Foreign Trade and FDI Stocks in British, US and French Industries: Complements or Substitutes?



Similar documents
Andrés López and Eugenia Orlicki 1

Modes of Delivery in Services

Do Currency Unions Affect Foreign Direct Investment? Evidence from US FDI Flows into the European Union

Using Administrative Data in the Production of Business Statistics - Member States experiences

Australia s position in global and bilateral foreign direct investment

United States-China Trade: Where are the Exports? Barry Bosworth Susan Collins Aaron Flaaen

FDI VERSUS EXPORTS: MULTIPLE HOST COUNTRIES AND EMPIRICAL EVIDENCE

A New Effective Exchange Rate Index for the Canadian Dollar

Foreign Direct Investment and the Domestic Capital Stock

Explaining Canada's Changing FDI Patterns

How To Understand The Economic Benefits Of Foreign Direct Investment In The United States

Impact of Foreign Direct Investment, Imports and Exports

Foreign Direct Investment and International Trade

Foreign Direct Investment and the Domestic Capital Stock

Tax planning may have contributed to high indebtedness among Swedish companies

Errors and omissions in the balance of payments statistics a problem?

Export Pricing and Credit Constraints: Theory and Evidence from Greek Firms. Online Data Appendix (not intended for publication) Elias Dinopoulos

U.S. International Business Travel: Its Impact on U.S. Merchandise Exports. by Maksim Belenkiy and David Riker

DOES FDI MATTER FOR TRADE IN BRAZIL? AN APPLICATION OF THE GRAVITY MODEL

D F o DI FDI i ness erv ces fo f ll lllow

National Money as a Barrier to International Trade: The Real Case for Currency Union Andrew K. Rose and Eric van Wincoop*

International Investment. Australia. Economic Diplomacy, Trade Advocacy and Statistics Section Department of Foreign Affairs and Trade September 2015

The labour market, I: real wages, productivity and unemployment 7.1 INTRODUCTION

Trends in Foreign Direct Investment the Austrian Perspective

FDI as a source of finance in imperfect capital markets Firm-Level Evidence from Argentina

Exchange Rates and Foreign Direct Investment

In 2012, GNP in constant prices increased by 1.8% compared with 2011.

International Comparisons of Australia s Investment and Trading Position

Theories of Exchange rate determination

Viewing the Current Account Deficit as a Capital Inflow

Domestic Multinationals and Foreign-Owned Firms in Italy: Evidence from Quantile Regression 1

Clustering or Competition? The Foreign Investment Behavior of German Banks

EXTERNAL DEBT AND LIABILITIES OF INDUSTRIAL COUNTRIES. Mark Rider. Research Discussion Paper November Economic Research Department

Currency Unions and Irish External Trade. Christine Dwane Trinity College Dublin. Philip R. Lane, IIIS, Trinity College Dublin & CEPR

Marketing Mix Modelling and Big Data P. M Cain

Economics 380: International Economics Fall 2000 Exam #2 100 Points

The Trade Balance Effects of U.S. Foreign Direct Investment in Mexico

Foreign direct investment

Explanation beyond exchange rates: trends in UK trade since 2007

THE NUMBER AND THE SIZE DISTRIBUTION OF FIRMS IN SWEDEN

Markups and Firm-Level Export Status: Appendix

Fewer net errors and omissions, that is a new format of the balance of payments

DETERMINANT FACTORS OF FOREIGN DIRECT INVESTMENT FLOWS IN CENTRAL AND EASTERN EUROPEAN COUNTRIES

Attracting Foreign Direct Investment through an Ambitious Trade Agenda: New Opportunities for the U.S. Economy and Workforce

A.2 The Prevalence of Transfer Pricing in International Trade

Definitions of Foreign Direct Investment (FDI): a methodological note Maitena Duce 1 Banco de España 2

ING International Trade Study Developments in global trade: from 1995 to United Kingdom

Multinational Firms, FDI Flows and Imperfect Capital Markets

Kiel. Policy Brief. The Importance of Investment Income and Transfers in the Current Account: A New Look on Imbalances. Rolf J.

Civitas: Online Report. The Costs and Benefits of the European Union

UNITED NATIONS CONFERENCE ON TRADE AND DEVELOPMENT. STUDY SERIES No. 63

Chapter 1. Introduction

Sources of Funds and Investment Strategies of Venture Capital Funds: Evidence from Germany, Israel, Japan and the UK

The Decline of the U.S. Labor Share. by Michael Elsby (University of Edinburgh), Bart Hobijn (FRB SF), and Aysegul Sahin (FRB NY)

Panel Data. Tomohito Hata. In the present paper, I will examine the determinants of grant-loan allocation in Japanese ODA,

How To Calculate Export Earnings

FOREIGN TAXES AND THE GROWING SHARE OF U.S. MULTINATIONAL COMPANY INCOME ABROAD: PROFITS, NOT SALES, ARE BEING GLOBALIZED.

Export Platforms and the Industry-Specific FDI-Trade Relationship

Qualitative analysis of a potential Free Trade Agreement between the European Union and India. Executive Summary

THE DETERMINANTS OF INTRA-INDUSTRY TRADE

The OLI or eclectic approach to the study of foreign direct investment (FDI) was

International Economic Relations

Commentary: What Do Budget Deficits Do?

The Employment Crisis in Spain 1

International and national R&D outsourcing: Complements or substitutes as determinants of innovation?

The Japanese economy?

Final Report: 7 February Stephen Bond and Lucy Chennells. The Institute for Fiscal Studies

GEOGRAPHY OF SPANISH SOVEREIGN DEBT. SOME SPATIAL CONSIDERATIONS FROM A POLITICAL PERSPECTIVE

Economic Commentaries

Stock prices are not open-ended: Stock trading seems to be *

FDI is investment that directly leads to productive activity within the host country

Chapter 4 Specific Factors and Income Distribution

DOES A STRONG DOLLAR INCREASE DEMAND FOR BOTH DOMESTIC AND IMPORTED GOODS? John J. Heim, Rensselaer Polytechnic Institute, Troy, New York, USA

The Global Operations of European Firms. Efige Policy Brief. Giorgio Barba Navaretti, Matteo Bugamelli, Gianmarco Ottaviano and Fabiano Schivardi 1

ENGINEERING LABOUR MARKET

Should U.S. Investors Hold Foreign Stocks?

