used by Girma et al. in this volume, is described more fully below. In the version we use, the accounts relate to 1995. The second data set consists of some 32,000 companies employing 8.6 million people; of these, some 22,000 companies employing 5.2 million people were in the non-manufacturing sector. In this sample 35.0 per cent of total employment is in foreign-owned companies. The book value of fixed assets is used as the measure of physical capital in the companies data set. Obviously it would be better to use the value of fixed assets at current replacement cost but such figures are not published in company accounts. For labour input, the only measures available are the number of employees and the average wage. The OneSource database covers both independent companies and subsidiaries. To avoid double-counting, for example including both the parent and the subsidiaries of which it is composed, we divide our companies into four groups: (1) subsidiaries owned by US companies; (2) subsidiaries owned by other foreign companies; (3) subsidiaries owned by UK companies; and (4) UK-registered independent companies which do not own any subsidiaries. So our breakdown by ownership is more detailed than in the ARD sample. The two samples each have their own strengths and weaknesses. The quality of the measures of human and physical capital is higher for the ARD. On the other hand, the ARD sample is confined to survivors in manufacturing only. The companies sample covers the whole corporate sector and has better information on ownership. The structure of the rest of this chapter is as follows. The following section describes our first data source, the ARD, setting out how a data set of survivors was constructed. For this sub-set we are able to make estimates of the physical capital stock. This is followed by an analysis of the determinants of differences in human and physical capital intensity between establishments, and in particular the role of foreign ownership. We also quantify the relationship between labour productivity and the measurable inputs, human and physical capital. An introduction to our second data source follows and we describe how the companies sample was derived from the OneSource database of company accounts. The analysis of the determinants of human and physical capital intensity and the causes of productivity differences is then repeated for this dataset. We discuss two hypotheses which are capable of explaining the findings and, in the final section, summarise our conclusions.
Why do foreign-owned firms have higher labour productivity*
125
THE ARD The ARD, or ABI Respondents Database to give it its full name, is an electronic database of the Annual Business Inquiry (Production), formerly known as the Annual Census of Production (ACOP). In principle, the ARD includes all the data collected under ACOP from 1970 to the present. It covers the whole of the production sector - manufacturing plus mining and quarrying and, for recent years, construction. Oulton (1997) provides a full description. The present chapter uses only the data for manufacturing. The most basic unit in the ARD is the 'local unit', defined as a plant or office at a single location. Above that is the establishment, which is the reporting unit. An establishment consists of at least one local unit (itself) and may consist of more. In recent years a bit under half of employment in 'selected' establishments 2 has been in establishments consisting of just one local unit and nine tenths in establishments with no more than ten local units. Most of the data in the ARD relate to the establishment as a whole. The establishment may or may not be a company in the legal sense. In 1987, 'company-based reporting' was introduced into the Census and the reporting units are now referred to as businesses, but the larger companies continued to be split up into smaller units.3 Three measures of output are available from the ARD. In descending order of size these are: gross output, net output and gross value added. Gross output and gross value added (GVA) are used, the latter mainly because it is additive across establishments and industries. Gross output and GVA, which are both reported in current prices, were converted to 1990 prices using producer price indexes for each Class of the 1980 Standard Industrial Clasification (SIC). There are 22 classes within manufacturing, Appendix 5A provides a complete listing.4 For the period studied here, industry is recorded in the ARD under the 1980 Standard Industrial Classification (SIC). Two measures of labour input are available from the ARD: the number of operatives and the number of administrative, technical and clerical employees (ATCs).5 No data on hours worked are available at the establishment level. Apart from the split into operatives and ATCs, there are no data on skills. However average wages for each group are given and these may be used as proxies for relative skills or human capital at a point in time. Real intermediate input is also required for some of the comparisons but the ARD contains only nominal intermediate input. We deflate the
126
Inward investment, technological change and growth
latter, defined as nominal gross output minus nominal GVA, by the Producer Price Index for materials and fuel for each SIC80 Class. Under the heading of investment, ACOP and the ARD distinguish four categories: (1) new building work; (2) land & existing buildings; (3) plant & machinery; and (4) vehicles. Each category except the first is measured as acquisitions less disposals. This means that each category of investment except the first can be, and frequently is, negative for an individual establishment. From 1992, the breakdown by type is no longer available at the establishment level. A further difficulty is that prior to 1979 the four categories are only recorded in the ARD for a minority of establishments; the reason for this is not clear. The upshot is that a breakdown by category is only available for the period 197991 inclusive. Total investment, the sum over these four categories, is therefore used as the investment measure. Total investment can also be negative at the establishment level. Investment is recorded gross of depreciation at current prices. From 1988, investment includes assets acquired under financial leasing. Prior to then financial leasing was omitted. It is important to note that though asset disposals are given as well as acquisitions, 'the figures for disposals exclude amounts written off for capital assets which are scrapped'. 6 In other words, scrapping is not recorded. Investment was converted to 1990 prices using the implicit price deflator for aggregate manufacturing investment from the UK National Accounts (the Blue Book). For the analysis of productivity, we are usually more interested in the capital stock, rather than in investment. Here we run up against a difficulty. The ARD contains no data on capital stocks, not even book values. So we must estimate stocks by cumulating investment flows. To do so, we have to assume an initial, year zero, capital stock. If the year zero is sufficiently far in the past, and the depreciation rate sufficiently high, the stock estimates for later years will not be too sensitive to the assumed initial stock. Thousands of establishments are entering and leaving the ARD every year and capital stock estimates based on only a few years of cumulated investment will be too unreliable to use. Our strategy for dealing with this problem is to create a data set of survivors, in this case establishments which are continuously present in the ARD from 1973 to 1993. We proceed as follows. Starting with a benchmark capital stock in 1973 for each establishment, we then estimate the stock K in subsequent years by the standard perpetual inventory method:
Kt=It+(l-8)Kt-i
(5J)
Why do foreign-owned firms have higher labour productivity*
127
where I is real investment and 5 is the depreciation rate. Depreciation measures the decline in the stock due to physical wear and tear, gradual obsolescence, and scrapping or retirement. US research suggests that depreciation rates much higher than those assumed by the ONS in generating the official 'net stock' estimates for the UK are appropriate (Fraumeni, 1997). In line with these results, an annual rate of depreciation of 7.5 per cent is assumed. To get the initial capital stock for each establishment, we multiply each establishment's 1973 employment level by the capital-labour ratio for the SIC80 class to which the establishment belongs.7
THE IMPORTANCE OF BEING FOREIGN-OWNED Characteristics of survivors There are 1,752 establishments which are present throughout the period 1973-93 (the latest available year when this research began).8 In 1973 employment in these survivors was nearly 1.8 million and constituted 29.6 per cent of all employment recorded in the ARD. In 1993, the corresponding figures were just over 1 million or 34.5 per cent of total ARD employment (ARD employment is about 80 per cent of total employment in manufacturing). These survivors exhibited a wide range of size. In 1973, the 28 establishments with 7,500 or more employees accounted for 24.9 per cent of employment in survivors; by 1993 there were only 8 such establishments and their employment share had halved to 12.8 per cent. The rates of growth of capital (and other inputs) vary across establishments. That these differences are cumulative and not just transitory can be seen from Table 5.1, which shows the standard deviation of the log of the capital-labour ratio by SIC80 class in 1993. Analysis of variance shows that most of the overall variation of this variable, 59 per cent, is due to variation within classes, not between them. This is rather remarkable given that the capital estimates assume that in 1973 the capital-labour ratio was identical for every establishment in a given class.9 The question then arises, are there any systematic factors behind these differences in capital stock growth rates and the eventual differences in levels to which they give rise? In this section we consider the possible role of foreign ownership. It has long been known from the published results of ACOP that foreign-owned firms tend to have higher labour productivity than
128
Inward investment, technological change and growth
Table 5.1. Capital intensity (K/L) amongst 1973-93 survivors: manufacturing, by SIC80 Class, 1993 Class
22 24 25 31 32 34 35 36 37 41 42 43 45 46 47 48 49 All classes(a)
N
Mean
Median
s.d. of \n(K
73 98 164 130 214 122 47 41 32 86 84 143 96 51 245
58,990 45,633 83,732 28,058 28,413 28,414 48,983 31,029 22,125 47,103 63,993 24,180 6,732 15,809 37,996 39,179 26,144 39,140
48,805 34,182 59,240 24,429 22,704 21,573 37,800 26,361 19,184 24,748 49,736 21,054 5,960 12,056 29,792 32,013 24,674 27,276
0.698 0.741 0.686 0.544 0.611 0.621 0.885 0.608 0.563 0.871 0.613 0.589 0.554 0.497 0.599 0.536 0.547 0.827
75
23 1,747
Source: ARD. Notes: Capital intensity is K/L where K is the capital stock in 1990 £ and L is total employment. Five classes omitted due to zero or small numbers. See the Appendix for the names of the classes. (a) Including omitted classes (23, 26, 33 and 44). domestically owned ones. Only a part of this disparity can be explained by a relative concentration of foreign-owned firms in high productivity sectors (Davies and Lyons, 1991). What has not been so clear is the source of the foreign-owned firms' advantage. Of the 1,752 surviving establishments, 176 were US-owned and a further 235 were owned by non-US foreign interests in 1993. 10 The USowned firms accounted for 16.6 per cent of total employment and the foreign, non-US ones for 14.1 per cent. Thus getting on for a third of employment amongst survivors was in foreign-owned firms. Table 5.2 first documents that the productivity gap between UK- and foreignowned firms applies to survivors, as well as to manufacturing as a whole. In 1993, value added per employee was 55 per cent higher in US-owned establishments and 25 per cent higher in non-US foreign-owned establishments than in UK-owned ones. These huge gaps are similar in size
Why do foreign-owned firms have higher labour productivity*
129
Table 5.2. Descriptive statistics in 1993 for 1973-93 survivors in manufacturing, by ownership ) Ownership
N
Mean
Ratio to UK-owned mean
S.D.
Median
Y/L
Foreign, non -US 235 US-owned 176 1,336 UK-owned
95,542 105,185 62,781
1.52 1.68 1.00
84,985 77,606 67,898
75,824 81,059 47,006
V/L
Foreign, non -US 235 US-owned 176 1,336 UK-owned
28,316 35,008 22,615
1.25 1.55 1.00
18,839 22,924 16,929
24,434 29,191 19,306
K/L
Foreign, non -US 235 US-owned 176 UK-owned 1,336
51,358 54,073 35,024
1.47 1.54 1.00
47,154 45,695 47,906
39,264 36,911 24,538
Foreign, non -US 235 US-owned 176 UK-owned 1,336
619 975 535
1.16 1.82 1.00
1,037 2,375 1,290
351 485 270
Foreign, non -US 233 US-owned 176 1,324 UK-owned
14,935 16,078 12,736
1.17 1.26 1.00
3,936 5,149 4,074
14,595 15,627 12,627
Foreign, non -US 235 US-owned 176 UK-owned 1,336
19,876 21,858 17,668
1.12 1.24 1.00
4,628 6,863 5,157
19,090 20,475 17,127
ATC/L
Foreign, non -US 235 US-owned 176 UK-owned 1,336
0.389 0.407 0.323
1.20 1.26 1.00
0.188 0.182 0.189
0.338 0.370 0.281
MIL
Foreign, non--US 232 US-owned 174 UK-owned 1,325
72,395 74,851 42,671
1.70 1.75 1.00
79,215 68,785 59,314
54,398 56,058 28,742
Source: ARD. Notes: Y= Gross output (1990 £k); V = gross value added (1990 £k); L = total employment (number); ATC = Administrative, technical and clerical employees (number); K = Capital stock (1990 £k); wQp - operative wage (£, current prices); wATC = ATC wage (£, current prices); M = intermediate input (1990 £k).
130
Inward investment, technological change and growth
Table 5.3. Growth rates of output, employment and capital, 1973-93: 1973-93 survivors in manufacturing, by ownership in 1993 (% p.a.) Ownership
N
Mean
S.D.
Median
NON-US US UK
232 173 1,326
2.19 2.23 1.76
3.09 3.30 2.92
2.15 1.93 1.82
NON-US US UK
235 176 1,336
4.60 4.27 2.85
3.17 3.41 3.36
4.56 4.38 2.84
NON-US US UK
235 176 1,336
2.23 2.86 1.04
3.90 4.06 3.89
2.20 2.40 0.63
NON-US US UK
235 176 1,341
-2.38 -1.41 -1.82
3.79 3.56 3.95
-2.52 -1.33 -1.66
NON-US US UK
232 173 1,326
-0.16 0.89 -0.06
4.46 4.64 4.78
-0.07 0.65 0.08
Source: ARD. Notes: L = total employment; V= gross value added (1990 prices); K = Capital stock (1990 prices). to the cross-country gaps which are estimated to exist between UK labour productivity in aggregate manufacturing on the one hand and US, Japanese or European productivity on the other (O'Mahony, 1999). We can also see from Table 5.2 that the average worker in a US-owned establishment had 54 per cent more capital to work with, and the average worker in other foreign-owned establishments 47 per cent more, than did their counterparts in UK-owned ones. We can also note that the proportion of employment which is white collar (ATCs) is 7-10 percentage points higher in foreign-owned establishments. White collar workers on average earn more than blue collar ones and we can also see that within each category wages are higher in foreign-owned establishments. For operatives, wages are 16-24 per cent higher and for ATCs, 12-20 per cent higher. Since companies do not pay higher wages out of the goodness of their hearts, this suggests that workers in
Why do foreign-owned firms have higher labour productivity*
131
foreign-owned establishments are more skilled. Thus it appears to be the case that foreign-owned establishments, particularly US ones, employ substantially higher physical and human capital per worker. Because the ARD allows us to look at the history of individual establishments, we can also compare growth rates by type of ownership (Table 5.3). Again the performance of foreign-owned establishments stands out. Value added per employee and capital per employee have both been rising much more rapidly in foreign-owned establishments. The capital stock of UK-owned establishments grew on average at only 1.04 per cent per annum over 1973-93, while that of US-owned ones grew at 2.86 per cent per annum and that of other foreign-owned ones at 2.23 per cent. However there is an important contrast between US and other foreign establishments. In US-owned establishments, value added has been rising, while in other foreign owned ones it has been falling, as it has too in UK-owned establishments. Furthermore, while employment has been falling in all types of establishment, it has done so most in other foreign-owned plants and least in US-owned ones. These differences cannot be dismissed as due solely to differences in industrial structure between foreign and domestic establishments. In thirteen out of seventeen classes foreign-owned establishments have substantially higher capital per employee and in twelve out of seventeen they have substantially higher value added per employee. UK-owned establishments have substantially higher value added per employee in only one class (35): see Oulton (1998b, Table 10). Analysing the foreign advantage We now take a more formal approach and test whether, after controlling for industrial structure, ownership is a significant determinant of input intensities. To test our hypothesis, we carry out a cross-section, multiple regression analysis of the following model: 5-1
Z = p 0 +p 1 US + p 2 N O N - U S + X P , - + 2 C ^ s 5 - + e
,sl)
where Z is alternately capital per employee (K/L), the operative wage (wQp), the ATC wage (wATC), the ATC proportion (ATC/L), intermediate input per employee (M/L), and value added per employee (V/L); all these variables are in logs. US is a dummy variable taking the value 1 if the establishment is US-owned. NON-US is a similar dummy variable taking the value 1 if the establishment is foreign-owned but not US-
132
Inward investment, technological change and growth
Table 5.4. Comparison of foreign and UK-owned establishments in manufacturing: cross-section regressions, 1973-93 survivors in 1993 \n(K/L) US
0.2643 **• (0.0526) NON- 0.2832**• US (0.0450) N 1747 R2 0.427
Dependent variable ln(ATC/L) ln(wop) HwATC) 0.1149** (0.0384) 0.0768* (0.0341) 1752 0.283
0.1499** (0.0241) 0.1050** (0.0195) 1733 0.330
ln(M/L)
0.1431** 0.4356** (0.0211) (0.0552) 0.0796** 0.4297** (0.0170) (0.0448) 1747 1752 0.324 0.170
ln(WL) 0.2756** (0.0443) 0.1362** (0.0379) 1736 0.269
Source: ARD. Notes: Model fitted by OLS is equation (5.2). Constant and dummies for 20 out of 21 SIC80 Classes included but not reported. Robust standard errors are in parentheses. Maximum possible number of observations is 1,752 but a few observations are lost for some dependent variables due to missing values. * Significant at the 5% level. ** Significant at the 1% level.
owned. The Class variables are dummies for membership in each of the 22 classes within SIC80 manufacturing. In practice S = 21 since there are no observations in Class 2 1 . Table 5.4 shows the results. All the major differences noted above between foreign and UK-owned establishments turn out to be statistically highly significant, even when we correct for industrial structure. Both US and non-US owned foreign establishments have higher productivity, a higher ATC p r o p o r t i o n , higher wages, and higher capital per worker than their domestic counterparts. By comparing the coefficients on the ownership dummies with the crude, percentage gaps in Table 5.2, we see that industrial structure does account for some of the differences between domestic and foreign firms. But, after adjustment, we still find large differences between foreign and UK-owned establishments. The largest adjusted difference is in intermediate input per employee. 1 1 However we should also note from Table 5.2 that foreign establishments, particularly US ones, tend to be larger than domestic ones. It is generally believed that larger establishments pay higher wages t h a n smaller ones and have higher labour productivity; this is true with our data too. So the foreign advantage might be an artefact. If size measured by employment is included, it is significant in the equations for wages, the ATC proportion and intermediate input, but has little effect on the coefficients on the ownership dummies. In any case, one might
Why do foreign-owned firms have higher labour productivity*
133
argue that size should be excluded since it is under the control of management.12 Turning to productivity, our hypothesis is that it may differ between establishments because (a) some establishments use more inputs per worker and (b) some establishments may have access to superior technology or superior business systems or may have superior products, that is products which can be sold at a higher price. Input intensity is measured by physical capital per worker and human capital per worker. We have no measures of superior technology or products but we can check whether, after controlling for input intensity, higher productivity is associated with ownership. Note that different types of ownership may be associated with greater or smaller input intensity. So ownership can have a direct effect on productivity, say if foreign-owned establishments have access to superior technology, and an indirect effect, say if they are more capital intensive. We therefore fit the following model: ln(V/ L) - p 0 + Pi HK/L)
+ p2
\n(ATC/L)
+ $3\n(wOP) + $4\n(wATQ) + p 5 U S + p 6 N O N - US+ JfrjjObssj
+e
(5 3)
'
;=i
This equation should not be interpreted as a structural relationship. Rather its aim is descriptive: how much of the variance of productivity can the included variables explain? The results of fitting this model for 1989 and 1993 appear in Table 5.5. 1993 is used because it is the latest year available and the influence of the starting values for the capital stock estimates will be minimised. On the other hand 1993 is a recession year while 1989 is a peak. The results for the two years are in fact very similar though slightly stronger for 1989. We can explain about half the variance of labour productivity across establishments. Capital per worker is highly significant though the size of the coefficient is sensitive to the other variables included. The variables measuring human capital per worker are also highly significant. Including the wage variables adds considerably to the explanatory power. But because the wage variables might be also picking up rentsharing or union power, results are shown as well with these variables excluded. The dummy for US ownership is significant but that for other foreign
134
Inward investment, technological change and growth
Table 5.5. Regressions explaining productivity differences amongst survivors in manufacturing: dependent variable is log of value added per employee, 1989 and 1993 1993 (1)
1993 (2)
1993 (3)
1989 (4)
1989 (5)
1989 (6)
0.1448* 0.2948* 0.1441 **• 0.1582* * 0.2937* = * 0.1569** (0.0244) (0.0221) (0.0242) (0.0199) (0.0251) (0.0199) \n(ATC/L) 0.0830* :* 0.1150* :* 0.0795* *• 0.1105* *0.1764* 5* 0.1094** (0.0235) (0.0242) (0.0235) (0.0213) (0.0218) (0.0214) 0.6342* :* 0.6315* *• 0.6380* * 0.6357** Mw o p ) (0.0597) (0.0598) (0.0427) (0.0427) 0.3053 * :* 0.2987* *• 0.2620* * 0.2601** \n(wATC) (0.0454) (0.0451) (0.0455) (0.0456) 0.0895* 0.1834* :* 0.1337* * 0.1889* :* US (0.0332) (0.0354) (0.0399) (0.0418) 0.0432 -0.0243 0.0074 NON-US -0.0012 (0.0325) (0.0355) (0.0368) (0.0399) 0.0054* US years 0.0081** — — — — (0.0022) (0.0023) NON-US 0.0026 0.0021 (0.0024) (0.0029) years N 1717 1731 1717 1739 1744 1739 0.512 0.475 0.476 0.415 R2 0.379 0.511 \n(K/L)
Source: ARD. Notes: Model fitted by OLS is equation (5.2). Constant and dummies for 20 out of 21 SIC80 Classes included but not reported. Robust standard errors are in parentheses. Maximum possible number of observations is 1,752 but a few observations are lost due to missing values. US years: number of years in US ownership. NON-US years: number of years in other foreign ownership. * Significant at the 5% level. ** Significant at the 1% level. ownership is not. In other words, non-US foreign ownership leads to higher physical and human capital, but no further effect on productivity. But US-owned plants seem to have some additional advantage, over and above greater capital per worker. This might be superior management, better process technology or better products (that is, products able to command a higher price in the market). Whatever the source, US ownership conferred an additional advantage of between 9 and 20 per cent in 1993 (14-21per cent in 1989).13 However, since our analysis is confined to survivors, we cannot allow for the impact of the more recent Japanese and Korean multinationals, who have made large,
Why do foreign-owned firms have higher labour productivity*
135
greenfield investments. If these could be included, we might need to take a more favourable view of non-US foreign ownership. A possible way of minimising the effect of foreign ownership is to argue that foreign companies are just particularly good at picking winners. According to this view, the foreign-owned establishments would have been successful anyway. Perceiving the likelihood of success, foreign companies took them over. The problem with this argument is that it requires remarkable prescience on the part of the foreign companies, since many of these establishments have been in foreign ownership for much of the period studied. For example, 176 of the 1,752 survivors were in US ownership in 1993. Of these, three-quarters had been USowned for fifteen years or more and over half for twenty years or more. (Other foreign ownership tends to be more recent. One-half of the 235 establishments in this category in 1993 had been so for six years or less and less than a third for fifteen or more.) As an alternative to a dummy variable for current ownership, we can also measure the impact of ownership by the number of years in US or other foreign ownership. This is a rather different concept from current ownership status since some establishments may have been in US ownership for part of our period even though currently they are not. In fact, 183 establishments ceased to be US-owned and 193 became US-owned at some point over 1973-93. Altogether 305 establishments experienced some period of US ownership. Changes in and out of other foreign ownership were of similar frequency: 156 establishments ceased to be in this category while 327 entered it. Instead of the ownership dummies, we can include years under US ownership and years under other foreign ownership in the regression. The third and sixth columns of Table 5.5 show the results. Years of US ownership are significant, while years of other foreign ownership are not. Each year under US ownership raises labour productivity by between 0.5 and 0.8 per cent. The conclusion is that more than just picking winners is involved in the superior performance of US-owned firms. We can now employ the estimates reported in Tables 5.4 and 5.5 to decompose the productivity gap between foreign and UK-owned establishments into an explained and an unexplained portion. From Table 5.4, we see that, after controlling for industrial structure, US ownership confers an advantage in value added per worker of (exp[0.2756]-l = ) 31.7 per cent. Other foreign ownership confers an advantage of 14.6 per cent, again after controlling for industrial structure. These figures should be compared with the unadjusted gaps of 55 per cent and 25 per
136
Inward investment, technological change and growth
Table 5.6. Contribution of measured inputs to explanation of productivity gap between foreign and UK-owned establishments: manufacturing in 1993 (1973-93 survivors) US Wage variables included Input K/L ATC/L
Contri- % of bution total
Non-US Wage variables excluded
Contribution
Wage variables included
% of Contri- % of total bution total
Wage variables excluded Contri- % of bution total
3.90 0.96 9.97 4.47
12.29 3.02 31.43 14.07
8.10 1.33 -
25.54 4.19 -
4.19 28.69 0.64 4.38 6.89 47.19 2.46 16.86
8.71 0.89 -
59.67 6.08 -
Total measured 19.30
60.82
9.43
29.73
14.17 97.12
9.59
65.75
20.13
63.44
0.00
0.00
0.00
29.56
93.16
14.17 97.12
9.59
65.75
2.17 6.84 31.73 100.00
0.42 2.88 14.59 100.00
W
OP
W
ATC
Non9.36 29.51 specific(a) Total explained 28.66 90.32 Unex3.07 9.68 plained 31.73 100.00 Total
0.00
5.00 34.25 14.59 100.00
Source: Coefficient estimates in Tables 5.4 and 5.5. See text for explanation. (a) The non-specific advantage is the estimated coefficient on the US or non-US ownership dummy in equation (5.2).
cent, respectively (Table 5.2). The contribution of each measured input to the US advantage is calculated as its estimated coefficient in the regression for ln(WL) from Table 5.5, multiplied by the estimated coefficient on the US ownership d u m m y in the regression with this inp u t as the dependent variable (from Table 5.4). For e x a m p l e , the contribution of capital to explaining the US advantage in 1 9 9 3 , if the wage variables are excluded, is 0.2948 x 0.2643 = 0.0779 which, expressed as a percentage, is 8.1 per cent. The contribution of each input to the other foreign advantage is calculated analogously. The estimated contributions appear in Table 5.6. If the wage variables are excluded, capital intensity explains 26 per cent of the US advantage and 60 per cent of the other foreign advantage. With wages included, the contribution of capital is halved, to 12 per cent and 29 per cent,
Why do foreign-owned firms have higher labour productivity*
137
respectively. Interpreting the wage variables and the ATC proportion as measuring labour quality, between them they account for 49 per cent of the US advantage and 68 per cent of the other foreign advantage. All told, the measured inputs, including labour quality, account for 61 per cent of the advantages of US-owned firms and 97 per cent of those of other foreign-owned firms.
T H E O N E S O U R C E DATABASE O F C O M P A N Y ACCOUNTS We now turn to the results employing company data derived from the OneSource CD-ROM entitled UK Companies, Volume 1 for December 1996. This contains the accounts of some 110,000 larger UK companies. The ultimate source is the accounts which companies are legally required to deposit at Companies House. The criteria for inclusion in Volume 1 is stated by OneSource to be: 'All public limited companies, all companies with employees greater than 50, and the top companies based on turnover, net worth, total assets, or shareholders funds (whichever is largest) up to a maximum of 110,000 companies'. Only 'live' companies are included. Companies which are dormant, dissolved, in liquidation, or in the process of being wound up are excluded. The database contains the latest available accounts and related information for each included company, including the date of the end of the accounting period. Though the CD-ROM which we used is dated December 1996, the data relate to a somewhat earlier period, which varies between companies according to the date of their accounts. For the great majority of companies, this date falls within 1995 (the average is about two thirds of the way through 1995). Hence for simplicity we refer to the year to which the data relate as 1995. 14 Companies are classified under the 1980 SIC. Output can be measured by either sales or value added (the latter defined as trading profit plus the wage bill). Physical capital is measured by the book value of fixed assets. Clearly, this is likely to be a very noisy measure of the true value, since it is in nominal terms and companies differ both in the time pattern of asset acquisition and in their depreciation practices. Employment is a headcount. There is no breakdown by type of labour or by skill but we can calculate the average wage which may serve as a proxy for the average level of human capital per worker.
138
Inward investment, technological change and growth
For each company, OneSource gives, first, the country of the holding company which owns the company in question and, second, the country of the ultimate holding company. Either or both of these may of course be missing; 'country' can include the UK. Foreign-owned companies are broken down into two groups, (a) US-owned companies and (b) other foreign-owned companies. A company is classified as USowned if either the country of the ultimate holding company is the US or, if this is missing, the country of the holding company is the US. Other foreign ownership is determined analogously. Amongst UK-owned companies we distinguish between subsidiaries of UK-owned companies and independent UK companies which do not own any subsidiaries. This is to avoid double counting.15 Avoidance of double counting leads to the elimination of 21,009 companies. A company is classified as a UK-owned subsidiary if it is not an ultimate holding company and either the country of its ultimate holding company is the UK or, if this is missing, the country of its holding company is the UK. A company is classified as a UK-owned independent without subsidiaries if it is not an ultimate holding company and it is not a subsidiary. These four categories should be mutually exclusive but unfortunately this is not the case in practice. There is an inconsistency in the OneSource database: some companies are classified as subsidiaries by one variable, the subsidiary indicator variable, but as independent by the type of ownership variable; the latter variable is the one used to exclude ultimate holding companies. We cannot resolve this inconsistency so we simply drop the companies which fall into more than one category. This leads to the elimination of 1,447 companies. A further 45,260 companies are lost due to missing or zero values. In summary, we start with 107,829 companies (after eliminating companies with out-of-date accounts), we eliminate a further 67,716 and the sample is then 40,113 companies which employed collectively 10.020 million people. Value added is available for a smaller number: 36,226 companies employing 9.391 million people. We also exclude loss-making companies, and so we arrive eventually at 31,954 companies employing 8.639 million people (see Appendix 5B, Table 5B.2).16 The private sector, a wider category than the corporate sector, employed just over 17 million in mid-1995 when self-employment is excluded (Economic Trends Annual Supplement 1997, Table 3.8), so over half of employment in the corporate sector is covered by our analysis.
Why do foreign-owned firms have higher labour productivity*
139
Table 5.7. Labour productivity and its determinants by ownership type, 1995: distribution of within class means across SIC80 classes (UK independents = 100) Ownership
US Other foreign UK-owned subsidiaries US Other foreign UK-owned subsidiaries US Other foreign UK-owned subsidiaries
25th
Quartiles 50th
75th
113.0 113.1 99.6 124.2 135.6 91.8 103.9 104.4 92.1
139.0 126.8 113.8 149.9 165.4 110.1 119.5 117.5 101.9
170.9 150.0 126.0 217.9 226.1 139.6 145.8 128.2 113.8
Source: Oulton (1998c, Appendix B, Table B2). Notes: 36 SIC80 Classes (35 for US-owned). Within-class means are employment-weighted. O W N E R S H I P A N D PRODUCTIVITY: RESULTS F O R C O M P A N Y DATA We start by considering some descriptive statistics (see Appendix 5B for the detail). Overall, UK subsidiaries account for 54.5 per cent of employment, US-owned companies for 13.0 per cent and other foreignowned companies for 22.1 per cent. The remainder, 10.5 per cent, is in UK independents. But this latter figure is an underestimate of the population proportion since our sample excludes many smaller companies. As a summary measure, we calculate the employment-weighted means of labour productivity (value added per employee, VIE) for each SIC80 class and for the four types of ownership. We also calculate the employment-weighted means of the determinants or correlates of productivity: physical capital intensity (K/L) and human capital intensity (w). That is, each mean is an employment-weighted average over the companies within a particular class. This information is summarised in Table 5.7, which shows quartiles of the distribution across classes of these employment-weighted means.17 Here the means are expressed as index numbers with the value for UK independents set equal to 100. The rank order for all three measures is foreign-owned first, UK subsidiaries second and last, UK independents. Clearly the distributions for
140
Inward investment, technological change and growth
Table 5.8. Ownership and the determinants of labour productivity in 1995 Manufacturing Dependent variable \n(K/L)
(D \n(age) US
\nw (2)
ln(WL) (3)
Non- manufacturing Dependent variable ln(i<:/L) \nw ln(WL) (4) (6) (5)
0.0601* * 0.0119* 5: "-0.0151** 0.1451 *'f -0.0035 -0.0464** (0.0104) (0.0042) (0.0051) (0.0106) (0.0042) (0.0055) 0.3544* * 0.1452* ::" 0.3038** 0.1215* 5:' 0.3429* ;* 0.4004** (0.0342) (0.0127) (0.0177) (0.0415) (0.0149) (0.0216)
NON-US
0.3906* * 0.1039*»:" 0.2091** 0.1226* : :" 0.2629* :* 0.3797** (0.0301) (0.0107) (0.0142) (0.0291) (0.0118) (0.0160)
UKSUB
-0.1050* * 0.0007 (0.0227) (0.0086)
N R2
10,061 0.183
10,061 0.151
0.0553** -0.2199* ! :" 0.0320* :* 0.0839** (0.0107) (0.0210) (0.0082) (0.0106 10,061 0.158
21,893 0.177
21,893 0.273
21,893 0.329
Source: OneSource. Notes: Companies included are either subsidiaries or independents which do not own any subsidiaries. Companies with negative profits excluded. Constant and dummies for the 60 SIC80 classes included but not reported. Estimated by OLS. Robust standard errors are in parentheses. * Significant at the 5% level. ** Significant at the 1% level.
