{"@context":{"@language":"en","Affiliation":"http:\/\/vivoweb.org\/ontology\/core#departmentOrSchool","AggregatedSourceRepository":"http:\/\/www.europeana.eu\/schemas\/edm\/dataProvider","Campus":"https:\/\/open.library.ubc.ca\/terms#degreeCampus","Creator":"http:\/\/purl.org\/dc\/terms\/creator","DateAvailable":"http:\/\/purl.org\/dc\/terms\/issued","DateIssued":"http:\/\/purl.org\/dc\/terms\/issued","Degree":"http:\/\/vivoweb.org\/ontology\/core#relatedDegree","DegreeGrantor":"https:\/\/open.library.ubc.ca\/terms#degreeGrantor","Description":"http:\/\/purl.org\/dc\/terms\/description","DigitalResourceOriginalRecord":"http:\/\/www.europeana.eu\/schemas\/edm\/aggregatedCHO","FullText":"http:\/\/www.w3.org\/2009\/08\/skos-reference\/skos.html#note","Genre":"http:\/\/www.europeana.eu\/schemas\/edm\/hasType","GraduationDate":"http:\/\/vivoweb.org\/ontology\/core#dateIssued","IsShownAt":"http:\/\/www.europeana.eu\/schemas\/edm\/isShownAt","Language":"http:\/\/purl.org\/dc\/terms\/language","Program":"https:\/\/open.library.ubc.ca\/terms#degreeDiscipline","Provider":"http:\/\/www.europeana.eu\/schemas\/edm\/provider","Publisher":"http:\/\/purl.org\/dc\/terms\/publisher","Rights":"http:\/\/purl.org\/dc\/terms\/rights","ScholarlyLevel":"https:\/\/open.library.ubc.ca\/terms#scholarLevel","Title":"http:\/\/purl.org\/dc\/terms\/title","Type":"http:\/\/purl.org\/dc\/terms\/type","URI":"https:\/\/open.library.ubc.ca\/terms#identifierURI","SortDate":"http:\/\/purl.org\/dc\/terms\/date"},"Affiliation":[{"@value":"Arts, Faculty of","@language":"en"},{"@value":"Political Science, Department of","@language":"en"}],"AggregatedSourceRepository":[{"@value":"DSpace","@language":"en"}],"Campus":[{"@value":"UBCV","@language":"en"}],"Creator":[{"@value":"Manger, Mark S.","@language":"en"}],"DateAvailable":[{"@value":"2009-12-17T20:53:41Z","@language":"en"}],"DateIssued":[{"@value":"2005","@language":"en"}],"Degree":[{"@value":"Doctor of Philosophy - PhD","@language":"en"}],"DegreeGrantor":[{"@value":"University of British Columbia","@language":"en"}],"Description":[{"@value":"In recent years, preferential trade agreements (PTAs) have proliferated rapidly, creating a spaghetti bowl of bilateral treaties. A growing share are North-South- FTAs between developed countries and emerging markets. Why the sudden proliferation of these agreements, when most-favoured-nation tariffs are at a historically low level? This dissertation argues that PTAs are the product of a competitive dynamic among states\u2014but not a competition over export markets. Rather, firms and countries compete over access to locations for FDI. Growth of the service sector and global integration of production networks motivate multinational firms to invest in emerging market countries. At the same time, manufacturing firms have strong incentives to seek strict rules-of-origin. In service industries, market structures work against late entrants. Preferential liberalization thus discriminate strongly against third parties. The combination of bilateral agreements and increased FDI flows has an unintended effect: excluded firms lobby for defensive agreements with host countries, triggering a rapid proliferation of FTAs. This study develops a model that explains these dynamics and tests it in several case studies. The in-depth case studies cover the North American Free Trade Agreement (NAFTA) and the two defensive agreements Japan and the EU signed with Mexico, followed by a range of survey cases, including the Japanese initiative for an FTA with Thailand and US and EU FTAs with Chile. The findings of the study imply that bilateral agreements come close to sectoral liberalization and threaten to undermine the multilateral trade regime. However, given the competitive dynamic of PTAs, preferential liberalization is going to become even more popular in the future.","@language":"en"}],"DigitalResourceOriginalRecord":[{"@value":"https:\/\/circle.library.ubc.ca\/rest\/handle\/2429\/16896?expand=metadata","@language":"en"}],"FullText":[{"@value":"Competitive Liberalization: The Proliferation of Preferential Trade Agreements by Mark S. Manger Dipl.-Pol., Universitat Hamburg, 1999 A THESIS SUBMITTED IN PARTIAL F U L F I L M E N T OF THE REQUIREMENTS FOR T H E D E G R E E OF DOCTOR OF PHILOSOPHY in The Faculty of Graduate Studies (Political Science) T H E UNIVERSITY OF BRITISH C O L U M B I A August 2005 \u00a9 Mark S. Manger, 2005 11 Abstract In recent years, preferential trade agreements (PTAs) have proliferated rapidly, creating a spaghetti bowl of bilateral treaties. A growing share are North-South-FTAs between developed countries and emerging markets. Why the sudden prolif-eration of these agreements, when most-favoured-nation tariffs are at a historically low level? This dissertation argues that PTAs are the product of a competitive dy-namic among states\u2014but not a competition over export markets. Rather, firms and countries compete over access to locations for FDI. Growth of the service sector and global integration of production networks motivate multinational firms to invest in emerging market countries. At the same time, manufacturing firms have strong incentives to seek strict rules-of-origin. In service industries, market structures work against late entrants. Preferential lib-eralization thus discriminate strongly against third parties. The combination of bilateral agreements and increased FDI flows has an unintended effect: excluded firms lobby for defensive agreements with host countries, triggering a rapid pro-liferation of FTAs. This study develops a model that explains these dynamics and tests it in several case studies. The in-depth case studies cover the North American Free Trade Agreement (NAFTA) and the two defensive agreements Japan and the E U signed with Mexico, followed by a range of survey cases, including the Japanese initiative for an FTA with Thailand and US and E U FTAs with Chile. The findings of the study imply that bilateral agreements come close to sec-toral liberalization and threaten to undermine the multilateral trade regime. How-ever, given the competitive dynamic of PTAs, preferential liberalization is going to become even more popular in the future. i i i Contents Abstract i i Contents i i i List of Tables vii List of Figures vii i List of Abbreviations ix Acknowledgements xi 1 Introduction 1 1.1 Unequal Partners: The Proliferation of North-South PTAs 3 1.2 Perspectives on Preferential Trade Agreements 8 1.3 Firms, States, and Competitive Liberalization: An Overview of the Argument 12 1.4 Methodology 15 1.4.1 NAFTA and its Effects 16 1.4.2 Survey Cases: Testing the Limits of Applicability 17 1.5 Caveats, Limitations, and Contributions 18 1.6 Precis of the Dissertation 19 2 Framework for Analysis 21 2.1 A Political Economy Approach 23 2.1.1 Assumptions about Firms 26 2.1.2 Emerging Markets and the Economics of Vertical Integration 27 2.2 Bilateral Agreements as Exclusive Clubs 30 iv 2.2.1 Closing the Backdoor: Rules of Origin and Export-Processing Zones 30 2.2.2 Getting a Head Start: First-Mover Advantages in Services 34 2.2.3 Relative Strength vis-a-vis Protectionist Forces 37 2.3 The Endogenous Dynamic of FTAs 39 2.3.1 Non-Member Firms'Reactions to Exclusion 39 2.3.2 Intensifying the Competition 42 2.4 Possible Objections 43 2.5 Networking the Globe 45 3 NAFTA\u2014The Original Sin? 47 3.1 Bilateralism in US Trade Policy 48 3.2 Mexico's Unilateral Liberalization 49 3.3 The Mexican Export Platform: FDI as Restructuring Agent 53 3.3.1 US FDI in Mexico before N A F T A 53 3.3.2 Retooling Mexican Factories for Export 54 3.3.3 Export Orientation in the Automotive Industry 56 3.3.4 Growth in the Maquiladora Sector 58 3.4 The N A F T A Negotiations: A Preference for Protectionism 60 3.4.1 Rules of Origin for Automobiles \u2022. 61 3.4.2 Banning M F N Rebates for Non-NAFTA Firms 62 3.4.3 Rapid Liberalization in Services 63 3.4.4 Preferential Aspects of N A F T A and Multilateral Commit-ments 65 3.5 Alternative Accounts: The Benign View of N A F T A 69 3.6 Conclusion: Raising the Barriers to Market Entry 70 4 A Beachhead in the North American Market 72 4.1 The Gravity Model\u2014An Overview 73 4.1.1 Establishing the Baseline Model 73 4.1.2 Estimating the Effects of Preferential Trading Arrangements 75 4.1.3 Model Specification 76 4.1.4 Data Sources 78 V 4.2 Estimation 80 4.2.1 Estimation Results 82 4.3 Conclusion: Getting a Foot into the Door 88 5 Countering the Effects of NAFTA 89 5.1 The EU-Mexico Free Trade Agreement 90 5.1.1 The Institutional Background: Trade Policy in the European Union 93 5.1.2 Moving Toward Free Trade with Mexico 95 5.1.3 Coalitions and Aims of European Lobbies 96 5.1.4 Outcomes: Beyond N A F T A Parity 101 5.2 The Japan-Mexico Free Trade Agreement 103 5.2.1 Japanese Foreign Direct Investment in Mexico 103 5.2.2 N A F T A affects Japanese Investment in Mexico 105 5.2.3 A Regional Trade Policy Option for Japan 107 5.2.4 Lobbying Efforts in Mexico and Japan 109 5.2.5 Negotiating Free Trade between Japan and Mexico . . . . 112 5.3 Alternative Accounts 115 5.4 Conclusions 116 6 The Unfolding Competition 118 6.1 The Far Side of the World: FTAs with Chile 120 6.1.1 Foreign Direct Investment in Chile 121 6.1.2 A Road Not Taken: Chile's Accession to N A F T A 125 6.1.3 Iberian Ties: The EU-Chile FTA 127 6.1.4 Catching Up: The US-Chile FTA 130 6.1.5 Outcomes: Parity Between the US and the E U 134 6.1.6 Politely Declining the Union: The Proposal for a Japan-Chile FTA 135 6.2 Japan's NAFTA? Moving Toward an FTA with Thailand 136 6.2.1 The Changing Character of FDI in Thailand 137 6.2.2 The Political Struggle over an FTA with Thailand 141 6.3 Bilateralism Comes into its Own 143 7 Conclusion 145 7.1 Findings: Proactive Agreements 147 7.2 Findings: Defensive Agreements 149 vi 7.3 Implications 152 7.4 Avenues for Further Inquiry 155 7.5 Outlook 156 Bibliography 157 Appendix I Estimating Trend Changes 187 Appendix II Investment Liberalization in Mexico 188 Appendix III Summary Statistics 191 vii List of Tables 2.1 Average Applied Unweighted Tariff Rates, Selected Countries . . 29 2.2 Preferential Tariffs with FTA and Rules of Origin 41 3.1 Liberalization of the Mexican Import Regime, 1985-1990 . . . . 52 3.2 Exports by US Firms Based in Mexico, 1983-1999 55 3.3 Sales of US firms back to the US, by Industry 56 3.4 Exports of Vehicles by Company, 1987 58 3.5 Passenger Car Sales by Principal Export Market 59 3.6 US Services Exports and FDI to Mexico 68 3.7 Average M F N Tariff Applied by Mexico 69 4.1 Variable Definitions 77 4.2 Countries in the Sample 79 4.3 Estimation Results (DGDP) 83 4.4 Estimation Results (DKPW) 84 4.5 Estimation Results (North-South Country Pairs) 86 4.6 Marginal Effects, DGDP and D K P W Specifications 87 4.7 FDI Stock from the EU-15 in Mexico, 1991-2001 87 I. 1 Numerical Minimization of Sum of Squares 187 II. 1 Comparison of Mexican Investment Regimes 188 III. 1 Descriptive Statistics 191 vii i List of Figures 1.1 Growth of North-South-PTAs 6 2.1 Preference Intensity of Firms 36 2.2 Vertical Integration without FTAs 40 2.3 With an FTA and Rules of Origin 41 3.1 US FDI Stock in Mexico and Time Trend 54 6.1 FDI Stock to GDP Ratio, Chile . . . 121 6.2 Shares of Individual Countries in Chile's FDI Stock 122 6.3 Exports of Passenger Cars and Parts from Thailand to Japan . . . 139 List of Abbreviations A C E A Association des Constructeurs Europeens d'Automobiles AICO A S E A N Industrial Cooperation A P E C Asia Pacific Economic Cooperation A S E A N Association of Southeast Asian Nations B A F T Bankers' Association for Trade and Finance B B V A Banco Bilbao Viscaya Argentaria BIT bilateral investment treaty B S C H Banco Santander Central Hispano C E E Central and Eastern Europe CEFIC Conseil Europeen de l'lndustrie Chimique C L E P A European Association of Automotive Suppliers COREPER Council of Permanent Representatives of the E U Member States CSI Coalition of Service Industries CUSFTA Canada-US Free Trade Agreement D L 600 Decree Law 600 (Chile) EP European Parliament EPZ export-processing zone ESF European Services Forum E U European Union E V S L Early Voluntary Sector Liberalization FDI foreign direct investment FTA free trade agreement FTAA Free Trade Area of the Americas GATS General Agreement on Trade in Services GATT General Agreement on Tariffs and Trade IMF International Monetary Fund JA Group Japan Agricultural Cooperatives Group J A M A Japan Automobile Manufacturers Association JETRO Japan External Trade Organization JMCTI Japan Machinery Center for Trade and Investment X JOI Japan Institute for Overseas Investment L D P Liberal-Democratic Party (Japan) M A F F Ministry of Agriculture, Forests and Fishery (Japan) M A I Multilateral Agreement on Investment M E P Member of the European Parliament M E T I Ministry of Economy, Trade and Industry (Japan) M F N most-favoured-nation MITI Ministry of International Trade and Industry (Japan) M N C s Multinational Corporations M O F A Ministry of Foreign Affairs (Japan) N A F T A North American Free Trade Agreement O E C D Organization for Economic Co-operation and Development PROSEC Programa de Promotion Sectoral PTA preferential trade agreement ROO rules-of-origin SECOFI Secretaria de Comercio y Fomento Industrial TPA Trade Promotion Authority U N C T A D United Nations Conference on Trade and Development U N C T C United Nations Conference on Transnational Corporations USITC United States International Trade Commission USTR United States Trade Representative WTO World Trade Organization xi Acknowledgements Completing this thesis would have been impossible without the support from var-ious people. First and foremost, the generous backing I have received from my supervision committee was invaluable. As my research supervisor, Max Cameron has offered plenty of constructive critique and support for my project from beginning to end. His graduate seminar was simply the best course I have taken in my whole tertiary education experi-ence. Once I embarked on my research, he generously spent his time on count-less reviews of draft chapters, despite taking on new administrative commitments and becoming a father. Later, he provided superb guidance on how to enter the job market. Max also taught me that while striving for methodological care, we should never lose sight of why we are doing political science in the first place. As a scholar and mentor, he will be an example for me for the rest of my life. Yves Tiberghien has been a source of inspiration and guidance since he came to U B C in 2001. I have benefited enormously from the breadth and depth of his knowledge and his detailed comments on various aspects of my work. Most im-portantly, Yves never hesitated to open doors for me during my field research in Japan, from arranging my affiliation at Todai to introducing me to senior bureau-crats and academics. His enthusiasm for both research and teaching come together perfectly in graduate student supervision. If Yves can be faulted for anything, then it is that he devotes too much time to his students! Without Mark Zacher's encouragement, I would never have taken on the study of international political economy. In his inimitable way, Mark began by giving me a reading list of over 200 titles \"that would be useful to know.\" His advice has been invaluable in pointing me towards the role of FDI in international relations, and I have benefited from his wisdom on the profession in many ways. I have to be most grateful that Werner Antweiler joined my supervision com-mittee. Werner pushed me to advance my quantitative skills and directed me towards reading frontier research in international trade. As a first-class trade economist, many of my questions must have seemed basic to him, but he never hesitated to take his time for thorough explanations. Xll During my stay in Japan, I have incurred immense debts. Gregory Noble offered to sponsor my affiliation with the Institute of Social Science at Todai, and provided crucial critique of my project. Nakagawa Junji opened important doors for me and provided great insights into international economic law. The German Institute of Japanese Studies hosted me as a dissertation fellow and provided other crucial resources. I am also grateful for the chance to present my research to the uniquely multidisciplinary Institute on several occasions. In both Tokyo and Brussels, I have to extend my gratitude to the officials and industry representatives who I had the honour to interview, but whom I cannot name here for reasons of confidentiality. In Belgium, my friends Patrick and Cecile made it possible to cram an incred-ible number of interviews into a much too short period of field research. Laura's and Esteban's hospitality made the trip a pleasure. I will never forget the days spent at the James' during the crucial interview period in Tokyo in December 2003. At U B C , I have to thank Fred Cutler, Angela O'Mahoney, Ben Nyblade, Pe-ter Dauvergne, Richard Price, and Alan Jacobs for offering advice when needed. I would also like to thank Brian Job and Paul Evans for getting me to U B C in the first place and offering guidance and support despite their busy schedules. Without Richard Johnston at the helm of our department, my graduate student ex-perience would surely have been much poorer. A special thanks goes to Josephine Calazan, Petula Muller, Irina Florov, and Dory Urbano. At the Centre for Japanese Research, Julian Dierkes set an example of support for graduate students. I would also like to thank my fellow students Mark Pickup and Scott Matthews, who were always ready to discuss the state of political science. Several other students deserve special mention: Matt Gillis created a stunning animation that greatly enhanced my job talk. Mark Holmes' advice on statistics was much ap-preciated. M y office mate Kaori Yoshida helped me out on various occasions when my Japanese language skills proved insufficient. Last not least, I thank my parents for instilling me with intellectual curiosity and giving support, love and encouragement. The most special thanks go to my wife Ivette, who has endured my whining over this dissertation for years and never failed to lift my spirits. This work is for her. Mark Manger, Vancouver, July 2005 Chapter 1 i Introduction Over the past fifty years, multilateral negotiations have been spectacularly suc-cessful in bringing down tariffs, liberalizing international commerce, and foster-ing economic growth. Yet in the last decade, we have witnessed an explosion of preferential trade agreements (PTAs). The World Trade Organization Secre-tariat counts over 250 PTAs as of 2002. More and more of these agreements are concluded between developed nations and emerging market countries, bringing together economies of vastly different sizes and levels of development. By themselves, however, PTAs do not create much trade. The commitments to lower barriers they embody are dwarfed by the unilateral steps taken by emerg-ing market countries to lower tariffs. Most-favoured-nation (MFN) tariffs are at historically low levels. Trade economists are divided over whether PTAs improve welfare at all (compare inter alia Freund 2000; McLaren 2002), but almost unan-imously judge them a second-best solution to multilateral and unilateral liberal-ization. Why then the sudden proliferation of these preferential trade agreements? Why do major economic powers sign agreements with partners that bring little market size and overall welfare benefits? And finally, will these agreements be beneficial for the ultimate goal of global free trade, as claimed by the supporters of competitive liberalization?1 This dissertation argues that North-South PTAs spread first and foremost be-cause of the growing importance of foreign direct investment (FDI) in the inter-national economy. FDI flowing from developed to liberalizing developing coun-tries changes the incentives for states regarding preferential trade agreements. In particular, it motivates states to pursue bilateral2 options because of the specific benefits they entail for multinational firms: Preferential agreements for trade and 'Note the explicit use of this term by the United States Trade Representative (USTR), http:\/\/www.ustr.gov\/releases\/2003\/01\/03-02.htm, accessed August 6, 2003. 2I prefer the use of \"bilateral\" over \"regional\" as antonym of \"multilateral\" in this dissertation because it emphasizes specific rather than diffuse benefits. Most North-South agreements are in fact bilateral, but not necessarily regional in a geographical sense. 2 investment offer the benefits of liberalization but can be used to raise the barriers for competitors. Raising barriers implies a discriminatory effect. Since nearly all recent PTAs are free trade agreements (FTAs) 3 in which the members set their own external tariffs, they require rules to determine the origin of goods. These rules of origin (ROOs) can be structured to the disadvantage of outsiders and to export protec-tionism. Moreover, a large proportion of FDI flows into the service sector, where market and regulatory structures penalize late entry and provide incentives for preferential liberalization. FTAs protect member country producers, but also at-tract investment from outsiders as beachheads in the market. North-South FTAs thus trigger a competitive liberalization: other countries conclude defensive bilateral agreements with the host country out of fear of being left behind. Yet each bilateral agreements requires its own rules of origin. Con-trary to the claims of its proponents, competitive liberalization is not just a beauty contest (Schott 1992) among developing countries over who is the most open to foreign trade. Rather, developed countries have strong interests in signing bilat-eral agreements that erect new barriers to trade and investment as they tear down others. To explain these developments, this dissertation offers a theoretical explana-tion based on a political economy approach, whereby governments decide their policies in response to pressures from organized societal groups. The interests behind recent North-South PTAs are analyzed across two levels with two pairs of explanatory and dependent variables: When developed countries seek preferen-tial trade agreements, they respond to firm demands for higher barriers for non-members. The agreements create a discriminatory effect, causing excluded firms to lobby for arrangements to counteract the negative effects. The theory is de-veloped in two stages, clarifying the preferences of firms to export protectionism first, then analyzing the choices for excluded countries. Although political vari-ables may shape the decision to pursue PTAs, I emphasize the economic incen-tives that cause their proliferation, since even a PTA concluded for non-economic reasons is likely to have redistributive effects within the member countries. To test the explanation against competing accounts and demonstrate the explanatory power of my framework, I construct analytic narratives focused on the sequence of events in several cases. The in-depth case studies cover the North American Free Trade Agreement (NAFTA) and the two defensive agreements Japan and the 3 In the following, I use the term PTA whenever an argument applies to all types of preferential agreements, and refer specifically to FTAs when warranted. 3 E U signed with Mexico. To demonstrate the key causal link\u2014investment from non-members of N A F T A that used Mexico as a beachhead\u2014I estimate the ef-fects of N A F T A on FDI using standard econometric techniques. I then test the applicability of the framework on a range of survey cases, including the Japanese initiative for an FTA with Thailand and US and E U FTAs with Chile. 1.1 Unequal Partners: The Proliferation of North-South PTAs Contrary to intuition, preferential trade agreements spread across the globe at a time when multilateral negotiations have lowered trade barriers to a minimum. Clearly, this trend does not herald a return to the protectionist blocs of the 1930s. A closer look at the institutional features of the global trade regime and the charac-ter of recent PTAs shows that today's agreements coincide with profound changes in the world economy. Developing countries seek to reintegrate into the global economy, causing changes in the character of investment in these \"emerging mar-kets\" and affecting the multilateral trade regime in turn. The experience of the interwar years, when retaliatory tariffs led to the cre-ation of protectionist blocs, provided the initial impetus for the US to support the creation of the General Agreement on Tariffs and Trade (GATT). Based on the constitutive norm of non-discrimination as expressed in most-favoured-nation (MFN) tariffs, the agreement covered trade in goods only. Following the failure of the 1947 Havana Charter of the International Trade Organization to win support in US Congress, the GATT was created as a temporary solution, lacking rules for foreign investment (Diebold 1952). Services were not considered tradeable at all. Until now, the global trade regime lacks the third pillar, first due to resistance from developing countries (Lipson 1985; Vernon 1971), but more recently because of the failure of the OECD members to come to an agreement on foreign direct in-vestment in the negotiations on the Multilateral Agreement on Investment (MAI) (Graham 2000). The counterpart to the GATT, the 1994 General Agreement on Trade in Services (GATS), provides for limited liberalization by granting the right to establish a \"commercial presence\" in the host country.4 Although the GATT was designed to protect the division of countries into competing blocs, it only provides limited disciplines for preferential trade agreements. 4This is referred to as \"mode 3\" of the provision of services in GATS terminology. 4 Created in light of the historical experience, Art. X X I V of the GATT stipulates that regional integration measures have to conform to two standards. First, they should cover substantially all trade.5 Second, they must not raise the barriers against third parties above the initial M F N level at which tariffs are \"bound\" by GATT members. However, many developing counties apply much lower tariffs than their bound rates, leaving much room for increases in tariff rates. Moreover, no similar clause exists with regard to non-tariff barriers such as rules of origin or the regulation of FDI. If trade liberalization is defined as the lowering of tariff barriers only, then the GATT was spectacularly successful. Negotiations cut down manufactured goods tariffs on M F N basis from an average of 53 percent to 8.6 percent.6 Most man-ufactured imports into the industrialized countries face near-zero or no tariffs. Regarding liberalization of other barriers, though, the record is mixed. The GATS liberalizes only those sectors that countries explicitly cite in their \"positive list,\" but has no built-in mechanism for further opening of economic sectors. The ex-clusion of manufacturing investment has spurred on the conclusion of numerous bilateral investment treaties (BITs). However, the vast majority of these treaties either do not provide for liberalization or merely lock in unilateral commitments. Consequently, their potential to attract FDI is limited at best (Hallward-Driemeyer 2003; U N C T A D 1998; see Biithe and Milner 2004, for a dissenting view). Only recent preferential trade agreements significantly lower barriers to investment be-yond unilateral steps, a move referred to as \"deep integration\" (Lawrence 1996). This reflects the close link between trade and investment in today's firm strategies, outlined in greater detail below. Besides an increase in sheer number, preferential trade agreements vary in their coverage of trade and factor flows, and in the economic characteristics of their member states. When counting PTAs, a more conservative measure than the previously quoted figure of over 250 agreements is advised. A considerable num-ber of agreements fall under the \"enabling clause\" of the GATT that allows devel-oping countries to sign agreements among themselves with generous timeframes for tariff reduction, resulting in little or no actual liberalization. Other agreements only reaffirm existing tariff-free trade between states that previously belonged to the same political entity, as in the 1992 FTA signed by Slovakia and the Czech Re-5Although no clear definition could be established so far, developed countries tend to adhere to a yardstick of 90% of trade or more on a value basis. 6Taking a simple, unweighted average. 