FDI Contributes to Output Growth in the U.S. Economy

Globalization & Economic Geography

Economic. The EU-Australia. relationship

mawer INSIGHT INFLATION: The Influence of Inflation on Equity Returns IN THIS ISSUE

OFFSHORING AND FOREIGN DIRECT INVESTMENT

ING International Trade Study Developments in global trade: from 1995 to Mexico

CHAPTER 11. AN OVEVIEW OF THE BANK OF ENGLAND QUARTERLY MODEL OF THE (BEQM)

THE U.S. CURRENT ACCOUNT: THE IMPACT OF HOUSEHOLD WEALTH

An Assignment Theory of Foreign Direct Investment

GDP and National Accounts: key concepts National Accounts (Part A)

Volume Title: The Effects of Taxation on Multinational Corporations. Volume URL:

KILM 16. Labour productivity

Does Foreign Direct Investment Transfer Technology Across Borders? #

Will Bulgaria Remain a "Quiet Place" for Higher Education?

Import Prices and Inflation

RENMINBI - A NEW SETTLEMENT CURRENCY WAS BORN

Chapter 4 Specific Factors and Income Distribution

Tests of Changes in the Elasticity of the Demand for M2 and Policy Implications: The Case of Four Asian Countries

The Effect of Mergers & Acquisitions and Greenfield FDI on Income Inequality

THE IMPACT OF EXCHANGE RATE VOLATILITY ON BRAZILIAN MANUFACTURED EXPORTS

Executive Summary, 21 January 2015

The Elasticity of Taxable Income: A Non-Technical Summary

Offshoring and ICT - Evidence for German manufacturing and service firms

Lars Osberg. Department of Economics Dalhousie University 6214 University Avenue Halifax, Nova Scotia B3H 3J5 CANADA

Transcription:

Foreign Trade and FDI Stocks in British, US and French Industries: Complements or Substitutes? Lionel Fontagné Michaël Pajot TEAM (University of Paris 1 and CNRS) lionel.fontagne@univ-paris1.fr pajot@univ-paris1.fr Abstract: The issue of complementarity-versus-substitution between trade and FDI is controversial. Specifically, the type of relationship obtained is highly dependent of the type of data used. At the aggregated level, a large part of the observed complementarity relationship is associated with spillovers between industries, whereas the complementarity is less ascertained at the disaggregated level. In addition, complementarity is generally more ascertained with FDI flows data, than with FDI stocks data. Accordingly, we tackle here this relationship at the disaggregated level using FDI stocks data. Considering three panels of industries (France, United-Kingdom and the United States), we highlight that transitory effects largely dominate for European countries under consideration: the permanent complementary effect of FDI stocks on trade flows is rather negligible. In contrast, in the US case, we point out a strong complementarity between trade flows and FDI stocks, associated with US net imports. JEL Classification: 421-441

Introduction Foreign Direct Investment (FDI) and international production are developing at a high pace. FDI flows have recorded a 19% increase in 1997, and again 10% in 1998, to reach roughly 440 billion US$. Around 50,000 parent and 450,000 affiliates operate worldwide (UNCTAD 1998). These affiliates account for 6% of world GDP to be compared with 2% in 1982 (Hummels et al., 1998). Intra-firm trade accounts for one third of world exports and foreign affiliates' sales grow faster than world exports. 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. These relationships have recently been addressed in the debate on the opportunity to implement a multilateral agreement on FDI. Drabek (1998) concluded that the links between FDI and trade were ascertained by the empirical literature and that such issue of the debate upon a multilateral agreement on FDI was no longer questionable. In the same way, the WTO working group on the relationship between Trade and Investment concluded that the relationship between trade and FDI, even if not restricted to the issue of complementarity, pointed out such complementarity (WTO, 1998). However, in contrast with such views, this complementarity is challenged in the literature. Indeed, at the level of the firm, such complementarity is questionable. Hence, the level of analysis matters: a substitution at the firm level may be associated with a deeper 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 macro-economic factors such as country size, per capita income etc., can be the very foundation of the observed macro-economic complementarity. From a more methodological point of view, it must be kept in mind that FDI flows, FDI stocks, local sales, and local production are different measures of foreign activity that may lead to 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, in an international comparative perspective. The rest of the paper is organised as follows. Mechanisms linking trade and FDI are examined in a first section based on a brief survey of the literature. A second section submits a new approach to the complementarity-substitution relationships, based on disaggregated trade and FDI stock data. This relation is tackled at the industry level for the United Kingdom, France and the United States in a third section. A last section concludes. Mechanisms linking trade to FDI 1 Following Stevens and Lipsey (1992), it has become usual to distinguish between financial and production implications of FDI. The formers, identifying a possible substitution between domestic and foreign investment, have a rather indirect impact on trade flows. We are interested in the latter: the question here is whether FDI displaces trade, output and employment. Two types of impacts of 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. It is however not necessarily the case at the industry level or at the macro-economic level. In addition, investing abroad induces intra-firms imports if labour intensive or resource intensive activities are relocated abroad. Hence, origin country's trade is not necessarily displaced by outward FDI. Reciprocally, host country's trade is not necessarily displaced either, since imports are likely to increase at least in the short term, while enhanced competitiveness is likely to boost exports too. Most empirical studies of the relationship between trade and FDI aim at estimating the direction and the magnitude of the impact of FDI on trade. Such estimates provide a direct measurement of substitution versus complementarity effects. Explaining exports (imports) as a function of FDI and different control variables, one is interested in the sign and magnitude of the parameter estimate associated with foreign investment. If a negative parameter on FDI is obtained in trade equations, FDI is considered displacing trade.