US- and other foreign-owned companies tend to lie above those for UK independents and subsidiaries. Though the largest differences are between UK independents and the rest, the differences between UK subsidiaries and the foreign-owned companies are also substantial. Median productivity is 18 per cent lower in UK subsidiaries than in USowned companies, 10 per cent lower than in other foreign companies. Foreign-owned companies have much higher capital intensity. Median capital per employee is 36 per cent higher in US companies, 50 per cent higher in other foreign companies. Foreign companies also pay much higher wages, indicating a considerably more skilled labour force. The median wage is 17 per cent higher in US companies, 15 per cent higher in other foreign ones. In order to see whether these impressions are statistically significant, we regress input intensity and other characteristics of companies on the ownership dummies and the controls, following equation (5.2). We do these regressions separately for manufacturing (SIC80 divisions 2-4) and non-manufacturing (divisions 0,1, and 6-9): see Table 5.8.18 Physical
Why do foreign-owned firms have higher labour productivity*
141
Table 5.9. Effect of ownership on labour productivity and its determinants (per cent) Ownership type
Manufacturing (1)
Non-manufacturing (2)
US NON-US UKSUB
+43 +48 -10
+13 +13 -20
US NON-US UKSUB
+ 16 + 11 0
+41 +30 + 3
US NON-US UKSUB
+35 +23 + 6
+49 +46 + 9
Source: Table 5.8. Notes: Each entry shows the percentage effect of a particular type of ownership on the dependent variable, relative to the omitted type of ownership, UK independents. Each entry is calculated as 100*[exp(p)-l] where P is the estimated regression coefficient on the ownership dummy.
capital intensity is measured by the log of the capital-labour ratio, while the log of the wage acts as a proxy for h u m a n capital per worker. Here, in addition to the d u m m y variables for the SIC80 class to which each company is assigned, we employ a number of other control variables. Since companies' accounts do not all relate to exactly the same period, the date of each company's financial year end is included. We also include company age since new companies may not yet have reached their optimal scale. The three ownership dummies are: US (=1 if US-owned), NON-US (=1 if foreign- but not US-owned), and UKSUB (=1 if owned by a UK company), with UK independents being the omitted category. 19 The foreign ownership dummies are highly significant in both manufacturing and non-manufacturing. Table 5.9, which derives from Table 5.8, shows the percentage effect of type of ownership on the determinants of labour productivity, capital intensity and skill (proxied by wages), and their effect on labour productivity itself. In general and relative to UK independents, foreign ownership, whether US or other, is associated with substantially higher capital intensity and higher skill (wages). UK subsidiaries tend to have lower capital intensity than even UK independents but similar skill levels. Foreign ownership has a much bigger effect on capital intensity in manufacturing than in non-manu-
142
Inward investment, technological change and growth
facturing. But for skill the foreign ownership effect is much larger in nonmanufacturing. Foreign ownership is therefore positively correlated with input intensity. So we shall obtain a maximum estimate of the effect of foreign ownership by regressing productivity on ownership plus controls only. These results are also summarised in Tables 5.8 and 5.9. In manufacturing, relative to UK independents, US ownership raises labour productivity by 35 per cent, while other foreign ownership raises it by 23 per cent. In non-manufacturing the US and other foreign effects are much larger, at 49 and 46 per cent respectively. UK subsidiaries also have an advantage over UK independents, but it is much smaller, 6 per cent in manufacturing and 9 per cent in non-manufacturing. Hence UK subsidiaries have substantially lower productivity than foreign-owned ones.20 The productivity gap in manufacturing using company data is remarkably similar to the gap estimated from the ARD sample. From the companies sample, US ownership raises labour productivity by 35 per cent after controlling for industrial structure, other foreign ownership raises productivity by 23 per cent (Table 5.9). The corresponding figures from the ARD sample are 32 per cent and 15 per cent (Table 5.6). Since the two samples are independent and employ different units (establishments versus companies), this adds confidence to the conclusions. It suggests that the use of survivors is not distorting the ARD-based results. In Table 5.10 we show the results of running the regression of equation (5.3) on the companies sample. It turns out that the regressions can explain some 67-71 per cent of the variance of log productivity. Despite the fact that both physical and human capital 21 are poorly measured in our data, each of these variables is highly significant. In both manufacturing and non-manufacturing, UK independents (the omitted category amongst the ownership dummies) are at a significant disadvantage vis-a-vis both UK subsidiaries and foreign-owned companies. Being a UK subsidiary seems to confer no disadvantage vis-a-vis non-US companies in manufacturing nor vis-a-vis US companies in nonmanufacturing. But US companies have a substantial productivity advantage over all other kinds of company in manufacturing, while other foreign-owned companies have a similar advantage in non-manufacturing.22 In the regressions of Table 5.10, the ownership dummies measure the direct effects of ownership; the indirect effects via input intensity must also be taken into account. We have already seen from Tables 5.8 and 5.9 that ownership has highly significant effects on input intensity.
Why do foreign-owned firms have higher labour productivity*
143
Table 5.10. Effect of ownership on labour productivity (value added per employee): manufacturing versus non-manufacturing companies in 1995 (dependent variable is ln(V/L)) Manufacturing
(D ln(K/L) ln(w) Mage) US NON-US UKSUB N R2
0.1067** (0.0054) 0.9063** (0.0190) -0.0323** (0.0033) 0.1345** (0.0120) 0.0732** (0.0090) 0.0658** (0.0063) 10,061 0.672
Non-manufacturing (2) 0.1234** (0.0040) 0.8530** (0.0116) -0.0612** (0.0039) 0.0928** (0.0154) 0.1403** (0.0113) 0.0838** (0.0072) 21,893 0.682
Source: OneSource. Notes: Companies included are either subsidiaries or independents which do not own any subsidiaries. Constant and dummies for the 60 SIC80 classes included but not reported. Estimated by OLS. Robust standard errors are in parentheses. * Significant at the 5% level. ** Significant at the 1% level. The productivity gaps associated with ownership which are revealed in Tables 5.8 and 5.9 - the 35 per cent lead of US companies over UK independents in manufacturing or their 49 per cent lead in non-manufacturing - are much larger than the direct ownership effects of Table 5.10. We therefore conclude that most of the effects of ownership on productivity are indirect, i.e. foreign ownership leads to higher h u m a n and physical capital intensity and this accounts for the productivity gap. 2 3 So in this respect too, the companies sample confirms conclusions derived from the A R D sample.
INTERPRETING THE FINDINGS: T W O ALTERNATIVES The substantial productivity lead of foreign-owned companies shown
144
Inward investment, technological change and growth
by our results is in line with a large literature stressing the productive effects of foreign investment in the UK (for example, Dunning, 1981; Barrell and Pain, 1997). The obvious question to ask now is, if foreignowned companies, located in Britain and employing British workers, use high human and physical capital intensity to achieve high productivity, why do not British-owned companies do the same? We explore two alternative, not mutually exclusive, hypotheses capable of explaining our findings: first, the cost of capital is lower for foreign-owned companies; and second, foreign-owned companies use superior technology. D o UK-owned companies face a higher cost of capital? The first explanation is that UK-owned companies face a higher cost of capital than foreign-owned ones, for a number of possible reasons. Financial constraints are now widely believed to be an important influence on investment (Caballero, 1997; Chirinko, 1993; Hubbard, 1998). Foreign companies are not presumably constrained to acquire funds for investment from the UK financial system, or at least not to the same extent as UK ones, so deficiencies in the UK system may be hindering investment by UK companies. Foreign companies may also have a lower cost of internal funds (Miles, 1993). An obvious objection to this is that large UK companies are themselves multinationals and face the same global capital market as foreign multinationals. However, the argument may have some force for smaller companies, and it is still possible that, when making investment decisions out of retained profits, even large UK companies are constrained by the short-termist views of the UK stock exchange.24 A second reason why UK firms may have a higher cost of capital is that they face a less favourable risk-return trade-off than foreign ones; consequently they may prefer less capital-intensive technologies. UK companies, even the large multinational ones, almost certainly make a higher proportion of their sales in the UK than do foreign companies. They may be heavily influenced in their investment decisions by the memory of bad experiences in the three long recessions of the past 25 years (the working lifetime of the people now running UK companies). If the UK is perceived as having greater macroeconomic instability than other countries, then, even if UK firms are no more risk averse than foreign ones, they will perceive their overall risk level as higher. By contrast, the large foreign multinationals which operate in UK manufacturing may be better able to balance the risk of poor outcomes
Why do foreign-owned firms have higher labour productivity*
145
in the UK against the chance of good ones elsewhere. Consequently, their preferred capital-labour ratio may be higher. This argument assumes that capital intensive technologies are riskier. This in turn may be justified if investment in physical capital is at least partially irreversible while labour and other inputs may be adjusted at relatively low cost. Empirically the second reason would be hard to distinguish from the first. In both cases, companies would act as if they faced a higher cost of capital, though the reason for the higher cost differs.
Capital-skill complementarity and the cost of capital A lower cost of capital for foreign-owned firms in conjunction with capital-skill complementarity could account for our findings. It has often been argued that capital and skilled labour are complements: see, for example, Griliches, 1965; Fallon and Layard, 1975; Berman et al., 1994; Goldin and Katz, 1998. Assume that competition prevails and that output (value added) is produced under constant returns to scale by means of three inputs: capital, skilled labour and unskilled labour. Assume too that capital costs are lower for foreign-owned firms, while wages for given skills are the same. If capital and skilled workers are complements, then foreign-owned firms will employ a higher ratio of skilled to unskilled workers. This is consistent with what we observe in manufacturing using the ARD sample. The capital-skill complementarity hypothesis implies too that the average wage in foreign-owned companies will be higher than in domestically owned ones, consistent with what we observe in the companies sample. If capital and skilled workers are complements, then capital and unskilled labour must be substitutes. This leads to foreign-owned firms using a higher ratio of capital to unskilled workers. But, even though firms with lower capital costs use more skilled workers, it can be shown that such firms will nonetheless have a higher ratio of capital to total labour (skilled plus unskilled). This is also consistent with observation. These propositions are derived rigorously under the assumption of a translog cost function in Oulton (2000). There it is also shown that the hypothesis implies that the share of capital costs in total costs will be higher for foreign-owned companies. This proposition is tested using the companies sample and support is found for manufacturing, but not for non-manufacturing. Therefore the ultimate reason why UK companies invest less than foreign-owned ones outside of manufacturing remains to be found.
146
Inward investment, technological change and growth
Foreign ownership and macroeconomic instability A lower cost of capital for foreign-owned firms has another implication. UK-owned companies which are damaged by a recession may be unable to invest as much as comparable companies which have been less damaged. A recession may drain a company's financial resources and make it more dependent on an unsympathetic capital market. But foreign-owned companies may be able to rely on the strength of their parent or on a supposedly more sympathetic foreign capital market. We can test this hypothesis with the ARD sample. The ARD does not contain any balance sheet variables which would enable us to measure a company's financial health directly, but we can measure the size of the shock that an establishment suffers in a recession by, for example, the change in output. This suggests testing the following equation: As ln(K,t/L,t)
= p 0 + P A InY^_s + Controls + zlt
<5-4)
The dependent variable is the growth in capital intensity during the course of a boom taking place from time t-s to t. The main explanatory variable is the growth of gross output over the preceding recession from t-s-r to t-s. The controls used are the SIC80 class dummies and a foreign ownership dummy (FOREIGN = 1 if foreign-owned). The growth of capital intensity is used in preference to the growth of capital, since a fall in output may indicate not just bad luck but poor prospects for the company. Such a company might rationally wish to invest less. But there seems no reason, other than financial difficulties, why a company which survives should use a less capital intensive technique just because its future growth prospects are not as good as they may once have appeared. Nevertheless results are presented also for capital stock growth as the dependent variable. The hypothesis to be tested is p 1 > 0: the greater the fall in output during a recession, the lower the rise in capital intensity in the subsequent boom. But we also allow for the possibility that p^ differs between foreign and domestically-owned companies by interacting the output growth variable with the foreign ownership dummy: that is, we assume Pi ~ [P2 + P3 ' FOREIGN]. If foreign-owned companies are less affected by recessions, then the coefficient on this interaction variable will be negative. The results of estimating this equation over two boom-recession pe-
Why do foreign-owned firms have higher labour productivity*
147
Table 5.11. Effect of recessions on capital stock growth in subsequent booms: manufacturing, 1973-93 survivors Dependent variable 1981-9 Growth of capital stock
1975-9
Growth of capital intensity
Growth of capital stock
Growth of capital intensity
FOREIGN
0.072* (0.029)
0.149** (0.031)
0.045* (0.018)
0.030 (0.018)
AlnY
0.297** (0.045)
0.110* (0.048)
0.220** (0.026)
0.179** (0.027)
FOREIGN*A\r iY s -0.047 ' S (0.095)
-0.010 (0.101)
-0.014 (0.058)
-0.100 (0.060)
1700 0.110
1700 0.080
1752 0.101
1752 0.077
N R2
s
Source: ARD. Notes: OLS estimates of equation (5.4). Constant and dummies for SIC80 class included but not reported. The output growth variable (A^lnY s) is the growth in output during the preceding recession, 1979-81 in the case of the 1981-9 boom and 1973-5 in the case of the 1975-9 boom. FOREIGN = 1 if foreignowned. For 1981-9, 52 establishments which switched their SIC80 class over this period were excluded. * Significant at the 5% level. ** Significant at the 1% level. riods are in Table 5 . 1 1 . The first b o o m is 1 9 7 5 - 9 , following a recession from 1 9 7 3 - 5 . The second is 1 9 8 1 - 9 , following the 1 9 7 9 - 8 1 recession. It will be seen that p^ is significantly positive so macroeconomic instability does indeed appear to damage investment, whether capital intensity or capital growth is the dependent variable. However the interaction variable in Table 5.11 is never significant. In other words, foreign-owned firms reduce their investment just as much as UK-owned ones as a result of bad experiences during a recession. This is evidence against the view that the difference in capital intensity between foreign and UK-owned companies is due to the UK financial system. 25 D o f o r e i g n - o w n e d c o m p a n i e s e m p l o y superior t e c h n o l o g y ? The second hypothesis to explain the findings is that foreign companies
148
Inward investment, technological change and growth
are using superior technology and business methods. For this hypothesis to be a complete explanation, these superior technologies must be more intensive in both capital and skilled labour. UK companies may just be slow to learn from and apply the best foreign practice, for several possible reasons. First, the relevant knowledge may be commercially confidential or located in the heads of foreign managers. Second, there may be workforce resistance to change. In the latter case, it might not pay for an established firm to adopt the superior technology because of factors such as the upfront cost of strikes. This will be all the more likely if the firm is a satisficer rather than a maximiser.26 Objections can be raised against this second explanation too. It would seem rather odd if superior technology is in general more intensive in both capital and skilled labour. Some superior business methods, e.g. just in time inventory controls, require less (inventory) capital not more. Also, the larger UK companies at least must be well aware of their foreign rivals' technology and could hire foreign managers if they so desired. And how potent is workforce resistance after the trade union reforms of the 1980s? While there are few areas of manufacturing which are not exposed to foreign competition, the same is not true of services. So the fact that the productivity gap seems to be about the same in manufacturing as in non-manufacturing argues against competition or the lack of it being the explanation. This explanation really boils down to the assertion that TFP levels differ between foreign-owned and UK firms. But we have already seen that, after controlling for differences in input intensities, ownership alone cannot explain very much of the productivity gap (Tables 5.6 and 5.10). So this hypothesis fails to convince.27
Multinationals: a third hypothesis? Doms and Jensen (1998), in their study of US manufacturing, were able to break down their domestically owned (that is, US) firms into those which are multinationals and those which operate only in the home market. They find that the real difference is between multinationals and non-multinationals, not foreign and domestically owned firms. This suggests a third explanation based on the theory of foreign investment. At any moment there is a range of capabilities amongst a country's firms. The better companies develop specific advantages. These allow them to compete successfully in foreign markets and consequently to go multinational (Dunning, 1981). The foreign-domestic productivity gap
Why do foreign-owned firms have higher labour productivity*
149
which we observe simply reflects this process. Indeed, the observed gap is, on this view, rather misleading since the performance of the more successful, domestically-owned multinationals is being obscured by their less successful colleagues who operate only in the home market. Whether this explanation works for the UK as well as for the US is unclear. It would require much more work beyond the scope of this chapter to identify the UK multinationals in the OneSource or ARD databases. But even if some British multinationals have high productivity, they must still represent a comparatively small proportion of UK employment, otherwise we would not find that the employmentweighted mean of productivity is generally lower in UK subsidiaries (Table 5.7). In any case, it is not clear that this third explanation is different from the other two since the specific advantages of multinationals have to show up in some measurable way (e.g. in technology or in the cost of capital).
CONCLUSIONS Results for manufacturing based on the ARD We have developed a dataset of 1,752 establishments which appear continuously in the ARD from 1973 to 1993 inclusive. For each of these survivors, we have been able to estimate the capital stock. Of course these survivors are atypical by virtue of the fact simply that they have survived. Nevertheless they make up about a third of the employment recorded in the ARD, they contain a wide range of sizes (the average is 590 employees), and have a foreign-owned proportion which is similar to the overall figure. Based on this data set, our main results are as follows: 1. There are large differences in capital intensity (the capital-labour ratio), between establishments located in the same 2-digit class (of which there are 22 in manufacturing). These current differences in capital intensity arise solely from differences in cumulated investment over the period 1973-93, since the capital intensity of all establishments in a given class was of necessity assumed to be identical in 1973. 2. The differences between establishments in productivity and capital intensity are not just random. There are systematic factors at work
150
Inward investment, technological change and growth
as well. Foreign-owned establishments, and in particular US-owned ones, substantially outperform UK-owned ones. On average, foreignowned establishments operate with 50 per cent more capital per worker and achieve 38 per cent higher value added per worker. Their labour forces are more white collar and considerably better paid. These disparities in performance cannot be dismissed as due simply to the concentration of foreign ownership in high productivity or high capital intensity sectors. 3. We found that physical and human capital intensity are significant determinants of productivity at a point in time. Our cross-section regressions, which control for industrial structure as well, can account for about a half of the variation across establishments in value added per worker. In addition, US-ownership is found to confer a productivity advantage of between 9 and 20 per cent, over and above the advantage conferred by higher capital intensity in US-owned establishments. 4. The total US advantage in value added per worker is 31.7 per cent, after controlling for industrial structure. The measured inputs, capital intensity and labour quality, can explain 61 per cent of this gap. In the case of other foreign-owned establishments, the labour productivity advantage is lower, 14.6 per cent after controlling for industrial structure, and the measured inputs account for 97 per cent of this gap. Results based on the companies sample The introduction asked, do foreign-owned companies have as big a lead in labour productivity in the rest of the economy as they do in manufacturing? The answer is yes. In fact the lead is larger. After controlling for industrial composition and other factors, US ownership was found to raise productivity by 35 per cent in manufacturing, relative to UK independents; other foreign ownership raises it by 23 per cent. In the rest of the economy, US and other foreign ownership raise productivity by even more, 49 per cent and 46 per cent respectively. In manufacturing, the companies sample and the ARD sample tell much the same story. For both manufacturing and non-manufacturing, the foreign lead over UK subsidiaries is not much lower than the lead over UK independents. The foreign productivity lead can very largely be explained, or at least accounted for, by higher capital per employee and a more skilled labour force.
Why do foreign-owned firms have higher labour productivity*
151
We have suggested that these findings are broadly consistent with the hypothesis that the cost of capital is lower for foreign-owned companies. If, in addition, capital and skilled labour are complements, foreign-owned companies would employ more physical capital and a higher proportion of skilled workers and thus enjoy higher labour productivity on both counts. Oulton (2000), using a cost function approach, finds empirical support for this proposition in manufacturing using the companies sample, but not in non-manufacturing. A further piece of evidence which tends against the cost of capital hypothesis is the investment behaviour of establishments in booms and recessions. Establishments which suffered more severely in the two recessions of 1973-5 and 1979-81 increased their capital intensity less in the subsequent booms than did more fortunate establishments. However, there was no difference in this respect between foreign and UK-owned establishments. In other words, macroeconomic instability seems to have had some adverse effects on investment, but no more so for domestic than for foreign companies. This is some evidence against the view that the shortfall in investment by UK-owned companies is due to the UK financial system. If we ask the question - why is labour productivity so much lower in UK subsidiaries and independents in non-manufacturing - the proximate answer is clear. These companies use substantially lower capital and skill inputs (Table 5.9). It is not surprising therefore that their labour productivity level is much lower. But we have not found an explanation as to why, faced apparently with the same input prices, these companies employ less capital and less skilled labour.
152
Inward investment, technological change and growth
APPENDIX 5A Divisions and classes of SIC80 Division 0 Agriculture, forestry and fishing 01 Agriculture and horticulture 02 Forestry 03 Fishing Division 1 Energy and water 11 Coal extraction and manufacture of solid fuel 12 Coke ovens 13 Extraction of mineral oil and natural gas 14 Mineral oil processing 15 Nuclear fuel production 16 Production and distribution of electricity, gas and other forms of energy 17 Water supply industry Division 2 Metals, mineral products and chemicals 21 Extraction and preparation of metalliferous ores 22 Metal manufacturing 23 Extraction of minerals not elsewhere specified 24 Manufacturing of non-metallic mineral products 25 Chemical industry 26 Production of manufacturing-made fibres Division 3 Metal goods, engineering and chemicals 31 Manufacture of metal goods not elsewhere specified 32 Mechanical engineering 33 Manufacturing of office machinery and data processing equipment 34 Electrical and electronic engineering 35 Manufacture of motor vehicles and parts 36 Manufacture of other transport equipment 37 Instrument engineering Division 4 Other manufacturing industries 41/42 Food, drink and tobacco manufacturing industries 43 Textile industry 44 Manufacturing of leather and leather goods 45 Footwear and clothing industries 46 Timber and wooden furniture industries 47 Manufacturing of paper and paper products; printing and publishing 48 Processing of rubber and plastics
Why do foreign-owned firms have higher labour productivity*
49
153
Other manufacturing industries
Division 5 Construction 50 Construction Division 6 Distribution, hotels and catering, and repair 61 Wholesale distribution (except dealing in scrap and waste materials) 62 Dealing in scrap and waste materials 63 Commission agents 64/65 Retail distribution 66 Hotels and catering 67 Repair of consumer goods and vehicles Division 7 Transport and communications 71 Railways 72 Other inland transport 74 Sea transport 75 Air transport 76 Supporting services to transport 77 Miscellaneous transport services and storage not elsewhere specified 79 Postal services and telecommunications Division 8 Banking, insurance and real estate 81 Banking and finance 82 Insurance, except for compulsory social security 83 Business services 84 Renting of movables 85 Owning and dealing in real estate Division 9 Social and personal services 91 Public administration, national defence and compulsory social security 92 Sanitary services 93 Education 94 Research and development 95 Medical and other health services: veterinary services 96 Other services provided to the general public 97 Recreational services and other cultural services 98 Personal services 99 Domestic services
154
Inward investment, technological change and growth
APPENDIX 5B The companies sample Table 5B.1 Summary statistics for variables and sample used in the regression analysis, 1995
Variable
Ownership
Manufacturing Mean N s.d.
Non-imanufacturing N Mean s.d.
ln(K/L)
954 US Other foreign 1,720 UK subsidiary 4,181 UK independent 3,206
2.725 2.812 2.289 2.346
1.008 1.078 1.101 0.983
1,265 3,448 7,687 9,493
2.302 2.396 2.157 2.303
1.473 1.609 1.600 1.365
\n(w)
954 US Other foreign 1,720 UK subsidiary 4,181 UK independent 3,206
0.337 0.341 0.359 0.417
1,265 3,448 7,687 9,493
3.158 3.008 2.716 2.632
0.533 0.630 0.569 0.602
ln(WL)
954 US Other foreign 1,720 UK subsidiary 4,181 UK independent 3,206 954 US Other foreign 1,720 UK subsidiary 4,181 UK independent 3,206
2.860 2.813 2.686 2.666 3.384 3.290 3.094 3.020 5.290 4.967 4.725 3.977
1,265 3,448 7,687 9,493 1,265 3,448 7,687 9,493
3.742 3.685 3.301 3.111 4.236 3.789 4.073 3.345
0.790 0.976 0.851 0.729 1.643 1.655 1.555 1.192
954 US Other foreign 1,720 UK subsidiary 4,181 UK independent 3,206
8.673 8.257 7.819 6.997
0.502 0.494 0.482 0.480 1.464 1.397 1.283 0.915 1.514 1.430 1.316 0.863
1,265 3,448 7,687 9,493
7.978 7.473 7.374 6.456
1.627 1.590 1.489 1.041
ln(L)
ln(V)
Source: OneSource. Notes: Value added (V), capital (K) and the wage (w) are in units of thousands of pounds; labour (L) is number of employees. The wage is calculated as the wage bill divided by the number of employees. Data are for 1995.
Why do foreign-owned firms have higher labour productivity*
155
Table 5.B2 Employment by ownership type, 1995 Manufacturing Ownership
Companies number
954 US Other foreign 1,720 UK subsidiary 4,181 UK independent 3,206 10,061 Total
Employment number % 596,559 17.5 846,286 24.9 1,690,324 49.7 269,871 7.9 3,403,043 100.0
Non-manufacturing Companies Employment number number % 1,265 3,448 7,687 9,493 21,893
522,501 1,058,736 3,014,726 639,990 5,235,952
10.0 20.2 57.6 12.2 100.0
Source: OneSource.
NOTES Fuller versions of the research reported in this chapter will be found in two discussion papers, Oulton (1998b, 1998c). A revised and expanded version of the second discussion paper is Oulton (2000). The research presented in the first discussion paper was financed by the Department of Trade and Industry, that in the second by the Leverhulme Trust as part of a wider project entitled Job generation in the corporate sector [F/59/AD]. To both of these I owe thanks. This chapter has benefited from helpful comments on earlier versions from Martin Baily, Peter Hart, Nicholas Owen, Nigel Pain, Martin Weale and participants at the NIESR Conference on Foreign Investment in September 1999; I am grateful to all of these. I would also like to thank the staff of the Office for National Statistics at Newport, in particular Wendy Fader, for assistance in using the ARD. This research was carried out at the National Institute of Economic and Social Research prior to my taking up an appointment at the Bank of England. The views expressed are my own and are not necessarily those of any of the above persons or of the Bank of England. A 'selected' establishment is one which is required by law to fill in a return. Apart from being in scope to the inquiry, an establishment must be of a certain size, which normally means employing twenty or more people, to be selected. It is not clear without further research how much difference this change made. For clarity, and because most of the data used here were collected before the change to company-based reporting, we
156
Inward investment, technological change and growth
continue to use the term 'establishment'. Note that in American usage an establishment is a plant and this should be borne in mind when comparing the present results with US ones. 4 For classes 2 1 , 36 and 44 no PPI exists, so the PPI for manufacturing as a whole was used. These deflators were obtained from Datastream. 5 In addition there is the small category of 'working proprietors'; where available this has been amalgamated with ATCs. 6 Introduction to the Summary Volume of the Census of Production, various issues. 7 I am grateful to my former colleague at the National Institute of Economic and Social Research, Mary O'Mahony, for supplying me with these capital-labour ratios. The initial capital stocks for each class have been derived by the perpetual inventory method using similar but more detailed assumptions about depreciation rates to the one in the text. 8 The earliest year in the ARD is 1970 but the data for the years 197072 were not in a form amenable to analysis when the research was carried out. 9 If we disaggregate further, industrial structure will of course explain more of the variation. The 1,752 survivors fall into 198 four-digit activity headings. Analysis of variance using activity headings instead of classes shows that industrial structure can explain 58 per cent of the variance of the log of capital intensity in 1993, but the average number of establishments per activity heading is now less than nine, so the additional explanatory power is rather spurious. If we could disaggregate still further, eventually everything could be explained by industrial structure. 10 Foreign-owned establishments include those deemed to be controlled by enterprises incorporated overseas, as well as those which are wholly owned. 11 Doms and Jensen (1998) and Globerman et al. (1994) report similar findings for the US and Canada respectively, namely that foreign-owned firms have higher labour productivity, higher capital intensity and use more skilled labour and that these differences remain after controlling for industrial composition and size. As regards the role of capital intensity, note that these authors are forced to use proxies: book value in the case of Doms and Jensen and energy input in the case of Globerman et al. In this respect, the results in the present chapter may be regarded as stronger, at least
Why do foreign-owned firms have higher labour productivity*
157
for manufacturing. 12 Results including employment size are in Oulton (1998b, Table 11). The role of size is discussed further below. 13 If size (log employment) is included as an additional regressor in equation (5.3), its coefficient is significant and positive except in 1989 when wages are included. Its inclusion has little effect on the coefficient on capital intensity. The US ownership dummy remains significant, except in 1993 with wages included, but reduced in size. This suggests that part of the reason for US success may be that USowned establishments are larger (see Table 5.2). But it could equally well be argued that high productivity companies are able to become larger than their less productive competitors. 14 We excluded 1,104 companies whose accounts predated 1994. A few other companies were also excluded since they claimed to have zero employees, even though supposedly actively trading. 15 The accounts for a holding company would normally be consolidated, i.e. they would incorporate the results of its subsidiaries. In some cases, the results of foreign subsidiaries may be included in the accounts of UK-based holding companies. Our procedure ensures that such results are also excluded. For example, if a UK-owned company owns five UK subsidiaries, we include the five subsidiaries but not the holding company. 16 The maximum sample for which employment is available is 49,009 companies. These collectively employed 10.775 million people: see Appendix B, Table Bl of Oulton (1998c). 17 The detailed data are in Oulton (1998c, Appendix B, Table B2). 18 Table 5.8 excludes companies which had losses in 1995; that is, companies where the wage bill exceeded value added; there were 4,272 such companies. For comparable results where such companies are included, see Tables 2 and 3 of Oulton (1998c). 19 US companies are significantly larger in employment terms than all other types of company in both manufacturing and non-manufacturing (Appendix 5B, Table 5B.2) and this remains true even after controlling for industrial composition. Other foreign-owned companies are larger than UK independents and UK subsidiaries in manufacturing; in non-manufacturing, while being considerably larger than UK independents, they are a little smaller than UK subsidiaries. Hence it could be argued that some of the advantages of foreign ownership which we find may really be advantages of size. We therefore tried to test for the presence of economies of scale by
158
20
21
22
23
24 25
26
Inward investment, technological change and growth
including two measures of size, employment and value added, both in logs. These two measures, while highly significant, have opposite signs: the coefficient on log employment is negative (except in the equation for K/L in manufacturing) while that on log value added is positive. Since economies of scale cannot be both increasing and decreasing, these variables cannot be measuring economies of scale. Employment is in the denominator of each of the dependent variables in Table 5.8, while value added is related to the numerator. So the effects captured by these variables are probably spurious (division bias). When companies making losses are included, the advantages of foreign ownership are reduced somewhat while still remaining very substantial, and the gap between UK subsidiaries and UK independents virtually disappears (see Tables 2 and 3 of Oulton, 1998c). As mentioned above, since human capital is proxied by the wage, there is the possibility of reverse causation here: high productivity may lead to high wages through rent-sharing or union bargaining power. Similar results were obtained with sales per employee as the dependent variable but the foreign ownership effects are larger. This suggests that foreign ownership affects the extent to which companies use intermediate input. In other words, foreign-owned companies tend to be more reliant on outsourcing. We also tested for scale effects by including either the log of employment or the log of value added. Once again, these variables had opposite signs, negative on employment and positive on value added. Hence they cannot be interpreted as measuring economies of scale. These results are in Oulton (2000, Table 5). Oulton (2000) takes this argument a bit further by calculating TFP differences between companies. These differences are found to be only weakly correlated with ownership. I owe this point to Steve Bond. The conclusion is not altered if separate dummies are introduced for US and non-US ownership. Note that during the 1979-81 recession the mean fall in output was about the same for foreign as for UK establishments. Baily and Gersbach (1995) argue that the crucial factor in inducing firms to adopt best practice technology is exposure to global, not just local or regional, competition. See also Nickell (1995, chapter 4, and 1996) on the beneficial effects of competition on
Why do foreign-owned firms have higher labour productivity*
159
productivity. 27 Oulton (2000) presents estimates of TFP levels for the companies sample, using two alternative methods. This study finds that TFP in UK subsidiaries is similar to that of all foreign-owned companies in non-manufacturing and to that of US companies in manufacturing; TFP is lower in UK independents. Even for the latter group TFP differences account for less than a third of the labour productivity gap with US-owned companies.
REFERENCES Baily, M.N. and Gersbach, H. (1995), 'Efficiency in manufacturing and the need for global competition', Brookings Papers on Economic Activity: Microeconomics 1995, pp. 307-47. Barrell, R. and Pain, N. (1997), 'Foreign direct investment, technological change, and economic growth within Europe', Economic Journal, 107 (November), pp. 1770-86. Berman, E., Bound, J. and Griliches, Z. (1994),'Changes in the demand for skilled labor within U.S. manufacturing: evidence from the Annual Survey of Manufacturing', Quarterly Journal of Economics, 109 (May), pp. 367-97. Caballero, R.J. (1997), 'Aggregate investment', NBER Working Paper No. 6264, Cambridge, Mass., National Bureau of Economic Research. Chirinko, R.S. (1993), 'Business fixed investment spending: a critical survey of modelling strategies, empirical results, and policy implications', Journal of Economic Literature, 31 (December), pp. 1875-1911. Davies, S.W. and Lyons, B.R. (1991), 'Characterising relative performance: the productivity advantage of foreign owned firms in the UK', Oxford Economic Papers, 43 (October), pp. 584-95. Doms, M.E. and Jensen, J.B. (1998), 'Comparing wages, skills, and productivity between domestic and foreign owned manufacturing establishments in the United States', in Baldwin, R., Lipsey, R. and Richardson, J. (eds), Geography and Ownership as Bases for Economic Accounting, Chicago, University of Chicago Press. Dunning, J.H. (1981), International Production and the Multinational Enterprise, London, George Allen and Unwin. Fallon, P.R. and Layard, P.R.G. (1975), 'Capital-skill complement-
160
Inward investment, technological change and growth
arity, income distribution, and output accounting', Journal of Political Economy, 83, 2, pp. 279-301. Fraumeni, B. (1997), 'The measurement of depreciation in the U.S. national income and product accounts', Survey of Current Business, July, pp. 7-23. Globerman, S., Ries, J.C. and Vertinsky, I. (1994), 'The economic performance of foreign affiliates in Canada', Canadian Journal of Economics, 27, 1 (February), pp. 143-56. Goldin, C. and Katz, L.F. (1998), 'The origins of technology-skill complementarity', Quarterly Journal of Economics, 113, 3, pp. 693-732. Griffith, R. (1999a), 'Using the ARD establishment level data to look at foreign ownership and productivity in the United Kingdom', Economic Journal, 109, pp. 416-42. (1999b), 'Productivity and foreign ownership in the UK car industry', IFS Working Paper no. 99/11. Griliches, Z. (1965), 'Capital-skill complementarity', Review of Economics and Statistics, 51, pp. 465-8. Hart, P.E. and Oulton, N. (1996), 'Growth and size of firm', The Economic Journal, 106, pp. 1242-52. (1999), 'Gibrat, Galton and job generation', International Journal of the Economics of Business, 6, 2, pp. 149-64. Hubbard, G. (1998), 'Capital-market imperfections and investment', Journal of Economic Literature, 36 (March), pp. 193-225. Miles, D. (1993), 'Testing for short termism in the UK stock market', Economic Journal, 103, pp. 1379-96. Nickell, S. (1995), The Performance of Companies, Oxford, Blackwell. (1996), 'Competition and corporate performance', Journal of Political Economy, 104, pp. 724-46. O'Mahony, M. (1999), Britain s Productivity Performance 1950-1996: An International Perspective, London, National Institute of Economic and Social Research. Oulton, N. (1997), 'The ABI Respondents Database: a new resource for industrial economics research', Economic Trends, 528 (November), pp. 46-57. (1998a), 'Competition and the dispersion of labour productivity amongst UK companies', Oxford Economic Papers, 50, pp. 23-38. (1998b), 'Investment, capital and foreign ownership in UK manufacturing', National Institute of Economic and Social Research Discussion Paper No. 141 (August).