5 public. Finally, some agreements are superseded by later PTAs; others suspended for political reasons. Counting only the PTAs in force and joint GATS Art. V and GATT Art. X X I V treaties (e.g. NAFTA) as a single agreement, we arrive at a cumulative figure of 115 PTAs in 2004. Taking a minimum difference of US$ 10,000 in per capita GDP as threshold, 52 are \"North-South-FTAs\"\u2014here used as a shorthand, although some economies such as Bulgaria would be better characterized as countries in transition. Furthermore, this figure is much smaller than to be expected based on the sheer number of countries involved, since the E U has a common external trade policy and the EFTA member states7 negotiate agreements jointly. This study focuses on this growing subset of PTAs between pairs of countries that are highly unequal in their level of development and sizes of the economies. Figure 1.1 graphs the growth of these agreements over time. Until 1991, North-South-FTAs were limited to a handful of agreements, mostly between the European Community and close neighbours, such as the EC-Malta FTA of 1971. The turning point came in 1991, when countries in Latin America and Central and Eastern Europe (CEE) began to seek FTAs. However, the first of these PTAs, the \"Europe Agreements\" between Central European countries and the EC, had very little practical impact. The E C did not change its tariffs substantially, while the unilateral liberalization undertaken by C E E countries reduced tariffs below EC levels (Beach 1997). More important were the agreements that entered into force in the mid-1990s, at which point the trend came into full swing. Given the growing importance of FDI, advanced developing countries like Chile or Mexico are preferred partners in today's bilateral agreements. Prior to liberalization in the developing world, commercial interests from the North were limited to resource extraction, \"tariff-jumping\" investment by multinational firms, or, in the case of many \"developmental states,\" closely circumscribed domains of export-oriented production. Liberalization creates new opportunities and thus the incentives for interested parties in the industrialized countries to lobby for agreements to secure preferential access. In Latin America, unilateral liberalization represents the first step to overcome the legacy of import-substitution industrialization.8 During most of the 1950s to 1980s, multinational firms produced outdated products, protected by high tariffs, in markets such as Mexico, Brazil and Argentina for domestic sales. Using various 7Iceland, Norway, Switzerland and Liechtenstein. 8Imports were to be substituted by domestic production, protected by high tariffs and quotas on imports. For a succinct description of these policies, see Krueger 1995, esp. Chapter 1. 6 60 50 40 1971 1976 1981 1986 1991 1996 2001 Year Figure 1.1: Growth of North-South-PTAs, 1971-2004. Source: WTO Secretariat, http:\/\/www.wto.org\/english\/tratop-e\/region -e\/regfac-e.htm; Dartmouth Trade Agreements Database, http:\/\/mba.tuck.dartmouth.edu\/cib\/research\/tradejagreements.html performance requirements, for example the sourcing of a percentage of inputs or mandatory export of part of the production, governments attempted to harness the benefits of foreign capital (Caves 1996; Greenaway 1992). To compensate multi-national firms for the high tariffs and host country requirements, governments struck deals that sheltered investors from competition and offered economic rents. (Evans 1979). While the provision of services remained in the hands of govern-ments, the high tariffs and restrictions made exports and investment by smaller firms from developed countries infeasible. FDI sought markets, but under the spe-cific conditions of the import-substitution policy of the host country. As Latin American countries began to liberalize in the late 1980s and early 1990s in the search for foreign capital, they became attractive for a different kind of invest-ment integrated with world markets. Despite important differences with Latin American countries, Asian \"devel-opmental states\" (Wade 1990) attracted similarly inward-oriented FDI. Multina-tional firms, in this case mostly from Japan, enjoyed exclusive market share ar-rangements for their products. Following the Asian Financial Crisis, liberalization based on the prescriptions of the IMF has reached this region as well. 7 As other sources of capital such as bank loans have dried up, countries in both regions have been forced to compete for investment (Graham 1996). In this competition, governments see direct investment as preferable to volatile portfo-lio capital flows. For developing countries, combined free trade and investment agreements offers an institutional package that locks in unilateral liberalization and provides guarantees for investors beyond WTO commitments (Fernandez and Portes 1998). In addition, even with only modest tariff reductions, a PTA with an industrialized partner gives developed countries an edge over competitors with similar factor endowments (Ethier 1998a, 1998b, 2001). Both benefits explain why developing countries queue up to sign bilateral agreements in highly unequal negotiations with developed countries. Less obvious, however, is why these developed countries should take up the offer. Most developing countries are negligible export markets. In terms of na-tional income, Mexico offered US firms barely six percent additional market size when N A F T A entered into force.9 Again, the real purpose of recent FTAs relates to FDI. Following the reintegration of many developing countries into the world econ-omy, they attract manufacturing FDI to serve as export platforms to (mostly) de-veloped country markets. In particular labour-intensive stages of production are relocated to developing countries. Because manufacturing FDI entails exports of machinery and inputs such as parts (capital and intermediate goods) to the FDI host, much of the trade liberalized by North-South FTAs is the result of FDI. Manufacturing FDI also creates a market for attendant services, for example insurance of exports or financing of direct investment. Moreover, since most de-veloping countries have only recently begun to open their general financial and telecommunications service markets, FDI in services represents a considerable share of the capital flows to emerging markets. Provisions for FDI in bilateral trade agreements therefore apply in large measure to these flows. Each bilateral agreement, however, necessarily entails regulatory and tariff discrimination for firms from third countries. Whether using the developing coun-try as export-platform, a market in its own right, or as a base for regional expan-sion, manufacturing firms from excluded parties face higher costs compared to their competitors from within a preferential trade agreement. For example, fol-lowing the conclusion of NAFTA, Japanese automobile firms in Mexico not only faced a 16 percent tariff on some capital goods and Inputs, but also a 62.5 percent 9In concrete numbers, a GDP of merely US$ 420 billion compared to the US GDP of US$6.9 trillion. 8 rule of origin, the percentage of an automobile that has to be sourced within North America. Service companies confront not only regulation that is structured for compatibility with PTA partners, but are sometimes entirely barred from entering significant markets. Increased flows of FDI, resulting trade in intermediate and capital goods, and investment in services change the parameters for trade policy. From this overview, it is evident that the recent wave of preferential trade agree-ments reflects a fundamental change in global trade relations. 1.2 Perspectives on Preferential Trade Agreements The scope of recent bilateral preferential trade agreements (PTAs), covering new issues beyond trade in goods, their character as partnership between countries of unequal level of development, and their often \"extra-regional\" geography set them apart from past regional arrangements. I argue that these particular features call for a modification of existing theories of regionalism. A more complete expla-nation requires solid theoretical foundations of firm-government interaction, in-corporating the domestic interests associated with economic transactions beyond trade in finished goods. Much of the literature on regionalism, especially in political science, has tended to adopt a state-centric focus. Recent arguments in International Relations in-creasingly point to the role of the WTO. Mansfield and Reinhardt (2003) posit that the growth of WTO membership, the recurrent negotiation rounds and in par-ticular the participation in trade disputes motivate states to seek PTAs as an insur-ance policy: Should WTO rounds fail or end up deadlocked, states secure export market access and increase their bargaining power based on market size. Like-wise, PTAs can serve as coalition-building strategy to increase bargaining power or to \"obtain countervailing market access\" (Mansfield and Reinhardt 2003, 830) in case of losses in WTO disputes. Yet, although PTAs improve market access, these gains are often circum-scribed: Both parties reserve the right to use the WTO as avenue to settle claims, even if the PTA contains elaborate dispute settlement mechanisms. PTAs with the US in particular do not offer protection from US trade remedy laws. Neither are PTA alliances of small countries emerging. Rather, these countries queue up to form bilateral agreements with the US, the E U and Japan in highly asymmetrical negotiations. Finally, this explanation offers little insight into the cross-regional variation in the spread of preferential trade agreements: PTAs have spread quickly 9 in the western hemisphere, but only recently arrived in Asia (Haggard 1997; Ravenhill 2003). But do PTAs at least offer an insurance against a stalled WTO, allowing like-minded countries to move faster? As Haggard (1997) has argued, the foremost reason for the impasse at the WTO table is the convergence or divergence of state interests. Bilateral agreements have not fared any better in resolving the intractable issue of agricultural liberalization. Most bilateral agreements avoid this contentious issue (Hoekman and Leidy 1993), perhaps because only multilat-eral rounds permit the necessary package deals (Davis 2003). The incentives to strike bilateral deals appear to lie elsewhere. Adopting a comparable state-centric perspective, the majority of economists has focused on the effects of PTAs on national and world aggregate welfare from a Pareto-optimality perspective. Most economists agree that multilateral liberal-ization is preferable to bilateral agreements. Just how preferable depends on the Pareto criterion: PTAs could still be desirable as long as they make some coun-tries better and none worse off. Welfare analysis, however, suggests avenues for inquiry, since many arrangements that decrease welfare concentrate gains in the hands of a few groups. The classic formulations by Viner (1950), Lipsey (1957) and Meade (1955) argued that PTAs can be welfare enhancing if more trade is created than diverted away from more efficient countries outside the arrangement. Kemp and Wan (1976) raised the prospect that PTAs can be constructed in a way that makes at least one member better off, but does not affect outsiders. Later contributions found that welfare effects tend to be ambiguous10. Only recently have analysts studied the actual motivations to form PTAs from a welfare perspective, albeit without consideration of effects on non-members (Baier and Bergstrand 2004). For the present analysis, two findings are important: First, FTAs can divert trade away from non-members to less efficient producers in member countries, and sec-ond, strict rules of origin can decrease welfare by concentrating gains among a few producers in the FTA (Duttagupta and Panagariya 2003; Krishna and Krueger 1995; Krueger 1993). With the second wave of regional trade agreements in the 1980s, the focus shifted from these \"static\" considerations to \"dynamic issues:\" Are PTAs \"stum-bling blocks\" or \"building blocks\" (Bhagwati 1991) towards global free trade? Krishna (1998) argues that trade diversion reduces the incentives for members of a bilateral agreement to reduce trade multilaterally. Levy (1997) even contends 10For overviews, see Panagariya (1999, 2000). 10 that bilateral PTAs can make multilateral liberalization politically unviable, while McLaren (2002) warns that PTAs can induce member countries to make relation-specific investments that inhibit future multilateral liberalization. Given these concerns, political economists have studied why states would con-clude PTAs at all. This in turn leads to the domestic sources of foreign economic policy. Government policy often reflects the interests of strong and well-organized societal groups (Grossman and Helpman 1994), although mediated by the domes-tic institutions that offer or restrict access to governments (Mansfield and Busch 1995; Nelson 1988). In Milner's framework, governments balance producer and consumer interests. Public officials therefore seek the support of industrial sectors, some of which will be in favour of liberalization. Firms support regional liberalization because it al-lows the mutual reduction of tariff barriers by \"trading scale economies across industries\" (Milner 1997, 91), in effect balancing costs and benefits between ex-porters. Along similar lines, Busch and Milner (1994, 270) put the growing im-portance of exports and intra-industry trade at the centre of their explanation of firm preferences. Producers that can achieve economies of scale in home mar-kets are more competitive globally, leading them to demand regional trade agree-ments. This demand wil l be most pronounced when firms are competitive in terms of technology and management, but lack the sizable home market to achieve the optimum efficiency. Milner cites the example of Canadian firms that supported N A F T A . 1 1 Most developing country markets, however, are too small to move firms down the cost curve towards greater efficiency. On the other hand, markets that are big enough, like China or India, are still well protected and induce tariff-jumping investment. While theoretically compelling, Milner's model offers little insight into North-South PTAs between highly unequal partners. In a series of papers, Chase (2003, 2004a, 2004b) addresses this problem, arguing that multinational corporations lobby for regional agreements with devel-oping countries because of the growth in offshore processing. Many firms, how-ever, face a legacy of sunk investments originally made under protectionist host country policies that need to be restructured. They therefore press for barriers to non-members as a defensive policy. Chase hypothesizes that ROOs are designed to prevent transshipment of goods into the US via Mexico, but ultimately finds lit-1 1 But see Thompson (1994, 23), who argues that Canadian firms were particularly opposed to further liberalization. Milner's case study ignores that Canada and the US had signed an FTA before NAFTA was proposed. Consequently, Canada's initial interest in NAFTA was limited to the protection of the gains made in the CUSFTA (Cameron and Tomlin 2000, 63-64). 11 tie support for this argument: Not only auto firms with FDI in Mexico, but also the highly protectionist textile sector obtained high rules of origin levels in NAFTA. M y explanation builds on and generalizes Chase's account: Rather than offer-ing temporary relief, rules of origin are a strategic policy instrument that firms try to manipulate to their advantage. Such barriers to entry are particularly attractive when the host country still retains high M F N tariffs, because they interact with ROOs to raise the costs for non-members firms. If member firms mainly source inputs from within the FTA, then ROOs are a costless device for insiders to extend protectionism to the host country and to gain the political support of intermediate goods producers. As outlined in the following chapter, ROOs therefore have the effect of diverting trade to producers in the PTA while inducing producers from non-members to relocate production into the PTA. If multinational firms are the main proponents of such PTAs, we need bet-ter answers for why these firms support bilateral agreements. By incorporating foreign direct investment in the explanation, my account therefore expands on po-litical economy models that predict that preferential agreements are most likely to be formed when they divert trade, because the gains to exporters then outweigh the costs to import-competing industries (Grossman and Helpman 1995; Panagariya and Findlay 1996). Contrary to the rhetoric of free trade, this trade diversion ex-plains much of the political attractiveness of preferential agreements, as predicted in a prescient essay by Hirschman (1981, 271): The larger the trade creating effects, that is, the greater the need to reallocate resources in the wake of tariff abolition, the greater wil l be the resistance to the union among various highly concentrated and vocal producers interests of the member countries. (...) Thus trade creation is a political liability. Trade diversion implies, on the con-trary, that concentrated producer groups of the member countries will be able to capture business away from their present competitors in non-member countries. The unintended consequence of the (realized or potential) trade diversion is the reaction by other countries. Excluded parties can counter PTAs in two possi-ble ways: They can either form PTAs among themselves, or they can attempt to join an existing agreement to prevent trade diversion. Baldwin (1996) calls this the \"domino effect of regional trade agreements.\" My framework explicitly recognizes the possibility of domino effects\u2014but since bilateral agreements are motivated by FDI, the reaction will be an agreement with the host country. 12 1.3 Firms, States, and Competitive Liberalization: An Overview of the Argument If free trade agreements represent a deliberate policy choice at the behest of firms, then they can be conceived of as a \"non-market strategy\" as defined by David Baron (1997, 1999). Firms use non-market, political strategies to compete by means of cost and market access rather than price or product characteristics. For example, a firm may use a political strategy to further its competitive position by pressing its government to pursue a certain action on its behalf. In contrast to the generic concept of lobbying, non-market strategies are conceived as proactive measures taken to open up or preserve commercial opportunities. Despite limited overall welfare effects, a country may therefore seek a free trade agreement that provides concrete economic benefits to a few firms: when the small size of the partner economies threatens few groups at home, the po-litical cost for the government is low, and protectionist measures can be used to assuage concerns. Reacting to firm demands, the pursuit of free trade agreements forms a deliberate strategy by the \"competition state\" (Cerny 1990) to further the economic position of its companies. Free trade agreements offer a variety of advantages to firms, depending on the sector in which investment takes place: In some economic sectors, the first to move to a PTA with an emerging market enjoys advantages. In others, they allow firms to lower the cost of an internationally fragmented production while raising it for competitors from non-member countries. These benefits are less evident at an aggregate economic level, but clearly visible at the level of individual firms. In services, the structure of many markets confers important first-mover ad-vantages. Many service markets are textbook examples of oligopolies or even natural monopolies. Efficient production requires a dominant market share. Con-versely, this means that first-movers with enough capital can buy up existing as-sets, such as branches in retail banking or telecommunication networks, and attain a commanding position. Preferential liberalization therefore threatens to shut out competitors. Because of the first-mover advantage, this applies even if host coun-try laws on FDI are subsequently made non-discriminatory, that is if liberalization is multilateralized later. In addition, first-mover firms may attempt to structure important regulations in a way that raises rivals' cost of market entry. As emerging market countries move towards the liberalization of services, the overhaul and, in some cases, creation of regulatory regimes becomes necessary, e.g. in the provision of financial services 13 and the protection of investor and intellectual property rights. With the primary exporters of these services located in developed countries, this process creates an interest to influence the regulation to adapt models used in the respective home country, thus lowering the barrier to entry for their providers and raising them for others (Wunsch-Vincent 2003). In manufacturing, liberalization creates the opportunity to use the advantages of specific locations in the production of goods. Capital-intensive production takes place in developed countries that offer access to high technology and research and design facilities and personnel. Labour-intensive stages of production are out-sourced to the developing world. This vertical fragmentation of production leads to an increase in intra-industry trade, or trade in the same industry in differen-tiated goods. Unlike intra-industry trade between developed countries, however, the traded goods are differentiated by \"quality:\" developing countries are more likely to export low-cost goods, while developed countries export high-cost, high quality goods. For example, while many car manufacturers produce their upscale vehicles in their home country, entry-level cars are produced in less developed countries. German car manufacturer Volkswagen assembles most of its high-end models in Germany, but produces its entry-level models Polo and Lupo in the Czech Republic and Brazil. While transport costs are less and less important, not every country is equally attractive as a location of production. Proximity to devel-oped country markets as well as other factors like available primary materials turn access to these locations into competitive advantages. Moreover, such investment implies an enormous increase of trade of intermediate goods: parts production of-ten takes place in specialized factories to achieve efficient production scales. The same parts are then shipped to different locations. Fragmented production thus creates a demand for tariff reduction in home and host countries. While unilateral or multilateral tariff liberalization could serve this purpose, it exposes the home market to foreign competition: Nothing would prevent firms from third parties to invest in the developing country and use it as an export base. Consequently, when two countries liberalize bilateral trade, firms try to raise the barriers for outsiders. The principal benefit of PTAs is that by means of rules of origin (ROOs), bilateral agreements can be structured to increase the cost of production for firms from non-member countries. Despite earlier liberalization, developing country tariffs are still higher than the M F N tariffs of developed countries. Sufficiently strict rules of origin in an FTA interact with the remaining tariff to extend protection: firms from within the PTA can produce in the host country, import intermediate goods from and re-export goods to the home market without costly tariffs. By contrast, firms from 14 non-member countries can only import parts from home at a high price\u2014if they exceed the rule of origin, they not only pay the initial tariff on the intermediate good, but also the M F N tariff upon export to the developed country! Theoretically, firms would be induced to switch to suppliers from within the FTA, the classic case of trade diversion. In practice, firms often cannot easily change suppliers with whom they have long-standing relationships, at times in-volving the contracting out of research and development. Principal suppliers within the FTA may have close ties to the competition. How do non-member firms react? Since they cannot influence the ROOs in the first FTA, non-member firms have to tackle the second part of the protectionist policy: the M F N tariff of the host country. The obvious reaction is to lobby the host government for unilateral re-ductions. But if the host government is seeking FTAs to attract investment, it now has a bargaining chip: the preferential margin offered to its first partner country. Non-member firms can offer little in return, unless they can convince their home governments to strike a deal with the host country. To retain their competitiveness, non-member firms therefore seek recourse with their home governments. The de-fensive reaction of these non-member firms is an unintended consequence of the first FTA. Since intermediate and capital goods producers in non-member countries are hurt by the same ROOs and tariffs, they become allies in this undertaking. Like-wise, excluded service firms that witness preferential liberalization will try to level the playing field. Even though firms from non-member countries may not have been interested in an FTA with the host country originally, they wil l now come to support it to prevent exclusion from an important host of FDI. In response, their home governments also negotiate FTAs with the host country\u2014the dominos fall, causing an endogenous proliferation of bilateral trade agreements driven by FDI to developing countries. In sum, two factors emerge as the driving forces behind today's bilateralism: concentrated interests in FDI-exporting countries have a strong incentive to lobby for bilateral agreements because they confer specific advantages over competi-tors. To be politically attractive, bilateral agreements must have a discriminatory effect on investment from non-members. This effect, however, manifests itself mainly for investors from other countries that are disadvantaged in their ability to use developing countries as export platforms, or that are excluded from services markets. These investors push their home governments for defensive agreements to remain competitive. 15 1.4 Methodology The aim of this study is to explain the emergence and proliferation of a phe-nomenon: bilateral agreements between developed and developing countries. If the hypothesis of an endogenous proliferation of FTAs holds, it presents a problem for the standard comparative method in political inquiry: endogeneity violates the crucial assumption of independent cases. As a remedy, I disaggregate the cases to the level of firms, engage in detailed process tracing, and test key causal links quantitatively when possible, using standard econometric procedures with a larger dataset of bilateral FDI flows. The following section discusses the individual cases in greater detail. In this combination of methods, the relative strengths of some counterbalance the weaknesses of others. I begin by developing an explicit model that establishes causal relations, but abstracts from concrete manifestations to ensure measure-ment validity (Adcock and Collier 2001). \"Model\" in this sense stands for a parsimonious simplification of reality that depicts systematic, regular causal re-lationships (Elster 1989, 3-10). M y cases are dyads of countries with different factor endowments in which foreign direct investment flows predominantly in one direction. The independent variables are the political demands made by sectors in the capital-exporting countries, while the actual institutional outcomes (pref-erential trade agreements) form the dependent variables. However, the dynamic nature of the model predicts that independent variables are partly endogenous: the conclusion of one PTA triggers several others. How can a model like this be tested without resorting to circular reasoning? As argued by Biithe (2002, 485) in a recent essay, modeling sequences of events offers a solution: Sequence \"(...) allows us to have causal feedback loops from the explanandum at one point in time to the explanatory variables at a later point in time only.\" Accordingly, I develop individual analytic narratives for each case study. Here, I emphasize process-tracing (George and McKeown 1985, 40-41; King, Keohane, and Verba 1994, 226) to identify the constraints and variables that matter to the actors in their decision-making. In developing these accounts, the model provides the cast of actors, their interests and strategies (McKeown 1999). Accordingly, I analyze separate cases to attenuate the problem of endogene-ity. First, holding host country tariffs and rules of origin constant, I compare two reactions to an initial agreement with a feedback from one case to the other: the EU-Mexico and FTA as a reaction to NAFTA, and the Japan-Mexico FTA as countermove to both prior agreements. Second, I include cases in which the com-16 petitive dynamic is restricted to only one sector: services only in the case of the FTAs of the US and Chile, including the \"failed case\" Japan-Chile, and manufac-turing only in the case of the initiative for a Japan-Thailand FTA. Although these cases differ in important respects, the violation of the assumption of comparable cases (Lijphart 1975, 688), variously referred to as \"most similar systems design\" (Przeworski and Teune 1970, 32-34) or \"unit homogeneity\" (King, Keohane, and Verba 1994, 91-94) is unproblematic: since the focus is on a convergence of poli-cies, any differences in the cases that causes variation loads the dice against my hypothesis. The case studies are based on qualitative data collected through the analy-sis of interest group publications, documented lobbying activity such as congres-sional hearings, and over 50 interviews with decision-makers in government and industry. Given different political systems, such data cannot be truly \"symmet-ric\" (King, Keohane, and Verba 1994, 48) in the sense that a uniform method of collecting information is used in all cases: lobbying in Japan leaves no visible paper trail, while evidence of comparable activities in the US is often publicly available. However, this problem is less acute in process-tracing than in small-N, comparative research. 