Exports are at least partially replaced by local sales on foreign markets, detrimental to the domestic industry of the investor. Production and employment can be negatively affected in home country. Reciprocally, host country's trade balance would benefit from this substitution effect if no additional imports from the parent company were induced. Trade and FDI are complements if a positive parameter is obtained. Investing abroad leads to an increase in exports of the investing country towards 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 competitors. In total, depending on the type of FDI data under consideration (firm, industry, manufacturing sector, total, flows vs. stocks), results will change dramatically. According to a simplistic micro-economic analysis, trade and FDI can be referred to as 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. Turning to the macroeconomic level, the classical Mundell s argument deserves much emphasis. However, either at the microeconomic level or at the macroeconomic level, the conclusions of empirical studies are far from being convincing. Firstly, the empirical literature addressing this relationship at the micro-economic level fails to reach clear-cut conclusions. The business literature surveyed in Blomström and Kokko (1994) makes a comparison between the effects of alternative ways for firms of exploiting their specific advantages. The main result is that the dominant effect is that FDI translates in an increase in the market share abroad. Second, exports of intermediate products towards foreign subsidiaries are compensating for crowded out exports. Lipsey and Weiss (1984) highlight that one dollar of local production induces 9 to 25 cents of additional exports from the investing country. Not surprisingly, the elasticity linking trade and local output is three times larger at the industry level than at the individual firms level, as a 1 The interested reader will find a good survey in Cantwell [1994].

result of externalities between firms within industries. Swedenborg (1979) concludes that Swedish FDI has no significant effect on the exports of the parent, local sales substituting these exports, while new exports are induced (intermediate goods or complements of supply) 2. This result was confirmed in 1982 by a study finalised by the same author: each dollar of local sales substitutes only 2 cents of exports but "creates" 12 cents of new exports, the net effect being a positive complementarity of 10 cents. Blomström, Lipsey and Kulchycky (1988) conclude to a limited effect of complementarity. Andersson (1993), and Blomström and Kokko (1994) suggest that the Swedish structure of exports might be more affected by outward FDI than the value of these exports. Rikker and Brainard (1997) stress the vertical splitting up of processes that is associated with FDI: FDI does not displace output, it splits the production process world-wide and it complements trade in intermediate products. Using individual company data for 14,096 French industrial firms in 1993, Chédor and Mucchielli (1998) show that French firms with offshore affiliates export more than the rest. The economic activities of offshore affiliates have a beneficial effect on exports from their parent firms. In sum, the empirical validation of the micro economic hypothesis of substitution is controversial. Swenson (1999) in her turn has drawn attention to the fact that the effect on trade to be expected from inward FDI differs greatly depending on the type of FDI under consideration 3. In view of the fact that there are more mergers and acquisitions than greenfield ventures, it is worth looking into the matter with the help of data from individual companies, breaking transactions down into greenfield ventures, expansions of capacity, acquisitions or joint ventures. Inward FDI stock 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. Later on it will be shown this effect is largely due to the sectorial composition of FDI and American trade. With regard to type of investment, expansion of capacity is the only one to have a positive and significant impact on American exports and imports. In contrast, mergers and 2 Sweden, to the benefit of the database provided by the Industrial Institute for Economic and Social Science Research (Stockholm), is generally a good case study. 3 This separation was made by the author using an ITA (International Trade Administration) database relating to foreign FDI in the United States over the period 1974-94. That made it possible to compile a series of FDI stocks of each type (greenfield, etc.) for each country investing in the United States.

acquisitions act as substitutes. Note that what is involved here is greenfield investment stocks, not their flows. Turning to 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. FDI stocks abroad amount to 3,500 thousand million dollars in the case of the United States, 1,000 thousand million in the case of Canada, 1,600 thousand million in the case of the European Union and almost 300 thousand million in the case of Japan. These figures are to be compared with 7,000 thousand million, 2,500 thousand million and 500 thousand million dollars respectively of exports and imports of goods and services. Such orders of magnitude suggest that FDI and trade share a number of determinants at macroeconomic level. This 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. Accordingly, Eaton and Tamura (1994), controlling for factors jointly determining trade and FDI, identify a large and positive relationship between outward FDI and exports, as well as imports, for Japan and the US. This relation is not verified for inward FDI. Using time-series analysis, Andersen and Hainaut (1998) find contrasted evidence of complementarity effects between exports and outward FDI flows: there is a complementarity for the United-States, Japan, and Germany, but not for the United Kingdom for which no significant relationship was found. Lastly, Pain and Wakelin (1998) challenge the issue of complementarity for 11 OECD countries over 1971-92. Estimates do not rely on bilateral exports and imports, but on total exports, controlling for world demand. The main result is that the relationship changes among countries and over time. However, inward FDI is demonstrated to be "pro-competitive", within the panel, in contrast to outward FDI, benefiting to the recipient country. These contrasted results are not surprising as long as the theoretical story does not point out the basic

mechanisms linking foreign activity and exports. FDI is only a second order effect of decisions to produce abroad taken by multinational firms competing on imperfect markets. According to such view, the new trade theory stresses that returns to scale limit the number of efficient plants, whereas transportation costs and more generally trade barriers act in the opposite direction. To summarise its conclusions, the new approach to international trade, which places imperfect competition in the centre of the analysis, highlights two types of major determinants of the relationship under review. Firstly, the way firms are organised is important, a firm arranged "vertically", locating various parts of its production process in different foreign affiliates to form what it conventionally termed a "global firm", will find its foreign trade and investment reinforcing each other (Head and Ries, 1997). On the other hand, a firm arranged horizontally will carry out most of its 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. This introduces the second category of determinants considered by recent literature on international trade. 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 useful number of plants, while transport costs, and more generally all obstacles to trade, have the opposite effect (Brainard, 1993). When the firm has high fixed costs and each plant has limited fixed costs, a multinational will locate its units of production close to its markets and FDI will substitute for trade if transport costs are not a negligible factor (Markusen and Venables, 1995; Swenson, 1999). Brainard has given proof of this by considering trade and local sales by American industries in relation to 27 partner countries. Where transaction costs (transport costs and trade barriers) are high, they have been shown to act as a brake on exports. The other side of the coin is that they promote local sales by affiliates. Another finding by Brainard is worth noting, namely that there is a positive relationship between the standard of living in the partner country and the net sales of American affiliates established there, whereas a negative relationship is observed between standard of living and net exports to the relevant

market. Hence in the case of high living standards in countries hosting American FDI, there is some degree of substitution of trade by local sales. On the other hand, there is no evidence that each dollar of local sales could have been replaced by a dollar's worth of exports. Bayoumi and Lipworth (1997) complement these explanations by clarifying the puzzling empirical issue of diverging results with series or FDI flows and stocks. They explain FDI outflows from Japan as a (positive) function of investment in Japan and the recipient country, real exchange rate (resp. negative) and the lagged stock of investment in the recipient country. This stock has a negative impact on FDI flows, highlighting a dynamic adjustment process. Turning to the relationship between trade and FDI flows (versus FDI stocks), this leads to a model in which the impact of direct investment on trade depends according to the term considered. It is expected that FDI flows lead in the short run to enhanced exports from the investing country, according to capital goods or components required for the launching of the subsidiary. In the long run, the subsidiary becomes more autonomous, the local content increases and the substitution effect dominates, as previous exports of the parent are being replaced by local production. Having such scheme in mind, the authors introduce simultaneously (lagged) FDI flows and stocks in exports and imports equations. They obtain a short run complementarity between trade flows and FDI flows, whereas no significant relationship is observed between trade and FDI stocks: "FDI has only a temporary impact on exports, consistent with the view that FDI influences exports largely through the short term need to equip new factories" (p. 24). In the following sections we tentatively concentrate on this "long-term" relationship using FDI stocks for three countries: the United-Kingdom, the United-States and France which provide the best bilateral data at the industry level concerning FDI stocks. This will be done using a framework designed to control for new trade theory variables such as economies of scale vs. transportation costs. A new approach based on sectorial bilateral data Our data set for United States, the United Kingdom and France is built on the basis of a common