Why do foreign-owned firms have higher labour productivity*
161
(1998c), 'Labour productivity and foreign ownership in the UK', National Institute of Economic and Social Research Discussion Paper No. 143 (September). (2000), 'Labour productivity and foreign ownership in the UK', mimeo. [A revised and expanded version of Oulton, 1998c]
6
Regional policy and the impact of FDI in the UK Nigel Driffield
INTRODUCTION The purpose of this chapter is to contribute to the evaluation of the impact of inward investment on the UK economy. In particular we aim to address the extent to which any spillovers from foreign firms to domestic firms are dependent upon the geographic locations of inward investors and domestic firms. As is well documented, inward investment can bring benefits to regions as well as to industries. This has been reflected in the structure of industrial and regional policies. Indeed it could be argued that the desire to attract foreign-owned firms has been the chief industrial and regional policy of the UK authorities over the past twenty years (Eltis and Fraser, 1992; Eltis, 1996). In addition to the obvious positive direct impact that new investments may have on employment, the reasoning behind the importance attached to inward investment is essentially based on the assumption that foreign-owned firms are in some sense 'superior' to domestic firms and that their proprietary technological or managerial capabilities can somehow be assimilated by the domestic sector. At a regional level, new foreign manufacturing investment can have additional beneficial economic consequences, particularly within disadvantaged or peripheral areas. In addition to job creation and resource transfer, foreign inward investments can also provide a conduit for new trading opportunities, and transfers of technology and skills to supplier and customer sectors. Dunning (1993) demonstrates how inward investment can improve both inter- and intra-sectoral allocative efficiency by shifting scarce resources to industrial sectors of above average comparative advantage and to sectors where comparative advantage is growing. The entry of new foreign
162
Regional policy and the impact of FDI in the UK
163
firms can also contribute to agglomeration economies at the industrial and regional levels, affecting both the extant foreign and indigenous manufacturing sectors (Cantwell, 1991). The significance of spillovers from multinational enterprises in the UK was demonstrated by Barrell and Pain (1997), who estimated that around 30 per cent of the productivity growth in UK manufacturing between 1985 and 1995 could be attributed to the impact of inward investment. The 'ripple through' effects of changes in production and working practices triggered by the presence of new inward investors have been particularly important. There are thus grounds to expect that there could be productivity gains at the industry and regional level connected with foreign investment in those industries and regions. However this must be investigated by controlling for other factors which may improve the productivity and comparative advantage of industries independently of inward investment. Of course it may be that the design of regional policies has not allowed the full potential benefits of inward investment to emerge. As Morgan (1997) notes, the policy of attracting FDI has been designed to address the symptoms of regional disparities, such as unemployment, rather than the underlying causes, such as low levels of technological development. An important question therefore is the extent to which inward investment will contribute to alleviating the underlying causes of regional underdevelopment, as well as simply generating employment. One criticism of policymakers in the UK and other countries is that new investments have contributed less to industry productivity growth than may otherwise have been the case because incentives have been given to encourage investors to locate in regions with high levels of unemployment rather than regions demonstrating significant locational advantages (Porter, 1996). This potential problem has seldom been addressed in the literature, as studies that have sought to investigate the regional impact of inward investments have often been unable to distinguish between inter-industry, and intra-industry effects. The research described here and in Driffield (2000b) overcomes such difficulties, and can therefore contribute to the policy debate by evaluating the scale of regional spillovers from inward investment and also the extent to which these spillovers are dependent on investment in the region, the industrial mix of the region, and on contiguity to other regions in which investment takes place. The extent to which spillovers are determined by the geographic location of the investments is largely a function of the types of policies that are pursued to attract inward investment.
164
Inward investment, technological change and growth
T H E P A T T E R N O F F O R E I G N I N V E S T M E N T I N T H E UK In this section we briefly summarise some of the important trends in the share of investment and employment accounted for by foreign-owned firms in the UK manufacturing sector. As in Chapter 3 in this volume, by Hubert and Pain, we use a data set covering 1983-92. In all, we consider eleven standard planning regions and twenty (two-digit level) manufacturing industries. By 1992 foreign-owned firms accounted for 17.9 per cent of manufacturing employment and 31.6 per cent of manufacturing gross fixed capital formation. Indeed net capital expenditure by foreign firms more than doubled through the 1980s, compared with an increase of approximately 60 per cent in expenditure by domestic firms. The annual distribution of total investment is shown in Figure 6.1. By 1992 foreign investment had increased to nearly half of the domestic level, compared with around a third just eight years previously. The rise in the share of investment undertaken by foreign firms was faster than the rise in their share of employment during this period. This is possibly because their greater capital intensity was associated with faster growth in their labour productivity, as discussed by Nick Oulton in Chapter 5 of this volume. The share of employment in foreign-owned firms is shown for sixteen industries in Figure 6.2. The importance of foreign firms is especially apparent in sectors such as office equipment (SIC 33), motor vehicles and components (SIC 35) and chemicals (SIC 25). The industrial concentration of foreign firms has remained fairly stable over time. All the industries with above average shares of employment in foreign-owned firms in 1992 also had above average shares in 1983. The regional distribution of gross fixed capital formation by foreignowned firms is shown in Figure 6.3. The most important host is the South-East of England.1 However, there was little growth in the absolute level of investment there during 1988-92. In contrast, investment by foreign firms grew dramatically in some other regions, most notably in the East Midlands and the North of England. Wales and Scotland were also able to attract levels of investment out of proportion to their size, and have the greatest levels of foreign ownership in the UK. This is shown in Figure 6.4, which reports the share of manufacturing employment in each of the eleven standard planning regions of the UK. The foreign share increased over the period shown in every region, with the exception of Northern Ireland. The rise in the foreign share is probably as attributable to a decline in the level of domestic manufac-
Regional policy and the impact of FDI in the UK
LOOOO
5 § ^
6000 5000 4000
1984 1985 1986 1987 1988 1989 1990 1991 1992 • domestic investment • foreign investment Figure 6 . 1 . Domestic
and foreign fixed capital
investment
0.7 0.6 h
0.5 0.4 0.3 h
0.2 L
1
IrJ-iB
Industry (SIC80) 11983 B 1 9 8 6 D 1 9 8 9 0 1 9 9 2 Figure 6.2. Share of foreign employment
by industry
(per cent)
165
166
Inward investment, technological change and growth
800 700 V
^
400
,
• 1984 D1988 01992 Figure 6.3. Foreign fixed investment expenditure by UK region
0.25 0.2 -
0.15 0.1
3
is
^
West
T3
X3 C o3 T3
Hum
0.05
x -s * z
o
<3
rt
2
>•
• 1984 D 1992 Figure 6.4. Foreign share of manufacturing employment by region (per cent)
Regional policy and the impact of FDI in the UK
167
turing employment as it is to an increase in inward investment. However, investment incentive schemes may also have been important. Wales and Scotland are commonly thought to have two of the most active development agencies. The increasing effectiveness of the Development Agencies in the West Midlands can also be seen.
THE C H A N G I N G PATTERN OF INWARD INVESTMENT A N D T H E P O T E N T I A L IMPACTS OF FDI Many regions of the UK, in common with most of those in other European countries, have Development Agencies. Their mission statements are generally very broad, but in practice such agencies have come to see their primary function as marketing the region in order to attract internationally mobile investments. Hill and Munday (1994) demonstrate that domestic regional inward investment incentives have influenced private investment rates across the UK regions. Wren and Taylor (1999) go one stage further, demonstrating that investment incentives have had the effect of changing the regional distribution of economic activity across the UK in general. Nijkamp and Blaas (1995) demonstrate that grants from the European Regional Development Fund have resulted in related effects for many European regions. Previous evidence on FDI and domestic productivity growth Given the changing pattern of fixed investment in the UK and the increasing provision of inward investment incentives, it is important to examine the benefits that may be generated by these investments. In addition to the potential for spillovers due to technology transfer, the work of Dunning (1993) suggests a number of additional sources of productivity gains at the industry level connected with intra-industry foreign investment. Non-technological advantages, such as managerial abilities, the exploitation of scale economies, or a superior coordination of resources, may all act to improve the performance of indigenous industries. Many of these can be imparted through so-called 'demonstration effects' (Caves, 1996). Local firms can learn better management techniques or develop coordination economies as a result of observing the behaviour of inward investors. The often cited 'Japanisation of UK industry' is a case in point, with changes in production and working
168
Inward investment, technological change and growth
practices having been triggered by the presence of new Japanese plants (Eltis and Fraser, 1992). Blomstrom and Kokko (1998) highlight the fact that technology transfers may be direct, via the licensing of a particular technology or through supplier networks and subcontracting arrangements, or indirect as knowledge becomes public. There is also a possibility that technology 'spills over' in a less formal manner to domestic firms, as the literature on technology externalities indicates.2 All these suggest that there could be productivity gains at the intra-industry and intraregional levels connected with foreign investment, as well as in other related industries and contiguous regions. The significance of spillovers from multinational enterprises in the UK was demonstrated by Barrell and Pain (1997), who estimated that around 30 per cent of the productivity growth in the UK manufacturing sector between 1985 and 1995 could be associated with the impact of inward investment. Perhaps the most optimistic results concerning region-specific benefits are reported by Young etal. (1988), and Neven and Siotis (1993), who contend that the multipliers associated with inward investment are extremely high, at least in some regions. The Barrell and Pain study and many regional studies focus on aggregate productivity growth and will thus include any 'batting average' effects that result from new entry, either by foreign or domestic companies. However there are studies which have shown that large inward investing groups can attract and promote the development of supplier clusters, and also play a role in the development of the social and physical infrastructure of the host region (Peck, 1996; Morgan, 1997). Equally inward investment can act to improve domestic performance, and improve revealed comparative advantage (Driffield and Munday, 2000). This suggests however, that for technology spillovers to be assimilated by the domestic firms, the domestic sector must be sufficiently technologically advanced. Although this is usually considered a problem when assessing the potential impact of inward investment in developing countries, there is no reason why it should also not matter for inward investment in industrialised countries such as the UK (Driffield, 2000a). Spillovers may be enhanced if agglomeration economies are present. Such economies arise from the geographical proximity of similar enterprises. The presence of multinational firms, as leaders in both technological and capital accumulation, will serve to stimulate further the possibility for agglomeration in such locations. This may in turn
Regional policy and the impact of FDI in the UK
169
increase the potential for technology transfer, and therefore improvements in the technological capabilities of domestic firms. Cantwell (1991) reports that there are significant agglomeration economies, both to foreign and domestic firms, related to the proximity of similar facilities. Braunerhjelm and Svensson (1996), in a study of Swedish outward FDI, suggest that the location decisions of Swedish firms are strongly influenced by the extent to which host countries had significant production in their industry. Agglomeration influences were found to be particularly strong in technology-intensive sectors. These findings suggest that the level of regional development gains from FDI will be linked to the presence of regional agglomerations. A strong foreign presence in an industry, particularly one which has a high level of regional concentration, is likely to increase the likelihood of knowledge spillovers and improve the comparative advantage of the industry. This is a common argument in the study of inward investment in developing countries (Rodriguez-Clare, 1996; Aitken and Harrison, 1999). H o w localised are spillovers? As Lovering (1999) shows, many Development Agencies in the UK regions appear to have fully embraced (for good or ill) the 'new regionalism' approach to development. This has not only resulted in a greater dependence on inward investment, but has also encouraged inter-firm links between foreign and indigenous companies within their region. Thus there is significant evidence that agencies are not only actively competing to attract international capital3 but are also attempting to contain any spillovers from FDI within their region.4 In effect the agencies are seeking to maximise agglomeration economies from inward investment, possibly to the detriment of other regions and the UK economy as a whole, although the success of these policies is open to question (Turok, 1993,1997; Phelps, 1993; Shirlow, 1995). As Porter (1996) shows, the potential costs of these policies are likely to be exacerbated where the location advantage has taken the form of an investment subsidy, rather than endowment based sources of comparative advantage. This is not to suggest that agglomeration economies from FDI do not help to generate productivity growth and regional development (Driffield and Munday, 2000). It is simply that maximising benefits at the regional level may have an opportunity cost which is felt elsewhere. This may occur either as a result of employment
170
Inward investment, technological change and growth
changes (Driffield, 1999), or in terms of limited investment spillovers. It may also be the case that spillovers are generic in the UK so that the effects of inward investment on productivity growth are independent of the location decisions of multinational enterprises. Regional agglomeration economies may also act to limit the extent to which spillovers are transferred across regions (Beeson, 1987; Englemann and Walz, 1995). The strategies employed by multinational firms in the management of their manufacturing processes will also influence linkages and therefore the extent to which such investments contribute to domestic productivity growth in the region (Brand, 1998; Young etal., 1994). Spillovers may also be limited if there are good reasons for firms to locate close to one another, which is the case for many buyer-supplier partnerships, as Twomey and Tomkins (1996) illustrate. Aitken etal. (1997) make a similar point concerning spillovers from FDI in Mexico. It is often the case that multinational firms either look to establish links with a high quality local supplier to maximise the benefits of modern production techniques or, in the absence of such a producer, will source components from established suppliers in other countries. In this case one of the potential channels for technology spillovers will be lost. In either case there is limited potential for interregional spillovers and the benefits of inward investment may be concentrated solely in host regions. The potential negative impacts of FDI on domestic productivity There may also be circumstances in which inward investment has an adverse effect on domestic productivity. For instance De Mello (1999) reports a negative relationship between domestic productivity growth and inward investment which he ascribes to a 'capital deepening' problem. Greater capital investment may raise labour productivity but reduce the growth rate of total factor productivity. Aitken and Harrison (1999) also highlight the potential for negative 'spillovers', but through a very different route. In fact they suggest that FDI has two simultaneous effects on productivity growth. The first is the standard transfer of technologies and skills to domestic sectors. The second is a 'competitive' effect, which can be described as follows. Positive spillovers occur, improving productivity and moving domestic firms on to a lower average cost curve; however at the same time, the increased
Regional policy and the impact of FDI in the UK
171
output of foreign firms has the effect of reducing the scale of operation of domestic firms. This potentially moves firms back along their (new) average cost curve, thus reducing productivity. Thus at the industry level inward investment may have a negative impact on productivity growth. Aitken and Harrison illustrate these effects using firm-level data for Venezuela. It is unclear whether such effects would be present in countries such as the UK. On the one hand local firms in the UK are likely to be more able to assimilate foreign technologies than in a developing country, so that the initial cost improvements may be greater. However it is also likely that domestic and foreign-owned firms in the UK are more direct competitors than is the case in a developing economy, so that the competitive effect may be expected to be stronger.
EMPIRICAL EVIDENCE O N R E G I O N A L SPILLOVERS The discussion above suggests a number of possible hypotheses that might usefully be tested. In particular: • • •
whether there are positive productivity spillovers from inward investment; whether any spillovers are confined to the host region; whether there is a negative impact at the industry level, based on increased competition.
There is a relatively large literature, following Caballero and Lyons (1989,1990,1992), which essentially seeks to encompass the spillovers at an industry level that occur as a result of an increase in external output expansions. The appropriate method for testing these phenomena is to use panel data, and adopt a dynamic framework. This involves estimating a standard production function model in which the output of domestic firms is modelled in terms of a vector of inputs and a vector of potential externalities, in this case various measures of inward investment. Driffield (2000b) provides a more detailed discussion and the full set of results which are summarised below.5 In order to evaluate the extent to which spillovers from FDI occur within and across particular industries, and within and across particular regions, we utilise five separate measures of inward investment, although it is clear that these measures should not be seen as represent-
172
Inward investment, technological change and growth
Table 6.1. Measure of 'inward investment' FDIrt_x FDIINDit_ FDIREGrt_1 FDIRI.t^ FDICR.rt t
Foreign fixed capital investment in the relevant industry (i) and region (r) in the previous year. Foreign investment in the relevant industry for the whole of the UK in the previous year. Total foreign investment in the host region in the previous year. Foreign investment in related industries for the whole of the UK in the previous year.(a) Foreign investment in the relevant industry in contiguous regions in the previous year.
Note: The related industries are determined using industry and occupational classifications. Industries that employ large numbers of similar occupation groups or employ similar technologies are classed as related industries.
ing mutually exclusive, or competing hypotheses. The five measures are described in Table 6 . 1 ; all use fixed capital investment data. The results are presented at length in Driffield (2000b). However, to summarise, there are four important findings: •
•
•
•
a significant positive effect on the productivity of domestic firms in the same industry and the same region as the investor. Thus there are intra-regional spillovers from inward investment; a significant negative effect on domestic firms from inward investment elsewhere in the UK in the same industry. This is consistent with the presence of a scale effect with greater competition causing output to decline, at least in the short term, and productivity to d r o p ; a marginally significant positive effect on domestic firms from inward investment in all other industries in the same region. Thus there are weak, indirect demonstration effects; there are positive but insignificant effects from inward investment in related industries and contiguous regions.
These results provide evidence that there are significant productivity spillovers from inward investment. However in the short term at least they are small and the externalities are assimilated only at the local level, suggesting that the local development agencies have succeeded in confining spillovers to the host region. There is no evidence of any general industry level productivity growth spillover from FDI. Indeed there is
Regional policy and the impact of FDI in the UK
173
expected negative effect at the industry level, confirming the results of De Mello (1999) and Aitken and Harrison (1999). The two investment variables that are positive and significant are the measure of foreign capital in the relevant industry and host region, and foreign investment in all other industries in the host region. An increase of 1 per cent in (the flow of) foreign capital investment in a particular industry and region will initially generate an increase of 0.042 per cent in the productivity of domestic firms in that region and industry. However it will initially reduce the productivity of domestic firms in that industry in other regions by 0.029 per cent. An increase of 1 per cent in foreign investment in all industries in a region will initially raise the productivity of domestic firms in that region by 0.07 per cent. All these effects will become somewhat larger over time as the model estimated includes a lagged dependent variable.
CONCLUSIONS The policy implications of these results are clear. From the perspective of the regional development agencies, the results can be seen as a justification of their approach to attracting inward investment. It is clear that there are gains at the regional level from attracting FDI, that spillovers occur, and that indigenous firms benefit from agglomeration economies generated by the presence of inward investors. There is also more tentative evidence that these spillovers are transferred across industries within the recipient region. However there is also evidence that these gains are not appropriated at the industry level. Indeed there is a loss in productive efficiency in the recipient industry throughout the rest of the UK. This 'competition' effect from FDI suggests that there are losses as well as gains in indigenous industry as a result of inward investment, concurring with the results reported by Driffield (1999). The interpretation of this is, however, open to debate. It may be argued for example that such an effect is in the long-term interests of the UK, as domestic industry is forced to become more efficient in order to compete with foreign firms. An alternative view is that the effect simply represents the 'crowding out' of otherwise efficient domestic producers as a result of competition from multinational firms with considerable market power and who have also benefited from public subsidies. As such, therefore, one may expect losses in productivity and employment in regions other than the host
174
Inward investment, technological change and growth
region, as a result of FDI. This suggests that, from the perspective of the individual regional development agencies, an optimal strategy is to attract as much FDI as possible. In contrast central government should perhaps be concerned that inward investment incentives are being used merely to move jobs around the UK These results also show that the concerns demonstrated by Porter (1996) among others have some credence. To the extent that the location of foreign investment is affected by investment subsidies, those subsidies have had the effect of changing the regional distribution of economic activity across the UK. Equally, the overall impact on UK productivity is reduced still, by a smaller, but significant, negative effect at the industry level. It clearly matters where the investment is located if the potential for spillovers or agglomeration economies is to be realised. Forcing new investments away from core related activities may stimulate development at the local level but reduce the opportunities for greater development elsewhere.
NOTES 1 This is also true for patenting, as shown in Chapter 8 in this volume, by Cantwell, lammarino and Noonan. 2 Griliches (1992) provides an excellent survey of this literature. 3 See, for instance, Lovering (1999), Young etal. (1994) and Gripaios etal. (1997). 4 Initiatives such as 'source Wales' are an explicit example of this, as Morgan (1997) outlines in some detail. 5 The data used is at the two-digit level for the manufacturing sector.
REFERENCES Aitken, B.J. and Harrison, A.E. (1999), 'Do domestic firms benefit from direct foreign investment? Evidence from Venezuela', American Economic Review, 89, 3, pp. 605-18. Aitken, B.J., Hanson, G.H. and Harrison, A.E (1997), 'Spillovers, foreign investment and export behaviour', Journal of International Economics, 43, 1-2, pp. 103-32. Barrell, R. and Pain, N. (1997), 'Foreign direct investment, technologi-
Regional policy and the impact of FDI in the UK
175
cal change, and economic growth within Europe', Economic Journal, 107, pp. 1770-86. Beeson, P. (1987), 'Total factor productivity growth and agglomeration economies in manufacturing, 19'59-1973', Journal of Regional Science, 27, 2, pp. 183-99. Blomstrom, M. and Kokko, A. (1998), 'Multinational corporations and spillovers', Journal of Economic Surveys, 12, pp. 247-78. Brand, S. (1998), 'Supply chains, material linkage and regional development: a comment', Urban Studies, 35, 4, pp. 769-73. Braunerhjelm, P. and Svensson, R. (1996), 'Host country characteristics and agglomeration in foreign direct investment', Applied Economics, 28, 10, pp. 833-40. Caballero, R.J. and Lyons, R.K. (1989), 'The role of external economies in US manufacturing', NBER working paper no. 3033. (1990), 'Internal versus external economies in European industry', European Economic Review, 34, 4, pp. 805-30. (1992), 'External effects in US procyclical productivity', Journal of Monetary Economics, 29, 2, pp. 209-25. Cantwell, J.A. (1991), 'The international agglomeration of R&D', in Casson, M.C. (ed.), Global research strategy and international competitiveness, Oxford, Basil Blackwell. Caves, R.E. (1996), Multinational enterprise and economic analysis, Cambridge, Cambridge University Press. De Mello, L.R. (1999), 'Foreign direct investment-led growth: evidence from time series and panel data', Oxford Economic Papers, 51, pp. 133-51. Driffield, N.L (1999),'The indirect employment effects of FDI into the UK', Bulletin of Economic Research, 5 1 , 3 , pp. 207-22. (2000a), 'The impact on domestic productivity of inward investment in the UK', Manchester School, forthcoming. (2000b), 'Regional and industry level spillovers from FDI', Cardiff Business School Discussion Papers in Economics, 2000:021. Driffield, N.L. and Munday, M.C. (2000), 'Industrial performance, agglomeration, and foreign manufacturing investment in the UK', Journal of International Business Studies, forthcoming. Dunning, J.H. (1993), Multinational enterprises and the global economy, Reading, Addison Wesley. Eltis, W. (1996), 'How low productivity and weak innovativeness undermines UK industrial growth', Economic Journal, 106, pp. 184-95.
176
Inward investment, technological change and growth
Eltis, W. and Fraser, D. (1992), 'The contribution of the Japanese industrial success to Britain and Europe', NatWest Bank Economic Review, pp. 2-19. Englemann, F.C. and Walz, U. (1995), 'Industrial centers and regional growth in the presence of local inputs', Journal of Regional Science, 35, 1, pp. 3-27. Griliches, Z. (1992), 'The search for R&D spillovers', Scandinavian Journal of Economics, 94, supplement, pp. 29-47. Gripaios, P., Gripaios, R. and Munday, M. (1997), 'The role of inward investment and urban economic development: the cases of Bristol, Cardiff and Plymouth', Urban Studies, 34, 4, pp. 567-603. Hill, S. and Munday, M.C. (1994), The regional distribution of foreign manufacturing investment in the UK, London, Macmillan. Lovering, J. (1999), 'Theory led by policy: the inadequacies of the "new regionalism" (illustrated from the case of Wales)', International Journal of Urban and Regional Research, 23, 2, pp. 379-95. Morgan, K. (1997), 'The learning region : institutions, innovation and regional renewal', Regional Studies, 31, pp. 491-503. Neven, D. and Siotis, G. (1993), 'Foreign direct investment in the European Community - some policy issues', Oxford Review of Economic Policy, 9, 2, pp. 72-93. Nijkamp, P. and Blaas, E. (1995), 'Comparative regional policy analysis: ex post valuation of the performance of the European Regional Development Fund', Journal of Regional Science, 35, 4, pp. 5 7 9 97. Peck, F. W. (1996), 'Regional development and the production of space: the role of infrastructure in the attraction of new inward investment', Environment and Planning A, 28, pp. 327-39. Phelps, N.A. (1993), 'Branch plants and the evolving spatial division of labour: a study of material linkage in the northern region of England', Regional Studies, 27, pp. 87-101. Porter, M.E. (1996), 'Competitive advantage, agglomeration economies and regional policy', International Regional Science Review, 1 9 , 1 , pp. 85-90. Rodriguez-Clare, A. (1996), 'Multinationals, linkages, and economic development', American Economic Review, 86, 4, pp. 852-73. Shirlow, P. (1995), 'Transnational corporations in the Republic of Ireland and the illusion of economic well-being', Regional Studies, 29 pp. 687-705. Turok I. (1993), 'Inward investment and local linkages: how deeply em-
Regional policy and the impact of FDI in the UK
177
bedded is "Silicon Glen"?', Regional Studies, 27, pp. 401-17. (1997), 'Linkages in the Scottish electronics industry: further evidence', Regional Studies, 31, pp. 705-11. Twomey, J. and Tomkins, J.M. (1996), 'Supply chains, material linkage and regional development', Urban Studies, 33, 6, pp. 937-54. Wren, C. and Taylor, J. (1999), 'Industrial restructuring and regional policy', Oxford Economic Papers, 51, pp. 487-516. Young, S., Hood, N. and Dunlop, S. (1988), 'Global strategies, multinational subsidiary roles and economic impact in Scotland', Regional Studies, 22, pp. 487-97. Young, S., Hood, N. and Peters, E. (1994), 'Multinational enterprises and regional economic development', Regional Studies, 28, pp. 65777.
7
A regional computable general equilibrium analysis of the demand and 'efficiency spillover' effects of foreign direct investment Gary Gillespie, Peter G. McGregor, J. Kim Swales and Ya Ping Yin1
The advantage applied modelling offers is that we are able to combine real world data with formal theory as part of the mix of inductive and deductive reasoning that makes up economic analysis . . . Having said this, the reader should also be aware that the results of numerical analysis can take on a life of their own when released into a policy environment. (Francois and Reinert, 1997, p. 20)
INTRODUCTION Despite government rhetoric concerning indigenous development, the attraction of foreign direct investment (FDI) remains a central plank of UK regional policy. The majority of studies of the regional impact of FDI concentrate on demand-side effects and use standard Keynesian or input-output multiplier methods. 2 However the benefits from FDI are no longer thought to be limited solely to a stimulus to employment from additional demand. Indeed the focus of much recent work, especially at the level of the national economy, has been on supply-side issues, particularly the identification of 'efficiency spillover' effects from FDI to the indigenous sector. This issue is considered directly in several of the papers in this volume, with previous research being summarised in the paper by Blomstrom et al. In Gillespie et al. (1998) we argue that the identification of the regional system-wide demand-side effects of
178
A regional computable general equilibrium analysis
179
assisted investment should incorporate capacity constraints and labour market displacement. This is not possible with the standard multiplier models. The argument for supply-side modelling holds a fortiori for the identification of the system-wide impacts of efficiency spillovers that operate primarily through changes in relative prices. In this chapter we report two sets of novel results. First, we calculate the traditional demand-side regional effects of inward investment using an ownership-disaggregated computable general equilibrium (CGE) model for Scotland (AMOSFDI). This model identifies separately the foreign-owned manufacturing sector and incorporates capacity constraints and endogenous wage and price competitiveness effects. Second, we use the same CGE modelling framework, together with the empirical estimates of efficiency spillovers given in Barrell and Pain (1997), to quantify the regional system-wide impact of the efficiency effects that have been attributed to FDI. We are interested in comparing the relative size and characteristics of these demand and supply effects partly because it has not been possible to quantify the supplyside impact of inward investment in previous formal evaluations of regional policy, even though this channel has been thought to be important. The chapter is structured as follows. A brief discussion of the efficiency spillover literature is followed by a description of the AMOSFDI model. Then we outline the adjustments that have been made to incorporate the foreign-owned manufacturing sector and simulate the impact of the standard demand shock associated with FDI in Scotland.We examine the system-wide impact of efficiency spillovers and present the combined effects of FDI on the regional economy, exploring the relative importance of demand- and supply-side effects. The final section offers a brief conclusion.
T H E SCALE OF EFFICIENCY SPILLOVERS F R O M FDI It is generally accepted that positive efficiency differences exist between multinational firms and indigenous plants and that, through linkages or information flows, the performance of the indigenous sector can be enhanced as a consequence. However quantifying the scale of such impacts is difficult. Within the UK manufacturing sector, efficiency spillover effects have been often cited (PA Cambridge Economic Consultants Ltd, 1993,1995), but these effects have begun to be quantified reliably only recently.3
180
Inward investment, technological change and growth
One approach to estimating the impact of inward investment on efficiency allows for endogenous technical change embodied within a dynamic factor demand function. This method avoids some of the problems associated with the growth accounting approach, such as the assumption of constant factor utilisation rates (for example, Coe and Helpman, 1995). Barrell and Pain (1997) is an influential example of such an approach. Their long-run labour demand relationship, derived on the assumption of CES technology, is given by:
ln(L) = ln(Q)-aln(W/P)-(l-a)^ + [aln{(l-s)/p}-(l-a)ln(y)] where L represents employee hours; Q is value added; Wis labour costs per employee-hour; P, the price of value added; a, the elasticity of substitution in production; (3 the mark up of value-added price over labour costs; Y and s are production function scale parameters and Xt is the rate of labour-augmenting technical progress in time period t. They further assume that technical progress in any given sector is dependent on the aggregate level of foreign-owned assets within that sector, together with an exogenous element proxied by a linear time trend: ** = ^ M E T I M E
+ ^FDI
ln(FDI);_4.
where TIME is the exogenous linear time trend and F D I ^ the stock of FDI within that sector, lagged four periods. Barrell and Pain (1997) embed their long-run labour demand function in a data-based dynamic equation and estimate this on quarterly time-series data for UK manufacturing and for non-manufacturing for the period 1972 to 1995. They find that a 1 per cent rise in the stock of FDI in UK manufacturing ultimately raises labour-augmenting technical progress in that sector by an estimated 0.27 per cent, but they find no evidence of a significant impact of the stock of FDI in the aggregate non-manufacturing sector. The impact of inward investment on the level of technical progress in the manufacturing sector takes some time to become fully established, with gradual adjustment to the steady-state impact. Hubert and Pain, in a series of papers (1999a, 1999b and this volume), investigate the robustness of this result by extending the analysis in a number of important respects. First, they explore the scale of efficiency spillovers using a variety of databases (including disaggregation of the services sector) and estimation techniques. Secondly, they inves-
A regional computable general equilibrium analysis
181
tigate the impact of generalising their approach to allow for other determinants of the rate of labour-augmenting technical progress, notably imports, research and development expenditures and human capital. Thirdly, they recognise the possibility that in an aggregate sectoral equation apparent efficiency impacts may simply pick up compositional effects. The distinctive characteristics of foreign-owned firms imply that an increase in FDI will raise average productivity, even in the absence of genuine spillovers. This issue, and that of whether the stock of FDI is a reasonable measure of the scale of inward investment, are investigated using a more disaggregated dataset for manufacturing industries. Whilst the point estimates of efficiency effects do vary as a consequence of the various modifications of the Barrell and Pain (1997) analysis, the finding of a statistically significant positive impact of the stock of FDI on efficiency in manufacturing appears robust. Indeed there is evidence that the impact of inward investment may even be larger than Barrell and Pain estimate. There is still considerable debate concerning the true scale of efficiency spillovers in the UK, with different methodologies generating different results. However in this chapter we propose to use the estimates derived in the original Barrell and Pain (1997) paper and apply these at a regional level. Our motivation for this decision is that these results appear to be fairly robust to extensive specification testing. Although the paper by Nigel Driffield in this volume identifies spillover effects at the regional level in the UK, his is the first paper to do so and therefore the robustness of his results remains untested. Accordingly, in the CGE model used here we apply a Harrod-neutral technical change that ultimately stimulates efficiency by 0.27 per cent for each 1 per cent increase in the capital stock generated by the inward investment. This affects both factor demand and price functions at the value-added level of production. However, this disturbance is applied gradually, reflecting the adjustment path implied by the Barrell and Pain (1997) estimates.