1.4.1 N A F T A and its Effects N A F T A has frequently been analyzed as a case of economic regionalism (Cameron 1997; Cameron and Tomlin 2000; Mayer 1998; Pastor and Wise 1994; Robert 2000; Ros 1994; Tornell and Esquivel 1995), but few analysts have put it in the perspective of a general move towards preferential trade agreements (Haggard 1997). The analysis of the lobbying process has been limited to the manufactur-ing sector (Chase 2003). Competing theories explain N A F T A on the basis of other factors, such as the agreement's role as commitment device for the Mexican gov-ernment, reassuring firms in the US and Canada of the trajectory of liberal reforms under future government following the Salinas presidency (Fernandez de Castro 2000; Tornell and Esquivel 1995). M y study contends that we can understand the interest of the US in signing N A F T A only in the context of US trade policy since the 1980s, which was more and more driven by the interests around direct investment and emerging sectors such as services (Haggard 1998, 320), as well as US manufacturers' search for low-cost production platforms. These interests were behind the 1987 \"Framework\" and the 1989 \"Understanding\" on trade and investment relations between the US and Mexico (Robert 2000, 27). 17 Providing evidence for my proposition that these new market sectors specif-ically lobby for bilateral agreements, I argue that these interests are decisive in motivating bilateral agreements within the multilateral context. I then show how N A F T A discriminated against non-member investors and exporters, leading to the conclusion of defensive agreements by the E U and Japan. In the realm of services, N A F T A went beyond Mexico's commitments under the GATS. In particular the case of the EU-Mexico Association Agreement and the time lag to the conclusion of the Japan-Mexico FTA strengthens the argument: given the stronger investment interests, the E U moved faster to counter the effects of NAFTA. The E U FTA in turn spurred Japan on to pursue its own FTA with Mexico. 1.4.2 Survey Cases: Testing the Limits of Applicabil ity Following King, Keohane and Verba (1994, 146), I include survey cases with comparable dynamics to strengthen the argument presented here. In particular, the plans for negotiating free trade agreements with Southeast Asian and other Latin American countries follow the causal mechanism described above. I focus on the case of the Japanese FTA initiative with Thailand, and the agreements of the US and the E U with Chile. The cases of Chile's agreements with the US and the E U demonstrate that even access to comparatively small markets drives the competitive dynamic in the proliferation of PTAs. While the traditional sectors of mining and agriculture still attract considerable investment, Chile began to liberalize its financial, energy and telecommunications sectors in the 1990s, leading to a flurry of foreign investment by US, Spanish and Canadian companies. Concluding a PTA with Chile presents the opportunity to open the remaining sectors to FDI, but also to serve regional markets. Given that the E U firms moved faster when investing in Chile than their US counterparts, US firms worried about a loss of competitiveness. As a result, the US-Chile agreement contains specific clauses to improve access for US investors to make up for perceived advantages of E U firms. The Japanese FTA initiative with Thailand shows important parallels to N A F T A . Just like Mexico, Thailand has become an important host of manufacturing FDI. In particular in the wake of the 1997 financial crisis, the country emerged as an important export platform. As the case study shows, Japanese automotive firms re-structured their operations in Thailand for exports to developed countries, includ-ing the Japanese home market. But the benefits of FDI in Thailand also attracted competitors from the US, Europe and Korea, prompting Japanese multinational firms to strongly support an initiative for an FTA. 18 1.5 Caveats, Limitations, and Contributions As with most political agreements, it is impossible to identify a single variable as decisive for the conclusion of an individual PTA. This applies to both the theo-retical and empirical dimensions of this study. In the case of NAFTA, concerns about the stability of the Southern neighbour and the resulting flows of immi-grants into the US played a role in motivating the free trade agreement. More importantly, N A F T A was a complex undertaking covering virtually every eco-nomic sector. Consequently, the central contribution of this work in this regard is to demonstrate the central causal relevance of political interests in the conclusion of N A F T A that have so far been overlooked. On a theoretical level, the explanation put forth in this thesis does not endeav-our to capture all of the forces at work in the current proliferation of preferential trade agreements. PTAs can also be used to express political support when strate-gic interests are at stake, as in the case of the US-Jordan FTA and the follow-up initiative to establish a regional free trade zone in the Middle East. Likewise, in cases of PTAs between countries of highly similar levels of development, other considerations such as achieving economies of scale will be of central importance (Haggard 1997, 79), while investment tends to flow in both directions. As a final limitation, this study does not try to explain the policy choices of the develop-ing country partner. This has been undertaken for the individual cases of Mexico and Chile as well as some Southeast Asian countries (Cameron 1997; Grugel and Hout 1999; Haggard 1995; Pastor and Wise 1994), although a general theoretical appraisal remains to be written. The study offers three specific contributions. First, it breaks new ground by showing the importance of the political interests that emerge around foreign direct investment and how these need to be considered when analyzing current global trade relations. The endogenous nature of PTAs has been hinted at, but not been explicitly theorized. Second, by explaining the interests behind a growing set of bilateral trade agreements with the same variables, it contributes to the theoretical analysis of regionalism in the world economy and sheds light on a set of trade policy outcomes unexplained by current theories. Finally, it connects the literature on firm strategy and preferences in emerging markets to the study of free trade agreements to provide a theoretical foundation for future research in this area. In addition to the cases analyzed here, the account can be applied to a growing number of cases: future Japanese PTAs with Southeast Asian and South Amer-ican countries, the anticipated EU-Mercosur agreement, as well as the various bilateral initiatives of the US, which include agreements with Morocco, South 19 Africa, and the A S E A N countries. Moreover, with some qualifications it can be usefully extended to cover initiatives between developing countries where foreign direct investment flows predominantly in one direction, such as Korea-Mexico and Korea-Chile. 1.6 Precis of the Dissertation The study is organized as follows. Chapter 2 establishes the theoretical argument for when and why preferential trade agreements between developed and devel-oping countries become viable, and why they trigger a round of countermoves towards bilateral agreements. The explanation is developed based on a simplified model of two countries exporting FDI and one seeking to attract it, building on theories of vertical integration of production across country borders. Chapter 3 focuses on N A F T A as the first case study. Under pressure from multinational manufacturing and service industries, the US begins to target in-dividual states for market opening, while promoting the same issues in the WTO from the mid-1980s on. Bilateral investment treaties and service negotiations form the templates that ultimately lead to the respective chapters in N A F T A , which in turn becomes the model for future FTAs. The section takes stock of the lobbying efforts of various industries before and during the negotiation, relating them to the selectively protectionist outcome and tracing the emergence of NAFTA's strict rules of origin to the demands of firms. Chapter 4 uses a standard gravity model to assess the investment diversion caused by NAFTA. Following the entry of N A F T A into force, Mexico attracted more FDI from non-members than average sample behaviour would predict, sug-gesting that excluded firms respond by establishing beachheads of FDI in NAFTA. In particular the pronounced increase of European investment raises the stakes for the firms involved in securing tariff-free access to Mexico. Chapter 5 shows that following the conclusion of NAFTA, both European and Japanese investors and service providers begin to lobby their governments to seek solutions against the discriminatory arrangement\u2014but to different degrees. Given the strong interest of European service providers, in particular the Spanish fi-nancial sector, the E U moves quickly to the conclusion of a framework agree-ment. Lobbying takes place around the Directorate General for Trade at the Euro-pean Commission and the Committee 133 (coordinating member states' interests and Community trade policy), in which Spanish interests clash with protection-ist forces from France. In the manufacturing sector, German automotive firms, 20 in particular Volkswagen, emerge as key supporters of a bilateral agreement with Mexico. By contrast, Japanese investment is concentrated in manufacturing of electron-ics and automobiles. The discriminatory effect of N A F T A is limited to a smaller number of companies, many of them located in Mexican export-processing zones. Japanese electronics and automotive firms first unsuccessfully exhaust all possi-bilities of liberalization between Mexico and Japan. The eventual conclusion of the EU-Mexico agreement convinces policymakers in the trade and foreign affairs bureaucracies to move towards a policy of free trade agreements. Chapter 6 surveys the US-Chile and EU-Chile agreements and the initiatives for a Japan-Chile and Japan-Thailand FTA. Although investment in Chile is mostly limited to the service sector, the competitive dynamic is sufficient to motivate the US and the E U to seek FTAs to secure equal access. Due to the lack of fast-track authority, the US is unable to conclude an agreement with Chile before the E U . As a result of the E U move, intense lobbying in the US drives the rapid conclusion of a defensive agreement. The Japanese initiative for an FTA with Thailand tests the applicability of the theory with only manufacturing FDI, demonstrating that once a country becomes viable as an export platform, important FDI home countries will quickly seek a NAFTA-style bilateral deal. Chapter 7 summarizes and concludes that bilateralism is a product of the greater importance of foreign direct investment. The current proliferation of pref-erential trade arrangement will not the emergence of protectionist blocs, but it may weaken support for multilateral negotiations by important industries that favour free trade, but whose demands are met in bilateral institutions. At the same time, it creates a patchwork of competing rules that mainly benefit the concentrated interests of a few multinational firms. 21 Chapter 2 Framework for Analysis Governments liberalize international trade for two economic reasons: they re-spond to the demands of export-oriented industries (Milner 1988), and hold causal beliefs that free trade increases the welfare of societies (Simmons and Elkins 2004). Yet these factors cannot explain why states would prefer bilateral over multilateral or unilateral liberalization. To account for the recent proliferation of preferential trade agreements, we need to consider which factors create incentives for bilateral agreements, and what determines the choice of partners. This work focuses on the economic interests of developed countries in concluding FTAs with emerging markets. Despite the label free trade, I argue, they are protectionist in disguise, and benefit producer interests in member firms at the expense of non-members. These protectionist features trigger the reaction by affected firms in non-members. Specifically, I make three claims regarding governments and multinational firms from developed countries and their interests in FTAs. FTAs erect new barriers. When a developed country is negotiating an FTA with a developing country, firms with FDI in this country and their suppliers will lobby for barriers that raise costs for competitors from non-member countries. Foreign direct investment rather than exports of finished goods assumes cen-tral place in North-South bilateral trade agreements. When developing countries become export platforms, they can be used by multinational firms to produce goods for exports to the home country. But since multilateral or unilateral lib-eralization would allow competitors from other countries to do the same, it would expose the home market to competition. Multinational firms therefore have an incentive to try to shut out competitors by means of an FTA. Interests linked to FDI, defined more precisely in section 2.1.1, describe firms that either invest themselves to benefit from the advantages of the location of the emerging market country, or that supply firms with such investments. In this sense, a strong interest would simply mean having made considerable investments in plants abroad, or selling substantial parts of a firms' production to such plants. 22 For example, General Motors has invested in Brazil, so exporters of machinery for the production of cars and parts suppliers such as Delphi have an interest in sales that derive from the initial investments made by the automotive company. If these firms succeed in lobbying for policies that impose costs on competitors from non-member countries, then the latter will try to counter those policies. FTAs trigger defensive agreements. If the developing country concludes its first FTA with a developed country partners, then firms from non-member countries with substantial FDI in the developing country and their suppliers will likewise call for an FTA. While firms that are already present in the host country wil l be the most vo-cal, newcomers that plan to invest will likely support such an FTA as well. The threshold at which firms from non-member countries become active is an empir-ical matter, but will be lower i f the barriers created by the FTA are particularly costly and a presence in the host country important to preserve commercial op-portunities. Service investment exacerbates the competitive dynamic, although the discriminatory effects of FTAs are the results of market structures rather than treaty clauses. Service firms have less of an incentive to raise barriers for com-petitors, because here the competition is over markets in the host country. But once the endogenous, competitive trend of bilateral FTAs comes into motion, it is hard to stop\u2014as competitors level the playing field by getting their own bilat-eral deals, it becomes more costly to stay at the sidelines. Consequently, at later stages, even much weaker investments and trade interests wil l suffice to trigger a reaction. Countries can no longer afford to rely on the multilateral system alone. FTAs become endogenous. Once countries begin to sign FTAs, the competitive dynamic will accelerate, forcing other countries to likewise sign FTAs. In other words, once the first FTAs with important host countries are signed, strategic interaction between the home countries of multinational firms becomes the driving force of bilateral deals. Again, the role of foreign direct investment is the central to this claim. Preferential trade agreements do not affect arms-length, traditional trade in finished goods enough to trigger a competitive dynamic. Only FDI and the resulting trade in intermediate goods creates a broad coalition of pro-ducers, and only FDI and intermediate goods trade is susceptible to the discrim-inatory effects of bilateral trade agreements. Accordingly, the model presented here is limited to country pairs in which FDI flows primarily from one to the other, driven by different factor endowments as explained in section 2.1.2. 23 As with any theoretical explanation, the account excludes part of the story. First, states conclude PTAs for a variety of reasons, like supporting allies (Gowa and Mansfield 1993; Gowa 1995) or sending signals that they are committed to an open trade policy (Fernandez and Portes 1998). However, any PTA of economic relevance will have distributive effects. Accordingly, even if governments decide to pursue a PTA for non-economic reasons, they cannot do so without support from important social groups. Governments will therefore be receptive to the economic interests of these groups. Second, developing countries have different incentives to pursue bilateral agree-ments than developed countries. In line with a growing body of literature (Ethier 1998a, 1998b, 2001; Grugel and Hout 1999; Oman 1994), I assume that their goal is to attract foreign direct investment. It is less clear why developed coun-tries would have economic reasons to seek bilateral deals with small and relatively poor markets. In the following five sections, this chapter presents a framework for the anal-ysis of these factors. The task is twofold: to clarify the assumptions of the model put forth here, and to derive and operationalize testable implications. The first section discusses basic assumptions, moving from the general underpinnings of the approach to specific reasons for firms to undertake FDI in emerging market countries. Section two outlines why firms prefer bilateral agreements, then de-velops propositions on firm preferences and lobbying. Section three ties these elements together to explain why FTAs trigger their own endogenous prolifera-tion. Sections four and five discuss possible theoretical objections and normative implications. 2.1 A Political Economy Approach M y approach follows Frieden's (1991, 15) definition of \"modern political econ-omy.\" Political economy in this sense assumes rational, goal-oriented, and utility-driven actors. Rationality, however, does not imply any normative judgement of the social outcome. In fact, much of this work is concerned with the unintended consequences of actions that are individually rational, but lead to a collectively suboptimal outcome. The approach rests on the specification of preferences of actors over govern-ment policies. Preferences can be derived from economic characteristics of actors, in particular the specificity of their assets. Asset specificity refers to the ease with which an asset can be transferred to a different use. Indirectly, asset specificity 24 determines the intensity of preferences: For example, if an economic asset such as machinery can only be put to a certain use, then actors have a strong interest to protect their ability to do so. Often, this calls for political action. If government policies can change the return on actors' assets, then actors are likely to lobby for policies that increase or protect the value of their assets. Actors operating in markets where policies directly influence the return on assets, for example by raising the barriers to entry of competitors, will be prepared to make greater efforts at lobbying. Conversely, the less their returns can be affected by a policy, the less incentive actors have to lobby for or against it. Assets also influence how economic actors organize themselves vis-a-vis the government. For example, firms with comparable assets will have similar prefer-ences over policy, even though they may be direct competitors in the marketplace.1 If actors choose to lobby to influence a government policy, the similarity in pref-erences may motivate them to form a coalition, if their common interest overrides the competitive element of the marketplace. For the purposes of this study, both firms and other social groups are assumed to be cohesive social actors, ignoring problems of organization such collective action as defined by Olson (1971). Based on these assumptions, I analyze the role of interest groups, in partic-ular firms, in demanding changes in government policies. The pursuit of these preferences takes place in political institutions in which decisions over foreign economic policy are made. Institutional differences could therefore lead to differ-ences in policy outcomes across cases. While this study takes into account the role of institutions in the political process, it does not address their effect on outcomes. As my focus is on policy convergence\u2014the rapid increase in FTAs\u2014rather than variation, I do not assume the existence of a force that affects polities equally, but leads to varying outcomes because of different unit characteristics. Institu-tional differences, however, cannot cause a convergence of policies independent of interests. Accordingly, the alternative explanations I engage either put the same emphasis on interest groups, or stress the relevance of systemic factors, but not of domestic institutions. How do government decision-makers formulate trade policy, given domestic actors with defined preferences? I follow classic public choice works in assuming that politicians, just like other actors, are utility-maximizers (Downs 1957) whose goal is re-election. When using trade policy to increase their election chances, they have three options (Grossman and Helpman 1994): they can increase con-1 Although rarely acknowledged by political economists today, this assumption was first made by Marx: the relation to the means of production determines the political interests. 25 sumer welfare, which would imply lower prices, raise firm profits, or improve the fiscal position of the state through greater tariff revenues. In an age of low M F N tariffs in the developed world, tariff revenues are negligible. Benefits in con-sumer welfare are likely to be small and highly dispersed. Under this assumption, politicians are susceptible to political pressure from concentrated groups. Con-sequently, politicians will prefer trade policies that raise firm profits, provided they do not negatively affect consumer welfare to the point where it changes the decisions of voters. An appealing aspect of these assumptions is that they help explain the choice of protectionist trade policies even though these are clearly not in the common interest (Alt et al. 1996). For example, Aggarwal and collaborators (1987) show how governments and producers collude in the negotiation of voluntary export restraints, a policy tool that benefits exporters in two countries at the expense of consumers. But even if consumers as voters understand the implications of such policies, their effect may not be ranked as important enough to change their vote choice. Accordingly, if protectionists are often successful, domestic support for free trade becomes a puzzle for analysis (Milner 1988). Like politicians, bureaucrats enter the picture primarily by responding to the demands of interest groups. A broad literature has addressed the role of bureau-crats in formulating economic policy, in particular in the context of the develop-mental state (Aoki, Kim, and Okuno-Fujiwara 1996; Hall 1986; Johnson 1982; Wade 1990; Woo-Cumings 1999). Since in many countries, bureaucrats are shel-tered from direct political pressures, they are seen as an independent political force. Yet a long tradition of public choice theory argues that bureaucrats are just like other rational actors, albeit with longer time horizons and different rewards (Downs 1967). Moreover, even within the developmental state tradition, some authors acknowledge that firms can counteract bureaucratic initiatives because of asymmetries of information (Okimoto 1989). Bureaucrats can expect little re-ward for policies that maximize national welfare, but are unresponsive or even costly for domestic interest groups. If, on the other hand, officials pursue policies that benefit firms, they can gain rewards, often in the form of lucrative positions after leaving office. Washington is notorious for its former officials turned lob-byists, but other capitals, in particular Brussels, have their share of \"consultants\" recruited from retired officials (Hayes 1993). In Japan, bureaucrats often move into important positions in private companies, following a brief period of retire-ment (Curtis 1999, 233-234). Finally, trade officials are constrained in how much time they can spend studying the potential effects of their policies (if these in fact could be reliably predicted). In consequence, they often resort to specific input 26 from industry in trade negotiations. This study therefore assumes that bureaucrats primarily act on behalf of their constituents, which in the case of trade policy moves firm preferences back into the centre of attention. 2.1.1 Assumptions about Firms Firms engage in the political process based on their preferences, which in turn can be derived from their characteristic assets and their position in the market and vis-a-vis political institutions. Often, firms compete with other interest groups over access to political decision-makers. My view of firms in the political process is based on Baron's (1999) concept of market and nonmarket strategies. Market strategies deal with pricing, quality enhancement and product development. By contrast, nonmarket strategies comprise lobbying and coalition building to influ-ence the regulatory framework. Both strategies can provide an advantage over competitors. Nonmarket strategies, however, often benefit all firms with similar assets i f regulatory frameworks are non-discriminatory, but not competitors from other countries who have more limited access to the domestic political process. To pursue such strategies, firms can form horizontal coalitions with companies with other firms in the same industry, or organize vertically along rent chains. Rent chains are composed of those who earn rents because of their interaction with a firm, for example suppliers and employees (Baron 1995). Baron (1999, 29-31) cites the example of the lobbying around the corporate average fuel economy bill (CAFE) in the US. Since strict C A F E standards would hurt US manufacturer producing big and inefficient vehicles, the automotive industry not only lobbied individually, but also mustered the support of suppliers, dealers, and even the National Cattlemen's Association\u2014a broad horizontal and vertical coalition. Possible coalitions depend on the nature of rent chains. In this study, I de-rive the goals that firms pursue with nonmarket strategies from their links with foreign direct investment in emerging markets. These links depend on the motiva-tions for FDI. Why would multinational firms invest in emerging markets, rather than export from their home countries? As reflected in the \"eclectic theory\" of the multinational enterprise (Dunning 2002), firms can invest overseas for several rea-sons. The eclectic theory divides the determining factors of FDI into ownership, location and internalization advantages. Different combinations of these advan-tages attract different types of FDI. The following refers to home country as the developed country in which the multinational firm is based, and host country as the developing country in which the firm invests. 27 Definitions. Resource-based international production seeks the ownership of specific resources that are generally bound to a certain location. Examples are ex-tractive industries such as mining and petroleum, whose location is determined by physical geography. Since trade in resources is rarely taxed, these industries are only occasionally the focus of trade policy-related lobbying. By contrast, import-substituting FDI reflects the location advantage of access to protected markets. Particularly during the 1950s to 1980s, developing country policies sought the substitution of imports with domestic products (Krueger 1995, 1-14), erecting high tariff and non-tariff barriers to protect infant industries. Multinational firms responded with \"tariff-jumping investment\" to serve protected markets (Evans 1979). While many developing countries still show the vestiges of import substitu-tion industrialization, most have opened their manufacturing and service markets to foreign investment in the 1980s and 1990s. In some of these sectors and industries, FDI is a precondition for market ac-cess. In particular in services, a commercial presence (\"mode 4\" in the GATS) 2 is necessary for business activities that require proximity to customers. Once coun-tries open their service markets for the first time, they attract distribution FDI of this kind, as witnessed by flows of direct investment to emerging market countries in the last two decades. Finally, multinational firms undertake vertically inte-grated FDI to specialize their manufacturing activities according to the benefits of the location, in emerging market countries primarily low labour cost. To inter-nalize this benefit, a division of labour across borders, but within the multinational firm, becomes a second important driving force of FDI in emerging markets. M y framework focuses on the last two kinds of foreign investment. 2.1.2 E m e r g i n g M a r k e t s a n d the E c o n o m i c s of Vert ica l Integrat ion Recent FDI in emerging markets had two main objectives: benefiting from differ-ent factor endowments and opening of new markets. To simplify, I assume that multinational firms primarily compete with firms from developed countries, but have a competitive advantage over developing country firms. In the manufacturing sector, multinational firms invest because of different factor endowments. Since labour is relatively abundant in developing countries and capital relatively scarce, 2The General Agreement on Trade in Services (GATS) divides the supply of services into mode 1-cross-border trade, mode 2- consumption abroad, mode 3-commercial presence, and mode 4\u2014 movement of natural persons. 