nomenclature for FDI stocks and bilateral trade flows. The matching has been done for each country separately, due to large differences in nomenclatures for 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. We consider separately each country facing its different partners, in different industries, over a period of 6 to 10 years. A selection in British data corresponding to 27 countries, 10 years (1987-96) and 13 industrial sectors has been implemented specifically for this paper 4. Concerning US investment abroad, we have BEA information on outward stocks for 1982-1994. Foreign investment in the United-States is however available only over 1980-1994. This has led us to exploit only partially the US database, as detailed in Appendix 2. We consider here 42 countries, 6 years (1989 to 1994) and 23 industries of which 12 are industrial sectors. Concerning French data, the Balance of Payments Appendices have been collected over 1989-1995 for FDI stocks. Contrary to US data, reinvested earnings are not reported until 1996. Hence, it has been impossible to extend our database to 1996 and to include data compiled with new principles. The French database, detailed in Appendix 1, entails 11,550 observations: 55 countries, 35 sectors of which 17 are manufacturing sectors and 6 years: 1989 to 1994. As highlighted by the literature review above, different data generally leads to different results concerning the relationship between trade and FDI. Our data set authorises to examine this relationship at the industry level and points out the national complexity of this relationship. In the next section, exports of individual industries will be explained by outward FDI in the

corresponding industries. This will answer to the question: "How does $1 of US (French/British) investment the automobile industry in Brazil affect US (French/British) exports of automobiles to Brazil?" To answer this question, we use gravity-type equations integrating FDI. Three sets of explanatory variables will be introduced in the estimates: country variables, sectorial variables, and FDI variables, as detailed in Appendix 7. Country variables: the size of markets, proxied by the average GDP of the declaring country and its partner (AVRGDP), the Balassa-Bauwens measure of differences in GDPs (DFGDP), the demand for variety and the standard of living proxied by the average income per capita of the declaring country and its partner (AVRGDPPC), the economic distance and the difference in human capital endowment both proxied by the difference in income per capita between the declaring country and its partner (DFGDPPC), transportation costs proxied by the geographical distance (DIST), the existence of a common border (adjacency, ADJ), the regionalisation proxied by the existence of preferential commercial schemes (dummy CPOL); Turning to sectorial variables, which have no country dimension, the concentration, the economies of scale, the share of white-collar workers in employment, the capital intensity, or the capital ratio (barriers to entry) can be used. Given the high level of industry aggregation, it has been decided to introduce economies of scale only. A "representative economy" has been built, pooling British, French, German and Italian firms, by size. Calculation is carried out at the three-digit level of the NACE. The relative productivity of larger firms (>500 employees) is calculated. Lastly, FDI is disentangled into outward and inward stocks and we consider bilateral relations between the reporting country k and its partner k', and relations of k with all other countries: accordingly the notations will be respectively OUT and IN, OUTOTH and INOTH. Our estimates integrate the proximity-concentration trade off introduced by Brainard, since both distance and scale economies are introduced as explanatory variables, or as various combinations of fixed effects. 4 We acknowledge financial support from the OECD and help from Mr Harrington.

Using such specification, British/French/US exports (imports) to (from) each trade partner are explained by a vector of macro-economic variables, plus 4 FDI variables: - Outward British/French/US FDI to the partner (OUT) ; - Inward FDI in the UK/France/USA from the partner (IN) ; - Outward British/French/US FDI to third countries (OUTOTH) ; - Inward FDI in the UK/France/USA from third countries (INOTH). We consider in the next section three panels of industries: British industries, French industries and US ones. Macroeconomic variables have generally the right sign in our equations when they are significant. The average size of markets (declaring country and partner), the average income per capita, the economic distance (intensity of the comparative advantage), the adjacency and the regional integration have a positive impact on the value of trade flows; and reciprocally for the difference in market sizes. Such results are extensively replicated in the literature and deserve much attention. We will focus in the following on the specific role of FDI, returns to scale and transaction costs. An analysis for three panels of industries The estimation procedure is the following: for each of our three countries we consider a panel of industries and partner countries over the period referred to above. The UK, France and the United- States are modelled separately. Given the large discrepancies in parameter estimates between these three reporting countries, it does not make sense pooling their data. We consider the panel as a sequence of (annual) periods. Each panel is three-dimensional: industry, partner, time. Not surprisingly, the Lagrange multiplier rejects OLS estimates. Turning to the choice between fixed effects and random ones, fixed effects have been preferred since one tries to control for structural determinants other than the ones associated with the explanatory variables. The Hausman test generally validates this choice. In addition, different combinations of fixed effects are

introduced in order to account for unobserved industry and partner effects 5. Lastly, fixed effects on periods are introduced, with the exception of the estimates for the UK, for which the Fisher test rejects such effects on periods. In sum, period plus partner fixed effects should be introduced systematically. However we must 6 maintain a fixed effect on periods, since we cannot add more than one fixed effect. The latter will alternatively be associated with partners or industries. Firstly, we estimate bilateral import and export equations (equations [1]), using the classical vector of macro-economic variables plus our 4 bilateral variables of FDI; in addition, a variable of economies of scale (having no time dimension) plus a fixed effect on partners account for the proximityconcentration trade off. Such fixed effects capture the bilateral distances for instance. But the existence of a common border, language culture etc. is also captured. Notice that according to estimates, some macroeconomic variables can be missing. [1] Z ijkt = α AVRGDP 1 + α DFGDP 2 + α AVRGDPPC 3 + α DFGDPPC 4 + α CPOL 5 + α SCALE 6 k + α OUT 7 ijkt + α IN 8 ijkt + α OUTOTH 9 ikt + α 10 INOTH ikt + e j + e t Z = X, M Secondly we estimate bilateral import and export equations (equation [2]), using the same vector of macro-economic variables plus our 4 bilateral variables of FDI. The industry dimension is no longer captured by a measure of economies of scale but by fixed effects on industries: hence, remaining industry specificity (e.g. capital intensity or market structures) is controlled for. Reciprocally, the fixed effect on the partner country is dropped out, and country specificity is captured by the traditional distance and adjacency variables. [2] Z ijkt = α AVRGDP α OUT 8 1 Z = X, M ijkt + α IN 9 + α DFGDP 2 ijkt + α 10 + α AVRGDPPC 3 OUTOTH ikt + α 11 INOTH + α DFGDPPC ikt 4 + e k + e t + α CPOL 5 + α DIST 6 ij + α ADJ 7 ij + 5 Hummels (1998) provides the theoretical framework for such fixed effects. 6 This is a binding constraint with LIMDEP-7.