T H E A M O S F D I GENERAL SIMULATION F R A M E W O R K AMOSFDI is a CGE modelling framework parameterised on data from a UK region, Scotland.4 A very brief description is presented in this section: more detail is available in Appendix 7A and a full listing of AMOS, the earlier model on which AMOSFDI is based, is provided in Harrigan
182
Inward investment, technological change and growth
et al. (1991). AMOSFDI has four transactor groups - households, foreign-owned and UK-owned corporations, and government; 5 four commodities and activities - foreign-owned and UK-owned manufactures and non-manufacturing tradables and non-tradables. 6 There are also two exogenous external transactors, the rest of the UK and the rest of the world (RUK and ROW). Throughout this chapter product markets are taken to be competitive. We do not explicitly model financial flows, our assumption being that Scotland is a price-taker in competitive UK financial markets. The AMOSFDI framework allows a high degree of flexibility in the choice of key parameter values, model closures and even aggregative structure. However a crucial characteristic of the model is that, no matter how it is configured, we impose cost minimisation in production with multi-level production functions, generally of a CES form but with Leontief and Cobb-Douglas being available as special cases. There are four major components of final demand: consumption, investment, government expenditure and exports. Of these, real government expenditure is exogenous. Consumption is a linear homogeneous function of real disposable income. Exports (and imports) are generally determined via an Armington link (Armington, 1969) and are therefore sensitive to relative prices. Investment is a little more complex, as we discuss below, although the initial inward investment is treated as exogenous throughout. In all the simulations in this chapter we impose a single Scottish labour market characterised by perfect sectoral mobility. We employ a regional bargaining wage curve in which the regional real consumption wage is directly related to workers' bargaining power, and therefore inversely to the regional unemployment rate (Minford et al., 1994). Empirical support for this wage-curve specification is now widespread, even in a regional context (Blanchflower and Oswald, 1994). The bargaining function is taken from the regional econometric work reported by Layard et al. (1991). The resulting wage curve has the form : wst =a-
0.068us + 0A0wst_x
where ws and us are the natural logarithms of the real consumption wage and the unemployment rate in Scotland, t is the time subscript and a is a calibrated parameter. When we introduce the Harrod-neutral improvement in technology, labour demand is determined in efficiency units. The number of efficiency units of labour supplied by each worker is raised
A regional computable general equilibrium analysis
183
as a result of the increase in technical efficiency. Therefore, for any given consumption wage, the cost of labour to the firm in efficiency units is reduced. The bargaining mechanism by which the wage to the worker is determined is not affected by these adjustments. Within each period of the multi-period simulations we report here, both the total capital stock and its sectoral composition are fixed, and product markets clear continuously. The exogenous inward investment stimulus in the manufacturing sector occurs in the first period. Each sector's capital stock is then updated between periods via a simple capital stock adjustment procedure, according to which investment equals depreciation plus some fraction of the gap between the desired level of the capital stock and its actual level. This process of capital accumulation is compatible with a simple theory of optimal firm behaviour given the assumption of quadratic adjustment costs. Desired capital stocks are determined on cost-minimisation criteria and actual stocks equal last period's stocks adjusted for depreciation and gross investment. The economy is assumed initially to be in long-run equilibrium, where desired and actual capital stocks coincide.7 For any increase in regional employment over a time period in which population is fixed, 75 per cent comes from the registered unemployed, while the remainder is supplied by increases in local labour-market participation. This is the assumption used in official government studies (Alexander and Whyte, 1995). Where migration is incorporated in the model, as it is in all the simulations reported below, population is also updated 'between' periods. We take net migration to be positively related to the real wage differential between Scotland and the rest of the UK and negatively to the unemployment rate differential, in accordance with the econometric results reported in Layard etal. (1991). This model was initially suggested by Harris and Todaro (1970), and is commonly employed in studies of US migration. Examples include Greenwood et al. (1991) and Treyz et al. (1993). The formal migration function we use is:
m=$-0.08(us-ur)
+ 0.06(ws-wr)
where m is the net in-migration rate (as a proportion of the indigenous population); wr and ur are the natural logarithms of the real consumption wage and unemployment rates in the rest of the UK, and p is a calibrated parameter. In the multi-period simulations reported below, the net migration flow in any period is used to update the population
184
Inward investment, technological change and growth
stock at the beginning of the next period, in a manner analogous to the updating of the capital stock. The regional economy is initially assumed to have zero net migration and, ultimately, net migration flows establish a new population equilibrium where this condition again holds. Unless otherwise stated (see below for the specification of foreign manufacturing) all sectors use a CES technology with 'best guess' elasticities of substitution of 0.3 (Harris, 1989) and Armington trade substitution elasticities of 2.0 (Gibson, 1990). The capital stock adjustment parameter (Xf) is taken to be 0.5 in each sector.8 Throughout, we typically discuss results by interpreting the conceptual 'periods' of our model as years. We believe this to be reasonable in that all of the data we have employed in the calibration (and, where applicable, estimation) of the model are annual. However, since the model is not (and cannot be) entirely econometrically estimated on annual data, this interpretation is suggestive, rather than definitive.
T H E F O R E I G N - O W N E D M A N U F A C T U R I N G SECTOR In the AMOSFDI model we have disaggregated the manufacturing sector by ownership. There are two important aspects to this disaggregation. First, foreign-owned and UK-owned manufacturing plants exhibit quite different structural characteristics, which are captured in the Social Accounting Matrix (SAM). Second, we expect behavioural relationships to differ across ownership types. Table 7.1 indicates the size of the foreign-owned manufacturing sector in Scotland, and some of its important structural characteristics. 9 First, foreign-owned plants make up a substantial share of Scottish manufacturing activity. They provide 33 per cent of manufacturing gross output, 26 per cent of manufacturing value added and 21 per cent of manufacturing employment. This is above their share in the aggregate UK manufacturing sector. They are also important variations in structural characteristics across ownership type. Wages, gross output per employee, value added per employee and intermediate inputs per employee are higher in foreign-owned manufacturing than in any of the other three Scottish sectors. Other important differences are that foreign-owned manufacturing has a low labour and domestic input intensity with low shares of wages and domestic inputs in gross output. This reflects the high proportion of (non-Scottish) intermediate inputs
A regional computable general equilibrium analysis
185
Table 7.1. Comparison of the structural characteristics of the ownershipdisaggregated database (1989) UK-owned Foreign-owned manumanufacturing facturing £l7,088m Gross output £5,538m Gross value added Total employment (full-time equivalents) 308,249 Gross output £55,434 per employee Value added £17,967 per employee Average sectoral wages and salaries £14,439 Scottish intermediate inputs per employee £21,367 Intermediate imports per employee £16,101
Non-manu- Sheltered facturin g traded
£8,283m £l,939m
£22,117m £l2,237m
£l7,204m £l4,038m
82,078
588,161
914,350
£100,918
£37,604
£18,816
£23,629
£20,806
£15,353
£17,338
£13,598
£12,169
£20,752
£9,873
£2,112
£56,537
£6,925
£1,351
Source: Gillespie (1998). in foreign-owned firms. Gross output is AXA times the size of value-added in these firms, compared with just over three times the size in the UKowned manufacturing firms. We n o w discuss differences in the key parameters a m o n g the sectors in the model. Together with the actual ownership-disaggregated SAM, these determine the calibrated form of the A M O S F D I model. There is clear evidence t h a t the behaviour of foreign-owned m a n u f a c t u r i n g plants is distinct from indigenous plants. 1 0 The key variables here are the substitution elasticities in the Armington demand equations and the substitution elasticities in the production function. For our specification of the export demand elasticity facing the foreign-owned manufacturing sector, we follow the empirical evidence of Bradley et al. (1993, 1995). We impose a high export d e m a n d elasticity in an attempt to approximate a 'law of one price' (LOOP) closure. The motivation for this type of closure is that, since multinational companies compete primarily in world markets, they have little influence over the price of their output. Also intra-firm benchmarking potentially renders the demand for the output of individual FDI plants extremely price-elastic. Essentially they are in competition with other plants owned
186
Inward investment, technological change and growth
Table 7.2. Functional forms and key parameter values for foreign and UKowned plants Functional forms
UK-owned
Technology Demands
CES (0.3) CES (2.0)
FDI manufacturing CES CES
(0.15) (9.0)
Note: CES denotes constant-elasticity of substitution. by the parent firm and capable of producing identical commodities. We also assume that there is little scope for adjustment in production technology in FDI plants. This we capture through imposing a low elasticity of substitution at all levels of the production function for the foreignowned sector. Table 7.2 summarises the specification and key parameter values for the sectors in the AMOSFDI model. This parameterisation renders demand more price-sensitive, but choice of technique less pricesensitive, in the foreign-owned sector than in the domestically owned sectors.
A N E X P O R T - O R I E N T E D FDI STIMULUS In this section we illustrate the regional impact of a stimulus equivalent to a 20 per cent increase in the stock of foreign direct investment in the base year, which is 1989. This involves an employment injection of 3,392 jobs and increases foreign-owned capacity and output in Scotland by 3 per cent. 11 We further assume that the total output of the incoming foreign-owned plants is sold outside the region. This represents a 3.97 per cent increase in FDI-related exports. 12 Table 7.3 summarises the nature of the disturbance. The precise specification of this demand-side simulation is as follows. In period 1 we introduce an exogenous 20 per cent increase in the stock of foreign direct investment in the manufacturing sector. In subsequent periods, investment is endogenous. In period 2 we permanently increase the FDI export-demand parameter by 3.97 per cent. This implies that, with unchanged relative foreign and domestic FDI prices, exports from the foreign-owned manufacturing sector will rise by 3.97 per cent. The export stimulus coincides with the increase in capacity generated by the investment in period 1. The model is then run forward 60 periods. Table 7.4 reports the proportionate changes in key economic variables
A regional computable general equilibrium analysis
187
Table 7.3. The impact of our standard export FDI shock in terms of capital stock, exports, output and direct employment
Capital stock Exports Output Employment
%
Value
3.0 3.97 3.0 3.0
£l82m £248m £248m 3,392
in the initial period and for subsequent periods 2 , 1 0 , and 2 5 . Table 7.4 also gives the long-run equilibrium results. Figure 7.1 plots the absolute total and sectoral employment changes over the first 60 periods. In period 1 the investment shock leads to a rise in total employment of 1,644 jobs, concentrated primarily in the sheltered sector, which includes construction. However there is capital-labour substitution in some sectors. The increase in the real wage resulting from the consequent fall in the u n e m p l o y m e n t rate, generates labour m a r k e t displacement in both the UK-owned and the foreign-owned manufacturing sectors. E m p l o y m e n t in indigenous manufacturing falls by a greater proportionate a m o u n t than in foreign manufacturing. This is despite the fact that export demand in the foreign-owned manufacturing sector is more sensitive to changes in the price of output produced in that sector. This result reflects the different structure of production in these t w o sectors. The indigenous manufacturing sector is more labour-intensive in the production of gross output, which accounts for the more adverse impact. The product price in the foreign-owned sector is cushioned against the increases in the domestic wage by three structural factors: the relative capital intensity of value added, the high import intensity of intermediate inputs, and the high intermediate intensity in the production of gross output. In period 2, the additional capacity in the foreign-owned manufacturing sector and the associated export demand comes on stream. The resultant expansion in aggregate output and employment in period 2 is over twice the value of period 1, with GDP rising above base levels by 0.18 per cent. From the very start, the operation of foreign-owned plants provides a much bigger boost to the local economy than their construction. However note that the stimulus is initially almost wholly restricted to the foreign-owned sector. O u t p u t and employment in UK-owned manufacturing falls as the rise in the nominal wage erodes export com-
4000 h 2000 h
"i
4
8
12
16
20
24
28
1
1
1
1
1
r
32
36
40
44
48
52
56
60
Period
-A -O—
Total employment Foreign manufacturing Sheltered
~—"•
UK manufacturing Non-manufacturing traded
Figure 7.1. Sectorally disaggregated employment change for a 100 per cent export FDI shock (20 per cent increase in investment and 3.97 per cent export demand shock) with regional bargaining and migration with LOOP approximation for foreign-owned sector
A regional computable general equilibrium analysis
189
petitiveness. Similarly activity levels in the non-manufacturing traded and the sheltered sectors are only marginally above the base-year value. Employment change in the sheltered sector, which had been boosted by the construction of the FDI plant in period 1, falls back in period 2, although it does subsequently expand, together with the other UK-owned sectors. A key factor in these simulations is the role of migration. After period 1, continuing in-migration occurs as the real wage and employment rates rise above their base-period values. This has both demand and supply-side effects on the regional economy. From period 2, the increased population augments the exogenous demand stimulus through higher consumption as a result of increased government transfers. 13 However, more importantly, with a given level of employment, in-migration increases the supply of labour within the region. This, in turn, increases the regional unemployment rate, which puts downward pressure on the real consumption wage through the bargaining function. Migration continues until the real wage and employment rates fall to their initial base-year levels. This restores the zero net migration condition. It can be seen from Figure 7.1 that in-migration helps to mitigate the adverse competitiveness effects arising from the initial increase in the nominal wage and employment eventually increases in all sectors due to the rise in intermediate and consumption demands. In period 2 the nominal wage is 0.19 per cent above its base level. By period 10, the increase is only 0.08 per cent, by period 25, 0.02 per cent. However this adjustment process is clearly slow and it is period 5, for example, before we observe positive employment growth in the UK-owned manufacturing sector. One characteristic of the impact of demand shocks where investment and flow-equilibrium migration are allowed is that in the long run the model generates extended (capital and population endogenous) inputoutput results. That is to say, price changes fall to zero, and in each sector output, employment, capital stocks and investment expand by the same proportionate amount (McGregor et al., 1996). These characteristics can be seen in the long-run results in Table 7. 4. With external prices and the interest rate held constant, migration and investment allow a perfectly elastic supply of capital and labour at their initial rental rates in the long run. Given linear homogeneity in production, there is then no cost incentive to change the factor or intermediate intensity of techniques.
190
Inward investment, technological change and growth
Table 7.4. The percentage change in key variables following a 20 per cent increase in FDI with 100 per cent export-intensity
GDP (income measure) Consumption Investment Nominal pre-tax wage Real pre-tax consumption wage Total employment UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Unemployment rate (%) Labour participation rate (%) Total population (000s) Price of value added: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Capital rental rates: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Consumer price index Value-added: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Sectoral output: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Capital stocks: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Exports: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Nominal income: Households' disposable income Firms' disposable income
Period 1
Period 2
Period 10
Period 25
Longrun
0.07 0.15 2.15 0.14 0.06 0.07 -0.07 -0.06 0.00 0.22 -0.56 0.02 0.00
0.18 0.26 0.39 0.19 0.11 0.15 -0.04 2.94 0.02 0.02 -0.98 0.03 0.02
0.30 0.33 0.48 0.08 0.04 0.27 0.07 2.98 0.17 0.12 -0.33 0.01 0.23
0.39 0.40 0.54 0.02 0.01 0.36 0.16 3.07 0.28 0.18 -0.09 0.00 0.35
0.44 0.43 0.58 0.00 0.00 0.41 0.20 3.20 0.34 0.22 0.00 0.00 0.41
0.09 0.03 0.14 0.28
0.17 0.08 0.21 0.20
0.09 0.08 0.11 0.09
0.03 0.06 0.04 0.03
0.00 0.00 0.00 0.00
-0.08 -0.23 0.14 0.88 0.07
0.05 -0.20 0.25 0.24 0.08
0.14 0.11 0.19 0.13 0.04
0.04 0.14 0.06 0.03 0.01
0.00 0.00 0.00 0.00 0.00
-0.05 -0.04 0.00 0.18
-0.03 2.96 0.01 0.01
0.07 2.98 0.16 0.12
0.16 3.07 0.28 0.18
0.20 3.20 0.34 0.22
-0.04 -0.04 0.01 0.19
-0.01 2.97 0.03 0.02
0.08 2.98 0.17 0.12
0.16 3.07 0.28 0.18
0.20 3.20 0.34 0.22
0.00 0.00 0.00 0.00
0.00 3.00 0.00 0.00
0.06 2.97 0.14 0.11
0.15 3.05 0.27 0.18
0.20 3.20 0.34 0.22
-0.10 -0.09 -0.20 -0.52 0.00 0.22 0.30
-0.18 3.74 -0.31 -0.37 0.00 0.34 0.39
-0.10 3.73 -0.16 -0.16 0.00 0.37 0.44
-0.03 3.82 -0.05 -0.05 0.00 0.41 0.46
0.00 3.97 0.00 0.00 0.00 0.43 0.47
A regional computable general equilibrium analysis
191
In this simulation, the relevant continuing injection is the increase in export demand in the foreign-owned sector that accompanies the increase in FDI in the first period. The initial one-off investment injection is neutralised by subsequent capital adjustments. In the long run, total regional employment increases by 9,233, which is equivalent to a longrun employment multiplier of 2.72. The adjustment process is however rather protracted, particularly if the focus is on the short- to mediumterm impact. By period 10, employment has only reached 68 per cent of the eventual long-run response. Previous attempts to identify the aggregate regional impact of FDI have used traditional demand-based input-output or Keynesian multiplier methods. Sometimes it is argued that the indirect and induced effects of FDI will be particularly limited (for example, Turok, 1993), but we do not find that to be true. For Scotland, the type-II employment multiplier value is large. 14 The CGE model that we employ produces input-output results in the long run as migration and investment relax resource constraints. However, in the short to medium run, which typically is the focus of policy, the employment generated falls well short of that predicted from an I-O model. The effect of capacity constraints and labour market crowding out significantly reduce the expansionary system-wide regional impact (Gillespie et al., 1998). However the estimated short- to medium-run effects on the Scottish economy are still substantial.
T H E I M P A C T OF EFFICIENCY SPILLOVERS F R O M FDI In this section we illustrate the regional system-wide impact of 'efficiency spillovers' arising from the presence of more productive foreign plants entering Scottish manufacturing. As discussed earlier, we take the central estimate for the scale of our shock from Barrell and Pain (1997). We therefore assume that a 1 per cent increase in foreign capacity leads ultimately to a 0.27 per cent increase in indigenous manufacturing efficiency. Following Barrell and Pain (1997) and Hubert and Pain in this volume, we interpret the efficiency stimulus as labour-augmenting in nature; that is, we assume that Harrod-neutral technical progress is induced by an increase in the stock of inward investment. Furthermore, for simplicity and comparability with the results of the preceding section, we focus here on the impact of a permanent step increase in the stock of foreign-owned plants. However the long-run results of this dis-
192
Inward investment, technological change and growth
0.9 n 0.8 -
5 6 Period
7
10
Figure 7.2. Build-up of efficiency spillover shocks - adjustment path implied by Barrell and Pain (1997) results turbance offer insights into the growth process that would be implied by any continuing additions to the stock of inward investment. Recall from the previous section that our export-oriented FDI stimulus increases foreign-owned manufacturing capacity by 3 per cent. The technical progress relationship implies that a 3 per cent increase in foreign capacity would lead to a 0.80 per cent labour-augmenting efficiency increase in indigenous manufacturing firms. We therefore introduce a 0.80 per cent Harrod-neutral improvement in technical progress to the UK-owned manufacturing sector in our model. Efficiency adjustments begin in period 2, together with the increase in FDI capacity and the export shock. However, the whole efficiency gain does not occur immediately and is distributed over time, as illustrated in Figure 7.2. The build-up of efficiency effects is relatively rapid, with over 90 per cent of the direct impact achieved by period 5 (that is, four periods after production comes on stream). Accordingly, we simulate the impact of the efficiency stimulus by adding these disturbances to the capacity and demand adjustments dealt with in the preceding section. However, in this section we focus solely on the additional effects generated by the efficiency shock. We therefore subtract the simulated impacts reported in the preceding section from the total simulated impact of combined disturbances, including efficiency effects.
A regional computable general equilibrium analysis
193
Table 7.5 reports the period-by-period proportionate impact of the efficiency stimulus on the same economic variables reported in Table 7.4. Figure 7.3 plots the employment change by sector attributable solely to this improvement in efficiency. The efficiency stimulus has no impact until production begins in period 2 and the initial effects are small. The increase in efficiency in UK-owned manufacturing takes the form of a reduction in the cost of labour, measured in efficiency units. This is reflected in the 0.21 per cent fall in the price of UK-owned manufacturing value added in period 2 and a corresponding 0.21 per cent stimulus to its exports. The expansion in intermediate demand from the UK-owned manufacturing sector generates a small rise in the output of the other two UK-owned sectors, so that Scottish GDP increases by 0.05 per cent. However, whilst employment rises in the non-manufacturing traded and sheltered sectors in period 2, employment falls in UK-owned manufacturing and there is no change in total employment. At a given wage rate per employee, a Harrod-neutral improvement in technology reduces unit costs. Under competition, there will be a corresponding reduction in the market price, an increase in output demand and an expansion in the derived demand for labour. However the efficiency gain is also likely to simultaneously reduce the labour input per unit of output, measured in natural units. The net change in employment is therefore the result of two offsetting forces. A Harrod-neutral increase in efficiency will generate an increase in employment with a given wage per employee only if the demand for labour, measured in efficiency units, is price-elastic. In the short run, with our default parameter values, the AMOS model generates an inelastic general-equilibrium demand for labour in the UK-owned manufacturing sector. Harrod-neutral improvements in technology thus produce a reduction in employment in this sector. In period 2, the capital rental rates in all sectors are above their initial values. This helps to push value added prices in the nonmanufacturing traded and sheltered sectors above their base levels, causing some crowding out of exports in these sectors. Investment relaxes capacity constraints, especially in the UK-owned manufacturing sector and, as output in this sector expands, there are further increases in output and employment in the other three sectors due to higher final expenditure through consumption and investment, and higher intermediate demand. The increased capacity in the UK-owned manufacturing sector continues to reduce value added prices in this sector, so that by period 10 export volumes have risen by 0.61 per cent. But the price
194
Inward investment, technological change and growth
Table 7.5. The impact of efficiency gains to UK manufacturing from FDI using the export FDI shock as the base, with regional bargaining and migration
GDP (income measure) Consumption Investment Nominal pre-tax wage Real pre-tax consumption wage Total employment UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Unemployment rate (%) Labour participation rate (%) Total population (000s) Price of value added: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Capital rental rates: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Consumer price index Value-added: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Sectoral output: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Capital stocks: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Exports: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Nominal income: Households' disposable income Firms' disposable income
Period 1
Period 2
Period 10
Period 25
0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
0.05 0.02 0.09 0.00 0.00 0.00 -0.12 0.00 0.03 0.02 0.01 0.00 0.00
0.18 0.09 0.18 0.01 0.02 0.06 -0.11 0.02 0.14 0.06 -0.17 0.01 0.04
0.25 0.14 0.22 -0.02 0.01 0.12 -0.03 0.08 0.23 0.10 -0.07 0.00 0.11
0.29 0.17 0.25 -0.03 0.00 0.16 0.00 0.17 0.27 0.13 0.00 0.00 0.16
0.00 0.00 0.00 0.00
-0.21 0.01 0.04 0.02
-0.58 0.03 0.05 0.02
-0.65 0.01 0.00 -0.01
-0.67 -0.03 -0.03 -0.03
0.00 0.00 0.00 0.00 0.00
0.53 0.03 0.12 0.07 0.00
0.26 0.05 0.13 0.05 -0.01
0.03 0.06 0.03 0.01 -0.03
-0.02 -0.01 -0.02 -0.03 -0.03
0.00 0.00 0.00 0.00
0.22 0.00 0.02 0.02
0.63 0.02 0.13 0.06
0.72 0.07 0.22 0.10
0.76 0.17 0.27 0.13
0.00 0.00 0.00 0.00
0.19 0.00 0.03 0.02
0.55 0.02 0.14 0.06
0.63 0.08 0.22 0.10
0.66 0.17 0.27 0.13
0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00
0.38 0.01 0.11 0.05
0.52 0.07 0.21 0.09
0.56 0.17 0.27 0.12
0.00 0.00 0.00 0.00 0.00 0.00 0.00
0.22 -0.03 -0.05 -0.03 0.00 0.02 0.11
0.61 -0.07 -0.07 -0.04 0.00 0.09 0.19
0.69 -0.02 0.01 0.02 0.00 0.12 0.19
0.71 0.08 0.04 0.06 0.00 0.13 0.20
Longrun
4000 3500 3000 2500 g B
2000
|
1500
w
1000 500 0 -500 -1000
Period i
i
4
—±— A
8
i
i
i
1
1
1
1
1
1
1
1
1
1—
12
16
20
24
28
32
36
40
44
48
52
56
60
Total employment Foreign manufacturing Sheltered
•
UK manufacturing Non-manufacturing traded
Figure 7.3. Aggregate employment gains from the efficiency spillover shock (using the results from the export FDI shock as the base), with regional bargaining and migration
^
196
Inward investment, technological change and growth
adjustment process is slow, so that, by period 10, output prices are still above their base values in the foreign-owned, non-manufacturing traded and sheltered sectors. By period 25, we observe small increases in exports in the non-manufacturing traded sector and the sheltered sector but there is still some crowding out in the foreign-owned sector. In long-run equilibrium, migration restores the original real wage which, combined with the efficiency improvement, means that all output prices fall. From Table 7.5 it can be seen that exports rise in all sectors, with the largest effect in the UK-owned manufacturing sector. Scottish GDP increases by 0.29 per cent and total employment by 0.16 per cent, although employment in UK-owned manufacturing does not rise above its base value. These simulation results illustrate the potential system-wide supplyside impact of an increase in efficiency in the UK-owned manufacturing sector as a result of an expansion in the stock of inward investment in Scotland. The rise in regional activity generated by the efficiency spillovers is driven by the improvement in regional price competitiveness and the subsequent increase in regional exports and import substitution. Not surprisingly, in terms of output, it is the sector experiencing the efficiency stimulus, UK-owned manufacturing, which shows the largest expansion in activity. On the other hand, the employment impact is largest in the non-manufacturing traded sector. The improvement in technical efficiency means that higher output in UKowned manufacturing is met entirely from a rise in labour productivity. Finally it takes time for the price effects generated by the efficiency improvements to work their way fully through the general equilibrium system. However the total impact on regional GDP and employment as a consequence of the efficiency spillover is substantial and ultimately equals 66 per cent and 39 per cent of the respective effects generated by the associated export stimulus.
THE COMBINED DEMAND AND SUPPLY EFFECTS OF FDI In this section we look at the combined simulation results for the model incorporating all the export demand, FDI capacity and UK-owned manufacturing efficiency spillover effects. The results are summarised in Table 7.6 and Figure 7.4. Two key points can be gained from these simulation results. First, the incorporation of efficiency spillover effects does have a marked impact
A regional computable general equilibrium analysis
197
Table 7.6. 20 per cent increase in FDI with 100 per cent export-intensity and 0.8 per cent Harrod-neutral efficiency shock to UK manufacturing with regional bargaining and migration
GDP (income measure) Consumption Investment Nominal pre-tax wage Real pre-tax consumption wage Total employment UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Unemployment rate (%) Labour participation rate (%) Total population (000s) Price of value added: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Capital rental rates: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Consumer price index Value-added: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Sectoral output: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Capital stocks: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Exports: UK manufacturing Foreign manufacturing Non-manufacturing traded Sheltered Nominal income: Households' disposable income Firms' disposable income
Period 1
Period 2
Period 10
Period 25
0.07 0.15 2.15 0.14 0.06 0.07 -0.07 -0.06 0.00 0.22 -0.56 0.02 0.00
0.24 0.28 0.48 0.20 0.11 0.15 -0.16 2.94 0.05 0.03 -0.97 0.03 0.02
0.47 0.43 0.66 0.09 0.06 0.33 -0.04 3.00 0.31 0.18 -0.50 0.02 0.27
0.63 0.54 0.76 0.01 0.02 0.48 0.13 3.15 0.51 0.29 -0.16 0.01 0.46
0.72 0.60 0.82 -0.03 0.00 0.57 0.20 3.37 0.61 0.34 0.00 0.00 0.57
0.09 0.03 0.14 0.28
-0.04 0.09 0.25 0.22
-0.49 0.11 0.16 0.11
-0.62 0.06 0.03 0.01
-0.67 -0.03 -0.03 -0.03
-0.08 -0.23 0.14 0.88 0.07
0.58 -0.17 0.37 0.31 0.09
0.40 0.16 0.32 0.18 0.04
0.07 0.20 0.09 0.04 -0.01
-0.02 -0.01 -0.02 -0.03 -0.03
-0.05 -0.04 0.00 0.18
0.19 2.96 0.04 0.03
0.70 2.99 0.29 0.17
0.88 3.14 0.50 0.28
0.96 3.37 0.61 0.34
-0.04 -0.04 0.01 0.19
0.18 2.97 0.06 0.03
0.63 3.00 0.30 0.18
0.79 3.15 0.50 0.28
0.86 3.37 0.61 0.34
0.00 0.00 0.00 0.00
0.00 3.00 0.00 0.00
0.43 2.99 0.25 0.15
0.67 3.12 0.48 0.27
0.76 3.37 0.61 0.34
-0.10 -0.09 -0.20 -0.52 0.00 0.22 0.30
0.04 3.71 -0.36 -0.40 0.00 0.37 0.50
0.52 3.66 -0.23 -0.20 0.00 0.46 0.63
0.66 3.79 -0.05 -0.03 0.00 0.52 0.65
0.71 4.05 0.04 0.06 0.00 0.57 0.67
Longrun
14000 12000 10000 a
8000
rt
4000
n
4
8
1
1
1
1
1
1
1
1
1
1
1
1
r
12
16
20
24
28
32
36
40
44
48
52
56
60
Period —A _ 0 —
Total employment Foreign manufacturing Sheltered
*
UK manufacturing Non-manufacturing traded
Figure 7.4. The employment impact of the combined effects of FDI (20 per cent FDI, 100 per cent export-intensity plus build-up of efficiency spillover effects)
A regional computable general equilibrium analysis
199
Table 7.7. Combined employment, value added and output multipliers from the inward investment expansion
Total employment Total value added Total output
Period 5
Period 10
1.51 1.98 1.53
2.19 2.61 1.83
Period 25 Long-run 3.21 3.48 2.24
3.78 3.98 2.49
on both regional GDP and total employment in the long run. However, over the short term (for example the first 10 periods), the supply-side effects are small compared with those generated by the initial export demand and capacity stimulus. It takes much longer for the efficiency shocks imposed on the UK-owned manufacturing sector than the combination of demand and capacity changes in the foreign-owned sector to work through fully. The second point to note is the combined impact of the FDI-related demand and supply shocks on activity within the UK-owned manufacturing sector. The incorporation of efficiency spillovers means that manufacturing is ultimately the UK-owned sector that receives the biggest stimulus in terms of increased value added. However the impact on employment in UK-owned manufacturing is negative for each time period over the policy-relevant first ten periods. This suggests that the concern sometimes expressed by spokesmen for domestic manufacturing firms relating to the adverse consequences arising from the incentives and aid given to inward investors might be misplaced, at least if efficiency spillover effects are of the scale suggested in Barrell and Pain (1997). However the fears of unions representing workers in the domestically owned manufacturing sector may be justified. Even so, it is clear that, taking all factors into account, the expansion of inward investment delivers a substantial injection to the local economy and generates significant benefits in all locally-owned sectors. The relevant multiplier values are given in Table 7.7. The five-year employment, value added and output multipliers are given as 1.51, 1.98 and 1.53. Over this period it is primarily the export demand stimulus, in combination with the expansion in capacity in the foreign-owned sector, which produces the bulk of the effects. Over time, the expansion of capacity in other sectors, plus the dissemination and incorporation of efficiency effects, means that the multiplier value grows. In long-run equilibrium, the export demand effects account for 61 per cent of the total increase in GDP and 72 per cent of the increase in employment.
200
Inward investment, technological change and growth
CONCLUSIONS In this chapter we have investigated the aggregate additional impact of inward investment on the Scottish economy using a computable general equilibrium model that explicitly disaggregates the manufacturing sector by ownership type. The model thereby encapsulates the major structural differences that are known to characterise the foreign-owned and indigenous sectors. We have also attempted to capture both the demand and supply-side impact of foreign direct investment. The results appear to be relatively reassuring from the perspective of current regional policy towards inward investment. The demand impacts dominate over all time periods and are large. However, using the Barrell and Pain (1997) estimates, efficiency spillover effects do prove ultimately to have a substantial impact on the regional economy, particularly on regional GDP. This is good news for those who wish to stress the wealthcreating character of regional policy, even though these gains are a rather long time coming. In fact the neglect of supply-side considerations in the conventional evaluation of regional policy proves to be comparatively unproblematic in the sense that its emphasis is on the short to medium run and its focus is on employment generation. Over the first ten periods, neglect of these supply-side impacts would result in the total employment impact of the FDI shock being underestimated by around 17 per cent. This result may prove to be sensitive, though, to the incorporation of forward looking expectations, which would increase the speed of adjustment to disturbances, while leaving steady-state impacts unaltered. Nonetheless the relatively short-term focus of conventional methods of evaluating the potential benefits of inward investment may mean that a potentially important source of policy-induced regional growth may not be fully taken into account.
A regional computable general equilibrium analysis
201
APPENDIX 7A A C O N D E N S E D VERSION OF AMOSFDI In this condensed version of AMOSFDI, we ignore intermediate inputs, tax, most transfers between transactor groups and the participation rate. We also express the wage setting function and the migration function in terms of the employment rate, rather than the unemployment rate.