28 it is more efficient to produce labour-intensive goods in developing countries, all else equal. As a result, developed countries export physical and human capital in-tensive, \"high quality\" goods, while developing economies produce \"low quality\" exports with greater input of low-skilled labour (Falvey 1981). The result is trade within industries, but in vertically differentiated goods, or vertical intra-industry trade (Greenaway, Hine, and Milner 1994). Firms engage in vertically integrated FDI when there is a gap in factor prices between host and home economy, when the market of the host economy is relatively small, and when transport costs be-tween home and host are low (Carr, Markusen, and Maskus 2001; Markusen 1995; Markusen et al. 1996).3 Markets for high-quality and luxury goods in the devel-oping country wil l be much smaller than for basic goods, so high-quality finished goods are likely going to be exported from the developed country in small quanti-ties. Accordingly, the developing country serves mainly as an export platform to developed country markets (Ekholm, Forslid, and Markusen 2003). Production of both high- and low-quality goods in different locations, how-ever, requires the shipping of goods across borders. Capital goods are mainly pro-duced in developed countries and have to be exported to the production site. Inter-mediate goods like parts, components and other inputs have an additional charac-teristic: their efficient production often requires achieving minimum economies of scale at the plant level. Automotive engine plants, for example, achieve efficiency at an annual production of at least 250,000, while the production of parts some-times requires figures of several million per year (Husan 1997; McAlinden 1997). Enjoying the benefits of different production sites while achieving economies of scale implies a specialization of plants and an increase in trade of intermediate goods, much of it intra-firm or with suppliers (Feenstra and Hanson 1996; Head and Ries 2001; Head, Ries, and Spencer 2004; Slaughter 2000). Accordingly, to make vertical FDI viable, three other conditions have to be fulfilled: neither the cost of transportion, the fixed cost of FDI, nor the transaction cost of moving goods and capital across borders should outweigh the benefits of location. Transaction costs are greatly increased by tariff and non-tariff barriers, making them potential aims for lobbying. Firms wil l try to reduce tariff and non-tariff barriers on intermediate, capital and finished goods as much as possible. This applies both to the multinational firm that produces the final good as well as to its suppliers of intermediate and capital goods, henceforth referred to as upstream producers. implicitly, these authors also make a political assumption: The differences in factor prices are not equalized because labour migration to the developed country is restricted. 29 When creating a framework suitable for vertical manufacturing FDI, devel-oping countries face a legacy of protectionist policies. While developed country M F N tariffs on manufactured goods are generally below 10 percent, most devel-oping countries still charge higher import taxes. Moreover, average tariffs often hide tariff peaks in certain commodity classes, some of which may be of particular interest to foreign firms. Table 2.1 gives an overview of the M F N tariff rates of selected developed and developing countries. Japan (2000) (industrial goods only) 6.5% European Union (2000) (industrial goods only) 6.4% Norway (2000) 8.1% Canada (2000) 4.4% U S A (2000) 4.6% Brazil (2000) 13.7% Mexico (2002) 16.5% Chile (2003) 6.0% Morocco (1996) 23.5% India (2002) 32.0% Malaysia (2000) 9.2% Table 2.1: Average Applied Unweighted Tariff Rates, Selected Countries Source: WTO Trade Policy Reviews, various years; Ministerio de Relaciones Ex-teriores de Chile, Direccion General de Relaciones Economicas Internacionales Besides erecting high tariff barriers, developing countries pursued policies to harness foreign investment to stimulate the development of local industry. Local content requirements promote a domestic base of supply industry, employment re-quirements create jobs, and technology transfer requirements and domestic equity participation quotas foster domestic human capital (UNCTAD 2001, 3). However, the measures impose costs on investors by forcing them to allocate their resources differently than it would in the absence of any measure. The costs are shifted to the home country or other markets (Caves 1996; Greenaway 1992, 146-147). Multinational firms therefore make considerable efforts to lobby both home and host governments to reduce these trade barriers. A preference for free trade, however, does not predict bilateral liberalization. Firms could also lobby home and host countries to unilaterally reduce tariffs on only the goods of their interest. Alternatively, offshore processing programmes allow such tariff reductions without true liberalization on an M F N basis. 30 Offshore processing programmes provide many of the benefits of vertically integrated production across borders. Under these programmes, firms ship inter-mediate goods to export processing zones (EPZs) in developing countries, where labour-intensive stages of the production process take place. Subsequently, goods are re-exported to the home or other countries. Often, more capital-intensive stages of production take place in the home country, leading to back-and-forth trade of the same good in different tariff categories. Home countries generally only tax the value added abroad. Furthermore, host countries often offer incen-tives and rebate M F N tariffs on capital goods. Offshore assembly programmes are used by the US and the E U (Feenstra, Hanson, and Swenson 2000). Numerous developing countries offer EPZs to attract manufacturing FDI (Mclntyre, Narula, and Trevino 1996; Warr 1989; 1990). Since these relatively simple policy options exist, the extra effort of bilateral trade agreements must promise specific benefits to its supporters. The next section addresses this question. 2.2 Bilateral Agreements as Exclusive Clubs Unlike multilateral negotiations in the WTO, bilateral trade agreements can be shaped to favour multinational firms from the capital-exporting partner. As out-lined above, trade and investment liberalization helps firms in lowering the cost of multinational production, particularly if a developing country functions as an export platform to the host market. Yet if liberalization took place multilaterally or unilaterally, nothing would prevent firms from third countries from enjoying the same benefits. European firms could invest in Mexico and export to the US. Japanese firms could build factories in Eastern Europe to serve E U markets. FTAs, by contrast, offer the benefits of liberalization, but can raise the barriers for these outsiders and protect the home market. Since most trade between developed and developing countries is vertical intra-industry in nature, protectionist policies tend to focus on these commodities. As part of a free trade agreement, two tools serve this end: the prohibition of tariff rebates, and the negotiation of tight rules of origin (ROOs). 2.2.1 Closing the Backdoor: Rules of Origin and Export-Processing Zones In both customs unions and FTAs, rules of origin define whether a good qualifies for tariff-free shipping across borders among the member countries. They are nec-31 essary because goods within a PTA are often either produced with input of parts imported from non-members, or produced abroad with parts originating in the PTA and the re-imported. In custom unions, the rules can be held simple. Often, they stipulate that a percentage of parts has to originate in the PTA. For exam-ple, if a good is produced in the E U , but of less than 50 percent parts originating there, the common M F N tariff is applied when the good is shipped to another E U member country. Rules of origin in FTAs are more complex. Since in an FTA, each member applies its own external tariff, rules are necessary to prevent transshipping. In the absence of such rules, goods would enter via the member country with the low-est external tariff. For example, trucks from the E U or Japan would be imported into Canada with a 6.1 percent tariff (presumably at a port close to the US to keep transport costs low) and transshipped to the US tariff free, rather than pay-ing the 25 percent tariff imposed by the US. Without rules of origin, FTAs would be highly liberalizing, because they would exert pressure to lower tariffs to the lowest level applied by any member. Weak rules of origin would thus clearly fur-ther the goal of \"open regionalism,\" or regional agreements that boost multilateral liberalization. In practice, this is rarely the case. Although many rules of origin are highly technical, they can be grouped into three categories according to the test of origin they apply (Falvey and Reed 2000, 1): \u2022 value added test\u2014rules that require that the last production stage taking place within the PTA adds a minimum value, or that a minimum value of local content is used; \u2022 change in tariff heading test\u2014rules that confer origin status if the final prod-uct is classified under a different tariff heading than its intermediate goods input; \u2022 technical tests\u2014rules setting out that certain production activities or specific parts confer origin status. Due to their ostensibly technical nature, rules of origin have only recently attracted the attention of political economists (Estevadeordal and Suominen 2004; Garay and Cornejo 2001; Hirsch 2002; Jensen-Moran 1996; Krishna and Krueger 1995; Krueger 1993). However, the intense lobbying around rules of origin demon-strates that they are important for firms. During the N A F T A negotiations, ROOs were among the last items on which negotiators reached agreement (Cameron and 32 Tomlin 2000, 168-175). More than a fifth of the final text is taken up by rules of origin. Consequently, it is now commonly accepted that ROOs can be an \"inde-pendent policy instrument\" (Falvey and Reed 2000, 4): Given different external tariffs, tight rules of origin can be used to export protection within an FTA. If ROOs stipulate a minimum local content, then they distort the use of inputs towards intermediate goods producers from within the FTA. A simplified version of the model developed by Krishna and Krueger (1995) illustrates how. For ex-ample, a firm produces a good for the whole FTA market in a country with a 20 percent external tariff on intermediate goods. The ROO is set at 50 percent cost basis. Goods cross borders within the FTA tariff-free if they fulfill this rule, otherwise they are taxed at 5 percent (MFN) tariff. Suppose intermediate goods producers from outside the FTA are 5 percent cheaper than intra-FTA producers, with negligible transport costs. Then it would be cheaper for the firm to switch to more costly intra-FTA intermediate goods, but ship to the other member countries tariff free. Dependent on the remaining M F N tariffs and the strictness of rules of origin, an FTA might therefore divert trade in intermediate goods away from non-members and towards member countries. Due to their trade-diverting effect, rules of origin can make FTAs politically viable in the first place, as shown formally by Duttagupta and Panagariya (2003). ROOs, seen as a policy affecting the return on assets of firms, are crucial in mus-tering the political support necessary for the conclusion of FTAs. Based on their trade-diverting effect, we can deduce the preferences that firms wil l have regard-ing rules of origin. Multinational manufacturing firms differ in their preferences over rules of ori-gin depending on their supply networks. Primarily, the share of regional content that a firm uses determines the desirability of ROOs. Hypothesis la. The more a firm sources regionally rather than globally, the higher the rules of origin it demands. High regional content characterizes incumbent firms with existing supplier linkages rather than newcomers (Graham 1994, 116). Often, firms maintain high levels of imports from their home country and other important production sites. But this may not always be the case: if the suppliers are outside the partner coun-try, firms wil l attempt to resist the negotiation of strict rules of origin. 33 By contrast, intermediate goods producers in the FTA have a strong incentive to lobby for strict ROOs if their main customers are in the FTA, because it creates an artificial barrier for suppliers from outside the FTA. 4 Hypothesis lb. If intermediate goods producers primarily supply companies within the FTA, then they will favour high rules of origin. It follows that intermediate goods producers and multinational firms may or may not lobby for the same rules depending on their relationship to regional pro-duction. Suppose a multinational firm sources only from within the proposed FTA, and all its suppliers are located in the two member countries. Then we would observe the extreme case of a coherent vertical coalition of the end-user firm and its suppliers. By contrast, if a multinational firm prefers to source from outside the proposed FTA, but there are potential, more expensive intermediate goods suppliers in the FTA, then the latter would demand higher rules of origin thresholds\u2014possibly preventing the emergence of a coherent vertical coalition. Following the same principle, firms will want to ensure that rules of origin are applied to goods produced in export-processing zones. In an FTA, export-processing zones are redundant for member country firms, since goods cross bor-ders tariff free. Just like FTAs in the absence of rules of origin, export-processing zones are a potential backdoor for non-member firms. As an incentive for for-eign investors, many developing countries (and some developed countries like Canada) offer rebates on M F N tariffs for the import of capital and intermediate goods that cannot be sourced in the host country. Common practice in EPZs, tar-iff rebates present a threat to the developed country partner, because non-member firms could invest in the EPZ and export to the home market. Consequently, home country firms wil l need to close this entry point by lobbying against tariff rebates and for an application of M F N tariffs and rules of origin. Hypothesis lc. Home country manufacturing firms will seek to ban exemptions from the MFN tariff for imports from non-members. Contrary to intuition and frequent press reports, conventional exporters of fin-ished goods are often the least interested in bilateral deals. Their return on as-sets is less likely to be favourably affected by bilateral agreements. Their pref-erences over preferential liberalization depend on the market size of the partner. But since differences in factor prices also imply differences in per capita income 4For the same reasons, labour employed by these firms should support the strictest rules of origin possible. 34 and purchasing power, exporters stand to gain less by supporting an FTA with a developing country than multinational firms. Occasionally, a firm may be highly dependent on one particular market to sell its products.5 Yet given the limited pur-chasing power of developing countries, they will rarely be such crucial markets. We can expect exporters to make less effort than firms that undertake FDI in the partner country. Hypothesis 2. Conventional exporters will not support an FTA, unless it offers access to sizeable markets. Manufacturing firms' preferences therefore depend primarily on their link to foreign direct investment. This also applies to the service sector, although the competitive advantages conferred by FTAs result from market structures as much as rules. 2.2.2 Getting a Head Start: First-Mover Advantages in Services Service firms occupy a middle ground between manufacturing investors and ex-porters. Often, service negotiators raise the same demands in bilateral and mul-tilateral negotiations. However, once preferential service liberalization occurs in an FTA, the market and regulatory structure of the service sector raises barriers to outsiders. The oligopolistic structure of many service markets also implies that even small markets are relatively more important for service providers than for exporters. In principle, services liberalization could be pursued bilaterally or multilater-ally. In the absence of specific benefits of bilateral liberalization, the choice of venues would be one of opportunity. As a result, service providers that supply host country markets, or market service providers, have no intrinsic preference for bilateral liberalization. They seek investment liberalization because it is a pre-condition for market access. For example, developing countries have liberalized telecommunications or financial services, in the past often provided by the state, and invited foreign investment. But if an FTA offers the chance for liberalization under GATS Art. V (free trade area in services), then service firms seize the oppor-tunity. Alternatively, if multilateral negotiations are more difficult because of the large number of players involved, bilateral negotiations could easier. Practitioners argue, however, that the main difficulty in services liberalization is in the domestic 5 I thank Stephan Haggard for making this important point. 35 work of adapting laws, regulations, decrees and procedures, making multilateral and bilateral liberalization equally hard to achieve (Stephenson 2002). This sug-gests that if PTAs are more successful in services liberalization, the reason lies in the specific trade-offs of the deal. Hypothesis 3a. Market service providers seek trade and investment liberalization in general, supporting bilateral deals when the opportunity arises. In most service industries, cost structures create first-mover advantages (Lie-berman and Montgomery 1988, 1998). Preferential liberalization grants these ad-vantages to firms from the partner country only. If previously protected banking markets in developing countries are opened for the first time, more efficient for-eign firms rapidly establish footholds (Claessens, Demirgiic-Kunt, and Huizinga 2001). In retail markets that require large numbers of branches, first entrants can achieve economies of scale in marketing and back office operations (Williams 2003) . Moreover, foreign firms can take over domestic firms and quickly achieve controlling market shares. In financial markets, dominant firms often enjoy market power and can price competitors out of the market and prevent entry (Berger et al. 2004) . Preferential liberalization therefore equals the profitable deals that some developing countries offered in the privatization of public utilities and telecom-munications firms (Ramamurti 2000). At other times, the regulatory structure directly determines whether an FTA will exclude outsiders. Often, a country will open its service markets bilaterally first, but then adapt its domestic laws and regulations to allow market entry on a most-favoured nation basis. However, in industries such as telecommunications, regulators often limit the number of licences granted to operators. In this situ-ation, late entry is not penalized, but completely blocked until the next licences are offered.6 Even though service firms may not be the first to push for bilateral liberalization, they cannot afford to let the opportunity pass. Even i f service providers show no interest in the small market of the develop-ing country, or the host country chooses not open its service markets, vertical FDI will demand attendant services. Multinational firms need service providers that re-duce transaction costs by offering financing and insurance for cross-border trade, shipping, and legal consulting. To distinguish these firms from those primarily offering services to host country customers, they are here referred to as intermedi-ate service providers. Once vertical FDI takes place on a sufficient scale, service firms have an incentive to seek the reduction of barrier to their entry into foreign 6I am indebted to Pascal Kerneis of the European Services Forum for this point. 36 markets. The pull exerted by vertical FDI therefore leads service firms to support liberalization. Hypothesis 3 b . Intermediate service providers will favour the reduction of barri-ers to entry into the host market in order to serve manufacturing FDI. If these propositions hold, they describe a key constituency for the liberaliza-tion of trade and investment between developed and developing countries. Sup-port for bilateral deals, however, presupposes either specific incentives in their favour or against multilateralism. Under the assumptions outlined above, FTAs do not supplant multilateral liberalization, but offer specific benefits to multina-tional firms. Hypothesis 3 c . Service firms will favour preferential liberalization if it offers a head start over competitors. However, unlike in the manufacturers sector, preferential liberalization does not add competitive pressure for service providers in their home markets, since investing in a developing country to provide services to a developed country mar-ket rarely offers benefits. We can now position the firms in the home country along a continuum ac-cording to the intensity of their preferences over bilateral rather than multilateral liberalization. Figure 2.1 aligns firms depending on the strength of their prefer-ences, summing up the predictions regarding firm lobbying. Finished goods exporters Intermediate service providers Multinational manufacturing firms weak preferences strong preferences Intermediate goods producers Figure 2.1: Preference Intensity of Firms 37 Unsurprisingly, if a potential FTA partner is a sizable market, the prospect motivates more lobbying by finished goods exporters and market service providers than access to a small country. The discussion so far shows that various firms have specific interests in bilat-eral trade agreements. The following section argues that host countries are likely to collude with these interests, making preferential agreements easier to achieve. By contrast, firms and economic groups that will resist liberalization efforts have no such allies. Although they are protectionist, they do not mount a principled op-position to preferential agreements. Rather, import-competing industries oppose liberalization in general. 2.2.3 Relative Strength vis-a-vis Protectionist Forces Success or failure of firm lobbying depends on relative political influence com-pared to protectionist forces. As in all trade negotiations, protectionists are found at home and in the partner country. In this sense, PTA negotiations do not differ from other deals across domestic and international levels, the subject of the sem-inal contributions by Evans, Jacobson and Putnam (1993) and Moravcsik (1993, 1997, 1998). According to the logic of two-level games, governments negotiate at two tables: one for the deal with their counterparts, the other for ratification by domestic constituents in the respective institutions. The preferences and rel-ative strength of domestic coalitions thus determine the leeway international ne-gotiators have for an agreement. With these considerations in mind, we can use evidence from the negotiations to test additional hypotheses. These propositions are only indirectly derived from the assumptions outlined above. However, they help explain why multinational firms are likely to succeed in pressing their points in bilateral agreements: firms are likely to make similar calculations about their influence on the negotiations. As a corollary to hypothesis lb, host countries will usually collaborate in ne-gotiating rules of origin to some extent, if these rules favour domestic producers of intermediate goods. At the same time, as outlined below, strict rules of origin will also act as a deterrent to outsiders. Host country governments will therefore have to balance competing aims. On the one hand, they want to attract FDI from more than one developed country, which weighs against strict rules of origin. On the other hand, domestic intermediate goods producers will call for tighter rules that would extend protection to them. Since rules of origin interact with other host country barriers to trade, the same logic applies to tariff liberalization. 38 Corollary 1 to Hypothesis lb. The more influential intermediate goods produc-ers in the host country, the higher the rules of origin to which a host country government will assent. Corollary 2 to Hypothesis lb. The more influential intermediate goods produc-ers in the host country, the higher the MFN tariff that the host country will apply. As incumbent multinational firms are the main buyers of intermediate goods produced in the host country, they indirectly create jobs and gain political influ-ence. In negotiations over tariffs and rules of origin, their views wil l therefore be represented at both sides of the table.7 Consequently, they will be more successful in their lobbying than conventional exporters. Corollary 3 to Hypothesis lb. Incumbent multinational firms will have greater influence in the negotiations than exporters from the home country. Note that this also implies that intermediate goods producers from the home country wil l lobby strongly, but will be less successful than investors if they have import-competing counterparts in the host country. The familiar contest between export-oriented and import-competing firms plays out in the negotiation of bilat-eral deals just like in multinational rounds. Resistance to negotiating a PTA wil l therefore come from the usual suspects: import-competing firms and industries. For them, a bilateral agreement is yet another threat to the protected home market. However, two factors may allow PTAs to be negotiated faster than multilateral agreements. First, not every partner country exports goods that compete with domestic producers. In this situation, negotiators have more freedom to strike a deal. Accordingly, the propositions put forth so far should capture the driving forces of an agreement. Second, although the institutional context in the form of GATT Art. X X I V limits the scope of exclusions, governments can still carve out a few sectors. Preferential trade agreements are supposed to cover substantially all trade, but given the right configuration of domestic coalitions, excluding a few sectors may suffice to reach agreement. Hypothesis 4. Import-competing firms will oppose a PTA if it opens the home market to competition from the partner country. If the arguments put forth so far hold, they explain why preferential liber-alization has become desirable for many firms. Foreign direct investment from 7Putnam (1988,459) briefly considers the effect of such transnational linkages on negotiations. 39 developed to developing countries has expanded rapidly in the last two decades, stimulating increasing flows of vertical intra-industry trade. But the accelerating proliferation of these agreements outpaces the growth of FDI and trade by far. As I argue in the next section, the discriminatory aspects of FTAs that make them po-litically attractive have unintended consequences: they motivate excluded parties to seek their own bilateral agreements with FDI host countries. 2.3 The Endogenous Dynamic of FTAs If firms are successful in pursuing their preferences in FTA negotiations, then they gain an advantage over competitors from other countries. Strict rules of origin, the abolishment of M F N rebates for producers in export-processing zones, and the structural features of service markets either impose costs on incumbents from non-member countries or raise the barriers for new entrants. Firms from non-member countries face an unpleasant choice: they can either forsake the host country as a market and location for investment, absorb the cost increase to stay in the market, or they can try to level the playing field to remain competitive. 