Then we estimate bilateral import and export equations excluding economies of scale, distance and adjacency, but we integrate alternatively industry fixed effects (equation [3]), or country fixed effects (equation [4]). [3] Z ijkt = α AVRGDP α OUT 6 1 Z = X, M ijkt + α IN 7 + α DFGDP 2 ijkt + α OUTOTH 8 + α AVRGDPPC 3 ikt + α INOTH 9 + α DFGDPPC ikt 4 + e k + e t + α CPOL 5 + [4] Z ijkt = α AVRGDP α OUT 6 1 Z = X, M ijkt + α IN 7 + α DFGDP 2 ijkt + α OUTOTH 8 + α AVRGDPPC 3 ikt + α INOTH 9 + α DFGDPPC ikt 4 + e j + e t + α CPOL 5 + On the whole, countries with which France or the United States trade the most are also countries with which bilateral links are the strongest: culture, history, language, geographic proximity. Hence, if such source of variance remains unobservable, this influence is captured by the FDI variable in absence of fixed effects on the partners: we get a biased estimator. Fixed effects on partners account for the whole bilateral unexplained variance. Hence, they are good substitutes for adjacency or transport costs variables. The British case Variables related to third countries have been dropped out of the British estimates as a result their low significance. This being done, one obtains parameter estimates reported in Table 2 below. Concerning British outward FDI, we observe in equation [1] a slight complementarity with bilateral British exports and imports; for each additional dollar invested abroad, one records a 2 cents increase in British exports and a 6 cents increase in British imports. These figures are small however, in comparison with the ones that would have been obtained using data for FDI flows. Estimates not reported here indicate that this is to be compared with 30 cents of additional exports and imports per dollar of outward FDI in the case of FDI flows. How this must be interpreted has already been discussed in the survey above, noticeably in relation with the Bayoumi and Lipworth paper. In

addition, this impact translates in bilateral British net imports (-4 cents). In contrast, estimates at the aggregated level would identify a slight positive net impact (2 cents), due to benefits for other industries than the one in which the foreign subsidiary is settled. Turning to inward FDI in the UK, a slight substitution effect with British imports is observed, whereas there is a slight complementarity effect with exports (respectively 6 and 3 cents). Two unexpected results have to be mentioned concerning this first equation as well as equation [4]. Firstly, the common partnership in the EU has no significant impact on trade. Such effect must be captured by the fixed effect on partners. Secondly, economies of scale are not significant either. This justifies turning to equation [2]: this equation models the industry dimension (economies of scale, production function, etc.) as a fixed effect. In addition, it reintroduces transportation costs, an important determinant in the proximity-concentration trade off. With the exception of a slight increase in the parameter for OUT, results are comparable as far as FDI related variables are concerned. More interestingly, transportation costs (negative impact) and the common partnership in the EU have the expected sign and are highly significant. This equation appears to be the more reliable specification. To check it, let us consider equation [3]: the parameter on distance is inflated since the belonging to the EU (narrow countries) is not controlled for. In total, our estimates deserve any relationship of strong complementarity or substitution between FDI stocks and trade at the sectorial level in the British case. As far as the net impact on the trade balance is concerned, we observe an "inward FDI-induced" slight improvement of the British sectorial trade balance and reciprocally a slight "outwardfdi-induced" worsening of the sectorial trade balance. Given the related magnitudes of the coefficients, these are certainly secondorder impacts however.

Table 2 United-Kingdom estimates, stocks, 1987-1996 Export Import effect partn. effect indus. effect indus. + Dist effect partn. + Scale effect partn. effect indus. effect indus. + Dist effect partn. + Scale Equation [4] Equation [3] Equation [2] Equation [1] Equation [4] Equation [3] Equation [2] Equation [1] OUT 0.024* 0.043*** 0.037*** 0.021* 0.064*** 0.066*** 0.060*** 0.060*** (1.910) (3.509) (3.083) (1.621) (3.465) (3.343) (3.085) (3.198) IN 0.034*** 0.031*** 0.031*** 0.034*** -0.061*** -0.068*** -0.068*** -0.062*** (3.477) (3.199) (3.157) (3.339) (-4.141) (-4.288) (-4.362) (-4.146) OUTOTH INOTH AVRGDP 0.0005*** 0.0004*** 0.0004*** 0.0005*** 0.0005*** 0.0005*** 0.0006*** 0.0005*** (4.206) (18.850) (20.137) (4.256) (2.764) (16.136) (16.173) (2.908) DFGDP 378.50) -439.130*** -433.60*** 455.97 684.37-686.560*** -681.52*** 788.17 (0.976) (-7.744) (-7.766) (1.104) (1.177) (-7.609) (-7.590) (1.277) AVRGDPPC 0.023** 0.007*** 0.005* 0.025** 0.028* 0.021*** 0.019*** 0.028* (2.273) (2.711) (1.882) (2.394) (1.832) (4.978) (4.490) (1.790) DFGDPPC -0.021*** -0.011*** -0.021*** -0.023*** -0.015 0.0008-0.008* -0.017 (-2.686) (-4.000) (-6.958) (-2.828) (-1.284) (0.180) (-1.664) (-1.371) DIST -0.024*** -0.021*** (-8.059) (-4.555) ADJ CPOL -12.860 407.520*** 267.56*** -32.597-15.521 529.190*** 401.97*** -16.342 (-0.117) (27.321) (8.357) (-0.287) (-0.094) (12.174) 7.806 (-0.095) SCALE -25.428-78.959 (-0.617) (-1.282) Nobs 2029 2029 2029 1908 2029 2029 2029 1908 Adj. R² 0.54 0.55 0.57 0.55 0.47 0.42 0.42 0.48 F value 76.27 133.02 133.64 70.66 57.39 77.36 75.26 54.60 LM test 4280 10908 11666 3940 4426 4463 4612 4366 Hausman test 17.22 4.44 3.60 17.41 13.92 3.82 3.08 13.81 Condition 13.10 (6.04) 13.10 (6.04) 14.84 (6.66) 18.25 (8.83) 13.10 (6.04) 13.10 (6.04) 14.84 (6.66) 18.25 (8.83) Number Student statistics in parentheses, significant at the level : 1% (***), 5% (**), 10% (*). Note : Last row, condition number without intercept in parentheses