Equations
Short-run =Pi(wen,wkl)
(1) Commodity Price (2) Wage setting
pt
(3) Efficiency wage
wen =wn /(1 + cp)
wn - wn(N 1L,cpi)
(4) Labour force
L=L (5) Consumer price index w^; V a ^ _L V (\RUK^.RUK , v aRow^-ROw (6) Capital supply
K- = K-
(7) Capital price index
kPt = I yto + lyfUKP*m
(8) Labour demand in efficiency units
N-d =
N-d(Qt,wen,wkl)
(9) Demand for employees N6/ = N-d(l +
K? =
(11) Total demand for ,
YNd
employees
^
K?(Qt,wen,wkl) = N
i
(12) Capital market clearing
K- = Kd Wk^^iwki
(13) Household income
Y = \\fnNwn +
(14) Commodity demand
Q = Q + It +Gj + Xj
i
(15) Consumption demand Ct =
Ct(pt,p^UK,p*OW,Y,cpi)
(16) User cost of capital
uck - uck(kpi)
(17) Desired capital stock
K] = K? (Qj, wen, uck)
(18) Investment demand
It = ^ ( p p p / ^ p f ° W , I b l f l f lf=Xj(K*-Ki)
+Ufowprw
202
Inward investment, technological change and growth
(19) Government demand
Gj = Gj
(20) Export demand
X, = X,(p,,pf UK
Multi-period model
-ROW^RUK^ROW}
Stock up-dating equations
(21) Labour force (22) (ZZ) Migration
—^-nmg(w nlcpi,wn
(23) Capital stock
I cpi
,N I L,N
IL
)
K,-t = ( 1 - 4 )*,-,-!+/g-
Notation Activity-commodities i, j are activity/commodity subscripts (There are four of each in AMOSFDI: UKowned manufacturing, foreign-owned manufacturing, non-manufacturing traded and sheltered.) Transactors RUK = Rest of the UK, ROW = Rest of World Functions nmg(.) p (.) uck(.) w(.) Kd(.), Nd(.) C(.), /(.), X(.)
Variables C D G I Id Kd, Ks, K L Ned, Nd, N Q
net migration function cost function user cost of capital formulation wage-setting equations factor demand functions Armington consumption, investment and export demand functions, homogenous of degree zero in prices and one in quantities consumption exogenous export demand government demand for local goods investment demand for local goods investment demand by activity capital demand, capital supply and capital employment labour force labour demand in efficiency units, demand for employees and total demand number of employees commodity/activity output
A regional computable general equilibrium analysis
X Y b cpi, kpi d nmg p uck wen, wn, wk \|/ 0 y X cp
203
exports household nominal income elements of capital matrix consumer and capital price indices physical depreciation net migration price of commodity/activity output user cost of capital efficiency wage, wage and capital rental share of factor income retained in region consumption weights capital weights capital stock adjustment parameter the Harrod-neutral efficiency disturbance
NOTES 1
2
3
4
The research reported here was partially funded by Scottish Enterprise N a t i o n a l . We are grateful to Nigel Pain for comments on an earlier version of this paper. We are also grateful for comments on related w o r k to participants in the Seminar on Regional Economic Modelling in H o n o u r of Philip Israilevich, Federal Reserve Bank of Chicago 1 9 9 8 , the N o r t h American Regional Science Association Conference, Santa Fe 1998, and the NIESR Conference on the Impact of Multinational Corporations on the UK Economy, L o n d o n 1999. Examples include Alexander and W h y t e (1995); Barkley and M c N a m a r a (1993); H M Treasury (1995); Hill and Roberts (1995); PA C a m b r i d g e Economic Consultants Ltd (1993); Peck (1990); Phelps (1993); T u r o k (1993). For reviews of the empirical w o r k see B l o m s t r o m (1991) and Blomstrom and K o k k o (1998). Chapter 2 by Blomstrom, K o k k o and Globerman in this volume provides a synthesis of the literature relating to the determinants of host country spillovers from FDI. A M O S is an a c r o n y m for a m a c r o - m i c r o model of Scotland. A M O S F D I is the ownership-disaggregated variant of the model, developed specifically to allow us to investigate the impact of FDI. For a g o o d survey of CGE models see Dervis et al. (1982) and Shoven and Whalley (1984, 1992): for good examples of previous
204
Inward investment, technological change and growth
applications in a regional context, see Harrigan and McGregor (1988), Higgse**/. (1988), Jones and Whalley (1988), Kimbell and Harrison (1984), Kraybill and Lugani (1992) and Morgan et al. (1989). Partridge and Rickman (1998) provide a critical review of regional CGE modelling. 5 In AMOSFDI, Scotland is treated as a self-governing economy, in the sense that there is only one consolidated government sector. Central government activity is partitioned to Scotland and combined with local government activity. 6 Manufacturing comprises sectors 12-89, non-manufacturing traded sectors 1-10, 91-97, 99-102 and 109-111 and non-traded sectors 11, 90 and 98, 103-108 and 112-114 in the 1989 Scottish InputOutput Tables (Scottish Office Industry Department, 1994). 7 Our treatment is wholly consistent with sectoral investment being determined by the relationship between the capital rental rate and the user cost of capital. The capital rental rate is the rental that would have to be paid in a competitive market for the (sector-specific) physical capital; the user cost is the total cost to the firm of employing a unit of capital. Given that we take the interest, capital depreciation and tax rates to be exogenous, the capital price index is the only endogenous component of the user cost. If the rental rate exceeds the user cost, the desired capital stock is greater than the actual capital stock and there is therefore an incentive to increase the capital stock. The resultant process of capital accumulation puts downward pressure on rental rates and so tends to restore equilibrium. In the long run, the capital rental rate equals the user cost in each sector, and the risk-adjusted rate of return is equalised between sectors. 8 This default value of the adjustment parameters is based on investment equations estimated for the Scottish manufacturing sector. This is, in fact, the only sector in AMOSFDI for which a time series of investment data exists. For other sectors information is available only for the years in which a Scottish Input-Output Table has been constructed. 9 The employment data in Table 7.1 relate to full-time equivalent (FTE) employment with one part-time worker equal to one-third of a full-time worker. Self-employed workers are not included. 10 The FDI literature points to the ability of foreign firms to prosper outside their own home market and explains this phenomenon in terms of transactional, structural, technological and ownership
A regional computable general equilibrium analysis
11
12 13
14
205
advantages (Enderwick, 1985; Cantwell, 1987; Davies and Lyons, 1991; Kay, 1991). Gillespie etal. (2000) explore the relative importance of structure and behaviour in the context of a demand disturbance. The employment injection in the foreign-owned manufacturing sector resulting from the FDI shock is equal to 3 per cent of the base employment in the model, which is 113,000. These data come from the 1989 Labour Force Survey. They are different from those in Table 7.1, which are derived from the Scottish Input-Output Table for 1989. The average export-intensity of the total foreign-owned manufacturing sector in Scotland is, in fact, 76 per cent. Government expenditure is not directly related to population in our model (and the demand-side effects of migration would be substantially increased if it were). However migration increases the level of unemployment associated with any level of employment. It therefore produces a stimulus to consumption financed by government transfers. The aggregate type I I I - O employment multiplier, which includes the direct, indirect and induced (consumption and income) effects, for the foreign-owned manufacturing sector in Scotland, for 1989, is 2.43. See Gillespie (1998) for a discussion of these results.
REFERENCES Alexander, J.M. and Whyte, T.R. (1995), 'Output, income and employment multipliers for Scotland', Scottish Economic Bulletin, 50, Winter 1994/95. Armington, P. (1969),'A theory of demand for products distinguished by place of production', IMF Staff Papers, 16, pp. 157-78. Barkley, D.L. and McNamara, K.T. (1993), 'Local input linkages: a comparison of foreign-owned and domestic manufacturers in Georgia and South Carolina', Regional Studies, 28, pp. 725-37. Barrell, R. and Pain, N. (1997), 'Foreign direct investment, technological change, and economic growth within Europe', Economic Journal, 107, pp. 1770-86. Blanchflower, D.G. and Oswald, A.J. (1994), The Wage Curve, Cambridge, Mass., MIT Press. Blomstrom, M. (1991), Foreign Investment and Spillovers, London,
206
Inward investment, technological change and growth
Routledge. Blomstrom, M. and Kokko, A. (1998), 'Multinational corporations and spillovers', Journal of Economic Surveys, 122, pp. 247-78. Bradley, J., Fitzgerald, J. and Kearney, I. (1993), 'Modelling supply in an open economy using a restricted cost function', Economic Modelling, 10, pp. 11-21. Bradley, J., Whelan, K. and Wright, J. (1995), 'Hermin Ireland', Economic Modellings 12, pp. 249-74. Cantwell, J.A. (1987), 'Technological advantage as a determinant of the international activity of firms', University of Reading discussion paper, Series B, no. 105. Coe, D.T. and Helpman, E. (1995), 'International R&D spillovers', European Economic Review, 39, pp. 859-87. Davies, S.W., and Lyons, B.R. (1991), 'Characterising relative performance: the productivity advantages of foreign owned firms in the UK', Oxford Economic Papers, 43, pp. 584-95. Dervis, K., De Melo, J. and Robinson, S. (1982), General Equilibrium Models for Development Policy, Cambridge, Cambridge University Press. Enderwick, P. (1985), Multinational Business and Labour, London, Croom-Helm. Francois, J.F. and Reinert, K.A. (1997), 'Introduction', in Francois, J.F. and Reinert, K.A. (eds), Applied Methods for Trade Policy Analysis: A Handbook, Cambridge, Cambridge University Press. Gibson, H. (1990), 'Export competitiveness and UK sales of Scottish manufactures', paper delivered to the Scottish Economists' Conference, The Burn. Gillespie, G. (1998), 'Foreign-owned multipliers: evidence from a Scottish input-output model', Strathclyde Papers in Economics 98/1, Department of Economics, University of Strathclyde, Glasgow. Gillespie, G., McGregor, P.G., Swales J.K. and Yin, Y.P. (1998), 'The displacement and multiplier effects of regional selective assistance: a computable general equilibrium analysis', Strathclyde Papers In Economics 98/3, Department of Economics, University of Strathclyde, Glasgow. (2000), 'The regional employment impact of foreign direct investment: an ownership-disaggregated computable general equilibrium analysis', Strathclyde Papers In Economics 2000/1, Department of Economics, University of Strathclyde, Glasgow. Greenwood, M.J., Hunt, G., Rickman, D.S., and Treyz, G.I. (1991),
A regional computable general equilibrium analysis
207
'Migration, regional equilibrium, and the estimation of compensating differentials', American Economic Review, 81, pp. 1382-90. Haddad, M. and Harrison, A. (1993), 'Are there positive spillovers from direct foreign investment? Evidence from panel data for Morocco', Journal of Development Economics, 42, pp. 51-74. Harrigan, F. and McGregor, P.G. (1988), 'Price and quantity interaction in regional economic models: the importance of openness and closures', in Harrigan, F. and McGregor, P.G. (eds), Recent Advances in Regional Economic Modelling, London, Pion. Harrigan, F., McGregor, P.G., Dourmashkin, N., Perman, R., Swales, J.K. and Yin, Y.P. (1991), 'AMOS: a macro-micro model of Scotland', Economic Modelling, 10, pp. 424-79. Harris, J.R. and Todaro, M.P. (1970), 'Migration, unemployment and development: a two sector analysis', American Economic Review, 60, pp. 126-42. Harris, R.I.D. (1989), The Growth and Structure of the UK Regional Economy 1963-85, Aldershot, Avebury. Higgs, P.J., Paramenter, B.R. and Rimmer, R.J. (1988), 'A hybrid topdown, bottom-up regional computable general equilibrium model', International Regional Science Review, 11, pp. 317-28. Hill, S. and Roberts, A. (1995), 'Inward investment, local linkages and regional development', paper presented at the Regional Science Association, British and Irish Section, Annual Conference, 13-15 September 1995, Cardiff Business School. HM Treasury (1995), A Framework for the Evaluation of Regeneration Projects and Programmes, London, HMSO. Hubert, F. and Pain, N. (1999a), 'Inward investment and technical progress in the United Kingdom', paper presented at European Association for Research in Industrial Economics Annual Conference, Turin, September. (1999b), 'Foreign direct investment, spillovers and technical progress in the UK', paper presented at the Money, Macro and Finance Research Group Annual Conference, Oxford, September. Jones, R. and Whalley, J. (1988), 'Regional effects of taxes in Canada, an applied general equilibrium approach', Journal of Public Economics, 37, pp. 1-28. Kay, N.M. (1991), 'Multinational enterprise as strategic choice: some transaction cost perspectives', in Pitelis C. and Sugden R. (eds,), The Nature of the Transnational Firm, London, Routledge. Kimbell, LJ. and Harrison, G.W. (1984), 'General equilibrium analy-
208
Inward investment, technological change and growth
sis of regional fiscal incidence', in Scarf, H.E. and Shoven, J.B. (eds), Applied General Equilibrium Analysis, Cambridge, Cambridge University Press. Kraybill, D.S. and Lugani, S. (1992), 'Multiregional computable general equilibrium models: an introduction and survey', mimeo, Department of Agricultural and Applied Economics, University of Georgia, Athens, GA. Layard, R., Nickell, S. and Jackman, R. (1991), Unemployment: Macroeconomic Performance and the Labour Market, Oxford, Oxford University Press. McGregor, P.G., Swales, J.K. and Yin, Y.P. (1996), 'A long-run interpretation of regional input-output analyses', Journal of Regional Science, 36, pp. 479-501. Minford, P., Stoney, P., Riley, J. and Webb, B. (1994), 'An econometric model of Merseyside: validation and policy simulations', Regional Studies, 28, pp. 563-75. Morgan, W., Mutti, J. and Partridge, M. (1989), 'A regional general equilibrium model of the United States: tax effects of factor movement and regional production', Review of Economics and Statistics, 71, pp. 626-35. PA Cambridge Economic Consultants Limited (1993), Regional Selective Assistance 1985-88, London, HMSO. (1995), Assessment of the Wider Effects of Foreign Direct Investment in Manufacturing in the UK, London, HMSO. Partridge, M.D. and Rickman, D.S. (1998), 'Regional computable general equilibrium modeling: a survey and critical appraisal', International Regional Science Review, 2 1 , pp. 205-48. Peck, F. (1990), 'Nissan in the North East: the multiplier effects', Geography, 75, pp. 354-7. Phelps, N.A. (1993), 'Branch plants and the evolving spatial division of labour: a study of material linkage change in the northern region of England', Regional Studies, 27, pp. 87-101. Scottish Office Industry Department (1994), Scottish Input-Output Tables for 1989, Edinburgh, HMSO. Shoven, J.B. and Whalley, J. (1984), 'Applying general equilibrium models of taxation and trade: an introduction and survey', Journal of Economic Literature, 22, pp. 1007-51. (1992), Applying General Equilibrium, Cambridge, Cambridge University Press. Treyz, G.I., Rickman, D.S. and Greenwood, M.J. (1993),' The dynam-
A regional computable general equilibrium analysis
209
ics of US internal migration', Review of Economic and Statistics, 75, pp. 209-14. Turok, I. (1993), 'Forward investment and local linkages: how deeply embedded is Silicon Glen', Regional Studies, 27, pp. 401-17.
8
Sticky places in slippery space - the location of innovation by MNCs in the European regions John Cantwell, Simona lammarino and Camilla Noonan
INTRODUCTION The changing nature of international production and business means that the competitive advantage of firms lies increasingly in their ability to create a coherent set of new capabilities and source technological knowledge internationally (Cantwell, 1989; Teece etal., 1994; Cantwell and Piscitello, 1999a). Although the majority of multinational corporations (MNCs) continue to locate a large proportion of their R&D in their home country, it is widely acknowledged that many MNCs are now making increasing use of international networks for technology development to augment technologies generated in their home base (Cantwell, 1995; Fors, 1998; Dunning and Lundan, 1998; Kuemmerle, 1999; Cantwell and Piscitello, 2000). The pertinent issue for host economies attempting to entice such high value added activity, is to understand what renders a location attractive or 'sticky in such slippery space' (Markusen, 1996). This chapter provides a comparative study of such issues by drawing together the findings of three recent studies in this area which look at regional innovation patterns in the UK and Italy (Cantwell and lammarino, 1998,2000) and innovation activity in Germany (Cantwell and Noonan, 1999). The main premise of the three papers is that the extent to which regional agglomerations of technological knowledge and capabilities in the EU attract foreign direct investment (FDI) in R&D varies across locations and industries according to a geographical hierarchy of regional centres. The precise research location chosen by an MNC will depend upon the number of regional centres available, their positioning in the geographical hierarchy and the degree to which the MNC has developed a strategy for new technological combinations
210
Sticky places in slippery space
211
and/or diversification through tapping into specific competencies across a range of centres. The three papers constitute a significant contribution to the analysis of sub-national innovative activity across the EU and represent a vital component in the formulation of a future EU-wide technology policy that aims to foster regional systems through the promotion of inward FDI in technological innovation. Drawing upon the results of these studies, this chapter is divided into five parts. The first reviews the recent literature on the globalisation of innovation, which highlights the importance of location in the globalisation process and therefore the need for a regional approach when analysing this phenomenon. After a short description of the data used for the empirical investigation, the third section analyses the degree to which common patterns of innovative activity characterise the three economies under investigation. Firstly, we report upon the geographical distribution of innovative activities of large firms, both indigenous and foreign-owned, across the regions of the UK, Italy and Germany during 1969-95. We then focus upon the locational patterns observed for foreign-owned firms. In particular, by isolating the areas in which activity is most concentrated, we examine the importance of the existing knowledge base for foreign-owned firms when locating their R&D abroad. The fourth section documents the characteristics of the hierarchies of regional research centres, which exist across the EU. We suggest that the identification of such regions serves as a prerequisite to unveiling the potential technological communication (or technological spillovers) that may be in existence within and indeed across these regions. Finally, the chapter focuses upon the differences recorded in the geography of MNC innovation across the three economies and the concluding section highlights the future research agenda that arises from such observations.
I N T E R N A T I O N A L I S A T I O N OF R & D The international business literature postulates a number of explanations as to why technological activities might be located outside the home economy. In the early literature particular emphasis was placed upon the necessity of locating some (small) degree of R&D abroad to facilitate the localisation and adaptation of products to local tastes and requirements (Vernon, 1966). From the viewpoint of the firm, this was commensurate with the exploitation of existing technologies that were
212
Inward investment, technological change and growth
predominantly developed in the home economy. However subsequent empirical work suggested that the internationalisation of R&D in large companies was considerably greater than had at one time been supposed (on the basis of evidence of US MNCs in the early post-war period), but confirmed that such internationalisation was essentially motivated by dissimilarities in the implementation of technologies between home and foreign locations (see, inter alia, Cantwell, 1995). More recent contributions in this field have seen a shift in emphasis towards recognising the necessity for corporate strategy to focus on enhancing extant capabilities and creating new competencies within the firm (Cantwell and Piscitello, 2000). While there is evidence that much of the technology developed abroad by large firms lies in their core areas of strength, suggesting that adaptation of technologies and technical support to foreign manufacturing plants continue to be important, MNC research in foreign locations is also increasingly associated with entry into new and more distantly related fields of technology. Such knowledge-seeking activity is undertaken to help define the future directions in the evolution of the corporations' sources of competitiveness (Pearce, 1999). While in the past the creation of new competencies was commensurate with increasing R&D allocations in the home economy, it is now increasingly associated with the commitment of research budgets to operations located abroad. Allied to this, the technical innovations of multinational companies occur increasingly as a result of internal and external networks. 1 Increased complexity and growing interrelatedness between technologies, coupled with increased costs of undertaking R&D, render research interaction both within the corporation and with foreign-owned counterparts (often competitors) an increasingly crucial element of corporate survival (Cantwell, 1989; Kuemmerle, 1997; Cantwell and Barrera, 1998; Cantwell and Janne, 1999; Patel and Vega, 1999; Zander, 1999). While establishment in a foreign jurisdiction facilitates the monitoring of developments in different technological fields, it also enables the extraction of local knowledge for MNC global networks (Cantwell, 1992,1995; Chesnais, 1992; Granstrand and Sjolander, 1992; Dunning and Wymbs, 1997). In the case of the former, the firm is likely to be active abroad in technologies where there are complementarities between its own activities and the strengths of the host economy. Regarding the latter, a firm locates its research facility abroad to exploit the technological advantage of the host region in order to alleviate technological weakness at home (Patel and Vega, 1999).
Sticky places in slippery space
213
This differentiation between types of R&D activity suggests that certain regions/locations will be favoured over others when R&D investment decisions are being undertaken. Following from this, it is therefore suggested that two types of agglomeration forces (or clustering) operate across regions. The first can be termed 'localisation economies', which are sector-specific and tend to intensify intra-border sectoral integration through local external networks between foreignowned affiliates, indigenous firms and local non-market institutions. The second are referred to as 'general external economies', which attract all kinds of economic activities and tend to strengthen inter-border intrafirm integration, allowing the feedback of knowledge, expertise and information within networks of affiliates. Arising from this differentiation of agglomeration forces, it becomes possible to distinguish between higher and lower-order sites or regional centres. Such centres arise 'as a consequence of the interaction and the intensity of general external economies and localisation economies, which in turn depend upon the characteristics of the regional system considered' (Cantwell and lammarino, 1998, p. 387). In lower-order research centres, knowledge in specific technological fields is accessed and injected into the multinational production network. In higher-order centres, MNC affiliates can enjoy a broader range of spillovers from the local environment. Differentiating between higher and lower-order centres of innovation across space thus highlights the attraction that various locations hold for MNCs or, in other words, the nature of the centripetal forces in play across regions.
Implications for the host economy The internationalisation of MNC R&D activity has highlighted a number of issues. These range from the short-run implications for corporate performance to the long-run ramifications for the economic growth rate of the host economy. The array of positive externalities generated by a foreign-owned research unit may mean that a region's long-term economic viability is strongly linked to its ability to entice R&D commitments from international sources. An important consideration for the host country is of course the precise nature and mix of the innovative activity located within its borders. Policies that aim to augment existing capabilities (that is, deepen the knowledge base) or, alternatively, diversify the extant aggregate technological profile (therefore widening the expertise that is potentially available to indigenous
214
Inward investment, technological change and growth
firms) help to firmly position the host economy in the international business arena. The ability to attract foreign-owned activity is first and foremost dependent upon the existing absorptive capacity of the region.2 Whilst large differences in absorptive capacity may give rise to a considerable degree of clustering across regions, knowledge will flow more easily and economic activity will be more evenly spread if high absorptive capability exists across space (Maurseth and Verspagen, 1998, p.4). Thus it is paramount to foster local innovative specialisation in areas of traditional strength. Indeed, whilst evidence indicates an overall trend at country level toward increased technological specialisation over time (Pavitt, 1988; Cantwell, 1989; Archibugi and Pianta, 1992; Patel and Pavitt, 1994), at the sub-national level the broadening of specialisation is one of the possible forms of incremental change in the composition of regional innovation (Cantwell and lammarino, 1999). Policy should also aim to promote research-conducive environments that facilitate interaction and knowledge exchange between the players.3 Once such knowledge-seeking activity is located in the host economy, 'each region finds itself increasingly integrated into an international division of labour for the development of new technological systems' (Cantwell and Piscitello, 1999b, p. 19).4
DATA Patent statistics are used to analyse the location of research activity across space. The data were obtained from the United States Patent and Trademark Office (USPTO) and cover all utility patents granted to large firms located in the UK, Italy and Germany over the 1969-95 period. Each patent was classified into one of 56 technological sectors derived by mapping from the primary classification of the USPTO and organising patents into common groups (see Appendix 8A for the resulting 56 sectors). To facilitate a sub-national analysis of location, the data were also regionalised according to the residence of the firstnamed inventor (research facility responsible). This was achieved by attributing the location of the principal facility responsible for the innovation to an area code obtained from Eurostat Nomenclature of Territorial Units for Statistics (NUTS). For further discussion of the data see Cantwell and lammarino (1998, 1999, 2000) and Cantwell and Noonan (1999).
Sticky places in slippery space
215
L O C A T I O N D Y N A M I C S OF T E C H N O L O G I C A L INNOVATION It is generally acknowledged that wide disparities in technological competencies exist across the economies and regions of Europe (Caniels, 1996; Paci and Usai, 1997; Verspagen, 1997). Table 8.1 reports on the distribution and penetration of foreign-owned innovative activity across the three economies under study. While the distribution figures (D) highlight the attractiveness of the various locations against one another, the penetration statistics (P) indicate the degree to which foreign-owned activity has infiltrated the aggregate local innovative activity within each of these economies. It is immediately clear that, while the UK hosted the largest proportion of foreign-owned activity in the early years, Germany has been the most attractive location (albeit declining since the mid-1980s) for foreign-owned patenting activity since 1973. It is also interesting to note the increase in the proportion of activity located in Italy since the mid1980s. In terms of penetration, the overall proportion of patenting in Europe accounted for by foreign-owned firms has on average increased over this time period with a most noticeable rise in recent years. In marked contrast to patterns observed in the UK and Italy, where foreign-owned firms constitute a substantial and rising proportion of aggregate activity (with shares of 45.2 per cent and 57.5 per cent respectively in 1991-5), the share of foreign-owned firms in Germany has hardly changed throughout the period considered. This highlights the historically strong indigenous technological base of the German economy. While the UK and Germany are similar in that they are amongst the leaders in terms of patenting activity in Europe, with Italy being further behind, considerable sub-national differences exist across these economies. Table 8.2 records the regional distribution of patenting activity by large firms located in the UK, Italy and Germany over the 1969-95 period. The UK and Italian cases are particularly remarkable. A strong geographical agglomeration of patenting from innovative activity is found in these economies, reflecting regional disparities and the clustering activity thesis outlined above. Even allowing for potential population and economic size effects, a very high proportion of patenting activity is concentrated in just one region in both economies - the South East in the UK (47 per cent) and Lombardia in northern Italy (53 percent).
216
Inward investment, technological change and growth
Table 8.1. Distribution (D) and penetration (P) of foreign-owned activity by location, 1969-95, per cent Host
1969-72
1973-77
1978-82
1983-86
patenting
1987-90 1991-95
UK
(D) (P)
29.3 27.7
26.8 30.8
25.0 31.3
22.6 36.0
21.0 35.4
21.2 45.2
Italy
(D) (P)
4.3 27.3
4.9 31.1
4.4 26.5
4.5 32.9
6.0 43.9
6.5 57.5
Germany (D) (P)
27.0 16.3
30.2 15.6
31.8 15.2
35.6 18.8
33.5 18.1
28.9 17.4
(D) 100.0 (P) 22.7
100.0 21.6
100.0 21.4
100.0 24.4
100.0 24.9
100.0 28.6
Europe
Source: Cantwell and Piscitello (1999b).
While in Germany agglomeration of innovation is also recorded, it is spread across a greater number of regions. Concentration is relatively strong in the regions of Nordrhein Westfalen and Bayern, which together host 50 per cent of total large firm patenting over this period. Coupled with these regions in Germany, substantial agglomeration of activity is also recorded in Baden Wuerttemberg and Hessen. Together, the four regions record over 80 per cent of total MNC research in Germany.5 While not as broad as Germany, some degree of dispersion is also observed in the UK, where large firms located in the North West and West Midlands account for an additional 26 per cent of total activity. In the Italian case, however, technological activity is concentrated in just one other region, Piemonte, which records approximately 25 per cent of total patenting. Hence, even a cursory look at the structure of R&D activity in these three economies suggests that they provide a unique comparative testing ground for analysing the location choices of large innovating firms in the EU and ascertaining the factors that shape such decisions. The economies are similar in that all three record high concentrations of innovative activity within their borders which allows for a generic classification of 'core' regions (that is, those which host the highest proportions of patenting activity over time). In the more detailed analysis that follows, we restrict our study to these regional centres of excellence. However observable variations, such as a differing intensity of activity across these regional economies, coupled with the differences
Sticky places in slippery space
217
Table 8.2. Regional breakdown of USPTO patent grants to large firms, 1969-95 (percentage of each group's total grants); population and GDP by region (1995), per cent Indigenous Foreign UK East Midlands East Anglia South East West Midlands North West Wales Scotland N. Ireland Others
6.3 1.5 40.2 16.4 17.0 2.4 2.0 0.1 14.1
4.5 5.3 60.8 3.6 7.8 3.4 4.0 0.3 10.4
Total
% Pop. % Output^
5.7 2.7 47.1 12.1 13.9 2.7 2.7 0.1 12.9
7.0 3.6 30.6 9.1 10.9 4.9 8.8 2.8 22.3
6.5 3.6 35.7 8.1 9.7 4.0 8.6 2.2 21.5
100.0
100.0
100.0
100.0
100.0
23,404
11,815
35,219
58.4
785,697
ITALY Piemonte Lombardia Others
31.8 50.3 17.9
11.3 57.1 31.6
24.4 52.8 22.8
7.5 15.6 76.9
8.5 20.0 71.5
Total Italy
100.0
100.0
100.0
100.0
100.0
Total (absolute nos) 4,490.0
2,540.0
7,040
57.2
810,036
29.0 25.0 16.0 13.0 3.0 9.0 5.0
19.0 14.0 31.0 14.0 5.0 5.0 12.0
27.0 23.0 19.0 13.0 4.0 9.0 5.0
22.0 15.0 13.0 7.0 10.0 5.0 28.0
22.5 18.8 16.2 11.2 10.1 4.9 16.3
100.0
100.0 15,523
100.0
100.0
92,058
81.5
100.0 1,420,439
Total UK Total (absolute nos)
GERMANY Nordrhein Westfalen Bayern Baden Wuerttemberg Hessen Niedersachsen Rheinland Pfalz Others Total Germany Total (absolute nos)
76,535
Source: Cantwell and lammarino (1998), Cantwell and lammarino (2000) and Cantwell and Noonan (1999). Notes: (a) Total population (millions), (b) Gross value added (million Ecu).
218
Inward investment, technological change and growth
in the geographical breadth of the pattern of locational concentration, render them ideal cases for the purposes of intra-EU comparisons. In Italy, the majority of the research in indigenous Italian firms and foreign-owned firms is concentrated in Lombardia. Outside this very striking geographical agglomeration foreign-owned research appears to be relatively more dispersed than that undertaken by their indigenous counterparts. While foreign-owned firms locate approximately 68 per cent of their R&D in the two core regions of Lombardia and Piemonte, 82 per cent of patenting by indigenous firms is located there. Substantial variations in the location patterns of foreign-owned and indigenous R&D are also observable in the UK. As in Italy, foreignowned firms are more highly concentrated in the core region (the South East), than their indigenous counterparts. However, while indigenous UK firms also locate a substantial proportion of their innovative activity outside the core region in the West Midlands and the North West, in contrast to the pattern in Italy, these regions have failed to attract much foreign-owned activity. Other regions such as East Anglia and Scotland have been able to attract a relatively high share of foreignowned firm innovative activity despite their low overall share in the UK-owned figure. A similar and indeed stronger result is found in the German case. Despite the fact that Baden Wuerttemberg is only the third most popular location for the research of German-owned firms, hosting approximately 16 per cent of their research, this region represents the prime location for foreign-owned firms, which undertake 31 per cent of their research there. The same pattern is apparent in the north-west region of Niedersachsen. It hosts a low overall share of total activity (approximately 4 per cent) but over 5 per cent of the patenting activities of foreign-owned firms. Indeed the German case contrasts with patterns recorded for both the UK and Italy on a number of fronts. Whilst the leading location for the research of foreign-owned and indigenous firms is the same in the UK (the South East) and Italy (Lombardia), this is not the case for Germany. Nordrhein Westfalen (which borders Belgium and the Netherlands in the west of the country) hosts the highest concentration of indigenous activity (29 per cent), but only represents the second most popular location for foreign-owned research. Foreign-owned firms, as noted above, record their highest concentration of innovative activity in the south-west region of Baden Wuerttemberg. This differing pattern for Germany, we believe, can be explained by considering the type of technological activity associated with Nordrhein Westfalen. This region is
Sticky places in slippery space
219
the traditional home of the German chemical/pharmaceutical industry and continues to record substantial technological advantage for indigenous firms that base their research there (see Table 8.4 below). This strength is further reflected in the research profiles of the universities and research institutes located in the region (Blind and Grupp, 1999, p. 461). We tentatively suggest, therefore, that because foreign-owned chemical firms may experience difficulty in trying to access the deeply entrenched technology networks and communication channels that have evolved through time, they disperse their research more widely and account for a relatively low share of total German research in chemicals. This contrasts quite significantly with the pharmaceutical industry in the UK where, despite historical indigenous strength in these technologies, the sector is now characterised by high foreign penetration (Cantwell and Piscitello, 1999b). Table 8.3 records the contribution of foreign-owned innovation to total regional patenting activity by technological sector over the period 1969-95. 6 As expected, the highest contribution of foreign-owned activity to the regional totals occurs in the South East of the UK (43.3 per cent), Lombardia in Italy (39.1 per cent) and Baden Wuerttemberg in Germany (28.2 per cent). In all cases these contributions are well above the national average (33.5 per cent in the UK, 36.1 per cent in Italy and 16.7 per cent in Germany). With the exception of Hessen and Niedersachsen, all other regions in Germany report foreign-owned patenting activity that is considerably lower than the national figure. Indeed the concentration of foreign-owned patenting activity has been increasing over time in both of these regions (Cantwell and Noonan, 1999). The relatively higher contribution of foreign-owned activity in Niedersachsen is rooted in chemical and pharmaceutical technologies (despite the fact that aggregate foreign-owned activity is low in this technology, as highlighted above). However, for the foreign-owned firms active in these technologies, the pattern recorded in Niedersachsen suggests the importance of locating close to the bordering indigenous chemical/pharmaceutical cluster in Nordrhein Westfalen. While location within the cluster itself may be too costly (due to, inter alia, congestion effects, see Cantwell and Noonan, 1999), the ability to access indigenous expertise and knowhow (however difficult) may still be an important element in the decision of foreign firms to undertake chemical/pharmaceutical research in Germany. Foreign-owned research located in Baden Wuerttemberg is also rela-
220
Inward investment, technological change and growth
Table 8.3. Average foreign-owned firm shares of US patents by macrotechnological sector and region, 1969-95 Macro sector
Region
Chemicals Metals Mech- Elec- Trans- Other anical tronic port
Total 56^
South East West Midlands North West Total UK
42.4 22.9 28.6 37.3
48.7 24.1 16.9 34.3
41.8 13.3 25.2 35.4
52.0 10.5 20.8 43.1
24.4 3.6 33.3 17.6
32.9 7.1 7.7 23.7
43.3 10.0 18.8 33.5
Piemonte Lombardia Total Italy
18.8 32.6 37.3
12.0 22.8 20.2
31.0 58.5 47.6
15.8 74.3 55.7
10.1 4.0 8.1
9.9 18.8 17.3
16.7 39.1 36.1
3.5 10.3
33.5 14.7
23.8 14.5
39.2 8.3
27.8 6.0
17.8 7.4
11.9 10.2
33.9 7.2 39.4 2.8 8.4
30.7 41.8 14.9 37.9 28.3
41.4 40.0 27.6 32.8 25.9
46.1 37.6 21.9 37.8 25.5
7.3 56.2 33.2 45.4 15.5
38.2 13.1 11.7 10.3 15.1
28.2 18.1 23.7 9.6 16.7
Nordrhein Westfalen Bayern Baden Wuerttemberg Hessen Niedersachsen Rheinland Pfalz Total Germany
Source: Cantwell and lammarino (1998), Cantwell and lammarino (2000) and Cantwell and Noonan (1999). Note: (a) This is the average across the total of all 56 sectors, as opposed to the more restricted number of technologies developed at a regional level.
tively pronounced in electronics technologies (accounting for 46.1 per cent of total patents), as well as in the mechanical sector (most noticeably in the textile and clothing machinery, and chemical, bleaching and dying technologies). In the UK foreign-owned firms have a particularly significant presence in electronics technologies in the South East, accounting for 52 per cent of the regional total research. Their contribution is even higher within this technological field in Lombardia in Italy, accounting for almost three-quarters of all research. Foreign firms also have a substantial presence in the mechanical sector, accounting for 58y2 per cent of total regional patents. At a more detailed industrial level, a commanding foreign presence is particularly noticeable in photographic processes, representing 88 per cent of total regional patents in the South East of
Sticky places in slippery space
221
the UK, and 67 per cent of total regional patents per cent in Lombardia. The foreign share of innovative activity at regional level corresponds to that in the country as a whole for both the South East in the UK and Lombardia in Italy. In both cases, innovative activity encompasses a relatively large number of technologies, which corresponds to the national pattern recorded in both economies. Interestingly, in the German case, a great degree of correspondence to the national model is found in the Nordrhein Westfalen region - the favoured location for indigenous activity as highlighted above. Whilst Rheinland Pfalz and Hessen also correspond somewhat to the aggregate sectoral picture, the other regions of Germany (following a similar pattern to Piemonte [in Italy!, or the West Midlands and North West [in the UK]), seem to indicate more focused and locally specific sectoral patterns in the distribution of foreign-owned participation in regional research.