2.3.1 Non-Member F i rms' Reactions to Exclusion Once the FTA is in place, how does the situation present itself to firms from non-member countries? Take the following stylized situation of three countries, two developed home countries A i and A2, and D, a developing host country. To sim-plify, country D imposes a uniform M F N tariff of 15 percent; countries A* and A2 levy a 5 percent uniform tariff. Under the assumptions outlined above, D attracts FDI in manufacturing and services from A i and A2 because of the differing factor endowments. In the absence of an FTA, all intermediate and capital goods from A i or A 2 are taxed at the same M F N rate. Host country D can rebate the M F N tariff on these goods for all foreign firms if they manufacture for export, and all firms pay the same 5 percent M F N tariff when they export finished goods to either A i or A2. Figure 2.2 presents this situation. As discussed in section 2.1.2, firms from either A would prefer tariff reduc-tion by D. Lowering the barriers to trade reduces the cost of vertically integrated production and makes FDI in D more profitable. Often, a host country will attract more FDI from one host than from the other, whether due to geographic proximity, lower barriers because of similar language and legal institutions, or differences in firm strategy. Geographically close coun-40 MFN tariff 15% levied by D on intermediate goods MFN tariff 15% levied by D on intermediate goods Country D (FDI host) Country A2 MFN tariff 5% levied by Ai on finished product Figure 2.2: Vertical Integration without FTAs tries have been labelled \"natural\" partners (Frankel, Stein, and Wei 1995) because of low transport costs. For these idiosyncratic factors, one country pair will ne-gotiate a first FTA. Compare the situation in figure 2.3, when firms from Ai have managed to secure a high rule of origin in an FTA. Now the excluded firms face different costs in a vertically integrated production across borders. Intermediate and capital goods pay the M F N tariff of the host country, which is now prevented from offering rebates. Moreover, firms from A2 must now fulfil the rule of origin to avoid paying a second tariff, the M F N tariff upon export to A i ! Table 2.2 sums up the differences in import taxes paid by firms from A i and A2. For A 2 ' s service firms, the first-mover advantages created by the FTA between A i and D penalize them when they seek to enter the market. Moreover, intermedi-ate service providers from A 2 will have fewer commercial ties with firms from A i and stand to gain little. Upstream producers from A 2 wil l be the worst affected, since multinational firms from A 2 are forced to switch to regional suppliers from within the FTA. What are the options for firms from A 2 ? The simplest choice would be to de-invest. Firms could forgo the opportunity of using country D as an export platform to supply A i ' s market, relocate production to a country with similar endowments, and export from there while paying M F N tariffs. But if the benefits of production in D, proximity to the market of A 2 combined with low labour cost, still outweigh the increased cost, non-member firms are likely to invest more rather than less. 41 Tariff-free access for intermediate goods MFN tariff 15% levied by D on intermediate goods FTA Country D (FDI host) Tariff-free access for finished products 5% for products of A2 produced in D Figure 2.3: With an FTA and Rules of Origin For sales in D For sales in A, Firms from country A| Zero tariff Zero tariff Upstream producers from A, Zero tariff Firms from A 2 15% M F N tariff levied by D on parts imports 15% M F N tariff levied by D on intermediate goods 5% M F N tariff levied by A, unless rule of origin is met Upstream producers from A 2 15% M F N tariff -Table 2.2: Preferential Tariffs with FTA and Rules of Origin 42 Since such a location of production creates a competitive advantage, non-member firms have a strong incentive to preserve this access. If firms from A 2 indeed chose to preserve or raise current levels of FDI, then they have a strong incentive to try and convince the host government of country D to lower its tariffs\u2014but they can offer little in return, other than the promise of staying in the host country. If complete unilateral liberalization was not feasible for D before, then the first FTA will not change this situation. The next-best response to the FTA between A i and D is to counter with an agreement between A 2 and D. The defensive FTA levels the playing field. Hypothesis 5. Upon conclusion of an FTA between Ai and D, firms from A2 will seek liberalization of tariffs between A 2 and D. To this end, firms need not demand further liberalization from the host country than conceded in the first FTA between A i and D\u2014achieving the same reduction of barriers to trade and investment suffices. Ideally, a defensive FTA wil l incor-porate a \"ratchet clause\" that ensures matching commitments if host country D accelerates its liberalization with A i . Hypothesis 6. Firms from A 2 will seek parity in the liberalization of trade and investment with A\\, but have no incentive to go beyond that. The propositions outlined above regarding the preferences of firms can be di-rectly extended to the defensive agreement. Barriers for non-members target im-mediate competitors, so the coalitions of firms lobbying for counter-agreements are often highly similar. Hypothesis 7. The coalitions of firms lobbying for a defensive FTA mirror those for the first FTA. The lobbying efforts by firms in A parallel the initial efforts made in A\u2014with one key difference: Higher barriers for non-members imply a lower threshold at which firms become active, because the potential losses provide stronger incen-tives to lobby. Because of this effect, the unintended\u2014and as I argue below, most undesirable\u2014consequence of bilateral deals between developed and developing countries is that they engender their own proliferation. 2.3.2 Intensifying the Competition While the first FTA resulted from FDI flows to the emerging market, based on strategic decision by firms about desirable production locations, the countermove-ment can result from much weaker trade and investment linkages. The initial FTA 43 offers the prospect of lower costs for production fragmented across borders and improves on the status quo. The decision to support an FTA therefore involves an element of uncertainty, because firms do not know ex ante whether their lobbying efforts will really pay off. For incumbent firms from non-member countries, however, the case is clear: If strict rules of origin are established and M F N rebates for non-members proscribed, they are worse off than before. These firms will immediately feel the effect of import taxes and administrative costs of compliance with rules of origin. Since the initial FTA raises the barriers, firms have greater incentives to seek a reduction of these impediments to trade and investment. Hypothesis 8. Defensive FTAs will be triggered by weaker trade and investment links than the initial agreement. The strength of the reaction depends on the restrictiveness of the rules of origin of the initial FTA and the remaining host country tariffs. The threshold levels of FDI, rules of origin and tariffs are an empirical matter that depends on the internal cost structures of the multinational firms and the importance of the market they seek to supply. At the margin, a sufficient number of FTAs may even motivate conventional exporters to lobby for compensatory FTAs, but as argued before, these firms are likely to be the least concerned about FTAs. But again, the more such FTAs are in place, the lower the threshold at which the excluded country will be pressed by its own firms to seek an FTA that defends its access to FDI hosts. Once a number of bilateral agreements have been signed, it becomes more costly for governments to stay at the sidelines. Given these circumstances, gov-ernment decision-makers will have an incentive to anticipate and pre-empt the initiatives by other countries by actively seeking FTAs. But this effect motivates even more countries, despite weak trade and investment linkages, to opt for bilat-eral deals. Once countries have gone down the road of FTAs, the proliferation of bilateral trade agreements becomes a self-enforcing dynamic. 2.4 Possible Objections On theoretical grounds, two objections can be raised against the argument put forth here. First, rules of origin can be seen as a reaction to protect sunk invest-ments made under more restrictive host country policies. In this sense, non-tariff barriers provide \"breathing room\" that give firms time to restructure their opera-tions (Chase 2003). Second, even though they create non-tariff barriers to trade, 44 FTAs could still be beneficial because they intensify pressures for competitive adjustment, bolstering the constituency in favour of free trade in long run. The case for a benign view of ROOs rests on whether investments in fact need to be restructured. If the operations of the same firms that lobby for ROOs are al-ready highly vertically integrated before the FTA is negotiated, then the argument for extending protection is much weaker. In the case of NAFTA, firms would not have begun restructuring in earnest before the agreement was an option. Ma-jor investments in Mexico and increasing vertical integration would follow the conclusion of the agreement. If, on the other hand, economic integration in the sectors with the strictest rules of origin was already far advanced before the agree-ment was contemplated, then there may have been few outdated factories left to restructure. These predictions are tested in Chapter 3. But do FTAs at least create a pro-free trade constituency? Again, much de-pends on the investment undertaken by firms. Take a situation in which two coun-tries with similar factor endowments conclude a PTA. These types of countries tend to trade in similar goods that are differentiated to suit different consumer tastes. Intra-industry trade will be horizontal rather than vertical in nature. In the standard example, Japan and the E U trade high-quality automobiles with each other. Production of these goods requires economies of scale to be competitive. As Milner (1997, 81) argues: \"firms would rather export from existing plants to attain optimal scale than invest abroad.\" But since firms also become more com-petitive within the PTA, they will want to export more to non-members. This requires reciprocal tariff bargaining. These predictions closely conform to the bargaining observed between the US and the E U . Since firms will want to export from home markets in the E U to the US markets, they wil l support liberalization efforts between the two, which in practice only occurs in the WTO. PTAs of this type could be stepping-stones to global free trade. By contrast, as McLaren has argued (2002), regionalism may be \"insidious\" because producers may make specialized investments in anticipation of regional agreements. In this sense, trade partners are only natural because their special-ization towards each other is a product of a PTA. Yet in bilateral agreements, multinational firms are undertaking much of the export-oriented investment. We can infer mutual specialization on the basis of trade flows: if vertically differenti-ated products of these multinational firms dominate trade between two countries, then a greater share of the trade wil l be vertical intra-industry trade. Yet as argued here, vertical intra-industry trade within a region creates incentives to extend pro-tection rather than to truly liberalize. If, at the same time, FTAs exclude protected sectors, then they are unlikely to bolster global free trade. 45 As the analysis of N A F T A in the following chapter shows, the \"silent inte-gration\" (Eden and Molot 1991) of the US market and the Mexican production platform was well under way before N A F T A was a realistic policy option. The evidence points to mutual specialization of Mexican and US production, driven by foreign direct investment. 2.5 Networking the Globe What implications for multilateral trade liberalization can be derived from the framework presented here? Although this study does not address the welfare ef-fects of PTAs, its findings suggest that the current proliferation of bilateral agree-ments may be less liberalizing than its proponents argue. The unfolding compet-itive dynamic creates a network of FTAs\u2014the \"spaghetti bowl\" effect (Bhagwati and Panagariya 1996)\u2014each with different rules that reflect the preferences of firms. FTAs therefore create an enormous burden for firms to comply with docu-mentation and tracing of origin rules. Especially firms from developing countries and small and medium enterprises face capacity problems in fulfilling these requirements. Anson and collaborators (2003) estimate that up to 40 percent of Mexico's preferential access gains to the US market in 2000 were absorbed by administrative costs related to N A F T A trac-ing rules. Depken and Ford (1999) show that N A F T A tracing requirements are strongly biased against smaller firms to the point where they may further indus-trial concentration within sectors and depress trade between the US and Mexico. Kunimoto and Sawchuk (2004) report that small Canadian exporters often rather pay the US M F N tariff than deal with NAFTA's complex ROOs. But such com-plex non-tariff barriers to trade by no means limited to NAFTA: Brenton (2003) argues that the European Union's \"Everything but Arms\" initiative to offer least-developed countries better access to its markets is much less effective than as-sumed, mainly because of complex ROOs. Given these considerations, the liber-alizing of bilateral agreements may be limited to a few, multinational firms. Yet if these firms actually benefit from ROOs, as argued in this study, then each PTA wil l create a group of supporters with vested interests in their continuation. The simplest countermeasure, at least in principle, is multilateral tariff reduc-tion. With lower M F N tariffs charged by the host countries of FDI, rules of origin lose their potential to divert trade. But if the propositions put forth here hold, then the host country should have little incentive to unilaterally reduce its tariffs, but would rather hold on to its preference margin to attract FDI. At the same time, 46 bilateral liberalization with the most important host countries of FDI will take away support for multilateral efforts by fulfilling key demands of multinational firms. Moreover, if states carve out protected sectors such as agriculture, then the multilateral trade institutions will be left with the most intractable problems of protectionism. The case studies in the next chapters evaluate existing agreements in the light of these concerns. 47 Chapter 3 NAFTA\u2014The Original Sin? Why did many important US industries come to support bilateral trade agree-ments, in particular NAFTA, in the late 1980s? What accounts for the specific form that N A F T A took regarding trade and investment liberalization and rules of origin? This chapter argues that contrary to the rhetoric of free trade, preferen-tial trade agreements became attractive when many US manufacturing industries were on the defensive, while liberalization of the more competitive service sec-tor made little progress in multilateral negotiations. For US multinational firms, N A F T A offered an opportunity to seize the benefits of the liberalization under-taken by Mexico, an important host country for FDI from the US. N A F T A did not, however, trigger the flows of FDI to Mexico. As the chapter shows, Mexican liberalization had proceeded far enough to attract considerable investment from the US. Key manufacturing industries began restructuring their operations toward export-oriented production well before N A F T A was on the agenda. Once underway, the N A F T A negotiations were decisively influenced by the preferences of US firms toward a selectively protectionist outcome in precisely those industries with export-oriented FDI in Mexico. Under increasing compet-itive pressures, US firms tried to secure the locational benefits of FDI in Mex-ico for themselves. In conjunction with Mexico's remaining barriers to imports and investment from the rest of the world, N A F T A therefore held the potential for significant trade and investment diversion. This trade and investment diver-sion, Chapters four and five show, triggered the defensive reactions by the E U and Japan. The account offered in this chapter therefore fills a gap in our understanding of the economic incentives for N A F T A in particular and FTAs between developed and developing countries in general. The most detailed analysis of the political struggle in the US to gain negotiating authority and to ratify N A F T A (Mayer 1998) clearly shows the motives of the protectionist opponents of NAFTA, but offers less insight into why the pro-free trade constituency would have supported the deal. Chase (2003) provides a partial explanation based on an inaccurate interpretation of firm preferences, as argued in detail below. 48 The chapter is divided into six sections. The first section gives a brief overview of the US turn in trade policy toward bilateralism, providing the background to connect the theoretical assumptions of Chapter 2 to the subsequent empiri-cal study. Mexico's unilateral liberalization, surveyed in section two, offered the prospect of a production location with low labour cost and an expanding service market. As shown in section three, US firms responded by turning Mexico into an export platform long before N A F T A was put on the agenda. When the option of an FTA arose, section four argues, the same firms seized the opportunity to press for rules to prevent Japanese and European manufacturers from following their example. When comparing the preferential liberalization under N A F T A with Mexico's multilateral commitments, the close correspondence of high barriers to the outside world with foreign direct investment interests from the US stands out. Section five engages competing explanation. Section six concludes. 3.1 Bilateralism in US Trade Policy Multilateralism and the GATT provided the foundations for US trade policy since 1948. While the US offered support for European regional integration, it abstained from participating in bilateral initiatives prior to NAFTA, with two exceptions: the US-Israel FTA, signed in 1985, and the 1987 Canada-US FTA (CUSFTA). Whereas the former was intended as a means to extend strategic support to Israel (Destler 1986, 226), the CUSFTA resulted from a turnaround in Canadian policy after almost a century of resisting US advances. Neither agreement, however, had much potential to affect outsiders. Israel's regional political isolation left it entirely dependent on European and US trade in any case. Canada lowered its M F N tariffs significantly during the Tokyo and Uruguay rounds and participated actively in the GATS, thus limiting the potential for trade diversion. US trade policy preferences moved away from unconditional support for mul-tilateralism with shifts in comparative advantage, in particular the rise of service industries and the concurrent decline of manufacturing in the US. Starting in the late 1970, services were pushed onto the negotiating agenda by US business groups (Freeman 1998, 184). In 1984, US Congress, stirred on by the 1981-82 recession and the uncompetitive exchange rate of the first Reagan administration that battered the industrial heartland of the US, passed a new Trade and Tariff Act. The 1984 Act expressed the Congressional reaction to growing concern among import-competing industries, a shrinking and specialized constituency for free trade (Richardson 1991), and limited success at pushing the services and in-49 vestment agenda at the 1982 GATT Ministerial meeting in Geneva (Destler 1986, 226; Frankel 1997, 5; Krueger 1995, 87; Schott 1983, 40-45; 1989, 4). In May 1985 the Advisory Committee for Trade Negotiations, an important body in which business interests and transmitted to the US Administration, issued a report on a potential future GATT round. Remarkably, support for a new multilateral round was only strong among the groups advocating the new issues of services, intel-lectual property, and investment. By contrast, most manufacturing industries ap-peared reluctant, instead requesting an aggressive trade policy in defence of US import competing industries (Ostry 1990, 27-28). The shift to a bilateral trade policy took place in the context of declining competitiveness. More successful companies, by contrast, supported multilateral efforts. Following the mandate of the 1984 Act, the USTR developed a series of bi-lateral instruments for trade and investment liberalization. One element was a programme of bilateral investment treaties (BITs). European countries, trying to provide their investors with a minimum of legal guarantees against expropriation by developing countries, had been active in signing BITs for over twenty years. BITs usually follow a common template, or model BIT. The US approach differed from European policy in that the 1987 and 1991 US model BITs also liberalized investment: they prohibited the use of performance requirements and scheduled previously closed sectors of the host country for opening to foreign investment (Vandevelde 1993, 621). The new BIT was first implemented with the 1991 US-Argentina treaty, which became the template for investment chapters in future free trade agreements. Nei-ther the US-Israel nor the CUSFTA, however, contained clauses equivalent to the US model BIT. N A F T A broke ground in creating a legal framework that closely reflected new economic realities: Investment and trade are linked in the economic integration with emerging markets, as Mexico's liberalization and the resulting changes in the character of FDI show. Yet as argued in the previous chapter, if the host country maintains high barriers to the outside world, preferential trade liberalization can cause the diversion of trade and investment. Mexico took major unilateral steps to liberalization during the 1980s; important tariff and non-tariff barriers, however, remained. 3.2 Mexico's Unilateral Liberalization Like many developing countries, until 1982, Mexico followed an economic pol-icy of import substitution. Foreign direct investment was restricted and burdened 50 by performance requirements like domestic content quota and export-balancing requirements. Importers required licences and paid high tariffs. The 1973 Law to promote Mexican investment and regulate foreign investment activities {Ley para Promover la Inversion Mexicana y Regular la Inversion Extranjera) reserved key sectors of the economy, in particular resource extraction or public utilities, for the state or exclusive Mexican ownership. A l l other activities had 49 percent ceiling on foreign participation, subject to approval by the National Foreign Investment Commission. The guiding principles were the protection of Mexican-owned com-panies, an increase in local employment, the use of local inputs, and the transfer of technology (UNCTC 1992, 13). Central to the Mexican development policy was the development of a domes-tic automotive industry. The government issued several decrees to guide foreign investment to this end. Between 1962 and 1983, four auto decrees mandated a 60 percent local content quota to stimulate the development of the local auto parts industry, while banning imports of assembled cars. Given the rapid growth of Mexico's economy that averaged 6 percent per year until 1982, the US Big Three automakers, Nissan and Volkswagen as well as (briefly) Renault invested in Mex-ico despite these policies (Peres Nunez 1990, 18). However, the protected market structure led to \"low production runs, high prices and poor quality\" (Calderon et al. 1995, 22). Transplant factories and especially parts manufacturers worked be-low efficient scales of production compared to plants in the developed world. Re-strictions of FDI, especially in the supply industry, prevented multinational firms from achieving economies of scale (Peres Nunez 1990, 22). Still, the sector gener-ated jobs, especially in the parts industry, and began to export successfully, espe-cially when the 1977 decree allowed 20 percent of the automakers' production to be shifted to the maquiladoras, or in-bond factories along the US-Mexican border (Robert 2000, 183-184). While not as efficient as automotive parts production in the developed world, the industrialization created the necessary human capital for the export boom of the second half of the 1980s. In contrast to the import-substitution industries, the maquiladora sector was open to 100 percent foreign ownership, provided that 80 percent of the produc-tion was exported. Under the former US tariff classifications 806.30 and 807.00, products could be exported to Mexico, processed, and re-imported into the US. Duty was only levied on the value added abroad. In addition, in a duty-drawback scheme, the Mexican government rebated the import tariff on intermediate goods (parts and other supplies) that were re-exported to the US. The maquiladoras at-tracted considerable investment from US firms, but also from Japanese and Euro-pean multinationals. These firms used their US subsidiaries to qualify for the US 51 export processing tariffs, but often imported capital and intermediate goods with M F N tariff rebates from outside of North America. Besides textiles, much of this investment was concentrated in auto parts and electronics manufacturing (Calderon et al. 1995, 23). Most prominent among these was the production of colour television sets, accounting for 4 million units in 1989. Combining sets and exports of subassembly parts, 65 percent of colour TVs in sold in the US in the same year were either wholly or partially produced in Mexico's maquiladoras (Koido 1991, 23). Unlike manufacturing, the service sector was completely closed to foreign investment. State-owned companies provided key services such as telecommuni-cations. Financial services were closed to outside investment, with the exception of Citibank, the only foreign bank that was allowed to operate since 1929. State-owned development banks like Nacional Financiera provided long-term lending for firms, while most other investment was based on the retained earnings of firms. The Mexican central bank mandated the holding of government bonds and set in-terest rate ceilings. In 1974, the government sought a consolidation into a univer-sal banking system along German lines, in contrast to the US system (Welch and Gruben 1993, 1-3). Finally, in the crisis year of 1982, President Lopez Portillo nationalized all Mexican banks and merged them into several larger institutions to stem the flight of capital (Ramirez 1989, 91). In that year, despite several decades of rapid growth, Mexico plunged into a severe crisis, brought about by the collapse of the oil price and the resulting difficulties in servicing its spiralling foreign debt. By 1983, the Mexican gov-ernment saw itself forced to open the foreign investment regime to attract capital from abroad. Facing severe balance-of-payments problems, the administration of Miguel de la Madrid (1982-88) agreed to an austerity programme designed by the IMF. Central elements were the opening of the economy to foreign competi-tion with the aim of joining the GATT, culminating in Mexico's formal admission on 25 July 1986, a promotion of non-petroleum exports through the expansion of maquiladoras, the liberalization of the FDI regime and the privatization of public companies (Ramirez 1989, 99-100). Over the course of the decade, Mexico embarked on a rapid and sweeping lib-eralization effort. In accordance with the 1983 National Development Plan, the National Foreign Investment Commission issued new guidelines that most sectors no longer required authorization for joint ventures with less than 49 percent for-eign participation. Beginning in 1985, many projects were approved with up to 100 percent foreign participation (Ramirez 1989, 106). The May 1989 decree on \"regulations of the law to Promote Mexican Investment and Regulate Foreign In-52 vestment\" changed the application of the 1973 law by allowing 100 percent own-ership in 73 percent of the 74 economic categories defined in the statute. Mexico's applied average M F N tariff was lowered from 100 percent to 20 percent, or 9.8 percent on a trade-weighted basis (Lustig 1998, 129-133) and 12.5 percent on a production-weighted basis, as table 3.1 shows. From a highly protectionist trade regime, Mexico quickly moved to a rapid opening of the economy. 1985 1986 1987 1988 1989 . 1990 Coverage of 92.2 46.9 35.8 23.2 22.1 19.9 import licences* Coverage of 18.7 19.6 13.4 0 0 0 reference prices* Maximum tariff 100 45 40 20 20 20 Production- 23.5 24 22.7 11 12.6 12.5 weighted average tariff * Percentage share of tradeable output. Table 3.1: Liberalization of the Mexican import regime, 1985-1990. Source: USITC (1990) Through a sweeping reduction of tariffs, Mexico had already unilaterally made the most important steps to attract manufacturing FDI before N A F T A was nego-tiated. By contrast, the service sector remained closed to foreign participation until much later. Mexican banks remained in state hands until 1991, although the de la Madrid Administration divested the non-bank assets of the nationalized banks. Finally, the government re-privatized the banks and returned to the uni-versal banking system with the Financial Groups Law of 1990. Bank shares were divided into different types of shares, with preferential stock of up 51 percent to be held only by Mexican individuals, \" B \" shares reserved for Mexican individu-als, firms, and institutional investors, and up to 30 percent \" C \" shares that could be held by foreigners (Gruben, Welch, and Gunther 1993). Although import-substitution industrialization failed, forcing an opening to the outside world upon Mexico, it created the industrial basis for the subsequent reorientation toward exports. Geographical proximity to the US and a pool of low-cost labour made Mexico attractive for manufacturing FDI that produced for the US home market. 53 3.3 The Mexican Export Platform: FDI as Restructuring Agent As multinational firms reacted to Mexico's liberalization efforts, the country be-gan to attract considerable inflows of FDI in manufacturing. Unlike earlier pe-riods, though, this FDI sought strategic assets and efficiency rather than market access. Mexico offered the central locational advantage of a pool of low-cost labour close to the US home market. Liberalization coincided with a change in the strategies of multinational firms, especially from the US, that sought to use developing countries as export platforms. The strategic shift, however, took place before a bilateral deal between the US and Mexico was a realistic option. As the following section argues, the turning point for the flows of US FDI into Mexico occurred several years before N A F T A in both quantitative and qualitative terms. Since this integration was in full swing before NAFTA, the evidence speaks against the alternative hypothesis advanced by Chase (2003b; 2004), that US firms needed rules of origin and other barriers as breathing room for restructuring. When Mexico unilaterally reduced its tariff, it threatened to become an export platform to the US market, whether the US and Mexico liberalized their bilateral trade or not, because US tariffs were already extremely low. Rules of origin could help close the backdoor to the US market. 