The French case On the whole, the diagnosis of slight complementarity between trade flows and FDI stocks is validated by our methodology, with the exception of the relationship between imports and outward FDI stocks. Noticeably, inward FDI stocks highlight a strong complementarity with French exports and imports. In both cases, the relationship associated with third-countries is rather negligible. Let us consider these results in detail. Considering equation [1], a one dollar outward FDI stock is associated with only 6 cents of additional exports, in the industry considered, vis-à-vis the country of investment. Hence, notwithstanding possible substitution between FDI and exports at the level of individual firms, the French panel of 19 industries validates the complementarity between exports and outward FDI stocks. It does neither mean that such complementarity is necessarily observed for each industry, nor that foreign affiliates production would lead the same result However outward FDI complements exports only. Outward FDI is in contrast associated with reduced imports. It could be the case that France imports more (is disadvantaged) in certain industries, 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. The only way to control for such outcome is to introduce fixed effects on industries: this however will impose to drop economies of scale out. Lastly, conditional to the previous remark, we get a provisional information on the magnitude, and not only the direction, of the relationship between net trade and FDI stocks within industries. Here, each dollar of French investment abroad is associated with a 15-cent trade surplus in the industry of the investment, vis-à-vis the destination country. Turning to inward FDI, equation [1] highlights a strong relationship of complementarity between trade flows and FDI stocks. One dollar of inward FDI is associated with 46 cents of additional imports and 28 cents of additional exports. Hence, there is in total 25 cents of additional net imports associated 16

with each dollar of inward FDI. Using equation [2], we control the individual specificity of the 19 industries in the panel: economies of scale, market structures, comparative advantage etc. Transaction costs are controlled by the adjacency variable, while geographic distance does not enter successfully in the estimates. When we control for the level of comparative advantage of individual industries, parameter estimates for imports are boosted to 0.18 for OUT and 0.65 for IN. In contrast, the parameter estimate for OUT in the export equation is no longer significant. Equation [3] includes fixed effects on industries and periods but does not control for transaction costs. Here, the parameter estimate for outward stocks recovers its significance. Equation [4] has fixed effects on periods and partners, and highlights results similar to equation [1]. Once again OUT is not significant in the export equation while the substitution effect between outward FDI and imports is assessed. How have these results to be interpreted? When we control for the fact that competitive industries export and invest more abroad, the positive impact of outward FDI on exports vanishes. To put it differently, as far as FDI stocks are considered, there is no evidence that outward FDI and exports are complementary if one introduces fixed effects on industries in the panel. Such result contrasts with the view that FDI and exports complement each other. Conversely, there is a large complementarity between inward FDI and trade flows: countries that trade the most with France are also those which invest the most in France. Hence, if one omits the fixed effects on partners, the parameter estimate associated with FDI variable is artificially inflated. Such outcome raises concern about the magnitude of the complementarity effect, if not about the complementarity itself. Such complementarity being much larger with imports than with exports, a large amount of additional net imports is clearly associated with inward FDI in the French case. Table 3 Export equations: France, 1989-1994 Effects on periods and partners industries industries +Dist 17 partners +Scale Equation [4] Equation [3] Equation [2] Equation [1] OUT 0.024 0.058** -0.020 0.063*** (1.097) (2.088) (-1.001) (2.819) IN 0.287*** 0.488*** 0.409*** 0.284*** (9.109) (12.447) (12.582) (8.585) OUTOTH -0.005** -0.002-0.0009 0.015***

(-2.044) (-0.184) (-0.123) (5.678) INOTH 0.003*** 0.011 0.013 0.014*** (8.730) (1.146) (1.589) (3.425) AVRGDP 0.003*** 0.0001*** 0.0002*** 0.0004*** (3.575) (4.992) (15.111) (4.152) DFGDP -568.012*** -190.59*** (-19.991) (-7.665) AVRGDPPC DFGDPPC -0.001*** -0.015*** -0.002** -0.011*** (-2.615) (-14.946) (-2.175) (-2.993) CPOL 331.831*** (19.102) SCALE 44.453* (1.823) ADJ 743.651*** (33.271) DIST Intercept 38.367 626.41*** 12.831-47.808 (0.452) (15.278) (0.350) (-0.512) Adj. R² 0.61 0.36 0.57 0.69 F value 122.58 101.25 213.18 141.83 Nobs 4896 4896 4896 3987 LM test 49023 4999 10298 44914 Hausman test 9.7 4.83 - - Condition Number 7.9 (5.6) 9.9 (6.2) 14.2 (11.3) 18.7 (7.4) Student statistics in parentheses, significant at the level: 1% (***), 5% (**), 10% (*); Note : Last row, condition number without intercept in parentheses Table 4 Import equations: France, 1989-1994 Effects on periods and partners industries industries +Dist 18 partners +Scale Equation [4] Equation [3] Equation [2] Equation [1] OUT -0.114*** -0.097*** -0.182*** -0.093*** (-4.252) (-2.881) (-6.270) (-3.381) IN 0.521*** 0.735*** 0.652*** 0.462*** (13.403) (15.516) (15.773) (11.307) OUTOTH -0.006** -0.0003 0.0005 0.005 (-2.105) (-0.028) (0.052) (1.508) INOTH 0.022*** -0.003-0.001 0.003 (4.608) (-0.267) (-0.135) (0.541) AVRGDP 0.0001*** 0.0002*** (5.719) (14.219) DFGDP -598.77*** -198.19*** (-17.444) (-6.281) AVRGDPPC 0.061 (0.665) DFGDPPC -0.017*** -0.003*** -0.007 (-13.953) (-2.507) (-1.416) CPOL 364.92*** (16.554) SCALE 53.207* (1.771) ADJ 779.93*** (27.5) DIST Intercept 205.05*** 685.85*** 29.105 171.761 (20.665) (13.847) (0.626) (0.931) Adj. R² 0.56 0.31 0.48 0.67 F value 102.91 79.94 151.68 128.04 Nobs 4986 4986 4896 3978 LM test 47096 1213 2181 42265 Hausman test 19.21 3.48 - - Condition Number 5.1 (4.6) 9.9 (6.2) 11.9 (6.4) 25 (12)