T E C H N O L O G I C A L PROFILES A N D T H E H I E R A R C H Y O F EU RESEARCH CENTRES One of the main drawbacks of using absolute numbers of patents is the resulting difficulty in making comparisons between the activity of heterogeneous areas of technological endeavour. Since the propensity to patent is higher in certain fields of activity (for example, pharmaceuticals), this poses potential problems when undertaking comparative analyses. This can be circumvented however, by employing the Revealed Technological Advantage (RTA) index, a technique first applied by Soete (1987) and subsequently developed by Cantwell (1989,1993). It is a proxy for technological specialisation and is calculated in the following way: RTA
=
7
'
where: Pjj = number of patents granted to technology i in region / PJM = number of world patents granted in technology i. The RTA for a given region in a specified technology will vary around unity. An index greater than one indicates a relative advantage (or specialisation) in this technology whereas an index less than one points to a relative disadvantage. It should be remembered that this is a proxy for relative advantage. A small region could demonstrate a high RTA in a particular sector but this could actually be associated with a rela-
222
Inward investment, technological change and growth
Table 8.4. Regional dispersion of technological specialisation, Region
Indigenous
1969-95
Foreign
South East West Midlands North West Total UK
0.81 1.08 0.87 0.47
0.69 0.97 1.08 0.49
Piemonte Lombardia Total Italy
6.40 1.10 2.10
1.60 2.10 1.20
Nordrhein Westfalen Bayern Baden Wuerttemberg Hessen Niedersachsen Rheinland Pfalz Total Germany
0.99 0.62 1.98 1.60 1.09 1.42 0.60
0.97 0.98 0.75 1.28 0.98 1.32 0.66
Source: Cantwell and lammarino (1998), Cantwell and lammarino (2000) and Cantwell and Noonan (1999).
tively low absolute patent count. W h a t emerges from an analysis of technological specialisation across the economies is a very clear dichotomy between higher and lower-order research locations. 7 It appears that the technological advantages of firms (both foreign-owned and indigenous) located in the UK are more widely dispersed at sectoral level than those of their counterparts located in either Italy or Germany, as measured by the standard deviation of the RTA index across fields of activity in Table 8.4. Whilst the crosssectoral variance of the RTA index for Germany does not diverge much from that observed in the UK, the Italian figure, albeit somewhat lower for the activity of foreign-owned firms located there, is in aggregate substantially higher relative to both the other t w o countries. The evidence is consistent with the UK and G e r m a n y being higher-order research locations, whereas Italy has the characteristics of a lower-order research centre, since both foreign-owned and indigenous firms are highly concentrated in their technological activities. The main explan a t i o n for this result is the smaller size of large-firm technological activity in Italy, given that there tends to be a good inverse relationship between technological size and the degree of technological specialisation (Archibugi and Pianta, 1992; Cantwell and Santangelo, 2000).
Sticky places in slippery space
223
Focusing specifically upon the differences in the specialisation patterns of foreign-owned and indigenous firms at an aggregate level, it is interesting to note that, in contrast to the Italian case, foreign-owned firms in the UK and Germany have a similar degree of technological specialisation to their indigenous counterparts. More notably still, technological activity is significantly more sectorally concentrated in indigenous firms in Germany than it is in the UK, even though the level of indigenous company patenting from Germany is over three times as high (see Table 8.2). In the light of the aforementioned expectation that there exists an inverse correlation between the standard deviation of the cross-sectoral RTA distribution and the overall size of innovative activity (see also Cantwell and Bachmann, 1998; Cantwell and Fai, 1999), this result is especially interesting, particularly so given that in Germany, unlike in the UK, the indigenous sector is considerably larger than the foreign sector - on average, approximately five times bigger. The finding that German-owned firms have a higher degree of technological specialisation in five of the six regions under study, is evidence of the very strong regional focus of these firms. Comparing the technological activity of both groups of firms across the sectors of the eleven European regions under study, it is apparent that technological activity is most broadly dispersed across sectors in the South East region of the UK. Full details are given in Appendix 8B. The Bayern region in Germany also displays a rather wide cross-sectoral distribution of large firm innovation, particularly for indigenous firms. This indicates that the general infrastructure and research environment fostered by these regions acts as a strong magnet to a wide range of corporate technological innovation. In contrast to the aggregate picture across the three economies, foreign-owned firms are generally more diversified than indigenous firms at a regional level.8 This diversity of foreign-owned innovative activity at regional level is particularly strong in the German case. This might reflect the fact that the German research infrastructure is developed to such a degree that proximity to any one cluster of activity (the indigenous hub for example) is not considered imperative in the location decisions made by foreign firms. Alternatively, it could be that activity is dispersed because firms cannot tap into all benefits from just one location. Additionally, whilst considerable regional differentiation of foreign-owned technological activity is observed in the UK and Italy, there is some similarity in the sectoral distribution of foreign-owned firm specialisation across the regions of Germany. Therefore, not only are
224
Inward investment, technological change and growth
Table 8.5. Regression results: I n R T A F O R j = a + Coefficient UK North West
Stn. Error
filnRTAINDIGj
+ 8;
T-Ratio
Prob.
Slope Intercept Slope Intercept Slope Intercept
0.791 -1.239 0.447 -2.125 0.154 -0.524
0.156 0.468 0.203 0.604 0.101 0.203
5.061 -2.646 2.200 -3.520 1.517 -2.589
[0.000] [0.011] [0.032] [0.001] [0.135] [0.012]
Slope Intercept Slope Intercept
0.111 0.429 0.541 0.150
0.159 0.106 0.109 0.091
0.703 4.058 4.950 1.645
[0.484] [0.000] [0.000] [0.105]
Slope Intercept Slope Intercept Baden Slope Wuerttemberg Intercept Hessen Slope Intercept Niedersachsen Slope Intercept Rheinland Pfalz Slope Intercept
-5.870 0.685 -0.134 0.672 -8.960 0.749 -0.169 0.809 0.173 0.522 -0.131 0.704
0.167 0.120 0.243 0.174 0.107 0.087 0.155 0.119 0.184 0.130 0.162 0.118
-0.351 5.710 -0.553 3.852 -0.838 8.648 -1.079 6.773 0.943 4.019 -0.806 5.988
[0.728] [0.000] [0.584] [0.001] [0.408] [0.000] [0.288] [0.000] [0.352] [0.000] [0.426] [0.000]
West Midlands South East
ITALY Lombardia Piemonte
GERMANY Nordrhein Westfalen Bayern
Source: Cantwell and lammarino (1998), Cantwell and lammarino (2000) and Cantwell and Noonan (1999). RTAFOR and RTAINDIG refer to the Revealed Technological Advantages of foreign and indigenous firms. the foreign-owned firms of a given industry more likely to disperse geographically their innovative activity internally within Germany than they are in the UK or Italy, they also tend to disperse themselves in a more similar cross-regional fashion than might have been expected. The notion that all German regions conform to higher-order locations is further supported by considering the degree to which the foreignowned RTA distribution in each of the regions correlates with that in G e r m a n y as a whole (see Cantwell and N o o n a n , 1999). This is in
Sticky places in slippery space
225
marked contrast to results obtained for the two other economies, which, notwithstanding the regions of the South East and Lombardia, otherwise demonstrate weak correlations with the aggregate foreign-owned specialisation patterns. To test the hypothesis that lower order regions are characterised by very similar foreign/indigenous technological specialisation patterns, a simple regression analysis was carried out for each of the regions studied. We regress the sectoral specialisation profiles of foreign-owned firms by region and country between 1969-95 on that of indigenous firms at both regional and national level. The results are shown in Table 8.5. In the South East region of the UK, Lombardia in Italy and all six German regions, we find little evidence of a statistically significant linkage between the sector and regional specialisations of foreign and domestic firms. This suggests that these regions attract foreign-owned firms for reasons other than their existing indigenous technological specialisation. Foreign firms may instead be attracted to these regions by their wider technological competencies and the available level of infrastructural supports (for example, the degree of openness to foreignowned investors, business climate, corporate and enterprise culture, language, and so on). The technological activity of foreign-owned and indigenous firms in these eight regions is typically broad-ranging in nature and extends across a spectrum of technologies. This is not to discard the possibility that intratechnological communication (or technology spillovers) takes place between firms. Indeed, while our results suggest that limited intraregional, inter-firm communication takes place in the same technological sector, knowledge spillovers in higher-order regions seem to operate mainly through exchanges in and around core technological systems. Relationships then form between actors in otherwise quite separate alternative fields of specialisation. Such core systems appear to be rooted in the background engineering, mechanical methods and electronics technologies - sectors in which foreign-owned and indigenous firm's technological advantages appear to overlap in these higher-order centres. It is suggested that foreign-owned firms' research in the biotechnology and pharmaceutical fields in the South East region of the UK for example, are attracted by this region's expertise in related disciplines (medical sciences for example),9 rather than by sector-specific indigenous capabilities. Similarly the technological profile of foreignowned firm activity in Baden Wuerttemberg in Germany seems to
226
Inward investment, technological change and growth
suggest that firms locate there not for any specialised expertise specific to this region but rather for the general capabilities of the region (particularly in mechanical engineering and information systems). Regression results for the other regions of the UK and Italy suggest the presence of lower-order research locations. Technological specialisation profiles of foreign-owned firms were found to be closely correlated with those of indigenous technological expertise so that knowledge spillovers are likely to be intrasectoral in nature. This is consistent with our hypothesis that lower-order regions attract the innovative activities of foreign-owned MNCs because of a very particular set of sector-specific expertise on offer in these regions. By basing research facilities in such locations, foreign-owned MNCs may be able to upgrade their own technological capabilities in particular technological fields which may be sub-sets of their own major areas of technological interest. For a similar analysis at a cross-country rather than an intra-country level of the locally specific pattern of activity in the leading European centres, see Cantwell and Janne (1999).
SUMMARY A N D C O N C L U S I O N S The recent acceleration of the internationalisation (international spread) and globalisation (international integration) of large firm R&D has resulted in a revival of the importance of location in the international business literature. While in the past, the exploitation of research by MNCs in their home country resulted in comparatively little feedback from the new knowledge being generated in foreign jurisdictions, more recently such firms have come to rely upon home-base augmenting activities to achieve their overall corporate growth strategy (Kummerle, 1999; Cantwell and Piscitello, 2000). Tapping into local knowledge sources through cooperative agreements and research alliances means that the ability of foreign locations to attract such high-value activities constitutes a vital component of the host economy's industrial policy and overall growth strategy. This chapter, in setting out to investigate the centripetal forces in operation, analyses the location of innovative activity across three selected European countries. While substantial agglomeration of activity is recorded across this space, differences in the geographical spread of such concentrations within national boundaries are highlighted, thus lending further support to the thesis that even within higher-income
Sticky places in slippery space
227
economies, significant regional disparities exist. While research is highly concentrated in all three economies, it polarises in just two regions in Italy, three in the UK and across six regions in Germany. This variety in the geographical agglomeration of technological activity offers a unique testing ground for our analysis of foreign-owned firm location and indeed provides a number of interesting observations. Firstly, this chapter provides support to the hypothesis that hierarchies of research locations exist across economies. In the UK and Italy just one region attracts foreign-owned activity for its overall business climate, support infrastructure and wider range of skills and expertise rather than for any specific intrasectoral knowledge base. However in Germany most foreign-owned activity is concentrated in such centres. This is confirmed by the fact that foreign-owned technological activity in each of the German regions is not correlated with that of the indigenous firms, as well as by the fact that foreign-owned technological specialisation for each of these regions correlates with aggregate foreign-owned specialisation in the relevant country. Indeed Germany is unique on a number of fronts. Whilst both indigenous and foreignowned innovation favours mainly the same region in the UK and Italy, a divergence of preferences is observed for Germany. It is suggested that foreign-owned firms may be especially attracted to Baden Wuerttemberg because of the highly developed Mittelstand (the innovative small and medium-sized industrial sector), although this finding requires further investigation. In addition, while the regional differentiation of indigenous large firms in Germany is greater than that of their foreign-owned counterparts, foreign-owned firms are less locationally selective in accordance with this distinct pattern of domestic innovation. This might be explained by foreign-owned firms' preference to avoid direct competition with the local science-based strengths of German industry (most notably the chemical/pharmaceutical cluster). Instead the main motivation for foreign-owned firms to locate their research in Germany is to access the more general core technological expertise available there. In contrast, outside the core regions of the South East and Lombardia, foreignowned firms in the UK and Italy are highly focused on accessing local indigenous expertise, for example in chemicals in the North West region of the UK, or vehicles in Piemonte. Potential knowledge spillovers in these regions are likely to be largely intrasectoral in nature. Our results highlight the distinctive nature of regional technological activity across Europe. Although still susceptible to further elaboration,
228
Inward investment, technological change and growth
they may help in the design of policy tools to influence the impact of globalisation upon regional locational hierarchies. This in turn would signal the opening of a new phase of interregional development across Europe.
Sticky places in slippery space
APPENDIX 8A: THE 56 FIELDS OF TECHNOLOGICAL ACTIVITY 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
Food and tobacco products Distillation processes Inorganic chemicals Agricultural chemicals Chemical processes Photographic chemistry Cleaning agents and other compositions Disinfecting and preserving Synthetic resins and fibres Bleaching and dyeing Other organic compounds Pharmaceuticals and biotechnology Metallurgical processes Miscellaneous metal products Food, drink and tobacco equipment Chemical and allied equipment Metal working equipment Paper making apparatus Building material processing equipment Assembly and material handling equipment Agricultural equipment Other construction and excavating equipment Mining equipment Electrical lamp manufacturing Textile and clothing machinery Printing and publishing machinery Woodworking tools and machinery Other specialised machinery Other general industrial equipment Mechanical calculators and typewriters Power plants Nuclear reactors Telecommunications Other electrical communication systems Special radio systems Image and sound equipment Illumination devices Electrical devices and systems Other general electrical equipment Semiconductors Office equipment and data processing systems Internal combustion engines
229
230
43 44 45 46 47 48 49 50 51 52 53 54 55 56
Inward investment, technological change and growth
Motor vehicles Aircraft Ships and marine propulsion Railways and railway equipment Other transport equipment Textiles, clothing and leather Rubber and plastic products Non-metallic mineral products Coal and petroleum products Photographic equipment Other instruments and controls Wood products Explosive compositions and charges Other manufacturing and non-industrial
Sticky places in slippery space
231
APPENDIX 8B Table 8B.1. RTA index (UK) relative to global technological specialisation, by sector and region, 1969-95 Sectors
Regions South East UK
1 3 5 6 7 9 11 12 13 14 16 17 20 23 25 28 29 33 34 35 36 37 38
39 40 41 42 49 50 51 53 56
2.74 0.65 0.87 0.23 0.66 0.75 1.31 3.48 0.75 0.68 0.90 0.62 0.54 0.18 0.34 1.09 1.28 0.98 0.93 3.46 0.57 1.96 0.85 0.58 0.34 0.48 1.35 1.19 0.69 1.23 1.12 1.03
Foreign 0.62 0.46 0.77 2.20 0.80 0.37 1.26 3.48 0.63 1.24 0.59 0.67 0.69 0.45 0.16 1.20 0.64 1.35 1.78 2.41 1.22 1.27 1.13 0.66 1.31 0.83 0.57 0.33 0.49 0.54 1.06 1.08
West Midlands UK 0.42 0.55 0.87 0.02 0.37 0.37 0.19 0.11 1.42 1.57 0.54 1.54 0.61 1.15 2.22 1.40 5.21 0.47 1.08 0.23 0.06 1.07 1.17 1.35 0.42 0.32 2.30 3.55 0.64 0.16 0.65 0.98
Foreign 0.58 1.13 1.12 0.32 0.50 0.25 0.21 0.92 3.95 3.17 1.83 2.67 1.43 0.76 0.00 2.82 1.88 0.20 1.10 0.41 0.21 0.50 0.69 0.46 0.47 0.28 0.78 1.60 0.87 0.00 1.02 0.24
Source: Cantwell and lammarino (2000).
UK North West UK 0.56 0.95 1.30 0.19 4.16 1.45 2.46 2.08 0.64 0.77 0.75 0.60 0.40 0.16 0.65 0.73 0.53 0.24 0.71 0.18 0.06 0.21 0.73 0.76 0.10 0.24 0.10 1.54 2.33 0.54 0.55 0.86
Foreign 1.34 2.00 1.40 0.95 5.16 1.42 2.48 3.21 0.80 0.42 1.08 0.17 0.48 0.35 1.75 1.44 0.58 0.19 0.51 0.00 0.24 0.15 0.34 0.21 0.60 0.26 0.22 0.46 0.50 0.07 0.45 0.33
UK Foreign 1.71 0.95 1.00 0.20 1.30 0.82 1.20 2.00 1.06 0.97 0.91 0.94 0.60 0.60 1.54 1.21 2.08 0.73 0.81 1.74 0.32 1.13 0.81 0.76 0.31 0.37 1.14 1.74 1.01 0.90 0.95 1.27
0.88 0.57 0.91 1.62 1.61 0.66 1.24 2.81 0.84 1.28 0.76 0.88 0.96 1.18 1.84 1.47 0.78 1.18 1.46 1.58 0.89 0.88 0.95 0.58 1.01 0.72 0.48 0.65 0.60 0.39 0.93 0.99
232
Inward investment, technological change and growth
Table 8B.2. RTA index (Italy), relative to global technological specialisation, by sector and region, 1969-95 Sectors
Regions Piemonte
3 5 6 7 9 11 12 13 14 16 17 18 20 25 28 29 30 33 34 38 39 40 41 42 49 50 53 56
Italian
Foreign
1.83 0.26 0.06 0.69 0.48 0.84 0.19 0.81 1.11 0.74 1.23 0.10 1.78 0.34 0.84 1.60 34.56 0.63 1.54 0.69 1.30 0.18 1.16 3.72 0.83 0.28 1.33 0.28
0.34 0.61 0.32 0.30 0.12 0.18 0.40 0.72 1.75 0.42 6.50 0.50 0.96 0.56 2.10 4.28 4.25 0.45 0.23 3.39 0.53 0.18 0.36 2.08 1.19 0.22 0.28 0.00
Italy Lombardia
Italian
Foreign
Italian
1.37 1.22 0.02 0.96 2.38 2.41 2.99 0.22 0.63 1.29 1.19 0.19 0.49 2.56 1.68 0.41 0.54 0.19 0.27 0.64 0.33 0.07 0.18 1.41 4.64 0.51 0.60 1.52
0.13 1.27 0.16 0.27 0.44 1.91 3.06 0.37 0.83 0.58 1.18 2.89 0.69 9.86 0.50 0.35 6.10 0.98 0.41 2.08 1.48 2.23 1.21 0.09 0.29 0.47 0.37 0.14
1.60 0.89 0.03 0.82 1.79 1.79 1.86 0.50 0.86 1.31 1.19 0.23 1.04 3.01 1.26 0.90 11.48 0.33 0.71 0.57 0.67 0.09 0.52 2.17 2.90 0.63 0.79 1.31
Source: Cantwell and lammarino (1998).
Foreign 0.34 1.09 2.08 0.47 0.65 1.49 2.04 0.76 1.04 0.81 1.73 3.02 0.96 5.82 1.26 0.84 3.96 0.69 0.39 1.90 1.23 1.73 0.84 0.34 0.84 0.35 0.41 0.20
I
ON ON NO ON
8 O
o
o o
x
s CO PQ oo
J2
O
CD
CQ
c/D
b>
"53 J*
O
O P-H
6 o X WO O PH
o
DC bX) O PH
bX)
B o DP bJD
O
DC bJD O PH
toJO
a
o X
O
O
O
O
O
O
T-H
O
(N
T-H c o
T-H
co
o
H
^j-
Sticky places in slippery space
o d d
co
T-H T-H ( N
O ^ t
^t"
CO T-H O
O
O
O
O
O
(N
d
O
T-H O C O T - H C \ 1 0 T - H T - H O C N T - H O T - H C O
T - H C O C O O C O C O ^ O N - ^ O O O O N O
N r n N t H H O \ H H C O O O ^ ) ( T \
^
t ^ °^ t ^ ^
T-H o
( N O O O N ^ H O O l O O O ^ ^ O N O N H d d d T-H' d d
T-HCNT-HOT-HT-HCOT-HT-H
o
NO T-H T-H T-H ( N
( N r H H r n N o o ^ N ( N ( N l \ r o r > j r n q r n \ 0
T-H o
O N OO iO ON IN co
^ ( N ^ O ^ O N C ^ ^ ( N O O ( N ^ I N O O M D O N O O [ \
O O N ^ N i o \ 0 ( N ( N r n ( N N ( ^ r n i 0 ^ r n O ( N ^t; ^t; IN i o t \ 00 (Nj ^ t (N o O ON IN. NO O IN ON CO
^
O
O
M
^
^
O
N
(
N
N
t
V
O
t
H
(
N
r
n
n
d d d T ^ d d d d d T - H c o O T - H d ^ t - c o c N c o
N
( N H H r H N r H N r i r i d d d d d d ^ o d
H
O
O
O
O
<
N
O
O
O
O
T
-
H
O
O
T
-
H
O
^
O
T
^
O
T
O
o
to O
T-H
T-H
O
H NO t
T-H CO T-H
O
-
O
O
T
0
O
T H ( N N r o r n ^ O O N ^ N O N O O ' t ( N O O ( N ^ O \
^
O
0
ro O oo l ^
0
O
0
N 0\
0
oo \c
0
O
0
t H
^
n o t o o \ O H ( s a \ , t N \.DrHT)-rH^t't\DOOOlO 0
O
-
H
O
O
O
O
O
O
NO
CO Tj"
I N N - O N - T H O N T ^ N . NO (N ^ "
00 lO r-H
iO NO IN ON
iO NO OO ON N (N M M
O c O C O N - ^ c N O N N - ^ H O c O O O O N O N c O r H O r H N ^ O O \ t N N f O H \ 0 ^ ^ 0 0 0 ^ CN co CN T-H O O O O O O
T ^ O T ^ O O O O O O O C O T - H T - H < N < N O < N
^ \ o ^ o o i o \ o N ( N ^ o o o o x m H \ o t o
'NI"
ON O
CO
233
234
O
'53 £
c
d)
T-H T-H* d o d o
T-H O
r \ ^ o o c o c o i ^ c N i c o N o o o o o i > o ' T - H O O O O O O O o
T-H O
O
O
O
H
O
O
T-H
O
o
r o o o T - H
o o o
O
T-H r-H T-H
l O l N r o c o c N ^ - T f - O N C N ^ H T f i O O N C N O O r H O O
CN IN NO CO CN ON rf «*• C O O O O O r H r - rHH O O Oc O
TH
OO CO NO IN NO Tf CO O IN O CN T-H O O O CN
O O O O c N O O O O O c N
i>0 ON CO NO NO ON T f T-H cN cN o o ON
^t- wo O O O r ^ c N O c N C O O N ^ O C N O O O N O T - H O O iO NO O ^ 0 ( N N 0 c 0 C 0 N 0 l \ r \ O N D N 0 N - W 0 C 0
r O t O M 3 \ ^ ^ ^ N t H \ O t \ H O O H O O H
en S T-H d T-^ d d
co NO r\ ^t- N- OO to co NO O ON CO (N O Tf NO O CN T-H NO T-H T-H N- T-H u^ rt• TJ- NO Tf CO CO NO ^ " CO d ^o ^ d d d T-H d
T-HT^T-HdOCNOOT-HT-HO
CO IN *>0 CO N- ^f C O O N ^ J - N - N D O O | \ O T - H I O N O N- ON CN CN O CO H H N O t r o r n o o O f N t
O M m ' O ' t O O O N N N V O i o O I - i o H f S CO T-H H » o TJ- cN T}- O T-H O O O OO O NO NO i>0
T-HONNDCO^|-(NNOIOOCN|COONOONO(NONO O c O c O ^ N O O O O C N l ^ v O t ^ c O c O O O i O C O T f T j -
0 0 ( N \ D O O H \ O O O N N H r O ( N O t O \ N N N O t ^ l N r ^ i O i O C O O c O O N O O O N O O N O N O
do
O t O O H i o o O t H H O O O N r O f N N ^ n
T - H O ^ J - O O c O N O N - T - H O N - O O i v O r t i - O O O C O O C N T - H O O O O C N O T - H T - H O T - H O N T - H T - H C N ^ -
( N O O O O O O O O O c o
r 0 M t 0 0 O \ N N ^ 0 0 \ D N ( N ( N ^ ^ O t t n i o H N ' O t N H r n t V O ^ N H N H
o d o d o d d c N
ON O I O NO CO O CN ^ | - O O C N O O ^ O O O I > O T - H O NO T T CO NO IN N- ^ CO O CN CN ON IN NO OO CN
T f N O ^ O O i > O l N C O c O C N | N O N - O O t o N - ^ O T - H i o 0 0 ( N OO ON T-H H 0 0 N - I N CN ^O N - ON 0 0 ^ f
Inward investment, technological change and growth
c o '5b
QQ
C
a
.S c .SP
-M c/3
O PH
"TJ J-I O
^3 Z£ o X
T - H C O ^ N O I N O O O N O T — I C N C O I N O N O C N C O N O r n r n r o r o r n r O r O T t T j - r j - T j - T j - r j - i ^ i ^ i o i r )
Sticky places in slippery space
235
NOTES 1 The globalisation of innovation refers, in fact, not only to the 'use' of MNC technological competencies but, even more, to the way they are generated. By 'globalisation' we refer to a high degree of interdependency among the unit, which constitutes the MNC. 2 Highlighted as being a 'necessary though not sufficient condition' (Cantwell and lammarino, 2000). 3 Which, as previously highlighted, is contrary to perceived wisdom regarding knowledge appropriation (Cantwell, 1999; Cantwell and Noonan, 1999; Cantwell and lammarino, 2000). 4 See also Cantwell (1991,1994), Teece etal., (1994), Loasby (1998) and Chandler etal.(1998). 5 The regions of Rheinland-Pfalz and Niedersachsen are also included in the analysis because of the relatively high absolute number of patents granted to firms located there. 6 As explained in Cantwell and lammarino (1998, 2000) and Cantwell and Noonan (1999), a number of technological sectors were dropped from the analyses. These correspond to technologies in which relatively low degrees of global patenting were recorded. Moreover the three studies restrict their analysis to the most prominent technological sectors, which vary across the three economies, although the total 56 column refers to the total number of patents for all 56 technological sectors. 7 The fact that a number of regions had to be dropped from the analysis, having inadequate numbers for statistical purposes, lends support, at least at first glance, to the hypothesis that internal geographical hierarchies exist in these economies. 8 There are, however, three exceptions - the North West in the UK, Lombardia in Italy and Bayern in Germany - in which indigenous innovation is characterised by activity across a broader range of technologies. 9 The regional scientific advantage is noticeably strong in this area (see Cantwell and lammarino, 2000).
REFERENCES Archibugi, D. and Pianta, M. (1992), 'Specialisation and size of technological activities in industrial countries: the analysis of patent
236
Inward investment, technological change and growth
data', Research Policy, 21(1), pp. 79-93. Blind, K. and Grupp, H. (1999), Tnterdependencies between the science and technology infrastructure and innovation activities in German regions: empirical findings and policy consequences', Research Policy, 28, pp. 451-68. Caniels, M. (1996), 'Regional differences in technology. Theory and empirics', Maastricht, MERIT Research Memorandum, 2/96-009. Cantwell, J.A. (1989), Technological Innovation and Multinational Corporations, Oxford, Basil Blackwell. (1991), 'The theory of technological competence and its application to international production', in McFetridge, D. (ed.), Foreign Investment, Technology and Economic Growth, Calgary, University of Calgary Press. (1992), 'The internationalisation of technological activity and its implications for competitiveness', in Granstrand, O., Hakanson, L. and Sjolander, S. (eds), Technology Management and International Business: Internationalisation of R&D and Technology, Chichester, John Wiley. (1993), 'Corporate technological specialisation in international industries', in Casson, M.C. and Creedy, J. (eds), Industrial Concentration and Economic Inequality, Aldershot, Edward Elgar. (1994) (ed.), Transnational Corporations and Innovatory Activities, London, Routledge. (1995), 'The globalisation of technology: what remains of the product cycle model?', Cambridge journal of Economics, 19(1), pp. 155-74; reprinted in Chandler, A.D., Hagstrom, P. and Solvell, O. (eds) (1998), The Dynamic Firm. The Role of Technology, Strategy, Organisation and Regions, Oxford and New York, Oxford University Press. (1999), 'Innovation as the principal source of growth in the global economy', in Archibugi, D., Howells, J. and Michie, J. (eds), Innovation Policy in a Global Economy, Cambridge and New York, Cambridge University Press. Cantwell, J.A. and Bachmann, A. (1998), 'Changing patterns of technological leadership: evidence from the pharmaceutical industry', International Journal of Innovation Management, 2(1), pp. 45-77. Cantwell, J.A. and Barrera, M.P. (1998), 'The localisation of corporate technological trajectories in the interwar cartels: co-operative learning versus exchange of knowledge', Economics of Innovation and New Technology, 6(2-3), pp. 257-90.