3.3.1 US FDI in Mexico before N A F T A US FDI into Mexico picked up in the late 1980s, with considerable flows into automobile production, electronics and rubber products. Yet these flows of FDI needed very little encouragement in the form of international agreements to \"lock in\" liberalization efforts. In a 1989 OECD survey among US CEOs, most respon-dents perceived Mexico's opening as driven by underlying structural forces rather than domestic politics, forces that precluded a rolling back of liberalization in any case (Peres Nunez 1990, 53). A second indication of the turning point for US FDI into Mexico can be found by analyzing the trend in total FDI stock over time. Replicating a study by Graham and Wada (2000) with more recent data, figure 3.1 below shows changes in the stock of US FDI in Mexico between 1966 and 2002, the most recent year for which data is available. Stock data is better suited than flow to analyze FDI trends, since flow data often shows an excessive volatility caused by individual, major investments. We can estimate the point(s) in time at which the trend turned by fitting a piecewise linear relationship. 54 5 2.5 I I 1965 1970 1975 1980 1985 1990 1995 2000 Figure 3.1: US FDI stock in Mexico and time trend, In of millions ofUSS A piecewise relationship with a breakpoint at 1988 fits the data better than any other breakpoint, suggesting that the change in the trend in 1988 was more important than any other trend change before or after.1 Before 1988, the FDI stock grew at an approximate rate of 3.2 percent per year, whereas after that year, the growth rate approximates 6.2 percent. Assuming that most projects take about two to three years of planning, it appears that unilateral liberalization, not the at the time unforeseeable free trade agreement, caused the FDI trend to turn. Changes in the character of FDI, analyzed in the next section, corresponded to the quantitative trend change. 3.3.2 Retooling Mexican Factories for Export As inflows into Mexico picked up, US FDI also underwent important qualitative changes. In the late 1980s, four factors coincided in Mexico: trade liberalization, an energy cost 50 percent than what foreign investors paid in Southeast and East 'While the authors find no evidence of a \"NAFTA effect,\" the analysis using data up to 2002 shows a slight dent in 1996. However, this appears to reflect that FDI flows picked up again after a brief decline following the 1994 currency crisis in Mexico (see Appendix I for a description of the procedure and the results). 55 Asia in 1987, very low labour costs following the depreciation of the peso in the wake of the debt crisis, and slumping domestic demand (Peres Nunez 1990, 64). US manufacturing firms therefore sought to restructure existing plants for the production for export. A crucial element of this strategy was the specialization of existing plants to achieve greater economies of scale (Calderon et al. 1995, 16). Given the changing circumstances in the host country, the new investment was different in character from earlier, market-seeking and tariff-jumping FDI. The extent to which US firms used Mexico as an export platform is striking. As table 3.2 shows, the share of total sales that was exported to the US grew from 14.3 to 26.7 percent from 1983 to 1999, with a peak of 40 percent in 1996 when the effects of the Mexican currency crisis depressed domestic demand. Virtually all of these exports were intra-firm trade or trade between close affiliates. Notably, the share of intra-firm trade did not change much between 1983 and 1999, suggesting that N A F T A only played a minor role in stimulating greater integration of US firms' operations. Sales to third countries from Mexico did more than double, but were still only a third of exports to the US. Total Sales Total Exports to the US Percentage Intra-firm Percentage of Total Sales Total Exports to Other Countries Percentage of Total Sales 1983 7,476 1,070 89.0 14.3 256 3.4 1984 9,607 1,526 94.4 15.9 460 4.8 1985 10,965 1,978 95.1 18.0 465 4.2 1986 9,140 2,287 95.9 25.0 545 6.0 1987 10,710 3,152 97.0 29.4 634 5.9 1988 13,987 3,832 97.5 27.4 616 4.4 1989 16,437 4,365 98.5 26.6 882 5.4 1990 19,307 5,066 98.4 26.2 803 4.2 1991 24,838 6,431 97.4 25.9 654 2.6 1992 30,137 7,451 98.5 24.7 766 2.5 1993 32,549 8,615 97.8 26.5 1,006 3.1 1994 39,421 11,197 95.4 28.4 1,201 3.0 1995 36,193 13,341 95.4 36.9 1,636 4.5 1996 46,402 18,627 94.5 40.1 3,195 6.9 1997 54,951 20,334 96.0 37.0 4,516 8.2 1998 64,089 20,970 94.3 32.7 4,868 7.6 1999 81,473 21,754 91.4 26.7 6,917 8.5 Table 3.2: Exports by US Firms Based in Mexico, 1983-1999. Source: United States Bureau of Economic Analysis, Author's calculations. The following discussion of the manufacturing sector focuses primarily on the automotive and electronics industries, since from 1985 to 1998, these two industries taken together accounted for between 71 and 93 percent of exports by US affiliates back to the home market, as shown in table 3.3. Since 1985, automotive exports dominated the sales of US firms back to the home market. This development resulted from the combination of Mexico's lib-56 Total Sales in millions Electric and electronic Transportation ofUSS Machinery equipment equipment Percentage share Percentage share Percentage share 1985 1,884 74.9 1986 2,194 1.7 17.0 76.2 1987 3,007 2.1 13.3 1988 3,553 3.5 15.6 73.7 1989 4,115 6.6 19.3 61.9 1990 4,794 5.5 1991 6,074 4.3 17.3 70.1 1992 7,106 3.8 1993 8,039 3.1 22.6 1994 9,966 4.7 18.7 1995 1996 12,347 4.6 18.6 65.7 1997 19,693 14.0 11.2 62.7 1998 20,381 9.6 61.7 1999 21,166 4.8 18.3 . = missing data Table 3.3: Sales of US firms back to the US, by Industry. Source: United States Bureau of Economic Analysis, Author's calculations eralization, the change in US firms' strategies, and the locational benefits of pro-duction in Mexico. The next section takes up these issues in greater detail. 3.3.3 Export Orientation in the Automotive Industry While manufacturing in the electronics and electrical equipment industry mostly takes place in maquiladoras, the automotive sectors is divided into parts producers in maquiladoras, assembly plants and a domestic Mexican supply industry. Given its central role in the Mexican manufacturing sector, achieving greater efficiency in the automotive industry was vital to overcome the economic crisis. At the same time, given the dominance of foreign multinational firms and the im-portance of the industry in providing jobs, the Mexican government tried to pro-mote the industry's export orientation. The 1983 auto decree required automakers to reduce the number of models per factory line and firm. Additional lines were only allowed if they were \"self-sufficient\" in generating foreign exchange, i.e. did not contribute to balance-of-payments problems, and more than 50 percent of production were exported (UNCTC 1992, 10). The 1989 auto decree reduced the local content requirement from 60 to 36 percent for assemblers and 30 percent for 57 parts producers. However, the auto parts industry retained enough political influ-ence to prevent a thorough liberalization and retained a ceiling of 40 percent for foreign ownership. Under the more liberalized host country policies, the production of cars in Mexico became rapidly more competitive. Technology consultants Booz, Allen and Hamilton estimated in a late 1980s study that car manufacturing in Mexico would be fairly competitive with around 50 percent foreign components; produc-tion with 70 percent would be highly competitive globally (cited in Peres Nunez 1990, 121). The currency depreciation made Mexican wages highly attractive just when productivity and quality workmanship began to reach international stan-dards (Womack 1991, 43, 52-54), although domestic demand bottomed out at the same time. The US Big Three manufacturers seized the opportunity to implement strategies \"aimed at defending their national market from import penetration by Japanese and other producers. They came to the conclusion that Mexico could be-come a low cost export platform for entry-level small cars\" (Calderon et al. 1995, 24). Starting in 1986, G M transferred virtually all of its production of wire har-nesses, upholstery, and a considerable share of subassembly work to maquiladora factories. By 1990, the process that was largely completed; G M had 30 maqui-ladora plants in operation (Womack 1991, 39). Ford expanded its stamping and assembling facilities as well as its production capacity for engines in Chihuahua. The production of the compact car model Fiesta was completely relocated from Germany and Korea to Mexico. Two new models began rolling off the lines in Cuautitlan. Chrysler began producing the Neon and smaller Dodge models in Toluca and transferred the production of light trucks to Saltillo (Moreno Brid 1996, 27). By contrast, the two other foreign car manufacturers, Volkswagen and Nissan, did not change their strategies proactively (Peres Nunez 1990, 119). V W only exported engines as a defensive move because of the 1983 auto decree requirements, and even entered the business of exporting honey to meet its for-eign exchange balancing requirements (Maxfield and Shapiro 1998, 89). Nissan did shift to overseas sales, but concentrated its efforts on other Central and South American markets (UNCTC 1992, 63). The different responses to the changed circumstances were reflected in the export figures in the following years. The automobile industry became Mexico's most successful exporter of manufactured goods. By 1992, automotive exports represented 16 percent of all Mexican exports to OECD countries, most of which went to the US. The Mexican operations of the US Big Three, Volkswagen and Nissan occupied ranks 6, 7, 12, 25 and 32 respectively of all exporting enterprises 58 from Latin America in 1993 (Calderon et al. 1995, 21). The difference in firm strategies stands out even more clearly when looking at table 3.4. Already in 1987, the US Big Three produced much more for export than both V W and Nissan, with export-oriented production exceeding 60 percent for Chrysler and Ford. Between 91 and 100 percent of these exports went to the US market. For US automotive firms, Mexico was the primary export platform for entry level vehicles for the home market. Firms Share of exports Car export (units) Trucks, buses, Main export Share of main in total production (%) truck-tractor and light truck exports (units) market export market in total exports (%) Chrysler 61.1 41,037 21,774 USA 91 Ford 60.4 51,773 - USA 99 GM 44.1 32,272 - USA 100 Nissan 18.7 10,325 5,782 Central America and the Caribbean 73 Volkswagen 0.1 74 11 - -Others 0.1 - 25 - -Total 39.7 135,481 27,592 - -Table 3.4: Exports of vehicles by company, 1987. Source: Peres Nunez (1990, 119) Moreover, the US Big Three specialized their plants in Mexico towards the production of a few models that could be sold in developed country markets. Out-dated models for the Mexican market only were phased out, while models for sale in the US were introduced. Table 3.5 shows that the sales of export or dual market models increased fourfold between 1978 and 1992, while sales of models for the Mexican market decreased to a fourth of their original share. In the automobile industry, US firms changed to a vertically integrated strat-egy during the 1980s. Entry-level models would be produced in Mexico, while the small demand for high-end vehicles could be satisfied through exports from the US. The low transport costs between Mexico and the home market attracted considerable stocks of vertically integrated FDI. Especially in the production of intermediate goods, much of this took place in the maquiladora sector. 3.3.4 Growth in the Maquiladora Sector Concurrent with the rapid growth in the automotive sector, the maquiladora sec-tor expanded rapidly. Employment increased tenfold between 1980 and 2000, 59 1978-82 1983-87 1988-1992 Units Mexican market 147.1 78.3 70.8 Dual market* 138.6 116.9 307.5 Export market 10.5 54.5 199 Total 296.1 249.7 577.4 Percentage share Mexican market 49.7 31.4 12.3 Dual market 46.8 46.8 53.3 Export market 0.4 21.8 34.4 100 100 100 * less than 50% for the Mexican market or less than 50% exclusively for export. Table 3.5: Passenger Car Sales by Principal Export Market, Thousands of Units. Source: Moreno Brid (1994) with about 65 percent of the capital of automotive parts maquiladoras coming from the US (Carillo 2000, 59). However, even in the case of the rapid growth of the maquiladoras, recent research suggests that N A F T A has either had no or even a slightly negative effect on foreign direct investment in the sector. In an ex-tensive econometric study, Gruben (2001) finds no connection between N A F T A and the growth of employment in the maquiladoras. Yang (1998, 272) compares maquiladoras with other manufacturing sectors and finds that the latter grew faster after N A F T A came into force. Rather, growth in the sector appears to be related to changes in Mexican wages relative to those in the US and Asia, in particular China, and to fluctuations in industrial production in the US. The considerable growth potential of the maquiladora sector ensured that the duty-drawback schemes would be linked to the question of rules of origin in the N A F T A negotiations. Most maquiladoras used very little Mexican inputs: In a comprehensive study of maquiladoras in the northern and central Mexico, Wilson (1992, 101-119) finds that factories in the electronics and auto parts industries sourced virtually all materials from outside of Mexico, usually the US. Surpris-ingly, maquiladoras owned by non-US firms and factories in low-technology in-dustries such as footwear, apparel, glass and toys sourced the most domestically, although they could have benefited from duty-drawbacks. This suggests that it was not the growth of off-shore processing in itself that determined what US firms de-manded during the N A F T A negotiations, as argued by Chase (2003, 145-146), but rather the chance to combine it with protection against firms from third countries. The example of the auto industry is reflected elsewhere. Research into corpo-rate strategies has shown that in the electronics and household appliances indus-60 tries, the integration of Mexico into a North American production network already picked up speed in the late 1980s. The primary aims were increase the volume of Mexican operations to achieve economies of scale, and to integrate Mexican op-erations into the US network to gain efficiency in R & D and marketing (Blank and Haar 1998, 34). In the N A F T A negotiations, the firms strategies described so far translated into often highly specific demands regarding the nature of the liberalization. US firms lobbied negotiators to ensure that the free trade agreement gave them highly preferential access to Mexico for trade and investment. 3.4 The NAFTA Negotiations: A Preference for Protectionism As the preceding discussion has shown, economic integration of the US and Mex-ico was well under way before N A F T A was a realistic prospect. As stressed by one chemical industry CEO in 1992: \"Business is so far ahead of politicians on this one that it almost makes the agreement secondary. To many in our industry, N A F T A is a fait accompli\" (Quoted in Blank and Haar 1998, 2). As a result of these corporate strategies, Mexico became an important export platform for US car manufacturers, while other industries integrated their production vertically by outsourcing labour-intensive stages of production. A free trade agreement would still be beneficial to these industries by locking in Mexico's liberalization. This commitment, however, was apparently unnecessary for attracting FDI from the US, as the discussion above has shown. While the Mexican government had an interest in achieving such a commitment vis-a-vis domestic political opposition (Pastor and Wise 1994), for US manufacturing industries it would theoretically have been easier to lobby the US government to unilaterally reduce the tariffs on the goods these firms wanted to import. Such efforts are common in US trade policy, accounting for a rather open trade policy and forming a counterweight to protectionist interests (Destler, Odell, and Elliott 1987, esp. 43-56). Reducing M F N tariffs on particular items, however, had the disadvantage that non-US man-ufacturers could follow the example set by the US Big Three and use Mexico as an export base to enter the US home market. N A F T A therefore offered a means to raise the barriers for non-NAFTA firms to investing in Mexico. By contrast, the US service industry was mainly interested in market access, since Mexico's outdated infrastructure in telecommunications and underprovision 61 with services in general offered enormous potential. In both sectors, Mexico's re-maining barriers to the rest of the world provided the preferential margin to secure the Mexican investment location. In particular firms from the US manufacturing sector spent considerable effort on lobbying for rules of origin that would provide the necessary protection. 3.4.1 Ru les o f O r i g i n for Automob i l es Negotiations over tariffs for automobiles, auto parts and the related rules of origin became one of the most contentious aspects of the N A F T A negotiations. US ne-gotiators were mainly concerned with protecting the Big Three and their auto parts suppliers from potential European and particularly Japanese competition (Cameron and Tomlin 2000, 91-92; Carlsen and McCarthy 1991). The respec-tive demands of different firms directly reflected their stake in the negotiation, based on their relation to the Mexican and, to a lesser extent, Canadian market and production location. As predicted, the more production was regionally concentrated in North Amer-ica, the higher the rule of origin that a firm sought (proposition la). Ford and Chrysler demanded a rule of origin of 70 percent, since they exported most of their Mexican production, but sourced parts virtually only from within North America. By contrast, General Motors preferred a 60 percent threshold because of its joint venture with Suzuki in Canada, which used Japanese parts (Eden and Molot 1992, 15; Molot 1993, 9). Intermediate goods producers demanded an even stricter ROO (proposition lb). Parts producers asked for a constraining rule of origin of 75 percent, or at a minimum a rule that guaranteed that the \"powertrain\" (engine and transmission) would be wholly produced from North American parts (Inside U.S. Trade, 23 September 1991; Motor and Equipment Manufacturers Associa-tion 1991; Robert 2000, 196; USITC 1991b, L-4). Testifying before the House Subcommittee on Trade in September 1992, Chris-topher M . Bates of the Motor & Equipment Manufacturers Association stated that \"since the outset, we have pushed for a N A F T A agreement structured to (...) pre-vent Mexico or Canada from serving as an export platform to the United States for vehicles and components assembled using high concentrations of non-North American content\" (U.S. Congress 1992, 297). In a 1991 proposal, the Big Three called for the N A F T A rules of origin to: E X P A N D employment and increase the international competi-tiveness of the North American automotive industry. 62 A S S U R E that the benefits of the full N A F T A only accrue to com-panies that have made meaningful manufacturing and research and development commitments in North America. There should be no op-portunity to inflate content levels or manipulate compliance through accounting, cost allocation, or pricing practices\" (Chrysler Corpora-tion 1991, 7-8; cited in Robert 2000, 196). The rule of origin question became closely tied up with a demand by the Big Three to differentiate the phase-out of tariff and non-tariff barriers between incumbent firms and newcomers. Similar to the rules of origin, the aim was to slow down the entrance of new firms that could invest in Mexico. Firms were to be divided into a tier I of incumbents (Chrysler, Ford, G M , Volkswagen and Nissan) and a tier II of newcomers (Inside U.S. Trade, 23 September 1991). Specifically, the Big Three lobbied US negotiators to press for a differential phase-out of the Mexican automotive decrees, an issue initially not well understood by the US trade nego-tiation team (Cameron and Tomlin 2000, 92): \"US negotiators must ensure that the evolving competitive environment under a N A F T A does not disadvantage the position of existing investors in Mexico relative to those who may wish to enter\" (Chrysler Corporation 1991, 3; cited in Robert 2000, 196). Mexico's remaining non-tariff barriers were to be phased out over a fifteen-year transition period, with a more rapid elimination for incumbent firms. US negotiators also pushed for an elimination of the trade-balancing require-ment in the 1989 auto decree, whereby a firm producing in Mexico would have to export 2.5 times the value of its imports (Robert 2000, 192). Finally, the US sought the complete elimination of the auto decrees with respect to parts sup-pliers. In their existing form, the auto decrees did not count the highly efficient maquiladora parts factories as national suppliers when calculating domestic con-tent (Johnson 1993a). This prevented US firms from integrating their Mexican production with its most important local partners, since Mexican auto parts and engines had to be mainly exported to the US. 3.4.2 Banning M F N Rebates for Non-NAFTA Firms For the rules of origin to work as effective protectionist devices, Mexico would also have to be barred from rebating its M F N tariff for non-NAFTA investors. Duty-drawback schemes of this type were an essential part of the Canada-US Auto Pact that was maintained under the CUSFTA, as well as of Mexico's maquiladora scheme. The US negotiation objective was to abolish the maquiladora benefits 63 as quickly as possible, while the Mexican side wanted to retain them as long as possible as an incentive for FDI from non-NAFTA countries. US firms had long criticized that Japanese investors in the maquiladoras, although nominally based in the US, imported most of their parts from Japan rather than sourcing them from the US or Mexican suppliers (Szekely 1991, 20). Mexican negotiators reported that just like in the automotive sector, US industries feared that outsiders would use Mexico as an export platform to enter the US market by using their US subsidiaries to process goods in Mexico (Silverstein 1992, 4). In particular the iron and steel industries, providers of the principal input for the parts used by US auto production in Mexico, \"spearheaded (...) a steady call (...) for the prohibition of duty-drawback programmes as part of NAFTA\" (Maxfield and Shapiro 1998, 87) This confirms the expectation that firms will use an FTA to close any back-doors in the form of M F N rebates (proposition lc), thereby harnessing the host country M F N tariff as measure to protect its foreign operations. 3.4.3 Rapid Liberalization in Services In the service sector, US negotiators pressed for market access at the behest of US firms organized in the Coalition of Services Industries (CSI) (Business Mex-ico Special Edition 1992, 25). Particular strong demands came from banks with a regional base in the southern US states as well as major credit card companies, who lobbied through the American Bankers Association, the Bankers Associa-tion for Trade and Finance (BAFT) and the Texas Bankers Association (United States Banker, June 1992; Sesit 1992). The Mexican retail banking market was \"underbanked,\" with only 21 million of the 48 million Mexicans aged 18 and over having a bank account in 1994, and highly attractive to US banks (USITC 1990, 2-21) who lobbied the negotiators directly and through the Senate Finance Committee (Inside U.S. Trade, 8 February 1991). Despite strong Mexican opposition, US negotiators regarded financial services as an essential component of N A F T A without which the deal would fail (Cameron and Tomlin 2000, 83). Mexico's banks had only recently been re-privatized in 1991 and were generally considered to be undercapitalized and uncompetitive, al-lowing US banks to enter the market with a clear advantage (Chant 1993). US and Canadian financial firms would be uniquely well positioned to offer finan-cial services for foreign investors setting up shop in Mexico and to offer import-export financing. By serving customers all across the continent, financial services providers could therefore achieve economies of scale with regard to back office 64 operations and marketing (Borrego 1991). Even in the case of remaining restric-tions to cross-border trade in services, US firms expected to enter the Mexican market through takeovers of established Mexican banks (Russell 1992). Central to market access was attaining national treatment for investment in services. Unlike in the manufacturing sector, however, the US financial services industry did not specifically prefer a bilateral solution to a sectoral deal or multi-lateral negotiations. In this sense, N A F T A only represented a chance to achieve objectives that US firms also lobbied for in multilateral negotiations. As per proposition 3a, service industries supported bilateral deals as opportunity for lib-eralization. Similar objectives drove negotiations with regard to telecommunications. Mex-ico's infrastructure was outdated, causing analysts to expect strong demand for an upgrading and the prospect of future investment (Frischkorn 1993). Market access consideration therefore guided lobbying efforts of US firms (USITC 1990, 2-9), although much of it was forward-looking toward the development of future tech-nologies (Messmer 1992), since the Mexican state telecommunications company T E L M E X had already been privatized in 1990. Directly linked to telecommunica-tions was the issue of government procurement, which would offer enormous op-portunities for the provision of services to the industries that remained in the hands of the Mexican government (Cameron and Tomlin 2000, 93; Messmer 1992). Investment liberalization in non-service sectors was closely tied to the ques-tion of guarantees for investors (Business Mexico Special Edition 1992,24). While the negotiations faced considerable difficulties over the language of Chapter 11 in terms of its expropriation and compensation clauses (Cameron and Tomlin 2000, 112), the more important issue in terms of NAFTA's preferential liberalization was the question of national treatment. Initially, the non-finance US business commu-nity was unclear whether it sought unconditional M F N treatment by US investors in Mexico, or whether preferential liberalization of sectors was more desirable. Except for sectors with differential liberalization targets as outlined before, US negotiations pressed for an schedule for the immediate liberalization of sectors that were previously closed to FDI (Business Mexico Special Edition 1992, 24). As argued in this section, US manufacturing firms with export-oriented FDI in Mexico showed a strong preference for protectionism in the disguise of an FTA. In particular the Big Three auto manufacturers and their rent chain of parts pro-ducers sought the strictest rules of origin. This reflected their strategic position vis-a-vis competitors: once Mexico became feasible as an export platform for the production of automobiles, it would likely attract investment from European and Japanese firms. Rules of origin would impose higher costs on these firms if they 65 chose to start up production in Mexico. But despite the differential phase-out clauses of the Mexican host country requirements for incumbents and newcom-ers, non-NAFTA incumbents like Nissan and Volkswagen would be hurt as well. Should they choose to export to the US market, strict rules of origin would impose higher costs on them if they decided to buy from suppliers outside of NAFTA. Through concerted lobbying efforts, US firms were able to achieve many of these goals in the actual negotiations. 3.4.4 Preferential Aspects of N A F T A and Multi lateral Commitments How much potential for trade and investment diversion was built into NAFTA? As the above discussion has shown, US manufacturing firms were intent on maintain-ing exclusivity. In the automotive sector, the final outcome in the negotiations was a 62.5 percent rule, a compromise between the US and the Mexican and Canadian negotiators who had sought a lower level. In addition to the restrictive rules, the US Big Three attained the requested two-tiered phase-out period of the remain-ing Mexican host country measures\u2014faster for incumbents than for new entrants. Firms that had existing operations in Mexico (the Big Three, Volkswagen and Nissan) were allowed to produce vehicles with a lower domestic content than new operations by competitors (Inside U.S. Trade, 23 September 1991; Robert 2000, 161, 194). Specifically, incumbents did not have to satisfy the 36 percent local (Mexican) content quota required under the 1989 auto decree as long as they ful-filled NAFTA's rule of origin, but could use the 1992 model year's content quota (Eden and Molot 1993). These factors jointly worked to the disadvantage of new entrants. N A F T A also included the phasing out of the maquiladora benefits. The in-tended side effect was raising the costs for non-NAFTA producers. N A F T A gran-ted a temporary extension of duty-drawback programmes, but limited this to the lower of the two M F N tariffs.2 Since the US tariff averages less than 4 percent, the higher Mexican tariff effectively raised the cost of importing intermediate goods by 10-15 percent. As of January 2001, the duty-drawback offered by Mexico has been completely eliminated and the M F N tariff is levied by the US on goods that do not meet NAFTA's rule of origin requirements (Maxfield and Shapiro 1998, 88). In 1994, maquiladoras could only sell domestically up to the equivalent of 55 percent of their exports. The ceiling was raised gradually by 5 percent per year, 2 NAFTA, art. 303(2). 