Student statistics in parentheses, significant at the level: 1% (***), 5% (**), 10% (*); Note : Last row, condition number without intercept in parentheses The US case Result obtained so far for our two European countries can be compared with the parameter estimates obtained with the US panel, which are much larger as far as outward FDI is concerned. Here, the complementarity between trade flows and FDI outward stocks is much higher at the industry level for the United States. It does not prove however that there is any benefit for the US sectorial trade balances. Particularly, when fixed effects on partners are introduced, the induced net imports associated with US outward FDI are large. Consider equation [1], [3] and [4]: there is roughly a "one to one" complementarity between US outward FDI stocks and US trade. At the industry level, each dollar invested abroad is complements one dollar of US exports. Such outcome has to be compared with the results obtained in the import equation. In the latter, we get a range of parameters between 1.5 and 1.7 for outward FDI. Hence, the complementarity is very large, as compared with results obtained for European countries, and leads to net imports. This may be due to imports from affiliates located in low wage countries. It must be stressed than the complementarity between outward FDI and exports is greater when one does not control for trade policies (equation [3] to be compared with equation [2]). Turning to inward FDI in the US, there is no scope for large complementarity with US exports: the huge US market attracts foreign investors and there is low chance that foreign countries invest in the US in order to ship products back home. In contrast, a significant reduction in US imports is expected as a result of tariff jumping strategies from foreign investors. Alternatively, investments in US industries can lead to imports of intermediate goods (parts, components). Hence substitution and complementary forces act in an opposite way and we will only observe the net impact of these forces on US trade flows. Our results confirm the weak impact of inward stocks on US exports: in equation [3] there is a weak substitution effect, turning to a weak complementarity when one introduce economies of scale and put the fixed effects on partners (and not on industries) as in equation [1]. Lastly, there is no 19

significant impact in equations [2] and [4]. Turning to imports, there is a 50 cents to 60 cents complementarity with inward FDI, as far as one introduces fixed effects on partners. Such complementarity effect however vanishes if one introduces effects on industries. To put it differently, the complementarity effect between inward FDI and US imports is largely due to the fact that US competitive industries import less and attract less entries, whereas US non-competitive industries attract simultaneously imports and foreign FDI. In total, it can hardly be argued that a complementarity effect between FDI stocks and imports exists: it is firstly a matter of import and inward FDI sectorial composition. Table 5 Export equations: United-States, 1989-1994 Effects on periods and partners industries industries +Dist partners +Scale Equation [4] Equation [3] Equation [2] Equation [1] OUT 0.937*** 1.191*** 0.719*** 0.923*** (12.026) (16.446) (10.333) (13.247) IN 0.041-0.161*** -0.017 0.106* (0.546) (-2.945) (-0.352) (1.753) OUTOTH 0.056*** 0.002 0.002 0.014 (5.741) (0.037) (0.439) (1.48) INOTH -0.051*** 0.002-0.006-0.014 (-3.921) (0.065) (-0.248) (-1.167) AVRGDP 0.001*** 0.002*** (3.641) (6.967) DFGDP AVRGDPPC DFGDPPC CPOL 4942.601*** (5.698) SCALE 21458*** (13.021) ADJ 3562.612** (5.135) DIST Intercept 241.45-3175.421-5966.647*** -25054*** (0.741) (-1.393) (-3.021) (-12.735) Adj. R² 0.55 0.52 0.64 0.64 F value 24.15 49.81 72.89 33.79 Nobs 715 715 715 715 LM test 448 914 1501.22 723 Hausman test 8.6 1.29 2.13 12.6 Condition Number 9.1 (7.6) 20.7 (9) 21.6 (9.4) 54.5 (9) Student statistics in parentheses, significant at the level: 1% (***), 5% (**), 10% (*); Note : Last row, condition number without intercept in parentheses Table 6 Import equations: United-States, 1989-1994 Effects on periods and partners industries industries partners 20

+Dist +Scale Equation [4] Equation [3] Equation [2] Equation [1] OUT 1.752*** 1.531*** 1.051*** 1.734*** (12.914) (11.707) (7.514) (13.669) IN 0.518*** -0.083 0.061 0.614*** (4.435) (-0.836) (0.618) (5.611) OUTOTH 0.086*** -0.047-0.032 0.027* (5.055) (-0.581) (-0.391) (1.608) INOTH -0.133*** 0.016 0.006-0.079*** (-5.645) (0.291) (0.121) (-3.508) AVRGDP 0.006*** 0.006*** (9.778) (11.423) DFGDP AVRGDPPC DFGDPPC 0.129*** 0.136*** (3.741) (4.000) CPOL 6918*** (3.875) SCALE 29808,102*** (9.931) ADJ 1908.221 (1.334) DIST Intercept 527.76-17152*** -20055*** 34623.210*** (0.929) (-4.125) (-4.998) (-9.673) Adj. R² 0.53 0.47 0.51 0.58 F value 22.53 37.17 39.77 27.68 Nobs 715 715 715 715 LM test 989.32 325.5 394.3 1271.48 Hausman test 11.97 2.92 3.22 13.27 Condition Number 9.1 (7.6) 22.5 (9.7) 23.3 (10) 54.5 (9) Student statistics in parentheses, significant at the level: 1% (***), 5% (**), 10% (*); Note : Last row, condition number without intercept in parentheses Conclusion FDI and international production are developing at a high pace. Intra-firm trade accounts for one third of world exports and foreign affiliates' sales grow faster than world exports. The question is whether FDI displaces trade, output and employment. Outward FDI may displace trade at the level of the firm, since foreign affiliates local sales substitute at least partially for exports. It is however not necessarily the case at the industry level: this paper has tried to tackle the latter issue in a systematic manner using data broken down by sector, country and partner, in a comparative perspective. Explaining exports (imports) as a function of FDI and different control variables, one is interested in the sign and magnitude of the parameter estimate associated with foreign investment. If a negative parameter on FDI is obtained in trade equations, FDI displaces trade, detrimental to the domestic industry of the investor. In this case, production and employment can be negatively affected in home 21