Sticky places in slippery space
237
Cantwell, J.A. and Fai, F.M. (1999), 'The changing nature of corporate technological diversification and the importance of organisational capability', in Dow, S.C. and Earl, P.E. (eds), Contingency, Complexity and the Theory of the Firm: Essays in Honour of Brian J. Loasby, Cheltenham, Edward Elgar. Cantwell, J.A. and lammarino, S. (1998), 'MNCs, technological innovation and regional systems in the EU: some evidence in the Italian Case', International Journal of the Economics of Business, 5(3), pp. 383-408. (1999), 'EU regions and multinational corporations: change, stability and strengthening of technological comparative advantages', paper presented at the Regional Science Association 39th European Congress, Dublin, August. (2000), 'Multinational corporations and the location of technological innovation in the UK regions', Regional Studies, 34. Cantwell, J.A. and Janne, O.E.M. (1999), 'Technological globalisation and the innovative centres: the role of corporate technological leadership and locational hierarchy', Research Policy, 28(2-3), pp.119-44. Cantwell, J.A. and Noonan, C.A. (1999), 'The regional distribution of technological development by foreign-owned firms in Germany', mimeo, University of Reading. Cantwell, J.A. and Piscitello, L. (1999a), 'The emergence of corporate international networks for the accumulation of dispersed technological competences', Management International Review, 39 (Special Issue 1), pp. 123-47. (1999b), 'Corporate diversification, internationalisation and location of technological activities by MNCs: differences between EU and non-EU firms in the European regions', mimeo, University of Reading. (2000), 'Accumulating technological competence - its changing impact on corporate diversification and internationalisation', Industrial and Corporate Change, 9. Cantwell, J.A. and Santangelo, G.D. (2000), 'Capitalism, profits and innovation in the new techno-economic paradigm', Journal of Evolutionary Economics, 10(1), pp. 131-57. Chandler, A.D., Hagstrom, P. and Solvell, O. (eds) (1998), The Dynamic Firm, the Role of Technology, Strategy, Organisation and Regions, Oxford and New York, Oxford University Press. Chesnais, F. (1992), 'National systems of innovation, foreign direct in-
238
Inward investment, technological change and growth
vestment and the operations of multinational enterprises', in Lundvall, B.A. (ed.), National Systems of Innovation, London, Pinter. Dunning, J.H. and Lundan, S.M. (1998), 'The geographical sources of competitiveness of multinational enterprises: an econometric analysis', International Business Review, 7, pp. 115-33. Dunning, J.H. and Wymbs, C. (1997), 'The geographical sourcing of technology-based assets by multinational enterprises', mimeo, Rutgers University. Fors, G. (1998), 'Locating R&D abroad: the role of adaptation and knowledge-seeking', in Braunerhjelm, P. and Ekholm. K. (eds), The Geography of Multinational Firms, Dordrecht, Kluwer Academic Publishers. Granstrand, O. and Sjolander, S. (1992), 'Internationalisation and diversification of multi-technology corporations', in Granstrand, O., Hakanson, L. and Sjolander, S. (eds), Technology Management and International Business: Internationalisation of R&D and Technology, Chichester, John Wiley. Kuemmerle, W. (1997), 'Building effective R&D capabilities abroad', Harvard Business Review, March/April, pp. 61-70. (1999), 'The drivers of foreign direct investment into research and development: an empirical investigation', Journal of International Business Studies, 30(1), pp. 1-24. Loasby, B.J. (1998), 'The organisation of capabilities', Journal of Economic Behavior and Organization, 35, pp. 139-60. Markusen, A. (1996), 'Sticky places in slippery space: a typology of industrial districts', Economic Geography, 72(3), pp. 293-313. Maurseth, P.B. and Verspagen, B. (1998), 'Knowledge spillovers in Europe and its consequences for systems of innovation', ECIS Working Paper 98-001. Paci, R. and Usai, S. (1997), 'Technology enclaves and industrial districts: an analysis of the regional distribution of innovative activity in Europe', CRENOS Working Paper, University of Cagliari. Patel, P. and Pavitt, K.L.R. (1994), 'Uneven (and divergent) technological accumulation among advanced countries: evidence and a framework of explanation', Industrial and Corporate Change, 3(3), pp. 759-87. Patel, P. and Vega, M. (1999), 'Patterns of internationalisation of corporate technology: location vs. home country advantages', Research Policy, 28, pp. 145-55.
Sticky places in slippery space
239
Pavitt, K.L.R. (1988), 'International patterns of technological accumulation', in Hood, N. and Vahlne, J.E. (eds), Strategies in Global Competition, London, Croom Helm. Pearce, R.D. (1999), 'Decentralised R&D and strategic competitiveness: globalised approaches to generation and use of technology in multinational enterprises (MNEs)', Research Policy, 28(2-3), pp.157-78. Soete, L.L.G. (1987), 'The impact of technological innovation on international trade patterns: the evidence reconsidered', Research Policy, 16, pp. 101-30. Teece, D.J., Rumelt, R., Dosi, G. and Winter, S.G. (1994), 'Understanding corporate coherence: theory and evidence', Journal of Economic Behaviour and Organisation, 23, pp. 1-30. Vernon, R. (1966), 'International investment and international trade in the rpoduct cycle', Quarterly Journal of Economics, 80, pp. 190207. Verspagen, B. (1997), 'European "Regional Clubs": do they exist and where are they heading? On economic and technological differences between European regions', paper presented at the Conference on Economic Growth and Change: A Comparative Perspective, Cagliari, June. Zander, I. (1999), 'Whereto the multinational? Evolution of technological capabilities in the multinational network', International Business Review, 8, pp. 261-91.
9
Foreign trade and FDI stocks in British, US and French industries: complements or substitutes? Lionel Fontagne and Michael Pajot
INTRODUCTION This chapter presents new empirical evidence about the impact of foreign direct investment on the level and composition of international trade. Foreign direct investment (FDI) and international production have both continued to expand rapidly in recent years. Global inflows of FDI rose by an average 19 per cent per annum between 1991 and 1997, and by an estimated 38 per cent in 1998, to reach roughly $644 billion. Around 50,000 parent companies, with 450,000 affiliates now operate worldwide (UNCTAD, 1998). The foreign affiliates alone account for 6 per cent of world GDP, compared with 2 per cent in 1982 (Hummels, 1998). Intra-firm trade now accounts for one-third of world exports, with the sales of foreign affiliates having risen faster than global trade in recent years. The impact of FDI on trade has recently been the subject of considerable scrutiny in the debate over whether to implement a Multilateral Investment Agreement (MIA) on FDI. Drabek (1998) suggests that the reluctance of many countries to push for an MIA arises from a concern that the effects of FDI on economic development are not fully understood. If FDI displaces trade, exports (imports) will be replaced by local sales on foreign markets. In contrast, if trade and FDI are confirmed as complements, investing abroad will lead to greater exports to (imports from) host countries. Drabek argues that in fact the existing empirical literature provides clear evidence of a complementary relationship between FDI and trade, although there are differences between countries and sectors. The recent WTO working group on the relationship between international trade and investment reached a broadly similar conclusion (WTO,1998).
240
Foreign trade and FDI stocks
241
However theoretical issues would suggest that there is no a priori reason why FDI should always be expected to be complementary to trade. Indeed, at the level of the firm, such complementarity is especially questionable. Hence the level of any empirical analysis matters. Substitution between exports and outward investment at the firm level may still be associated with complementarity at the industry level, since spillovers between firms are accounted for in the latter case. In the same way, spillovers between industries, within the manufacturing sector, provide an additional basis for complementarity. In addition, unobserved macroeconomic factors, such as country size and per capita income, can help to generate any observed macroeconomic complementarity. It must also be kept in mind that FDI flows, FDI stocks, local sales and local production are different measures of foreign activity that may generate different results. To our knowledge, these complex relationships have never been addressed in the literature in a systematic manner, using data broken down by sector, country and partner, and in an international comparative perspective. This chapter seeks to provide such a perspective and is organised as follows. The potential mechanisms linking trade and FDI are examined in the next section, based on a brief survey of the literature. A full description of our disaggregated trade and FDI stock data follows. This data set is used to investigate whether trade and FDI are complements or substitutes at the industry level in the United Kingdom, France and the United States. The empirical results are then presented and the final section offers some concluding comments. M E C H A N I S M S LINKING T R A D E T O FDI 1 Following Stevens and Lipsey (1992), it has become usual to distinguish between the financial and production implications of FDI. The former highlight the potential for substitution between domestic and foreign investment if firms have limited financial resources, but have only indirect implications for trade flows. Here we focus on the direct production implications of FDI and ask whether FDI displaces trade, output and employment. Two types of impact from FDI can be identified. Outward FDI may displace trade at the level of the firm, since foreign affiliates' local sales substitute at least partially for exports. However this is not necessarily the case at the industry level or at the macroeconomic level. In
242
Inward investment, technological change and growth
addition, investing abroad can induce intra-firm imports if labour-intensive or resource-intensive activities are relocated abroad. Hence trade by the home country is not necessarily displaced by outward FDI. Equally the trade of the host country is not necessarily displaced, since imports are likely to increase at least in the short term, while enhanced competitiveness is likely to boost exports as well. Most empirical studies of the relationship between trade and FDI attempt to estimate the direction and the magnitude of the impact of FDI on trade. Such estimates provide a direct measurement of substitution or complementarity effects. Explaining exports (imports) as a function of FDI and different control variables, such studies focus upon the sign and magnitude of the parameter estimate(s) on the foreign investment variable(s). If a negative parameter on FDI is obtained in equations explaining the level of trade, then FDI is a substitute for trade and exports are at least partially replaced by local sales on foreign markets. Home country production and employment are then negatively affected. The trade balance of the host country would benefit from this substitution effect provided no additional imports from the parent company were induced. Trade and FDI are complements if a positive parameter is obtained. In this case investing abroad leads to an increase in exports from the home country to the host country. Accordingly the impact on the host country trade balance should be negative, even if some of the investor's gains were balanced by market share losses for third-country competitors. Depending on the type of FDI data under consideration (firm level, industry level, the manufacturing sector, the whole economy total, flows or stocks) the results could change dramatically. It might be thought obvious that trade and FDI are alternative strategies. Firms can alternatively produce at home and export, or produce abroad and substitute foreign affiliates' local sales for exports. In the latter case, local sales partially substitute for previous exports of the investing company, even if local sales and local production require imported inputs. At the macroeconomic level the equalisation of factor prices across countries can be brought about either through international trade or through the international mobility of factors of production, with factor mobility acting as a substitute for trade. Despite such considerations, the conclusions of empirical studies at the microeconomic level and the macroeconomic level are far from clear cut. The business literature compares the effects of alternative ways through which firms choose to exploit their specific advantages. A useful
Foreign trade and FDI stocks
243
survey is provided by Blomstrom and Kokko (1994). It is clear that FDI helps to bring an increase in the foreign market share. There is also evidence that higher exports of intermediate products to foreign subsidiaries help to compensate for the crowding out of exports of finished goods. Lipsey and Weiss (1984) highlight the fact that one dollar of local production induces 9 to 25 cents of additional exports from the investing country. Not surprisingly, the links between trade and local output are found to be three times as large at the industry level than at the firm level due to externalities between firms within industries. Swedenborg (1979) concludes that any loss of exports of finished goods due to FDI from Sweden was more than offset by higher intermediate exports. 2 This result was confirmed in a subsequent study in 1982 by the same author; each dollar of local sales substitutes for only 2 cents of exports and 'creates' 12 cents of new exports, the net effect being a positive complementarity of 10 cents. Blomstrom etal. (1988) also find weak evidence in favour of complementarity. Andersson (1993), and Blomstrom and Kokko (1994) suggest that outward FDI has had more effect on the structure of exports from Sweden than on their total value. Riker and Brainard (1997) stress the vertical splitting up of processes associated with FDI. FDI does not displace output but splits the production process worldwide, thus generating trade in intermediate products. Using individual company data for 14,096 French industrial firms in 1993, Chedor and Mucchielli (1998) show that French firms with offshore affiliates export more than other firms. The economic activities of offshore affiliates have a beneficial effect on exports from their parent firms. Thus, overall, it is far from clear that the empirical evidence supports the microeconomic hypothesis of substitution between FDI and trade. Swenson (1999) argues that different types of foreign direct investment may have different effects on trade. 3 Investments can be broken down into greenfield ventures, expansions of capacity, mergers and acquisitions and joint ventures. The stock of inward FDI in the United States was found to complement imports. The inference is that foreign affiliates in the United States use many components sent to them by their parent companies or import other models of their range for resale. In the empirical work below it will be shown that this effect is largely due to the sectoral composition of FDI and American trade. With regard to the type of investment, Swenson found that the expansion of capacity is the only one to have a positive and significant impact on American exports and imports. In contrast, mergers and acquisitions act as sub-
244
Inward investment, technological change and growth
stitutes. It should be noted that this empirical evidence relates to the stock of greenfield investments rather than to flows. At the macroeconomic level, trade and FDI relationships vary according to the type of data, econometric methodology and country. The principal difficulty in the search for macroeconomic evidence is to distinguish between complementarity associated with FDI and complementarity associated with general macroeconomic conditions. Like trade, FDI is a phenomenon largely concentrated in North America, Europe and Asia, and it is to be expected that both will be positively associated with economic growth and the level of income in these regions. The co-determination of the two modes of internationalisation, for instance by size of economy and level of income, should not of course be confused with any complementarity that might exist between them. Eaton and Tamura (1994), controlling for factors that jointly determine trade and FDI, identify a large, positive relationship between outward FDI and exports, as well as imports, for both Japan and the US. However there is less evidence of a relationship with inward FDI. In a study using aggregate time-series data, Andersen and Hainaut (1998) find evidence of complementarity between exports and outward FDI flows for the United States, Japan, and Germany, but not for the United Kingdom. Pain and Wakelin (1998) challenge the issue of complementarity for eleven OECD countries over 1971-92 in a study of total merchandise exports in which they control for world demand. They focus on export performance, in effect market share, rather than on the absolute level of exports. Their results indicate that the relationship between FDI and exports varies across countries and over time. However inward FDI is found to have a positive effect on the exports from most of the host countries, with the opposite effect found for outward FDI. These contrasting findings from different empirical studies are not surprising since theory does not provide a clear guide to the mechanisms linking foreign activity and exports. New theories of trade under imperfect competition stress two important factors - the way in which firms are organised and the dual effect of economies of scale and transport costs. Horizontally integrated firms will carry out most of their production of a given commodity at a given location, where possible close to the market if transport costs are high and minimum plant size not too large. In contrast vertically integrated firms will locate different parts of their production process in different foreign affiliates to form what is conventionally termed a 'global firm'. Foreign trade and
Foreign trade and FDI stocks
245
investment are likely to reinforce each other in this case (Head and Ries, 1997). The second major determinant of the relationship between FDI and trade relates to the dual effect of economies of scale and transport costs. Increasing returns place a limit on the optimal number of plants, while transport costs and other obstacles to trade have the opposite effect (Brainard, 1993). When the firm has high fixed costs, each plant has limited fixed costs and there are non-negligible transport costs, a multinational will locate its units of production close to its markets and FDI may substitute for trade (Markusen and Venables, 1995; Swenson, 1999). Brainard (1993) demonstrates this by considering trade and local sales by American industries in 27 partner countries. Where transaction costs (transport costs and trade barriers) are high, they are shown to act as a brake on exports, but promote local sales by affiliates. Brainard also finds that there is a positive relationship between the standard of living in the partner country and the net sales of American affiliates established there. In contrast there is a negative relationship between the standard of living and net exports to the relevant market, and thus again some evidence of substitution of local sales for trade in countries with high living standards. Bayoumi and Lipworth (1997) complement these results by clarifying the puzzling empirical issue of the different results found using FDI flows and stocks. Outflows of FDI from Japan are found to be positively related to fixed investment in Japan and the recipient country, and negatively related to the real exchange rate and the lagged stock of investment in the recipient country. The negative impact of the lagged stock indicates that a dynamic stock adjustment process is at work. This can explain why the relationship between trade and FDI can differ according to whether flows or stocks are used. In the short term, flows of new investments will enhance exports from the investing country, reflecting the supply of capital goods required to establish the subsidiary. However over time the subsidiary gains greater autonomy, the share of local content in production increases and the substitution effect dominates, with exports from the parent company being replaced by local production. Bayoumi and Lipworth test this hypothesis by including both (lagged) FDI flows and stocks in their equations for exports and imports. They find a short-run complementarity between trade flows and FDI flows, but there is no significant relationship observed between trade and FDI stocks. They conclude that 'FDI has only a temporary impact on exports, consistent with the view that FDI influences exports
246
Inward investment, technological change and growth
largely through the short-term need to equip new factories' (p. 24). In the following sections we examine the robustness of this conclusion by testing for a possible long-term relationship using data on FDI stocks for three countries, the United Kingdom, the United States and France. All three countries have good bilateral industry-level data. The model we use attempts to control for 'new trade theory' variables such as economies of scale and transportation costs. These results complement those using flow data in Fontagne and Pajot (2000).
T H E DATA SET Our data set for the United States, the United Kingdom and France is constructed so as to ensure a common nomenclature for FDI stocks and bilateral trade flows. The matching has been done for each country separately, owing to large differences in the coverage of the national FDI data. In the French case the classification is done according to the sector of the subsidiary for inward investment stocks and refers to the sector of outward investment stock otherwise. In the US and British cases, however, it systematically refers to the sector the subsidiary belongs to. Hence comparisons have to be cautious. We match these data sets with trade data in order to estimate gravity-type equations at the industry level. For the UK our data set covers 27 countries and 13 manufacturing sectors for a ten-year period (1987-96). It has been constructed specifically for this chapter.4 For the US our data set covers 42 countries and 23 sectors, of which 12 are manufacturing ones, for a six-year period (1989-94). For France we consider 55 countries and 35 sectors, of which 17 are manufacturing sectors, also over 1989-94. 5 All FDI data are converted into US dollars. Further details are provided in the Data Appendix. 6 However there are a number of missing observations, so that the eventual number of observations used in estimation is considerably smaller. The main problem lies in obtaining data for direct investment stocks at the same level of detail as available for trade. Thus for France we eventually consider 17 sectors for 48 partner countries, giving 4,896 observations in all. In general it is easier to obtain comprehensive data on direct investment flows than it is for direct investment stocks. Our objective is to explain bilateral exports and imports by industry using outward and inward FDI in the corresponding industries. This will
Foreign trade and FDI stocks
247
provide an answer to questions such as, 'How does $1 of US (French/ British) investment in the automobile industry in Brazil affect US (French/British) exports of automobiles to Brazil?' We use a gravity model which includes our FDI measures as well as country-specific and industry-specific variables. We consider four separate FDI stocks for each pair of trading countries, the bilateral outward and inward stocks between the reporting country / and its partner /', and the aggregate outward and inward stocks of FDI for country i with all other countries. These variables are denoted OUT, IN, OUTOTH and INOTH. The country-specific variables include market size, proxied by the average GDP of the trading countries (AVRGDP), the Balassa-Bauwens (1987) measure of differences in their GDPs (DFGDP), the average income per capita of the two countries (a proxy for the demand for variety and the standard of living, AVRGDPPC), the difference in their income per capita (a proxy for economic distance, DFGDPPC), geographical distance (DIST), common borders (ADJ), and regionalisation as proxied by the existence of preferential commercial schemes (CPOL), such as the EU and NAFTA. There are several possible industry-specific variables that might be used, such as concentration, economies of scale, the share of white collar workers in employment or capital intensity. It is difficult to construct data on all of these given the high level of industry disaggregation. Thus we introduce only economies of scale. A representative economy was constructed by pooling British, French, German and Italian firms by size. The relative productivity of larger firms (with 500 or more employees) was then calculated using NACE three-digit-level data. We assume that the productivity differential is positively correlated with scale economies and also assume that it is the same for each industry k in each reporting country i. The need to construct this data is one factor which limits the number of industries that can be included in the empirical exercise. Our estimates integrate the proximity-concentration trade-off introduced by Brainard (1993), since both distance and scale economies are introduced as explanatory variables, or as various combinations of fixed effects. In most cases the familiar macroeconomic variables have the expected sign when they are significant. Average market size (declaring country and partner), average income per capita, economic distance, adjacency and the regional integration proxy are all found to have a positive impact on the value of trade flows, while the difference in market size has
248
Inward investment, technological change and growth
a negative effect. These findings are common to many other studies and so we focus primarily on the 'new trade theory' variables - FDI, returns to scale and transaction costs.
EMPIRICAL RESULTS The empirical model For each of the three countries we consider a separate panel of industries and partner countries. Given the large discrepancies in parameter estimates we find between these three countries and their different industrial coverage, it is not appropriate to estimate a pooled relationship. We consider the panel as a sequence of (annual) periods, so that the panel has three dimensions - industries, partners and time. Hausman tests favour the choice of fixed effects for each panel against the alternative of random effects. In each model a Lagrange Multiplier test due to Breusch and Pagan decisively rejected the use of a single intercept against the alternative of either fixed or random effects. 7 We also include time dummies (denoted et below) in the French and US panels, but not in the UK panel since they are rejected by a Fisher test. We estimate four different specifications in order to ascertain whether our results are sensitive to the form of fixed effects used in estimation. We alternate between fixed effects for partner countries and fixed effects for industries. Variables that were insignificant in all the regressions are not reported in the tables of results. Our first model consists of bilateral import and export equations that allow for the standard vector of macroeconomic variables plus our four bilateral FDI variables. It also includes the measure of economies of scale plus fixed effects for each partner country (ej) to account for the proximity-concentration trade-off. As the measure of scale economies does not vary over time we cannot also include industry-specific fixed effects. The dependent variable Z-k denotes exports from (imports in) country / to (from) country / in industry k. Zljkt = a.AVRGDP^
+a2DFGDPljt+a3AVRGDPPCtjt
+a4DFGDPPCljt + a5CPOLljt + a6SCALEk +a7OUTlfkt+a8INtlkt+a9OUTOTHlkt +a10INOTHikt+ej+et
Z=XM
(9.1)
Foreign trade and FDI stocks
249
In the second model we replace the scale economies variable with industry-specific fixed effects. This will control for all factors that vary between industries but do not vary across time. The fixed effects for partner countries are also dropped and replaced by variables for distance, common borders and membership of a regional integration scheme. These do not vary over time and so they cannot be included with country-specific fixed effects. Zijkt =
aiAVRGDfyt+^DFGDPijt+a^VRGDPPCjjt
+a4DFGDPPCtjt+a+5CPOE a5CPOL tjt ijt + a6DISTtj +a7ADJljt + asOUTtjkt+a9INt]kt I +alQOUTOTH lkt Z = X,M
+ anINOTHtkt
'2)
+ek+et
The final two models omit either the scale economies variable or the distance and adjacency variables. Equation (9.3) is the same as (9.2), but without the distance and border measures, so that there is nothing picking up country-specific fixed effects. Equation (9.4) is the same as (9.1), but without the scale economies effect, so that there is nothing picking up industry-specific fixed effects. However, as it is not possible to derive our scale economies measure for all industries, the sample used in estimating (9.1) is smaller than that used to estimate (9.4). Zijkt = axAVRGDPljt
+a2DFGDPljt
+a3AVRGDPPCljt
+a4DFGDPPCljt+a5CPOEljt+a6OUTljkt +a7INljkt + asOUTOTHlkt
+ a9INOTHlkt
+ek+et
(9.3)
Z = X,M Zijkt =
a1AVRGDPijt+a2DFGDPijt+a3AVRGDPPCijt
+a4DFGDPPCtjt+a5CPOLljt+a6OUTljkt +a7INtjkt + asOUTOTHikt
+ a9INOTHtkt
+ ef + et
(9.4)
Z = X,M This procedure provides a means of assessing the impact of omitting key variables on the FDI estimates. On the whole, countries with which France or the United States trade and invest the most are also countries
250
Inward investment, technological change and growth
with which bilateral links are the strongest in terms of culture, history, language and geographic proximity. Without country-specific dummies (fixed effects or distance) some of these effects may be attributed incorrectly to the FDI variables, resulting in biased estimates.
The British case The empirical results for UK exports are reported in Table 9.1 and those for UK imports are reported in Table 9.2. It is clear that there is a significant positive relationship between the o u t w a r d FDI stock and both the level of exports and the level of imports. Each additional dollar invested abroad raises exports by 2 - 4 cents, and raises imports by 6-6V2 Table 9.1. Export equations, United Kingdom Equation (9.1) OUT IN AVRGDP DFGDP AVRGDPPC DFGDPPC
0.021* (1.621) 0.034*** (3.339) 0.0005* (4.256) 455.97 (1.104) 0.025** (2.394) -0.023** (-2.828)
DIST CPOL SCALE Nobs Adj. R2 F value Hausman test
-32.597 (-0.287) -25.428 (-0.617) 1908 0.55 70.66 17.41***
Equation (9.2) 0.037*** (3.083) 0.031*** (3.157) 0.0004** (20.137) -433.60*** (-7.766) 0.005* (1.882) -0.021*** (-6.958) -0.024*** (-8.059) 267.56*** (8.357)
2029 0.57 133.64 3.60
1987-96 Equation (9.3)
Equation (9.4)
0.043*** 0.024* (3.509) (1.910) 0.031*** 0.034*** (3.199) (3.477) 0.0004**'= " 0.0005*** (18.850) (4.206) 439.130*** 378.50 (-7.744) (0.976) 0.007*** 0.023** (2.711) (2.273) -0.011*** -0.021*** (-4.000) (-2.686)
407.520* (27.321)
2029 0.55 133.02 4.44
-12.860 (-0.117)
2029 0.54 76.27 17.22***
Note: ^-statistics in parentheses, significant at the level: 1 per cent (***), 5 per cent (**), 10 per cent (*).
Foreign trade and FDI stocks
251
Table 9.2. Import equations, United Kingdom 1987-96 Equation (9.1) OUT IN AVRGDP DFGDP AVRGDPPC DFGDPPC DIST CPOL SCALE Nobs Adj. R2 F value Hausman test
Equation (9.2)
Equation (9.3)
Equation (9.4)
0.064*** 0.060*** 0.060*** 0.066*** (3.198) (3.085) (3.343) (3.465) -0.062*** -0.068*** -0.068*** -0.061*** (-4.362) (-4.141) (-4.146) (-4.288) 0.0005*** 0.0006* *' 0.0005** * 0.0005*** (2.764) (2.908) (16.173) (16.136) 788.17 -681.52*** -686.560*** 684.37 (1.277) (-7.590) (-7.609) (1.177) 0.028* 0.019*** 0.021*** 0.028* (1.790) (4.490) (4.978) (1.832) -0.017 -0.008* 0.0008 -0.015 (-1.664) (-1.371) (-1.284) (0.180) -0.021*** (-4.555) -16.342 401.97*** 529.190*** -15.521 (12.174) (-0.095) 7.806 (-0.094) -78.959 (-1.282) 1908 0.48 54.60 13.81**
2029 0.42 75.26 3.08
2029 0.42 77.36 3.82
2029 0.47 57.39 13.92**
Note: ^-statistics in parentheses, significant at the level: 1 per cent (***), 5 per cent (**), 10 per cent (*). cents. These figures are small, however, in comparison with the ones that are obtained using data for FDI flows. Estimates not reported here indicate that an extra dollar of outward FDI flows is associated with approximately 30 cents of additional exports and imports. The differences between the stock and flow effects are similar to those found by Bayoumi and Lipworth (1997). There is also a significant positive relationship between the stock of inward investment and the level of exports to the country of the investor, with a one dollar increase in inward investment raising exports by 3-3Vi cents. This suggests that inward investment has indeed helped to improve British export performance. In contrast to outward investment, inward FDI appears to be associated with a significantly smaller level
252
Inward investment, technological change and growth
of bilateral imports, with a one dollar rise in the stock reducing imports by 6V4-6V4 cents. This is consistent with the hypothesis that the assembly of finished products has been transferred to the UK from the investing country, in contrast to the results found for France and the US below. However it is not possible to test this hypothesis formally within our present data set. Overall a rise in inward investment has a positive impact on the balance of trade, more than offsetting the negative impact from outward investment. However we do not consider that our results point to a strong relationship between FDI stocks and trade at the sectoral level, as the magnitudes of the coefficients suggest that direct investment can explain only a small part of the overall level of bilateral trade. One other notable feature of our UK results, which again is in contrast to the findings for France and the US, is that direct investment stocks in or from third countries (the OUTOTH and INOTH variables) were found to be insignificant. This suggests that affiliates in different countries may not be especially integrated with each other. Looking at the other variables in the model we find, as expected, significant positive effects from the levels of GDP and GDP per capita, while distance appears to have a significant negative effects on trade levels. However the scale economies measure is not significant. The dummy for membership of a regional integration arrangement (in this case for membership of the EU) is significant only when industry-specific fixed effects are included in equations (9.2) and (9.3).
The French case The results for France are reported in Tables 9.3 and 9.4. On the whole they also point to a weak complementarity between FDI stocks and the overall level of trade, although the magnitude of the individual effects from outward and inward investment varies across the four different models we estimate. The most robust finding is that inward investment is associated with a significant positive impact on the level of imports into France. This might reflect either a growth in the level of intermediate inputs or the extent to which inward investment helps to expand the available market for finished goods. A one dollar rise in inward investment raises imports by between 46-73 cents. Once again we find that inward investment has a significant positive impact on the level of exports, although it is smaller than the impact on imports, so that inward investment is associated with a deterioration in the trade balance.
Foreign trade and FDI stocks
253
Table 9.3. Export equations, France 1989-94 Equation (9.1) OUT IN OUTOTH INOTH AVRGDP
0.063*** (2.819) 0.284*** (8.585) 0.015*** (5.678) 0.014*** (3.425) 0.0004*** (4.152)
DFGDP DFGDPPC
-0.011*** (-2.993)
CPOL SCALE
Equation (9.3)
-0.020 0.058** (-1.001) (2.088) 0.409*** 0.488*** (12.582) (12.447) -0.002 -0.0009 (-0.184) (-0.123) 0.013 0.011 (1.589) (1.146) 0.0002*** 0.0001** * (4.992) (15.111) -190.59*** -568.012*** (-7.665) (-19.991) -0.002** -0.015*** (-2.175) (-14.946) 331.831*** (19.102)
Equation (9.4) 0.024 (1.097) 0.287*** (9.109) -0.005** (-2.044) 0.003*** (8.730) 0.003*** (3.575)
-0.001*** (-2.615)
44.453* (1.823)
ADJ Nobs Adj. R2 F value Hausman test
Equation (9.2)
743.651*** (33.271) 3987 0.69 141.83 -
4896 0.57 213.18 -
4896 0.36 101.25 4.83
4896 0.61 122.58 9.7
Note: ^-statistics in parentheses, significant at the level: 1 per cent (***), 5 per cent (**), 10 per cent (*). The complementarity between inward FDI and trade flows reflects the fact that countries that trade the most with France are also those which invest the most in France. If the gravity model is estimated without including fixed effects for partners, there is a danger that the parameter on the inward FDI variable may be artificially inflated. This possibility raises concern about the magnitude of any complementarity effect, even if the underlying direction of the relationship is accepted. There is a significant negative relationship between the stock of outward investment and the level of bilateral imports in all the estimated models. In contrast, outward investment has a significant positive re-
254
Inward investment, technological change and growth
Table 9.4. Import equations France, 1989-94 Equation (9.1) OUT IN OUTOTH INOTH
-0.093*** (-3.381) 0.462*** (11.307) 0.005 (1.508) 0.003 (0.541)
AVRGDP DFGDP AVRGDPPC DFGDPPC
0.061 (0.665) -0.007 (-1.416)
CPOL SCALE
Equation (9.3)
-0.182*** -0.097*** (-6.270) (-2.881) 0.652*** 0.735*** (15.773) (15.516) 0.0005 -0.0003 (0.052) (-0.028) -0.001 -0.003 (-0.135) (-0.267) 0.0002** 0.0001** (14.219) (5.719) -198.19*** -598.77*** (-6.281) (-17.444)
-0.003*** (-2.507) 364.92*** (16.554)
Equation (9.4) -0.114*** (-4.252) 0.521*** (13.403) -0.006** (-2.105) 0.022*** (4.608)
-0.017*** (-13.953)
53.207* (1.771)
ADJ Nobs Adj. R2 F value Hausman test
Equation (9.2)
779.93*** (27.5) 3978 0.67 128.04 -
4896 0.48 151.68 -
4986 0.31 79.94 3.48
4986 0.56 102.91 19.21***
Note: ^-statistics in parentheses, significant at the level: 1 per cent (***), 5 per cent (* *), 10 per cent (*).
lationship with the level of exports apart from when industry-specific fixed effects and the adjacency variable are included. The latter is preferred to distance as a proxy for transaction costs. The findings for exports suggest that either the positive effects from outward investment are largely concentrated in adjacent countries (that is, other EU economies), or they arise in industries with time-invariant advantages that can otherwise be picked up by the fixed effects. For instance, it could be the case that France imports more (is disadvantaged) in certain industries,
Foreign trade and FDI stocks
255
while France mainly invests in advantaged industries. Advantaged industries invest abroad and export more and import less. Conversely, disadvantaged industries import more and invest less abroad. Averaging across the four models, an extra dollar of outward investment is estimated to improve the trade balance by approximately 15 cents, even though it may be associated with an overall reduction in the absolute amount of trade. An extra dollar of inward investment does, however, stimulate additional trade, although it is on balance associated with a decline in the trade balance of approximately 23 cents. Of course it should be remembered that these are average effects across all nineteen industries; they do not necessarily hold for every industry in the panel. There is some evidence of a significant relationship with stocks in third countries, although this varies across the different models. Inward investment by third countries has a significant positive effect on exports to partner countries except when industry-specific fixed effects are included, suggesting that it may be picking up some particular factors that do not vary over time. Again it appears that inward investment acts to stimulate trade, while outward investment to third countries acts to substitute for trade. However, in both cases, the effects associated with third-country investment are small.
The US case Results for the United States are reported in Tables 9.5 and 9.6. The most notable feature of these results is that the effects obtained from the stock of outward investment are far larger than those found for either the UK or France. There appears to be a strong positive relationship between the outward stock and bilateral exports and imports, which overall is associated with a decline in the trade balance. The net imports associated with US outward FDI are large, particularly when allowance is made for fixed effects for partner countries. There is an approximate 'one-to-one' complementarity between the outward FDI stock and the level of exports. At the industry level, each dollar invested abroad is associated with an extra dollar of exports. The results obtained for the imports equation indicate that it might also be associated with up to $1.70 of extra imports. The most likely explanation for these large coefficients is the extent to which US multinationals have been able to restructure their operations to outsource to affiliates in labour-abundant developing economies with low labour costs. It is interesting to note that the effect from outward investment on exports
256
Inward investment, technological change and growth
Table 9.5. Export equations, United States 1989-94 Equation (9.1) OUT IN OUTOTH INOTH
0.923*** (13.247) 0.106* (1.753) 0.014 (1.48) -0.014 (-1.167)
AVRGDP CPOL SCALE
0.719*** (10.333) -0.017 (-0.352) 0.002 (0.439) -0.006 (-0.248) 0.002*** (6.967) 4942.601*** (5.698)
Equation (9.3)
Equation (9.4)
1.191*** (16.446) -0.161*** (-2.945) 0.002 (0.037) 0.002 (0.065) 0.001*** (3.641)
0.937*** (12.026) 0.041 (0.546) 0.056*** (5.741) -0.051*** (-3.921)
21458.0*** (13.021)
ADJ Nobs Adj. R2 F value Hausman test
Equation (9.2)
3562.612** (5.135) 715 0.64 33.79 12.6**
715 0.64 72.89 2.13
715 0.52 49.81 1.29
715 0.55 24.15 8.6*
Note : ^-statistics in parentheses , significant at the level: 1 per cent (***), 5 per cent (**), 10 per cent ('=").
is lowest in equation (9.2) which includes a dummy for the NAFTA economies. An expansion in inward investment is also found to be associated with a deterioration in the trade balance in each of the four models considered. However the effects are generally much smaller than obtained for outward investment. The huge US market attracts foreign investors and there are few instances of foreign firms investing in the US in order to ship products back home. It seems likely that the larger effects found from inward investment in models (9.1) and (9.4) reflect factors in particular industries, since the coefficients become much smaller when industry-specific fixed effects are included in models (9.2) and (9.3). Again this suggests that US industries with particular competitive advantages import less and receive less inward investment.