66 allowing US manufacturers to increasingly use maquiladoras as suppliers for as-sembly in Mexico (Robert 2000, 163). US manufacturers could therefore benefit from access to a world-class supply industry in Mexico. The potentially negative effect of rules of origin on outsiders was by no means unexpected. In a 1991 report to Congress, the USITC clearly cited the implica-tions for foreign producers based in Mexico, especially automotive and electron-ics firms, as well as their concerns about the possible discriminatory effect of an FTA (USITC 1991a, 4-19, 4-21, 4-26). The same report noted that because of the small size of the Mexican market, the trade creating effect of an FTA would be concentrated in intermediate goods serving US investment in Mexico (United States International Trade Commission 1991a, 4-21, 4-25). A second USITC re-port pointed out that the investment liberalization clauses of N A F T A 3 enabled US automakers to wholly own parts suppliers in Mexico, which should further increase their competitiveness vis-a-vis firms from third countries (USITC 1993, 4-7). Ironically, the automotive firms with the greatest unrealized economies of scale were Nissan and Volkswagen. Meeting the demands of the US Big Three and their suppliers in the negotiations of N A F T A implied that the agreement would discriminate not only against newcomers, but also against incumbents that im-ported parts and capital goods from outside of NAFTA, unless Mexico could be motivated to lower its remaining tariff and non-tariff barriers. In contrast to regionally integrated industries, US industries that preferred to source globally opposed tight rules of origin. Most prominently, the computer industry procured most inputs from Asia and thus preferred free trade to regional integration (Wonnacott 1993, 14). I B M specifically opposed the inclusion of hard disks as a requirement for computers to qualify as North American (Cameron and Tomlin 2000, 90), since most hard disks were manufactured in Singapore. However, it did not oppose the requirement that cathode-ray tubes for monitors be produced in North America\u2014Japanese firms assembled monitors in Mexico with imported cathode-ray tubes (Johnson 1993b, 12). The outcome was selectively protectionist, requiring high North American content for some products, but not for others. The USITC states: \"although the rules of origin are complex, they reportedly reflect the needs and desires of the domestic [US] industry\" (USITC 1993, 5-2, footnote 5). 3 NAFTA, Annex I, Reservations for Existing Measures and Liberalization Commitments, Schedule of Mexico, I-M-32. 67 In the services sector, the potential for investment diversion appeared as an unintended consequence of the preferential liberalization, since it depended pri-marily on the remaining barriers imposed by Mexico. In the financial services in-dustry, N A F T A guarantees national treatment for financial intermediaries, allow-ing US and Canadian banks to establish subsidiaries in Mexico. Mexico reserved the right to approve on a case-by-case basis the ownership of Mexican banks or securities firms by U.S. and Canadian commercial or industrial corporations in accordance with its 1993 law on foreign investment (see Appendix II). N A F T A also guarantees the right to purchase financial services in another N A F T A country. Market access was originally limited to some extent by a cap on the share of the Mexican banking system that may be owned by US and Canadian banks: When N A F T A entered into force, the aggregate capital of U.S. and Canadian banks op-erating in Mexico was limited to 8 percent of the capital of the Mexican banking system as a whole. The limit was scheduled to increase gradually until reaching 15 percent in 1999 and completely abolished on January 1, 2000, although Mex-ico reserved the right to restrict increases beyond 25 percent until 2006, when all restrictions are eliminated. Caps on the capital share of individual banks were allowed until 2000, with some limitations on acquisitions. Despite these remain-ing limitations, US banks did not hesitate to enter the market on a grand scale, from retail banking offered by Bank of Boston, Bank One and Citibank to spe-cialized financial services for certain industries provided by Chemical Bank and Nationsbank (Ioannou 1994). Even these limited caps fell in the wake of the 1994 peso crisis, as the Mex-ican government was forced to open the sector fully to FDI. In February 1995, the schedule for opening the financial services sector was accelerated to allow full takeovers of Mexican banks, provided the assets represent less than six percent of the entire financial system (Santfn Quiroz 2001, 221). The preferential liberal-ization under N A F T A contrasts with Mexico's unilateral and multilateral (GATS) commitments\u2014even the liberalized 1993 Mexican law on foreign investment re-tained limits of 30 percent ownership of financial groups and 49 percent owner-ship of banks, brokerages, and insurance companies (see Appendix II), compared to 100 percent under NAFTA. Non-NAFTA banks are restricted in their operation of Mexican banks because of a limit of 10-20 percent on individual holdings and the requirement of effective control in Mexican hands (Stephenson 2002, 200). In tourism as well as a range of professional services, N A F T A allows 100 per-cent foreign ownership compared to 49 percent under Mexico's law. Investment in services and exports from the US to Mexico increased considerably, as shown in table 3.6 below for the period 1982 to 1999, the most recent year for which data 68 is available. In particular financial services investment and increased their share of total investment. Manufacturing investment continued unabated since the late 1980s, as shown earlier, with the exception of a brief period following the 1994 Mexican currency crisis. Services exports* FDI stock in financial services FDI in other services 1985 259 62 1986 1903 250 232 1987 1577 310 271 1988 1222 120 342 1989 1787 279 1990 2393 619 291 1991 2998 670 317 1992 3354 795 335 1993 3743 2,106 233 1994 4475 2,243 387 1995 4469 2,101 545 1996 4629 2,612 609 1997 5267 3,911 1998 5527 4,842 1999 6107 4,135 618 . = Data unavailable * Services exports less travel, passenger transportation, royalty payments and licence fees. Table 3.6: US Services Exports and FDI to Mexico. Source: United States Bureau of Economic Analysis, Author's calculations In the manufacturing sector, the interaction of high rules of origin with Mex-ico's tariff rates implied highly discriminatory trade barriers by OECD country standards. Since Mexico's tariff rates under the WTO are usually bound at 50 percent and, with the failure of the Millenium round of the WTO, no further mul-tilateral liberalization occurred after N A F T A and the Uruguay Round. Instead, Mexico even raised some M F N tariffs to higher levels, as shown in table 3.7 be-low. This compares with a gradual phasing out of tariffs in N A F T A during the same period. By 1999, Mexico had eliminated tariffs on 65 percent of good originating in the US or Canada. The phase-out is to be completed by 2009, at which point over 95 percent of Mexican imports from the NAFTA countries will be tariff-free. 69 1994 1995 1996 1997 1998 1999 Capital goods 11.7 11.7 11.5 11.4 11.4 14.5 Intermediate 11.4 11.8 11.3 11.2 11.2 13.9 goods Consumer goods 17.2 24.8 25 24.9 24.5 29.3 Total 12.4 13.7 13.3 13.3 13.2 16.1 Table 3.7: Average MFN Tariff Applied by Mexico. Source: Preufie (2000) 3.5 Alternative Accounts: The Benign View of NAFTA Did the US Big Three primarily lobby for N A F T A because it would offer them the ability to fully integrate their production in North America, as argued by Chase (2003) and Johnson (1993a)? The export figures for US car manufacturers and the increase in investment prior to N A F T A suggest otherwise. If the Mexican host country measures had been really that onerous, it is unlikely that US firms would have increased their investment at all. The early turn of the trend toward more FDI in Mexico indicates that US firms may have found the tariff reduction in N A F T A beneficial, but did not see it as a precondition for production in Mexico. Moreover, although the volume of re-exports of US firms to the home market kept growing, the composition of these exports did not change much following the conclusion of the agreement. It seems more likely that US manufacturers chose to refer to their onerous legacy of investment in Mexico because it did not appear outright protectionist. Conveniently, the same rules would apply to Japanese factories in Canada. Here US manufacturers sought to avoid a repetition of intrusions like Honda's production of Civic models in Ontario that sold well in the US market.4 Did N A F T A offer protection against the loss of the main export market, in line with the explanation put forth by Mansfield and Reinhardt (2003)? For the US, the Mexican market is simply to small for this argument to be credible. Yet even for Mexico, the importance of FDI from the US undermines this argument.5 4In 1992, US Customs had ruled that Honda Civics assembled in Ontario between January 1989 and March 1990 did not fulfil the origin requirement of the CUSFTA, set at 50 percent. New York Times, 17 June 1991. 5If securing market access had indeed been Mexico's main objective, NAFTA was a mixed bag. Just like the CUSFTA, the agreement preserves the application of domestic laws on anti-dumping and countervailing duties. Following the CUSFTA precedent, NAFTA created the institution of binational panels to settle disputes over the propriety of the application of domestic trade remedy laws. But Mexico did not even have such laws in place at the time of negotiation, and thus had 70 Given the importance of US FDI as source of Mexican exports, it is unlikely that secure market access for conventional exports was high on Mexico's list of priorities. Since US firms in Mexico produced the most important goods for export to the home market, labour remained as constituency whose product competed with imports from Mexico. Yet only firms, not labour unions, can raise dumping cases in the US that are legally actionable. Securing market access against US trade remedy laws would have been much more likely in multilateral negotiations, in which both Canada and Mexico would have found allies. As the Doha round negotiations on these issues show, both countries pursued this route later. 3.6 Conclusion: Raising the Barriers to Market Entry Mexico's reintegration into the world economy and the subsequent negotiation of N A F T A offered US firms broad opportunities to create an export platform for manufacturing industries that benefited from a low wage, educated labour force in close proximity to the home market. As the case study shows, the restructur-ing of US direct investment in Mexico towards exports was well underway before N A F T A was a serious policy option, suggesting that it would have continued even without a formal free trade agreement. The negotiation of NAFTA, however, of-fered US industries the chance to lobby for rules of origin that would potentially raise the costs for outsiders considerably when they invested in Mexico to serve the US market. Unless Mexico could be motivated to reduce its tariff applied to the intermediate and capital goods imports of these investors, they would be at a severe competitive disadvantage vis-a-vis US and Canadian firms. The case also shows that in markets with high tariff and non-tariff barriers, or markets in which investment is a requirement for access, the investment and ser-vices provisions of a free trade agreement can be highly discriminatory in nature. The preferential liberalization of the Mexican banking market therefore offered significant first-mover advantages. However, it is unlikely that either lobbyists or negotiators in the US anticipated that within a few years of the free trade agree-ment between entering into force, both the E U and Japan would seek both relief for their multinational firms active in Mexico and market access comparable to NAFTA. to commit to an overhaul of its domestic legislation, although it did press for changes in US trade laws during the negotiations (Cameron and Tomlin 2000, 47-49). 71 Since Mexico did not lower its M F N tariffs, the rules of origin negotiated in N A F T A could exert their full force. Investors from non-member countries and their intermediate goods suppliers were targeted to prevent Mexico from becom-ing an export platform to the US market. To be attractive to US firms, N A F T A had to divert trade and investment. If European and Japanese firms wanted to use Mexico as an export platform to produce for the US market, they would incur a higher cost than their US competitors. 72 Chapter 4 A Beachhead in the North American Market For the Mexican government, a central objective of N A F T A was to attract foreign direct investment\u2014not just from the United States, but also from other countries whose firms would produce goods in Mexico for the US and Canadian market. As shown in Chapter 3, if this objective were to be attained, the worst fear of US com-panies would come true. Mexico would function as a beachhead within NAFTA. To forestall or at least delay such an outcome, strict rules of origin became one of the key negotiating goals of the US. Once strict rules of origin were in place, outsiders could react in at least three ways. First, they could de-invest and pull out of Mexico. Manufacturing outdated, import-substitution era goods at inefficient scales for the Mexican market alone would become infeasible with US firms producing and selling similar products in Mexico and the US. Second, outsider firms could fulfill Mexican hopes and invest more, jumping over rules of origin instead of tariffs, but primarily buy parts from North American suppliers. Finally, they could invest more, secure Mexico as a beachhead, but still choose to procure their supplies from outside NAFTA, whether for reasons of firm strategy, choice of technology, or cost. If the latter turned out the be the case, it would have two observable implica-tions. First, Mexico would attract more FDI from non-NAFTA countries after the agreement came into force than other countries, all else being equal. Second, an increase in the FDI stock of non-NAFTA firms in Mexico would raise these firms' stake in tariff-free access to Mexico for the goods necessary for their production, either through a reduction of Mexican M F N tariffs or a bilateral agreement. In turn, it would motivate non-NAFTA firms with FDI in Mexico to action. This chapter conducts an econometric test of the first implication. Chapter 5 turns to the second in two case studies. The first section of the chapter presents an overview over the model used in the analysis, the standard gravity model of international trade, and its application to foreign direct investment. It then discusses model specification, how to esti-73 mate the effect of membership in NAFTA, and data sources. The second section covers estimation procedures and presents the results. Section three concludes. Appendix III contains summary statistics. 4.1 The Gravity Model\u2014An Overview As the workhorse of empirical trade economics, the gravity model goes back over four decades to Tinbergen's (1962) work. The model is based on a simple anal-ogy to the law of gravity in physics: A l l else equal, trade between two countries should be positively related to the size of their economies and negatively to the distance between them. By adding control variables that have been empirically shown to influence trade, the gravity equation becomes a baseline model that pre-dicts typical trade flows between countries. Anderson and Wincoop's work (2003) represents the current state of the art specification of the gravity model. Gravity models are consistent with various theories of international trade. Baier (2001) derives a gravity model from a model of monopolistic competition; Deardorff (1995) from perfectly competitive markets; Feenstra et al. (2001) from a model of reciprocal dumping with homogenous goods. Most importantly, grav-ity models have recently been expanded to analyze flows of foreign direct invest-ment (Di Mauro 2000; Loungani, Mody, and Razin 2002; Blomstrom and Kokko 1997; Levy Yeyati, Stein, and Daude 2003; Balasubramanyam, Sapsford, and Griffiths 2002). Although a complete derivation of the gravity model from theo-ries of FDI is beyond the scope and purpose of this study, the next section outlines the key elements, drawing on the theory of vertical FDI on which this work rests as well as other frameworks. 4.1.1 Establishing the Baseline Model As mentioned in Chapter 2, the traditional or \"eclectic\" theory of foreign direct in-vestment focused on the puzzle of exports versus FDI. Investing overseas imposes costs and creates risks: different linguistic and cultural backgrounds, secondment of personnel abroad, unfamiliar legal and regulatory frameworks, exchange rate changes and host country governments that may suddenly decide to expropriate foreign firms. Dunning's (1977, 2000) original theory focused on the \"OLI\" ad-vantages of ownership of patents and trade secrets relative to local firms, the lo-cational advantages of market access and different factor endowments, and the internalization advantages of affiliated production over licensing to protect in-74 tellectual property and prevent asymmetries of information. Unfortunately, this theory relies on many practically unobservable factors, making it difficult to build empirical models. More recently, Helpman (1984) and Helpman and Krugman (1985) have fo-cused on the ownership and locational advantages, theorizing that firms make in-vestment decisions based on capital intensity and relative factor prices\u2014the \"fac-tor proportions hypothesis.\" This leads to a split of production processes between capital-intensive activities like research and marketing that take place in capital-abundant countries, and more labour-intensive production in plants abroad, as de-scribed in Chapter 2. Although highly useful in explaining flows of manufacturing FDI from de-veloped to developing countries, the theory of vertical FDI employed in this study can only explain a limited share of FDI. Almost 90 percent of global foreign direct investment flows from one OECD country into another, suggesting that it is pri-marily market seeking. Markusen and Venables (1998) focus on such relocation of production to facilitate distribution, hypothesizing that FDI between two countries will grow with an increase in the size of their economies and world income and as they become more similar in terms GDP per capita, reflecting the importance of intra-industry trade. In their model, two kinds of economies of scale motivate foreign direct investment. Firm-level scale economies give multinationals an ad-vantage over domestic firms. But the choice between exports and FDI depends on the relation of trade costs (including tariffs) to plant-level economies of scale. As trade costs increase, firms prefer FDI, as long as plant-level scale economies exist, but do not outweigh trade costs. Brainard (1997) has found empirical support for this model, sometimes referred to as \"proximity-concentration hypothesis.\" Note that combined, these factors may have an influence that reaches all the way to corporate strategies: firms may invest in a developing country to benefit from dif-ferent factor endowments, but wil l still want to achieve economies of scale at the plant level, as in the case of FDI in Mexico in the automobile industry described in Chapters 3 and 5. In addition to these basic driving forces, a number of intuitive factors influence FDI, in particular the role of language, similarities in regulatory frameworks, and distance. Likewise, tariff and non-tariff barriers may induce FDI, for example in circumventing \"voluntary\" export restraints, when FDI is market-seeking. In sum, the specification of the gravity models of FDI and trade are nearly identical, since the principal independent and control variables are the same as those presumed to influence international trade. 75 In analogy to mass, gross domestic product (GDP) enters the equation in one of two forms: as product of the GDP of both countries, if total trade or FDI be-tween the partners is taken as dependent variable (Rose 2004), or as sum of both countries GDPs if flows of trade or investment from one country to the other are taken. When estimating FDI flows, this variable reflects the framework of Markusen and Venables (1998). GDP per capita proxies for different factor en-dowments as per Helpman and Krugman (1985) in the absence of reliable data for more direct measures for enough countries, such as capital per worker or gross fixed capital formation in relation to working population. This simple measure assumes that there are only two factors of production, capital and labour, as in the stylized theory of vertical FDI. Distance is usually measured as between the biggest or capital cities. Physi-cal distance increases transport costs and entails a higher costs of communication with firm affiliates abroad. On the other hand, FDI could substitute for trade if the cost of exporting is much higher than the cost of management from a dis-tance (Loungani and Razin, 2002). Since multinational operations have become increasingly integrated globally, this substitution effect is probably limited and in turn obscured by the attendant trade in intermediate goods created by FDI. Many authors (Boisso and Ferrantino 1997; Fink and Primo Braga 1999; Soloaga and Winters 2001) include variables for a common language. Rose (2004) adds variables for adjacency and colonial linkages. The former follows the notion of neighbouring countries as natural economic partners, whereas the latter controls for spheres of influence as well as similarities among far-away countries such as Canada and New Zealand with a common colonizer. Having established the base-line model that takes economic, geographic and historical or path-dependent vari-ables into account, the effect of political institutions on trade and FDI becomes discernible. 4.1.2 Estimating the Effects of Preferential Trading Arrangements The simplest approach to estimate PTA effects is to to add a dummy variable for membership in a particular agreement (Aitken 1973). A significant coefficient on the variable implies that the PTA caused them to trade differently than what can be expected, based on their economic characteristics and the average behaviour of the other countries in the sample. Frankel (1997) and Bayoumi (1997) add two dummy variables, one if both countries are member of a PTA, a second if only 76 one country is a member. A negative coefficient on the second dummy variable suggests that the PTA has led to trade diversion, implying that a country pair in which only one member is participating in a PTA is trading less with each other than expected based on the average behaviour of countries in the sample. Finally, PTAs could create or divert exports and imports differently. To estimate this effect, Soloaga and Winters (2001), Adams et al. (2003) and Rose (2004) add three dummy variables, one dummy if only the exporting country is a member but the importer is not, one dummy if only the importing country is a member but the exporter is not, and a third if both are members. A negative coefficient on the first dummy variable equals a diversion of exports, i.e. the PTA-member exports less to non-members, while a negative coefficient on the second dummy implies diversion of imports. A positive coefficient on the third dummy variable suggests that total trade within the PTAs is more than a country pair with the same characteristics is estimated to trade without a PTA. When estimating trade flows, this approach is problematic, because treaty clauses are not applied immediately and tariff and non-tariff barriers phased out over time after then entry into force of a PTA. In NAFTA, many tariffs are only completely eliminated after ten years. By constrast, when estimating the effect of PTAs on foreign investment, using a simple dummy variable is likely to be ade-quate: investment decisions are not based on a immediate response to tariff-related changes of prices, but rather on longer term effects of a trade agreement. Since this study focuses primarily on the diversion of investment from non-NAFTA mem-bers to Mexico, I use the simple dummy variable approach. 4.1.3 Model Specification To avoid misspecification, the gravity model employed here includes as many con-trol variables as possible that past studies have identified as significantly affecting FDI flows patterns. The gravity model used in this study is specified as follows, with variable definitions given in table 4.1.3: ln FDIijt = di + ij + Xt + 0o + 0i In GDPSUMi3 + 02 In DGDPcapij +\/?3 In distij + 0iComcol + d^colonyij + fi\u00a7borderid- + Q^ang^ +08ECtoMEX + faJPNtoMEX + 0wROWtoMEX + e 77 In is logarithmic transformation FDIijt is the value of the stock of outward FDI from country i in coun-try j in year t is unobserved fixed effects in country i TJ is unobserved fixed effects in country j \\ is unobserved fixed effects in year t A ) is the common intercept SUMGDPijt is the sum of real GDP of i and j in year t DGDPcapijt is the absolute value of the difference in real GDP per capita between i and j in year t DKPWijt is the absolute difference in capital per worker, used instead of DGDPcapijt in an alternative specification is the distance between the two capital cities of i and j colonyij is a dummy for former colonial relationship comcolij is a dummy for common colonizer borderij is a dummy for a common land border langij is a dummy for a common language TRADEOPENi is the revealed trade openness of country i e x P o r ^ + m i P o r t s TRADEOPENj . . . i J \u2022 J t t \u2022 exports+imports is the revealed trade openness of country j \u2014 c \u2014 G D P -ECtoMEXijt is a dummy that takes on the value of 1 if the FDI home country is in the European Community, the host country is Mexico and NAFTA is in effect JPNtoMEXijt is a dummy that takes on the value of 1 if the FDI home country is Japan, the host country is Mexico, and NAFTA is in effect ROWtoMEXijt is a dummy that takes on the value of 1 if the FDI home country is in the rest of the world excluding the EU, Japan, Canada and the US, the host country is Mexico, and NAFTA is in effect Sijt is an error term, assumed to be normally distributed Table 4.1: Variable Definitions 78 As shown in table 4.1.3,1 include three different dummy variables to estimate the effect of NAFTA. The first dummy (ECtoMEX) captures the effect of N A F T A on FDI from the European Community member countries to Mexico. The sec-ond dummy (JPNtoMEX) measures the effect on Japanese FDI to Mexico. For comparison purposes, I also include a dummy to capture the effect of N A F T A on FDI from the rest of the world to Mexico. A positive and significant coeffi-cient on these variable indicates that N A F T A indeed attracted more FDI from the EC, Japan, or the rest of the world than could be expected based on the average behaviour of the countries in the sample. 4.1.4 D a t a Sources Data on bilateral FDI stocks and flows is difficult to obtain for many countries and suffers from various inconsistencies in measurement. FDI data for this study is drawn from the SourceOECD database on outward FDI stock from all OECD members and supplemented by data from the U N C T A D World Investment D i -rectory (2004). The latter is available electronically, but at an exorbitant price, requiring the manual collection from the paper edition. Stock data is generally superior to flows, which often show wild swings caused by individual projects. The quasi-standard OECD defines direct investment as a minimum 10 percent stake and operational control, but does not allow for any depreciation. Moreover, not all O E C D members adhere fully to this definition. The U K , for example, does not include the substantial investment in financial services in its official data. The ideal dependent variable for this study, production of affiliates of multinational firms, is only measured by the US and Sweden. Sectoral FDI data is only available for very few countries, making it impossible to separate investment in services and manufacturing for a sufficient number of countries. These data limitations qualify any study of FDI covering many countries. Availability of FDI data also restricted the sample in this study to the countries listed in table 4.2, although the sample size of 108 countries is still markedly bigger than in most previous studies. Ideally, the study would also have included tariff data, but it is only available for a very short time period for a sufficient number of countries. GDP is measured as national income in constant 1995 US$, obtained from the World Bank's World Development Indicators (2004). GDP per capita is calculated by dividing GDP by population, the latter sourced from official estimates of the United Nations Population Division. Population estimates for 2002 are middle scenario projections. Alternatively, I use the absolute difference in capital per worker to capture different factor proportions, taken from Baier et al. (2005) and 79 Albania Germany Nigeria Algeria Ghana Norway Angola Greece Oman Argentina Guatemala Pakistan Australia Guinea Panama Austria Guyana Papua New Guinea Bahamas Honduras Paraguay Bahrain Hong Kong (China) Peru Bangladesh Hungary Philippines Barbados Iceland Poland Belgium-Luxembourg India Portugal Belize Indonesia Saudi Arabia Bermuda Iran Seychelles Bolivia Ireland Singapore Brazil Israel South Africa Bulgaria Italy Spain Burkina Faso Jamaica Sri Lanka Cameroon Japan Sudan Canada Jordan Suriname Chile Kenya Sweden China Korea Switzerland Colombia Kuwait Syria Congo Lebanon Tanzania Costa Rica Luxembourg Thailand Cote d'lvoire Malawi Trinidad and Tobago Cyprus Malaysia Tunisia Denmark Mali Turkey Dominican Republic Malta Uganda Ecuador Mauritius United Arab Emirates Egypt Mexico United Kingdom El Salvador Mongolia United States Ethiopia Morocco Uruguay Fiji Nepal Venezuela Finland Netherlands Vietnam France New Zealand Zambia Gabon Nicaragua Zimbabwe Table 4.2: Countries in the sample. Countries in bold are both FDI home and host countries. 