country. If a positive parameter is obtained trade and FDI are complements, to the benefit of investing country's exports and industry. The result of such estimates is however largely dependent on the type of FDI data used. Permanent and temporary effects of FDI on trade must be disentangled. According to Bayoumi and Lipworth, the former can be identified using FDI stocks, whereas the latter are associated with FDI flows. Focusing on permanent effects, the current paper aimed at modelling bilateral export equations as a function of 1) macroeconomic variables, 2) variables associated with the so called "proximityconcentration trade off" and 3) FDI stocks. These equations have been estimated using three panels of industries, hence using FDI and trade data broken down in the same sectorial nomenclature. Estimates referred to in the current paper deserve any systematic permanent complementarity effect at this industry level. In the British case, outward FDI only slightly complements trade flows: we record a 2 cents increase in British exports and a 6 cents increase in British imports associated with each dollar of FDI stock. Such permanent effect is slightly detrimental to the British sectorial trade balance. However, these figures are rather negligible in comparison with the temporary effects: roughly 30 cents of additional imports and exports if one had considered FDI flows. Inward FDI in the UK has a slight permanent substitution effect on British imports, and conversely slightly complements British exports. In total, the impact of inward FDI on the British trade balance is positive. In the French case, results are similar at a first glance: FDI stocks slightly complement exports. However, such outcome is largely due to the fact that French competitive industries export and invest largely abroad. Having controlled for such structural composition, by introducing fixed effects on industries, we conclude that there is no significant permanent effect of FDI on French exports. Hence, the increase in exports associated with FDI flows appears to be only transitory. Conversely, it proves at least that there is no permanent trade displacement when industries invest abroad, in contrast with popular fears. Turning to inward FDI, the complementarity is ascertained and of a large magnitude. This translates in permanent net imports (roughly -20 cents per dollar when controlling 22

for the geographic composition of French exports, and alternatively 25 cents when controlling for the geographic structure). These results can be compared with the transitory impact: controlling for economies of scale, Fontagné and Pajot (2000) record that a one-dollar outward FDI flow is associated with 54 cents of additional exports and with 24 cents of additional imports, in the industry considered vis-à-vis the country of investment. The US case leads to very different conclusions. Firstly, a strong complementarity between outward US FDI and trade is ascertained. Such complementarity is associated with sizeable net imports, ranging from 30 cents to 80 cents according to the specification considered. These are striking orders of magnitudes since permanent effects are involved. Not surprisingly, inward FDI has no significant impact on the US exports and leads to additional imports only when one does not control for the sectorial composition of imports. These results can be once again compared with transitory effects. With fixed effects on periods only, Fontagné 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: 35 cents of additional exports and twice as much additional imports. In total, European countries considered here largely validate the conjecture of dominating transitory effects of FDI on trade flows. The long-term effect is quite negligible, with the exception of inward FDI in the French case. Such result contrasts with the large complementarity obtained when considering US outward FDI, indicating that even in the long run there must be a positive impact of FDI on trade flows. Some limitations of this analysis must be underscored. First, correlation does not mean causality; hence, we do not consider that FDI causes trade imbalances. Second, our estimates misses long run benefits (technology transfer etc.) generally associated with FDI, noticeably with inward FDI. Lastly, the dynamics of the relationship has not been tackled since no lagged variables were introduced. 23

References ANDERSEN P.S., HAINAUT P. [1998], Foreign Direct Investment and Employment in the Industrial Countries, Bank for International Settlements Working Paper, 61. ANDERSSON T. [1993], "Nya villkor för ekonomichi och politik. Ekonomikommissions förslag" SOU, 1993:16, bilaga 3 (s85-107). BALASSA B., BAUWENS L. (1987), Intra-Industry Specialization in a Multi-Country and Multilateral Framework, The Economic Journal, (97), 923-939. BAYOUMI T., LIPWORTH G. [1997], "Japanese Foreign Direct investment and Regional Trade", IMF Working Paper, WP/97/103. BRAINARD S.L., [1993], "An Empirical Assessment of the Proximity-Concentration Tradeoff between Multinational Sales and Trade", NBER, Working Paper, 4583. BLOMSTRÖM M., KOKKO A. [1994], "Home Country Effects of Foreign Direct Investment : Evidence from Sweden ", CEPR discussion paper, 931, April. BLOMSTRÖM M., LIPSEY R., KULCHYCKY K. [1988], " US and Swedish Direct Investment and Exports", in R. BALDWIN ed. Trade Policy Issues and Empirical Analysis, Chicago, Chicago University Press. CANTWELL J.A. [1994] "The Relationship between International Trade and International Production", in D. Greenaway and L.A. Winters (eds.), Surveys in International Trade, Oxford: Basil Blackwell, 1994, 303-328. CHEDOR S., MUCCHIELLI J.L. [1998], Implantation à l étranger et performance à l exportation. Une analyse empirique sur les implantations de firmes françaises, Revue Economique, 49, (3), 617-628. DRABEK Z. [1998], "A Multilateral Agreement on Investment: Convincing the Sceptics", WTO Staff Working Paper, ERAD-98-05. EATON J., TAMURA A. [1994], "Bilateralism and regionalism in Japanese and US trade and direct foreign investment patterns "NBER working paper 4758. FONTAGNE L., PAJOT M. [2000], "Relationships between Trade and FDI Flows within Two Panels of US and French Industries", in R.E. LIPSEY and J.L. MUCCHIELLI eds. Multinational Firms and Impacts on Employment, Trade and Technology. New Perspectives for a New Century, forthcoming. HEAD K., RIES J. [1997], Overseas Investment and Firm Exports, University of British Columbia Working Paper. HUMMELS D. [1998], "Toward a Geography of Trade Costs", Miméo, University of Chicago. LIPSEY R.E., WEISS M.E.. [1984], "Foreign Production and Exports of Individual Firms" Review of Economics and Statistics, (63), 4, pp. 488-494. MARKUSEN J.R., VENABLES A. J. [1995], "Multinational firms and the new trade theory" NBER working paper, 5036, February. PAIN N., WAKELIN K. [1998], "Export Performance and the Role of Foreign Direct Investment", NIESR Discussion Paper, 131. RIKER D.A., BRAINARD L.S. [1997], U.S. Multinationals and Competition from Low Wage Countries, NBER Working Paper, 5959. STEVENS G., 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, United Nations, Geneva. WTO [1998], Report of the Working Group on the Relationship Between Trade and investment to the General Council, WT/WGTI/2, December. 24