Foreign trade and FDI stocks
257
Table 9.6. Import equations, United States 1989-94 Equation (9.1) OUT IN OUTOTH INOTH
1.734*** (13.669) 0.614*** (5.611) 0.027* (1.608) -0.079*** (-3.508)
AVRGDP DFGDPPC CPOL SCALE
Equation (9.2) 1.051*** (7.514) 0.061 (0.618) -0.032 (-0.391) 0.006 (0.121) 0.006*** (11.423) 0.136*** (4.000) 6918.0*** (3.875)
Equation (9.3)
Equation (9.4)
1.531*** (11.707) -0.083 (-0.836) -0.047 (-0.581) 0.016 (0.291) 0.006*** (9.778) 0.129*** (3.741)
1.752*** (12.914) 0.518*** (4.435) 0.086*** (5.055) -0.133*** (-5.645)
715 0.47 37.17 2.92
715 0.53 22.53 11.97***
29808.1*** (9.931)
AD] Nobs Adj. R2 F value Hausman test
1908.221 (1.334) 715 0.58 27.68 13.27***
715 0.51 39.77 3.22
Note: ^-statistics in parentheses, significant at the level: 1 per cent (***), 5 per cent (**), 10 per cent (*). CONCLUSION FDI and international production are continuing to expand rapidly. Intra-firm trade now accounts for one-third of world exports and the sales from foreign affiliates are rising faster than global exports. A key empirical question is whether FDI displaces trade, output and employment. This chapter contributes to that debate. Outward FDI may displace trade at the level of the firm, since affiliate sales may substitute directly for exports. However this need not necessarily be the case at the industry level; this chapter has tried to tackle the latter issue in a systematic manner, using data broken down by sector, country and partner, in a comparative perspective.
258
Inward investment, technological change and growth
In models in which exports (imports) are related to FDI and different control variables, interest lies in the sign and magnitude of the parameter estimate associated with foreign investment. A negative parameter on FDI indicates that FDI displaces trade. In this case, production and employment can be negatively affected in home country by outward investment. If a positive parameter is obtained, trade and FDI are complements, which is likely to stimulate production in the investing country. The results from models of this type seem to vary according to the type of FDI data used. Permanent and temporary effects of FDI on trade must be disentangled. According to Bayoumi and Lipworth (1997), the former can be identified using FDI stocks, whereas the latter are associated with FDI flows. The focus of this chapter has been on the potential size of the permanent effects. Accordingly we have sought to model bilateral exports in terms of macroeconomic variables, variables associated with the socalled 'proximity-concentration trade-off and FDI stocks. These equations have been estimated using three panels of industries, with the FDI and trade data carefully matched by industry. In the British case, outward FDI has only a small complementary effect on trade flows. We estimate that each additional dollar on the FDI stock is associated with an increase of 2 cents in British exports and an increase of 6 cents in British imports. There is thus an overall deterioration in the balance of trade in that sector. These figures are rather small compared with the temporary effects from flows of foreign direct investment, with each additional dollar on the flow associated with roughly 30 cents of additional imports and exports. The stock of inward investment in the UK is found to substitute for imports, but to stimulate exports, implying a net positive effect on the balance of trade. At first glance the results for France are similar, with the stock of outward investment being positively associated with exports. However this outcome is largely due to the fact that those industries that are relatively export-intensive are also the ones that tend to invest overseas. If this structural relation is controlled for, by introducing fixed effects on industries, there does not appear to be a significant permanent association between outward investment and exports. However we fail to find any evidence that suggests that there is a permanent trade displacement when industries invest abroad, contrary to popular fears. Inward investment in France does appear to have a well-determined and comparatively large complementary association with trade flows. A rise of one dollar is associated with a permanent increase in net imports of 20 cents
Foreign trade and FDI stocks
259
when controlling for the geographic composition of French exports and 25 cents when controlling for the geographic structure. These results can be compared with the transitory impact of FDI. Controlling for economies of scale, Fontagne and Pajot (2000) record that a one-dollar outward FDI flow is associated with 54 cents of additional bilateral exports, and with 24 cents of additional imports, in the industry considered. The US evidence points to very different conclusions. First, we find a strong complementarity between outward FDI and trade. Each additional dollar on the FDI stock is associated with sizeable net imports, ranging from 30 cents to 80 cents according to the specification considered. These are striking orders of magnitude since permanent effects are involved. In contrast the inward investment stock is not found to have any significant impact on US exports and leads to additional imports only if controls are not included for the sectoral composition of imports. These results can be compared once again with the transitory effects. With fixed effects for periods only, Fontagne and Pajot (2000) obtain a 2.7 dollar complementarity for each dollar of US outward FDI flows. Turning to inward FDI, they obtain a significant transitory impact, with 35 cents of additional exports and twice as much additional imports. The evidence for the UK and France largely validates the conjecture that the primary effects of FDI on trade are transitory. The long-term effects, from FDI stocks, are small, with the exception of inward FDI in the French case. However this does not appear to be a conclusion that can be applied to all countries, since there appear to be large permanent effects from the stock of US outward FDI, although it is reassuring that they point to a positive long-run association between foreign direct investment and the level of trade. Finally, some limitations of this analysis should be noted. Correlation does not mean causality, and it may be factors other than FDI which are ultimately behind trade imbalances. We also do not take into account the potential long-run benefits, such as technology transfer, generally associated with inward FDI. It may also be useful to consider jointly the transitory and permanent effects of FDI by including lagged variables in the empirical analysis.
260
Inward investment, technological change and growth
DATA APPENDIX X (M): British, US or French exports (imports) to (from) individual partners ($ million). OUT: British, US or French outward FDI stocks to individual partners ($ million). IN: British, US or French inward FDI stocks from individual partners ($ million). OUTOTH: Total British, US or French outward FDI stocks excluding bilateral stocks with the partner country ($ million). INOTH: Total British, US or French inward FDI stocks excluding bilateral stocks with the partner country ($ million). AVRGDP: Average British, US or French GDP and partner country's GDP($million). AVRGDPPC: Average of British, US or French per capita income and partner country per capita income ($million). DFGDPPC: Economic distance between / and / proxied by the absolute value of the difference in per capita incomes. DFGDPPC = \GDPPQ - GDPPCj\; u n i t * US $. DFGDP: Difference in size between country i and country/, proxied by the difference in GDPs between the two countries. The absolute value between these GDPs is replaced by a normalisation proposed by Balassa and Bauwens (1987). The latter indicator is not sensitive to the absolute size of the countries, in contrast to the former. It takes values ranging from 0 to 1. The more the countries are similar, the more w tends to 0.5, the more the indicator tends to 0. [w\nw+{l-w)\n(l-w)] \ ' ln(2)
DFGDP = 1 +x
>. i
_
with w= ™
VV1, X1
GD?t GDPj+GDP;
DIST: Geographic distance between the capitals of countries / and /; unit : kilometre; source : Pc-Globe@.
Foreign trade and FDI stocks
261
ADJ: Dummy for adjacency between / and /; if i and / have a common frontier, then ADJ = 2, or else ADJ = 1. EU: Dummy assessing for the common participation in a regional integration scheme of /' and / (2,1). SCALE: economies of scale are estimated by the productivity of large firms (500 plus employees) relative to that of small firms, using a hypothetical economy constructed by combining NACE three-digit-level data for the UK, France, Germany and Italy. CPOL: Dummy for common involvement of countries / and / in a regional integration scheme.
NOTES 1 The interested reader will find an earlier detailed survey in Cantwell (1994) and a complementary review of the empirical literature in Pain and Wakelin (1998). 2 The experience of Sweden has been studied widely because of the detailed database on Swedish multinational companies maintained by the Industrial Institute for Economic and Social Science Research in Stockholm. 3 The author uses an ITA (International Trade Administration) database relating to foreign FDI in the United States over the period 1974-94. This database permits the construction of stocks of each type of FDI (greenfield and so on) for each country investing in the United States. 4 We acknowledge financial support from the OECD and help from Simon Harrington at the UK Office for National Statistics. 5 It is difficult to extend this data set as there is a definitional change in 1996 when reinvested earnings are included for the first time. They are included throughout in the UK and US data. 6 Full details on the industries and countries included are available on request from the authors. 7 Large values of the Hausman tests favour fixed effects against random effects. Large values of the LM test favour models with specific effects against the alternative of an OLS model.
262
Inward investment, technological change and growth
REFERENCES Andersen, P.S. and Hainaut, P. (1998), 'Foreign direct investment and employment in the industrial countries', Bank for International Settlements Working Paper, 61. Andersson, T. (1993), 'Nya villkor for ekonomichi och politik. Ekonomikommissions forslag', SOU, 6, 3 pp. 85-107. Balassa, B. and Bauwens, L. (1987), 'Intra-industry specialization in a multi-country and multilateral framework', Economic Journal, 97, pp. 923-39. Bayoumi, T. and Lipworth, G. (1997), 'Japanese foreign direct investment and regional trade', IMF Working Papers, WP/97/103. Blomstrom, M. and Kokko, A. (1994), 'Home country effects of foreign direct investment: evidence from Sweden', CEPR Discussion Paper no. 931. Blomstrom, M., Lipsey, R. and Kulchycky, K. (1988), 'US and Swedish direct investment and exports', in Baldwin, R. (ed.), Trade Policy Issues and Empirical Analysis, Chicago, University of Chicago Press. Brainard, S.L. (1993), 'An empirical assessment of the proximity-concentration tradeoff between multinational sales and trade', NBER Working Paper no. 4583. Cantwell, J.A. (1994), 'The relationship between international trade and international production', in Greenaway, D. and Winters, L.A. (eds), Surveys in International Trade, Oxford, Basil Blackwell, pp. 303-28. Chedor, S. and Mucchielli, J.L. 1998), 'Implantation a l'etranger et performance a l'exportation. Une analyse empirique sur les implantations de firmes franchises', Revue Economique, 49, 3, pp. 617-28. Drabek, Z. (1998), 'A multilateral agreement on investment: convincing the sceptics', WTO Staff Working Paper, ERAD-98-05. Eaton, J. andTamura, A. (1994), 'Bilateralism and regionalism in Japanese and US trade and direct foreign investment patterns', NBER Working Paper no. 4758. Fontagne, L. and Pajot, M. (2000), 'Relationships between trade and FDI flows within two panels of US and French industries', in Lipsey, R.E. and Mucchielli, J.L. (eds), Multinational Firms and Impacts on Employment, Trade and Technology. New Perspectives for a New Century, forthcoming. Head, K. and Ries, J. (1997), 'Overseas investment and firm exports',
Foreign trade and FDI stocks
263
University of British Columbia Working Paper. Hummels, D. (1998), 'Toward a geography of trade costs', mimeo, University of Chicago. Lipsey, R.E. and Weiss, M.E. (1984), 'Foreign production and exports of individual firms', Review of Economics and Statistics, 63, 4, pp.488-94. Markusen, J.R. and Venables, A. J. (1995), 'Multinational firms and the new trade theory', NBER Working Paper no. 5036. Pain, N. and Wakelin, K. (1998), 'Export performance and the role of foreign direct investment', The Manchester School Supplement, 66, pp. 62-88. Riker, D.A. and Brainard, S.L. (1997), 'US multinationals and competition from low wage countries', NBER Working Paper no. 5959. Stevens, G. and Lipsey, R. (1992), 'Interaction between domestic and foreign investment', Journal of International Money and Finance, 11, pp. 40-62. Swedenborg, B. (1979), The Multinational Operations of Swedish Firms, Stockholm, Almqvist & Wicksell International. Swenson, D. (1999), 'Foreign investment and international trade linkages', mimeo, University of California, Davis. UNCTAD (1998), World Investment Report: Trends and Determinants, Geneva, United Nations. WTO (1998), Report of the Working Group on the Relationship Between Trade and Investment to the General Council, WT/WGTI/ 2, December.
This page intentionally left blank
Index
absorptive capacity 214 acquisitions and mergers 243-4 adoption, technology 45-7 after-care programmes 28 agglomeration economies 4 - 5 , 162-3, 213 R&D in UK, Italy and Germany 215-21, 226-7 regional policy 168-9, 169-70 Aitken, B.J. 106, 112-13, 170-1, 173 Alexander, J.M. 183 AMOSFDI 179, 181-4, 201-3 Andersen, P.S. 244 Andersson, T. 243 appropriable technology 35-6 appropriation, risk of 37-8 ARD (ABI Respondents Database) 12, 123, 124, 125-7 foreign ownership 127-37, 142, 149-50; analysis of foreign advantage 131-7; characteristics of survivors 127-31; macroeco-nomic instability 146-7 see also Census of Production Armington, P. 182 Arrowsmith, J. 6 Baden Wuerttemberg (Germany) 216,217,218,220,227 technological specialisation 222, 224, 225-6, 233-4 Balassa, B. 247 Bank for International Settlements 2 Barrell, R. 5, 22, 23, 66, 74, 78, 80, 82,90, 179, 181, 191, 192, 199, 200 productivity growth 163, 168 technical progress 14, 16, 42, 67, 70, 71, 72, 180
Barrell continued US affiliates in UK and Germany 27-8 Barry, F. I l l Basant, R. 51 Bauwens, L. 247 Bayern (Germany) 216, 217, 220 technological specialisation 222, 223, 224, 233-4 Bayoumi, T. 52, 245-6, 251, 258 Bellak, C. 10 Belot, T. 112 Bernard, A.B. 21 Bernstein, J.I. 49, 50, 51 Blaas, E. 167 Blake, A.P. I l l Blanchflower, D.G. 182 Blind, K. 219 Blomstrom, M. 40, 43, 45, 55-6, 73, 168, 243 booms and recessions 146-7, 151 Borensztein, E. 25-6 Bradley, J. I l l , 185 Brainard, S.L. 243, 245, 247 Braunerhjelm, P. 169 Caballero, R.J. 171 Cabral, S. I l l Cameron, G. 85 Canada 19 technology policy 56-7 Cantwell, J.A. 10, 36, 210, 212, 226 agglomeration economies 163, 169 R&D 212, 213, 214; location dynamics 216, 217, 219, 220; technological specialisation 221, 222, 224, 231-4 capacity, expansions of 243-4 capital, cost of see cost of capital
265
266
Index
capital deepening 170 capital expenditure 164, 165, 166 capital intensity: and labour productivity 12, 86-8, 93-5, 122-61 capital-skill complementarity 145, 151 capital stock 126-7, 131, 183 Caves, R.E. 42, 44, 73, 167 Census of Production 68 technical progress in manufacturing 83-96 see also ARD central coordination 28 Centre for Research in Globalisation and Labour Markets 104-5 centres of excellence, technological 40 CES production function 70-1 Chedor, S. 243 chemical industry 219 Chen, E.K.Y. 46 City of London 5, 77 Cobb-Douglas production function 68-9, 69-70, 73 Coe, D.T. 5 1 , 7 0 companies sample 11, 12, 123-4, 137-8, 154-5 ownership and productivity 13943, 150-1 competition negative spillovers 170-1, 172, 173 product market 17-18, 39-40, 112, 114, 115-18 Competitiveness White Papers 1 complementarity capital-skill 145, 151 FDI and trade 240-63 technology 4 1 , 42, 43, 53-4 computable general equilibrium analysis see regional computable general equilibrium analysis concentration industrial 47 proximity-concentration trade-off 247, 248, 258
core technological systems 219-21, 225-6, 227 cost of capital 144-5, 151 capital-skill complementarity and 145 macroeconomic instability 146-7 Crafts, N. 67 Davies, S.W. 105, 122, 128 De Bresson, J.N. 45 De Mello, L.R. 170, 173 decision to export 113, 115-16 demand determinants of spillovers 36-7, 38-9, 39-41 labour demand 70-2, 180-1 regional CGE analysis 178-209; combined demand and supply effects 196-9; demand shock 186-91 demonstration effects 45-7, 167-8, 172 export spillovers 112, 114, 11518 determinants of FDI spillovers see spillovers development agencies 28, 167, 169 Devereux, M.P. 5 distribution effects of spillovers 5 8 9 distribution services 74-83 Doms, M.E. 122, 148 Drabek, Z. 240 Driffield, N.L. 163, 168, 169-70, 171, 172, 173 wages and productivity 105-6 DTI 1 Dunning, J.H. 47, 66, 96, 148, 162,167 EAG6 Eaton, J. 244 economic rent 57-8 economies of scale 245, 247 efficiency spillovers see spillovers efficiency wages 36, 59 electronics sector 219-20 Eltis, W. 168 employment 13-14, 66-7
Index
employment continued foreign-owned manufacturing firms 130, 131, 164-7 and productivity 74-6 productivity and wages spillovers 110-11 regional CGE analysis 187-9, 193, 195, 198, 199 trade and FDI 241-2, 257-9 endogenous technical progress 72, 89 energy sector 8-9 Engelbrecht, H.J. 52 establishments sample 12, 123, 124, 125-7, 146-7 importance of foreign ownership 127-37, 142, 149-50 European Regional Development Fund 167 European Union (EU) 6, 21-2 hierarchy of research centres 2 2 1 6 SMP 6, 96 exogenous technical change 72 expansions of capacity 243-4 expatriate managers 36 exports decision to export 113, 115-16 FDI and 21-2, 23-4, 248-59 propensity to export 113-14, 116-18 regional CGE analysis 185-6, 193-9; export-oriented FDI stimulus 186-91 spillovers 111-18, 119; modelling export spillovers 113-15; results 115-18 see also trade Fikkert, B. 51 financial services 74-83 firm-specific advantages 12-13, 148-9 fixed effects panel 78-9, 80, 81-2 flows of FDI 245-6 UK inward 7, 8 Fontagne, L. 246, 259 foreign direct investment (FDI) 1-2
267
FDI continued flows of 7, 8, 245-6 growth of global FDI 3-6, 240 policy 54-6 stocks of see stocks of FDI types of FDI 25, 54, 243-4 foreign ownership 122-61 explanations for higher labour productivity 143-9 manufacturing establishments 123, 124, 125-37, 142, 146-7, 149-50 non-manufacturing companies 123-4, 137-43, 150-1 Fors, G. 50 France 23, 246, 252-5, 258-9 Francois, J.F. 178 Fraser, D. 168 Fraumeni, B. 127 general external economies 213, 215-26, 227 Germany 20, 27-8, 215-28 location dynamics of R&D 2 1 5 21 technological specialisation 2 2 1 6, 233-4 Gibson, H. 184 Gillespie, G. 178-9, 185 Girma, S. 106-10, 110-11 global FDI 3-6, 240 globalisation 104 of R&D 211-14; implications for host economy 213-14 Globerman, S. 48, 52, 73, 106, 110, 122 Gorg, H. 27 government contracts 37 greenfield ventures 243-4 Greenwood, M.J. 183 Griffith, R. 5, 95, 106, 110, 122 Griliches, Z. 73 Grupp, H. 219 Hainaut, P. 244 Harrigan, F. 181-2 Harris, J.R. 183 Harris, R.I.D. 184 Harrison, A.E. 170-1, 173
268
Index
Hart, P.E. 123 Harvey, A.C. 72 Hay, D.A. 17 Head, C.K. 5, 245 Helpman, E. 51, 70 Hessen (Germany) 216, 217, 220, 221 technological specialisation 222, 224, 233-4 hierarchies of research locations 221-6, 227 higher-order research centres 20, 213, 222-6 Hill, S. 167 Hood, N. 28 horizontal integration 244 host country effects 104-21 MNCs and export spillovers 111— 18, 119; modelling export spillovers 113-15; results 115-18 MNCs, wages and productivity spillovers 105-11; productivity and wage differentials 106-10; productivity and wage spillovers 110-11, 118-19 Hubert, F. 180, 191 human capital 25-6, 93-5 determinants of spillovers 48-9, 52 foreign ownership and labour productivity 122-61 Hummels, D. 240 lammarino, S. 210, 213, 214, 217, 220,222,224,231-2 Imbriani, C. 42-3 imports FDI and 248-59 openness to 52 see also trade incentives, investment 13, 26-7, 163, 167, 174 industrial concentration 47 industrial policy 27, 162 information spillovers 112, 114, 115-18 innovation see research and development
input intensities 122-61 input-output results 189-91 instability, macroeconomic 146-7, 151 intellectual property protection 57, 59 see also patents interdependence 54 inter-industry spillovers 24-5, 678, 77-83, 89-90, 97 intermediate inputs 2 2 - 3 , 88 international trade see trade internationalisation see globalisation intra-industry spillovers 68, 83-96, 97 investment incentives 13, 26-7, 163, 167, 174 inward investment 1-33 and international trade 21-4 policy implications 26-8 product market competition 1 7 18 productivity spillovers 13-16 and R&D 18-21 research agenda 24-6 in the UK see United Kingdom Italy 20, 215-28 location dynamics for R&D 2 1 5 21 technological specialisation 2 2 1 6,232 Janne, O.E.M. 226 Japan 109-10 Jensen, J.B. 2 1 , 122, 148 joint ventures 55-6, 243-4 Kokko, A. 43, 45, 168, 243 Kremers, J.J.M. 77 labour demand 70-2, 180-1 labour mobility 25 see also migration labour productivity 11-13, 122-61 explanations for higher productivity in foreign-owned firms 143-9; cost of capital 144-7; firm-specific
Index
labour productivity continued explanations continued advantages 5, 148-9; superior technology 147-8 manufacturing establishments 123, 124, 125-37, 149-50; analysis of foreign advantage 131-7; characteristics of survivors 127-31; importance of foreign ownership 127-37 models of 73-4, 93-5 non-manufacturing companies 123-4, 137-43, 150-1 technical progress and 74-6, 8 0 1, 85-8, 93-5 wages and productivity spillovers 107-11 law of one price (LOOP) closure 185-6 Layard, R. 182, 183 licensing 44 linkages, vertical 44-5 Lipsey, R. 241, 243 Lipworth, G. 245-6, 251, 258 Liu, G.S. 17 living standards 245 local development agencies 28 localisation economies 213, 2 1 5 16, 227 location 2 0 - 1 , 163, 174, 210-39 dynamics of technological innovation 215-21 internationalisation of R&D 2 1 1 14; implications for the host economy 213-14 patent data 214 RTA indices 221-6, 231-4 technological profiles and the hierarchy of EU research centres 221-6 tournaments 4 Lombardia (Italy) 215, 217, 218, 220-1 technological specialisation 222, 224, 225,232 London, City of 5, 77 Lovering, J. 169 lower-order research centres 213, 222-6
269
Lyons, B.R. 105, 122, 128 Lyons, R.K. 171 macroeconomic instability 146-7, 151 Mansfield, E. 2 - 3 , 46 Mansfield, W. 57 manufacturing CGE analysis 184-200 labour productivity and foreign ownership 123, 124, 125-37, 142, 146-7, 149-50 Standard Industrial Classifications 84, 98, 125, 152-3 technical progress in 14-15, 67-8, 74-96, 97; impact of inward investment in services and 77-83; Census of Production data 83-96 UK inward FDI 8-9, 11 market structures 47-8 Markusen, A. 210 Maurseth, P.B. 214 McFetridge, D.G. 47 McGregor, P.G. 189 mean group estimates 78-81 mergers and acquisitions 243-4 'micro' focused analysis of spillovers 58 migration 183-4, 189-91, 196 Miles, D. 144 Minford, P. 182 Mittelstand 227 mobility, labour 25 see also migration mode of international business 25, 54, 243-4 Mody, A. 5 Mohnen, P. 50, 51 Morgan, K. 163 Mucchielli, J.L. 243 Mudambi, R. 28 Muellbauer, J. 85 Multilateral Investment Agreement (MIA) 240 multinational corporations (MNCs) 148-9 multipliers 199 Munday, M.C. 167, 168, 169
270
Index
Nadiri, M.I. 42 national income 14, 66-7 nationality of investors 9, 27, 95-6, 109-10 negative regional spillovers 170-1, 172-3 Nelson, R.R. 70 Neven, D. 168 New Earnings Survey 84 Nickell, S.J. 17, 89 Niedersachsen (Germany) 217, 218, 219, 220 technological specialisation 222, 224, 233-4 Nijkamp, P. 167 non-operatives 11, 12, 86-8 Noonan, C.A. 210, 214, 217, 219, 220, 222, 224, 233-4 Nordrhein Westfalen (Germany) 216,217,218-19,220,221 technological specialisation 222, 224, 233-4 North West (UK) 216, 217, 218, 220 technological specialisation 222, 224,227,231 OECD 2, 23 O'Mahony, M. 84, 130 OneSource database 11, 12, 123-4, 137-8, 154-5 ownership and productivity 13943, 150-1 openness 21-4, 52, 53 Oswald, A.J. 182 Oulton, N. 122, 123, 145, 151 output FDI and trade 241-2, 257-9 regional CGE analysis 16, 193-6, 199 technical progress 82-3, 86-96 outward investment 4 and trade 241-2, 244, 250-7 PA Cambridge Economic Consultants 179 Pain, N. 5, 22, 23, 66, 74, 78, 80, 8 2 , 9 0 , 111, 191,192, 199, 200, 244
Pain continued efficiency spillovers 179, 180-1 productivity growth 163, 168 technical progress 14, 16, 42, 67, 70, 71, 72, 180 US affiliates in UK and Germany 27-8 Pajot, M. 246, 259 panel data models 78-80 Park, W.G. 52 Patel, P. 212 patents 19, 214-26 Pearce, R.D. 212 Pesaran, M.H. 8 1 , 9 1 Piemonte (Italy) 20, 216, 217, 218, 220 technological specialisation 222, 224, 227, 232 Piscitello, L. 212, 214, 216, 219 policy implications of inward investment 26-8 implications of spillovers 54-8 industrial 27, 162 regional see regional policy restrictions on foreign ownership 40-1 Pooled Mean Group (PMG) estimator 91-3 Porter, M.E. 163, 169, 174 prices 193-6 private sector 138 private services 14, 67-8, 74-83 impact of inward investment 7 7 83,97 product market competition 17-18, 39-40, 112, 114,115-18 production 241-2, 257-9 production functions 68-9, 69-72, 73, 82 productivity 172 direct spillovers 13-16, 42-3 growth in the UK 167-9 host country effects 105-11; differentials 106-10; spillovers 110-11, 118-19 labour see labour productivity potential negative impacts of FDI on 170-1
Index
productivity continued technical progress 68-9, 69-96; employment and productivity 74-6 total factor productivity 69-70, 108-11 profiles, technological 221-6 propensity to export 113-14, 11618 protectionism 56 proximity-concentration trade-off 247, 248, 258 public services 67-8, 74-83 impact of inward investment 7783 recessions 146-7, 151 Reganati, F. 42-3 regional computable general equilibrium (CGE) analysis 178-209 AMOSFDI 179, 181-4, 201-3 combined demand and supply effects 196-9 export-oriented FDI stimulus 186-91 foreign-owned manufacturing sector 184-6 impact of efficiency spillovers 191-6 scale of efficiency spillovers 17981 regional development agencies 28 regional policy 15, 162-77 changing pattern of inward investment and potential impacts of FDI 167-71; domestic productivity growth 167-9; localised nature of spillovers 169-70; negative impacts on domestic productivity 170-1 empirical evidence on spillovers 171-3 pattern of foreign investment in the UK 164-7 Regional Selective Assistance (RSA) 13 regions 26, 215-28 location dynamics of innovation 215-21
271
regions continued technological specialisation 2 2 1 6, 231-4 Reinert, K.A. 178 remitted earnings 8 rent, economic 57-8 research agenda 24-6, 58-9 research and development (R&D) 4, 210-39 fields of technological activity 214, 229-30 internationalisation 211-14; implications for the host economy 213-14 location dynamics 215-21 RTA indices 221-6, 231-4 spillovers 18-21, 49-52 technological profiles and the hierarchy of EU research centres 221-6 resource allocation 162 Revealed Technological Advantage (RTA) index 221-6, 231-4 Rhee, Y.W. 112 Ries, J. 245 Riker, D.A. 243 risk of appropriation 37-8 Rodrik, D. 70 Romeo, A. 2 - 3 , 46 Romer, P. 25, 69 Ruane, F. 27 scale economies 245, 247 Scotland 167 CGE analysis 16, 179, 181-200; efficiency spillovers to UK manufacturing 191-6; exportoriented FDI stimulus 187-91; foreign-owned manufacturing sector 184-7 Sectoral Exports Index (SEI) 114, 115, 117 services 14, 67-8, 74-83 impact of inward investment 7783,97 Single Market Programme (SMP) 6, 96 Siotis, G. 168 Sjoholm, F. 55-6, 73
272
Index
skilled labour 25-6, 86-8 complementarity with capital 145, 151 see also human capital Soete, L.L.G. 221 sources of UK inward FDI 9, 27, 95-6, 109-10 Sousa, N. I l l , 113 South East (UK) 20, 215, 217, 218, 219, 220-1 technological specialisation 222, 2 2 3 , 2 2 4 , 2 2 5 , 231 specialisation, technological 213, 221-6,227,231-4 spillovers 2, 13-24, 34-65, 104-21 determinants of FDI spillovers 35-41; demand 36-7, 38-9, 39-41; supply 36-8, 39-41 empirical evidence on spillover determinants 41-54; demonstration effects 45-7; direct productivity measures 4 2 - 3 ; FDI and market structures 47-8; investments in human capital 48-9; licensing 44; R&D 49-52; vertical linkages 44-5 exports 111-18, 119 indirect 17-24; international trade 21-4; product market competition 17-18; R&D 1 8 21 inter-industry 24-5, 67-8, 77-83, 89-90, 97 intra-industry 68, 83-96, 97 modelling technical progress and 69-74 policy implications 54-8 productivity see labour productivity; productivity regional 163, 167-74; empirical evidence 171-3; localisation 169-70; negative 170-1, 172-3 regional CGE analysis see regional computable general equilibrium analysis research agenda 24-5, 58-9 wages 110-11, 118-19 Srinivasan, K. 5
Standard Industrial Classifications (SIC) 84, 98, 125, 152-3 Stevens, G. 241 stocks of FDI 68, 90-1 global 5-6 inward to UK 7, 10; composition 8-9, 76-7 and trade 245-59 substitution: FDI and trade 240-63 superior technology 147-8 supply determinants of FDI spillovers 36-8, 39-41 shock in regional CGE analysis 191-6, 200; combined demand and supply effects of FDI 196-9 Svensson, R. 169 Swedenborg, B. 243 Swenson, D. 243-4 Tamura, A. 244 Taylor, J. 167 technical progress 14-15, 66-103, 180-1, 191-2 Census of Production data for manufacturing 83-96 data issues 74-7; employment and productivity 74-6; foreign direct investment 76-7 impact of inward investment in manufacturing and services 7 7 83 modelling technical progress and spillovers 69-74 models of labour productivity 7 3 4, 93-5 nationality of investors 95-6 technological centres of excellence 40 technological profiles 221-6 technological sectors 219-21, 2 2 3 6, 229-30 technological specialisation 213, 221-6, 227, 231-4 technology appropriable 35-6 competence in host country 4 1 , 42,53
Index
technology continued complementarity between host country and investor 4 1 , 42, 43, 53-4 diffusion 45-7 foreign ownership and superior 147-8 licensing 44 policy 56-7 see also research and development technology consortia 37 technology transfer 2 - 3 , 167-8 policies 57-8 Todaro, M.P. 183 Tomkins, J.M. 170 Toniolo, G. 67 total factor productivity (TFP) 6970, 108-11 see also productivity trade, international 4, 21-4, 104, 240-63 data 246-8, 260-1 empirical model 248-50 France 23, 246, 252-5, 258-9 mechanisms linking trade to FDI 241-6 policy 56 UK 21-2, 23, 246, 250-2, 258 US 23, 243-4, 246, 255-7, 259 transport and communications 7 4 83 transport costs 245 Treyz, G.I. 183 Twomey, J. 170 types of investment 25, 54, 243-4 UNCTAD 240 United Kingdom (UK) inward investment 1-3, 6, 7-13; composition of stock of 8-9, 76-7; and output growth 82-3; pattern of 164-7; sources of investment 9, 27, 95-6, 109-10 R&D 1 8 - 1 9 , 2 0 , 2 1 5 - 2 8 ; location dynamics 215-21; technological specialisation 221-6, 231 trade 21-2, 23, 246, 250-2, 258
273
UK continued US affiliates in UK and Germany 27-8 United States (US) 22, 52 affiliates in Germany and the UK 27-8 and Canada's technology policies 56-7 importance of inward investment in UK 9, 96 productivity and wage differentials 109-10 trade 23, 243-4, 246, 255-7, 259 US-owned firms and labour productivity 128-43,148-9, 150 value added per employee 130, 131,150 value added multipliers 199 Vega, M. 212 Vernon, R. 211 Verspagen, B. 214 vertical integration 244 vertical linkages 44-5 wages 105-11, 182-3 differentials 106-10 efficiency wages 36, 59 foreign-owned firms 130-1, 133 spillovers 110-11, 118-19 Wagner, K. 84 Wakelin, K. 244 Wales 167 Wang, Y. 40 Weiss, M.E. 243 West Midlands (UK) 216, 217, 218,220 technological specialisation 222, 224,231 Wheeler, D. 5 Whyte, T.R. 183 World Trade Organisation 2, 23, 240 Wren, C. 167 Young, G. 16 Young, S. 28, 29, 168 Zejan, M. 55