80 extrapolated to fill in missing years. Another possible measure would have been GDP per worker, but data on working population in developing countries is limited and notoriously imprecise. A l l variables measured in US$ are taken in logarithmic form to obtain co-efficient estimates from which percentage changes can be easily calculated as %Ay = 100 \u2022 [exp0Ax2) - 1] (Wooldridge 2003, 187). This also mitigates prob-lems of skewed and heteroskedastic conditional distributions of these variables. FDI stock, exports, imports, GDP and GDP per capita are converted to millions of US$ using IMF exchange rate data when necessary, although most U N C T A D data is already reported in current US$. A l l dollar values are deflated to 1995 dollars using the US consumer price index. Werner Antweiler provided data on common language, adjacency and distance calculated on the basis of the Geo-Data Atlas (Kurian 1989). Data on colonial re-lationships and common colonizer were coded by hand based on the Encyclopedia Britannica, counting both actual colonies and protectorates such as Afghanistan as part of the British Empire. Ideally, the model would also incorporate a measure of openness to FDI. Unfortunately, such measures are not available for a suffi-cient number of countries and years. The World Bank has published investment climate reviews for a broad range of countries, but the data only goes back to 2000.1 Its \"governance data\" covers fewer countries and reaches back to 1996.2. The best-known index, the World Competitiveness Yearbook (2004), only covers about 40 mostly developed countries and is inconsistent in its survey methodol-ogy before 1994. Given these circumstances, the best alternative is to include \"revealed trade openness,\" total trade divided by GDP. Openness to trade directly influences the profitability of foreign direct investment by affecting the cost of intermediate goods exports, as outlined in Chapter 2. Moreover, revealed trade openness is likely to be highly correlated with openness to FDI. 4.2 Estimation A particular prevalent problem of gravity models is that according in official statis-tics, most countries report an outward FDI stock of zero in the partner country. This often reflects values that are too small to be registered. As a result, 30^10 percent of the values of the dependent variable are zero. Even this figure probably underestimates the unequal distribution of FDI between different countries: many 'http:\/\/www.ifc.org\/ifcext\/economics.nsf\/Content\/IC-InvestmentClimateAssessments 2http:\/\/www. worldbank.org\/wbi\/governance\/data.html 81 countries that do not report any FDI figures to UNCTAD probably do not attract any foreign investment at all. If the dependent variable were taken in level from, this would present a problem for standard regression techniques: OLS would pre-dict negative FDI values for some partner countries. In addition, large dollar val-ues in level form lead to considerable heteroskedasticity problems that are largely resolved by taking the log. Yet in log form, the zeros simply drop out, omitting possibly important information. I therefore follow Eichengreen and Irwin (1995)3 in expressing the dependent variable as ln(l + value). Zero observations are still zeros, since Inl = 0, but for small values of FDI, ln(l + FDI) \u00ab FDI, whereas for large values ln(l + FDI) \u00ab InFDI. This specification preserves the useful semi-elasticity interpretation of the log-log model. Since the dependent variable is still censored, the proper specification is a tobit model. The tobit model assumes the existence of a latent variable, which captures the aforementioned \"threshold effect\" in official trade and FDI statistics: x'iP + Si 0 if y* < 0 Vi if y ; > o . Tobit models for panel data with random effects are commonly implemented in statistical software packages, although they require an extremely strong assump-tion: the random effects have to be uncorrelated with all included variables for the estimator to be consistent. Often, a fixed effects estimator is therefore prefer-able. When specifying the gravity model, Matyas (1997, 1998) proposes the use of three separate fixed effects for home and host country as well as years. Using country-specific fixed effects also proxies for the \"multilateral resistance term,\" (Anderson and Wincoop 2003, 180), which captures the fact that trade is more important for smaller than for bigger countries, and in turn affects the importance of trade barriers. Since nonlinear fixed effects models were generally held to be biased because of the \"incidental parameters problem\" identified by Neyman and Scott (1948), Adams et al. (2003) is so far the only gravity model study that used this estimator. The bias of fixed effects tobit models, however, is directly depen-dent on the size of T, as shown by Greene (2002). With low, fixed values for T, the bias is substantial, but it quickly drops and is already very small (a factor of 1.008) when T = 10. Moreover, the estimator appears to quickly lose any inefficiency 3See also Anderson and Smith (1999); Amin, Hamid, and Saad (2004); Lewer and Terry (2002); and Levy Yeyati, Stein, and Daude (2003). 82 introduced by the fixed effects as T and N grow. With T between 16 and 21 for on average over 700 annual cross-sections in this study, the use of a fixed effects tobit model appears permissible. Hence, I include the full set of dummy variables for countries and years. For comparison purposes, I also estimate the model in a random effects spec-ification. Although the coefficient estimates were broadly similar, the findings could not be seen as reliable. First, since random-effects tobit models are calcu-lated by a Gauss-Hermite quadrature approach, they are sensitive to large panel sizes and the within-panel correlation. Re-running the regression with different number of quadrature points resulted in marked changes in the coefficient. Sec-ond, a Hausman test (1978) indicated that the random effects assumptions were not met. Finally, I use scaled OLS regression to double-check the result. Scaled OLS obtains an approximation of the tobit estimates by dividing the coefficient esti-mates by the share of non-zero observations in the sample (Greene 2003, 766). Naturally, the fixed effects dummies can be included without any problems. 4.2.1 Estimation Results Table 4.3 shows the actual estimation results for the coefficients. In the baseline specification (1), all coefficients have the expected signs. The bigger two coun-tries in terms of combined income, the more bilateral FDI is undertaken, with a particularly strong effect of the GDP of the home country. By contrast, vertical FDI seems to be much less prominent than assumed in the Helpman and Krug-man (1985) framework. Differences in GDP per capita in fact discourage FDI according to the estimation results shown here, pointing to the overwhelming im-portance of horizontal FDI in global terms. The dummy variables for colony, common colonizer and common language have positive signs. The negative sign of the coefficient for a common land border indicates that firms may chose to ex-port rather than invest in neighbouring countries.To save space, the fixed effects dummy coefficient are not shown. Specification (2) introduces the dummy variables for the effect of NAFTA. Clearly, the agreement had a strong effect on FDI from the European Community. The results for Japanese FDI to Mexico after N A F T A are unclear; the coefficient is insignificant with a sizeable standard error, but this may simply reflect an insuf-ficient number of data points. Interestingly, the coefficient of the dummy variable for FDI from the rest of the world is negative and insignificant. If N A F T A at-tracted beachhead FDI, then most of it seems to have come from the European 83 Community. Models (3) and (4) reestimate the model as in rescaled OLS form. The results do not differ substantively from the tobit estimates, except for the ef-fect of the joint GDP of both countries, which becomes negative and insignificant. Table 4.4 shows the results of re-estimation of the model with differences in cap-Dependent variable: InFDI (1) Fixed effects tobit (2) Fixed effects tobit (3) Scaled OLS (4) Scaled OLS InRGDPl 4.361*** 4.435*** 3 910*** 3.970*** (0.975) (0.975) (0.822) (0.822) lnRGDP2 1.054* 1.102* 0.903* 0.942* (0.609) (0.610) (0.519) (0.520) InDGDPcap -0.220*** -0.223*** -0.237*** -0.240*** (0.053) (0.053) (0.045) (0.045) lndist -2.161*** -2.172*** -1.866*** -1.876*** (0.091) (0.091) (0.077) (0.077) comcol 0.740** 0.718** 0.690*** 0.672*** (0.308) (0.308) (0.258) (0.258) colony 0.582* 0.589* 0.555* 0.561* (0.354) (0.354) (0.302) (0.302) border -1.062*** -1.064*** -0.648*** -0.650*** (0.249) (0.249) (0.211) (0.211) lang 2.431*** 2 427*** 2.081*** 2.078*** (0.244) (0.244) (0.207) (0.207) TRADEOPEN1 0.035*** 0.035*** 0.022*** 0.022*** (0.010) (0.010) (0.008) (0.008) TRADEOPEN2 0.014*** 0.014*** 0.012*** 0.012*** (0.004) (0.004) (0.003) (0.003) ECtoMEX 2 949*** 2.524*** (0.994) (0.851) JPNtoMEX 1.305 1.055 (2.261) (1.944) ROWtoMEX -0.390 -0.378 (0.954) (0.800) Constant -109.798*** -112.540*** -95.142*** -97.244*** (27.506) (27.512) (22.147) (22.153) N 13183 13183 13183 13183 R-squared 0.58 0.58 Standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% Table 4.3: Estimation results for tobit and scaled OLS model with differences in GDP per Capita, coefficient estimates. Fixed effects dummy variables not shown. 84 ital per worker instead of GDP per capita. Columns (5) and (6) show the results for the fixed-effects tobit, (7) and (8) for the scaled OLS specification. Although the sample size is somewhat smaller, none of the coefficient estimates changes substantially. Finally, I re-estimate model in which I limit the sample to country Dependent variable: InFDI (5) (6) (7) (8) Fixed effects tobit Fixed effects tobit Scaled OLS Scaled OLS InRGDPl 4.637*** 4.751*** 4139*** 4.234*** (0.936) (0.936) (0.810) (0.810) lnRGDP2 1.273** 1.308** 1.126** 1.157** (6.586) (0.587) (0.511) (0.512) InRDKPW -0.178*** -0.181*** -0.190*** -0.192*** (0.049) (0.049) (0.042) (0.042) lndist -2.072*** -2.084*** -1.865*** -1.875*** (0.087) (0.087) (0.075) (0.075) comcol 0.172 0.143 -0.558*** -0.566*** (0.303) (0.303) (0.205) (0.205) colony 0.696** 0.702** 0.188 0.164 (0.350) (0.350) (0.261) (0.261) border -0.819*** -0.829*** 0.701** 0.706** (0.237) (0.237) (0.306) (0.306) lang 2.525*** 2.528*** 2.250*** 2.252*** (0.236) (0.236) (0.204) (0.204) TRADEOPEN1 0.041*** 0.041*** 0.028*** 0.028*** (0.010) (0.010) (0.008) (0.008) TRADEOPEN2 0.013*** 0.013*** 0.012*** 0.012*** (0.004) (0.004) (0.003) (0.003) ECtoMEX 2.811*** 2.479*** (0.931) (0.815) JPNtoMEX 1.400 1.196 (2.104) (1.847) ROWtoMEX -1.506 -1.207 (0.950) (0.811) Constant -119.110*** -122.369*** -103.883*** -106.623*** (25.257) (25.258) (21.895) (21.898) N 11872 11872 11872 11872 R-squared 0.58 0.58 Standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% Table 4.4: Estimation Results for Tobit and Scaled OLS model with differences in capital per worker, coefficient estimates. Fixed effects dummy variables not shown. 85 pairs in which the home country is a high-income country (9), the home country is a high-income country and the host country is not (10), and repeat the estimation with scaled OLS, shown in columns (11) and (12). The results are reported in table 4.5. The coefficient on differences in GDP per capita loses its significance\u2014 perhaps indicating that the proximity-concentration hypothesis does not hold for FDI in mostly developing countries\u2014but all other coefficient estimates are sim-ilar to the full sample. This indicates that the results hold even when comparing Mexico only with other middle- and low-income countries. Since the tobit estimator is non-linear, the coefficient estimates tell us little about the actual magnitudes of the effect. To obtain these values, the marginal ef-fects have to be calculated. Table 4.6 shows the marginal effects at the mean of the independent variables. Although magnitude of the marginal effects is surprising at approximately 290 percent, it is not entirely unexpected, given the rapid growth of the European FDI stock in Mexico shown in table 4.7. This result confirms that European (and possibly Japanese) firms made major investments in Mexico to gain a beachhead in the North American market. Notably, the effect of N A F T A is pronounced despite the 1994 currency crisis in Mexico that depressed domestic demand. This suggests that the market sought by this investment was that of the US, but that the efficiency gains were to made in Mexico. Moreover, since the FDI stock also includes considerable FDI in services, the result appears in line with the expectations. 86 . (9) (10) (11) (12) Dependent variable: InFDI Fixed effects tobit Fixed effects tobit Scaled OLS Scaled OLS InRGDPl 7.135*** 7.969*** 6.261*** 7 517*** (1.543) (2.763) (1.270) (2.420) lnRGDP2 1.434* 1.815** 1.269* 1.608** (0.869), (0.815) (0.720) (0.716) InDGDPcap 0.030 -0.077 0.008 -0.139 (0.157) (0.283) (0.127) (0.246) lndist -2.709*** -2.650*** -2.228*** -2.438*** (0.160) (0.210) (0.130) (0.184) comcol 0.913* 0.323 0.937** 0.358 (0.495) (0.863) (0.403) (0.761) colony 0.321 0.070 0.393 0.197 (0.419) (0.447) (0.340) (0.388) border -0.995* -0.338 -0.781* -0.309 (0.548) (0.490) (0.454) (0.430) lang 3.876*** 2.418*** 3.216*** 2159*** (0.375) (0.419) (0.305) (0.365) TRADEOPEN1 0.060*** 0.126*** 0.039*** 0.110*** (0.017) (0.020) (0.013) (0.017) TRADEOPEN2 0.018*** 0.008* 0.015*** 0.007* (0.005) (0.004) (0.004) '(0.004) ECtoMEX 3.348*** 2.610*** 2.823*** 2.293*** (1.107) (0.966) (0.924) (0.852) JPNtoMEX 2.413 2.520 1.804 2.171 (2.512) (2.170) (2.108) (1.922) ROWtoMEX 0.204 -0.093 -0.004 -0.363 (1.074) (1.155) (0.871) (0.991) Constant -170.664*** -200.499*** -149.608*** -211.507*** (41.180) (63.960) (33.924) (70.937) N 6823 5045 6823 5045 R-squared 0.53 0.53 Standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% Table 4.5: Estimation results for tobit and scaled OLS model for North-South country pairs, coefficient estimates. Fixed effects dummy variables not shown. 87 (1) Dependent , \u201e . . , \u2022 ui i T-T-.T Fixed effects tobit variable: InFDI (2) Fixed effects tobit (5) Fixed effects tobit (6) Fixed effects tobit InRGDPl 4.361 4.435 4.637 4.751 lnRGDP2 1.054 1.102 1.273 1.308 InDGDPcap -0.22 -0.223 InRDKPW -0.178 -0.181 lndist -2.161 -2.172 -2.072 -2.084 border -1.062 -1.064 -0.819 -0.829 comcol 0.74 0.718 0.172 0.143 colony 0.582 0.589 0.696 0.702 lang 2.431 2.427 2.525 2.528 T R A D E O P E N 1 0.035 0.035 0.041 0.041 T R A D E O P E N 2 0.014 0.014 0.013 0.013 E C t o M E X 2.949 2.811 JPNtoMEX 1.305 1.4 R O W t o M E X -0.39 -1.506 Table 4.6: Marginal Effects, DGDP and DKPW Specifications. Year FDI stock (US$ Millions) Annual Growth (%) 1991 2,533 1992 2,987 17.9 1993 2,886 -3.4 1994 3,612 25.2 1995 3,571 -1.1 1996 4,841 35.6 1997 7,987 65.0 1998 7,693 -3.7 1999 15,478 101.2 2000 20,619 33.2 2001 25,945 25.8 Table 4.7: FDI stock from the EU-15 in Mexico, 1991-2001. Source: EUROSTAT, OECD 88 4.3 Conclusion: Getting a Foot into the Door The results of the gravity model strongly suggest that European (and possibly Japanese) firms reacted to N A F T A by fulfilling Mexican hopes: they invested massively to establish a beachhead in the North American market. As most of this investment took place shortly after the Peso crisis, it is safe to assume that short-term profits in the Mexican market were not its main objective. Rather, the buyers of European products would be consumers in the US. While N A F T A was merely the icing on the cake for US firms that used Mexico as an export platform, as shown in Chapter 2, the agreement seems to have exerted a strong pull on European firms. Unless these firms sourced all their supplies, parts and machinery in North America, these firms would have high stakes in tariff-free access to Mexico, in particular once NAFTA's rules of origin were tightly applied and duty-drawbacks faded out. The stronger their ties to the home country, the more they would be pressed to ensure this access. If Mexico were to delay further reduction of its M F N tariffs or even raise trade barriers again, these firms would have to act to pursue their plans. As the next chapter shows, within a few years of the conclusion of NAFTA, both the E U and Japan sought their own FTAs with Mexico. 89 Chapter 5 Countering the Effects of NAFTA Why did both the E U and Japan seek bilateral trade agreements with Mexico within the space of a few years? Mexico was not an obvious partner for either party. Neither the E U nor Japan had signed a free trade agreement outside their geographic regions before. In Europe, government attention was focused on fu-ture E U enlargement. The E U signed association agreements with Central and Eastern European countries to prepare them for membership in the Union, but neglected relations with Latin America. Japan had relied on multilateralism as the foundation of its trade policy for four decades, only concluding its first FTA with Singapore in 2001. Finally, multilateralism had just received a strong boost: the successful Uruguay Round established the WTO. Preparations for the M i l -lennium round were underway. There was little indication that the E U or Japan would shortly embark on negotiations with a Latin American country. NAFTA's entry into force and the 1994 financial crisis, however, changed the parameters for foreign direct investment in Mexico. As the previous chapter has shown, European firms did not hesitate to invest considerable sums to use Mexico as a beachhead. A depreciating currency depressed domestic demand, but made Mexico more attractive as an export platform. A weakened financial sector, in which the government lifted ceilings on foreign participation, offered the chance to enter a new market. Yet to the detriment of outsiders, N A F T A ensured that only US and Canadian firms could fully benefit from these developments. By ex-erting pressure to switch from overseas suppliers to producers based in Mexico, NAFTA's rules of origin raised the cost of production for European and Japanese firms. Preferential access to the Mexican service market put European service providers at a disadvantage. In financial services, N A F T A threatened the long-term exclusion from the market, should US and Canadian firms manage to estab-lish market dominance in the Mexican banking sector. As the following two case studies show, N A F T A prompted European and Japanese firms to lobby their governments for agreements with Mexico to achieve parity with NAFTA. Yet the cases also offer strong evidence for an endogenous 90 dynamic: Crucially, it took the move by the E U to motivate changes in Japanese policy toward Mexico and to overcome the resistance of protectionist forces. In addition, the two cases allow for useful comparisons. Whereas the Mexican position remained the same throughout both negotiations, the reactions by the E U and Japan varied over time. Mexico sought FTAs with all major trade partners in order to attract foreign direct investment, but retained and twice raised M F N barriers to trade. By contrast, the E U and Japan changed their policies towards Mexico. Given the higher stakes of European companies in Mexico, the E U was motivated to move faster. Manufacturing firms from the E U had made major investments in Mexico. At the same time, several major banks had set their eyes on the Mexican market. By contrast, the coalition in support of an FTA in Japan was smaller, especially since services were not a major Japanese interest. Moreover, Mexico's agricultural exports were less threatening to European producers than to their Japanese counterparts. The two cases therefore support the model put forth in this work. In both cases, multinational firms with investment in Mexico formed a coalition with their upstream suppliers, but their relative strength vis-a-vis domestic resistance differed. The effects of N A F T A stand out most clearly at the level of individual firms. To gain explanatory leverage, the following sections therefore disaggregate both cases to individual firm behaviour and lobbying. The case studies first explore the interests of major firms with FDI in Mexico, followed by an analysis of these firms' reactions to NAFTA. I then trace the processes of lobbying, decision-making and negotiations that determined E U and Japanese strategies towards an FTA with Mexico. The case studies conclude with an assessment of the agree-ments in relation to the predictions by my model. At the time of writing, however, a full analysis of the Japan-Mexico FTA is not yet possible, since the schedules of the agreement have not been published. 5.1 The EU-Mexico Free Trade Agreement When President Salinas travelled to Europe in late January 1990 to advertise Mex-ico as an attractive investment location, he returned disheartened. European cor-porate and political leaders, absorbed in the economic potential of Eastern Europe emerging from communist rule, showed little interest in Mexico (Cameron and Tomlin 2000, 1-2). Yet in 1999, the E U concluded its first extra-regional FTA with Mexico. Three years later, E U Trade Commissioner Lamy referred to Mex-ico as a beautiful bride between two lovers, whom the E U \"would like to tempt 91 (...) back closer to the centre of the bed, and invite (...) not to sleep right on one edge of the mattress!\" (2002). What had made Mexico so attractive? Mexico first retained and later even increased the margin between N A F T A and M F N trade barriers by lifting the applied rate to the bound rate, the maxi-mum legally possible under the GATT. In addition, within a few years, exporters lost market share. Between 1990 and 1994, E U exports to Mexico grew by 64 percent, but then began to level off, first as a result of the Peso crisis, but later because of competitive disadvantages created by N A F T A . At the same time, the EU's share in Mexican imports fell from 17.4 percent in 1990 to 9 percent in 1997. FDI from the E U , however, began to rise rapidly, as shown in Chapter 4. The concurrence of both developments suggests that European firms reacted to N A F T A by establishing or protecting beachheads for production within the free trade zone. In doing so, they incurred a higher cost than their competitors from within NAFTA: Intermediate and capital goods from the E U faced a 10-20 percent higher tariff than those from the US and Canada. With the conclusion of the agreement, European firms with investment in Mex-ico in the automotive industry were among the first to be affected. In particular Volkswagen faced increasing competitive pressure. In contrast to the US Big Three, Volkswagen's manufacturing and supply base was concentrated in Mex-ico's central region. Volkswagen had been an incumbent since 1964, when it established a manufacturing plant in Puebla, 90km east of Mexico City. Between 1980 and 1991, Volkswagen dominated the domestic passenger car market with an average share of 30 percent (Asociacion Mexicana de la Industria Automotriz 1993). In response to the 1977 auto decree, it built an engine plant and aluminium foundry, concentrating production on a single model (the Beetle, or vocho) and achieving higher domestic content than the US manufacturers. In 1981, Volk-swagen began exporting engines to Germany, starting with about 15,000 units and reaching a peak of 340,000 in 1990. Still, the domestic production remained well below efficient scales for individual models, motivating Volkswagen to invest US$ 1 billion in the Puebla plant in 1991 and to focus it on the production of Golf and Jetta models for the US market. NAFTA's rules of origin posed a challenge to this strategy. While North Amer-ican companies had access to their supplier networks to fulfill the regional content quota, Volkswagen did not. In 1992, Volkswagen acquired half of its inputs from Germany, 40 percent from Mexico and 10 percent from the US. However, this percentage differed considerably from model to model. Whereas the old Bee-tle was 80 percent Mexican, the third generation Golf\/Jetta was to be assembled of up to 90 percent imported parts. Compliance with the rule of origin required 92 an enormous strengthening of the local supply industry in terms of technology, quality and efficiency. While the decision to do so predated NAFTA, Volkswa-gen estimated that it would take ten years to develop adequate supplier networks (Hanson and Shapiro 1994, 12-13). At the same time, neither Volkswagen nor its European suppliers could invest in Mexico's parts industry without restrictions, since Mexico still maintained the 49 percent ownership ceiling for non-NAFTA firms. Volkswagen therefore faced the competition by the US Big Three with a burdensome inheritance of the pre-liberalization era, but with a much higher cost of restructuring. The major decisions about restructuring the Mexican operations coincided with the N A F T A negotiations. Volkswagen initially opposed higher rules of ori-gin, but then shifted its position, possibly because N A F T A granted differential phase-in periods for incumbent firms, including Volkswagen and Nissan, and new entrants (Cameron and Tomlin 2000, 134-135). However, following the 1994 Mexican peso crisis that depressed domestic demand, exports from Mexico be-came all the more imperative, including exports of Jetta models back to Germany. Volkswagen therefore again shifted its position to be among the first to lobby for a compensatory arrangement. Two other German automotive firms also looked at the Mexican market. B M W entered the market by setting up a small assembly facility in Mexico in Toluca in 1994, but aimed at an annual output of only 10,000 vehicles (Asociacion Mexi-cana de la Industria Automotriz 2004; B M W de Colombia 2004). Mercedes Benz set up a manufacturing plant for large trucks in northern Mexico in Derramadero, but did not produce passenger cars in Mexico (Moreno Brid 1996, 26nl7). The second principal force for an FTA was the service sector. With the pros-pect of a growing Mexican market, European firms began to eye an expansion into Latin America. European financial service companies, however, at first faced additional costs because they had to use their US subsidiaries to enter the Mexi-can market ( A F X News 19 August 1994). Spanish banks made major investments in Mexico after the initial steps of liberalization undertaken by the Mexican gov-ernment. Following a process of concentration via mergers and acquisitions in the Spanish domestic market, Banco Bilbao Vizcaya (BBV), Argentaria, Banco Santander and Banco Central Hispano (BCH) emerged as major players.1 Using their competitive advantage due to familiar legal and cultural backgrounds, these 'Further mergers in 1999 resulted in the current groupings Banco Bilbao Vizcaya Argentaria (BBVA) and Banco Santander Central Hispano (BSCH), the two major banks representing the main Spanish investors in the Mexican financial sector. 93 banks invested heavily in Mexico, where they held 33 percent of the total capital of all foreign banks. In 1996, BJ3V took over the Mexican bank Probursa, making it the fifth-biggest retail bank in the country, and bought a 30 percent share of the administration of the pension fund Profutura (Banco Bilbao Vizcaya 1998). San-tander already entered the market in 1996, holding 71 percent of Banco Santander Mexicano, and administered the second-biggest and twelfth-biggest pension funds in the country, Afore Santander Mexico and Afore Genesis (Calderon and Casilda 1999, 20-30). Hong Shanghai Bank Corporation (HSBC), based in the U K , was the third European bank with major investments in Mexico. Like their US counterparts active in the region, these banks' interests in re-gional trade agreements and service sector liberalization have to be seen in a broader context. B B V A and B S C H hold shares in the telecommunications firm Telefonica, a major investor in both Mexico and Chile, as well as in other Spanish utilities and energy companies such as Repsol that have focused on Latin Ameri-can markets (Calderon and Casilda 1999, 22). Moreover, they also provide financ-ing for. foreign direct investment by European companies in the region. B B V A and B S C H are the only banks in the global market that hold greater investments in Latin America than in their home country (Calderon and Casilda 1999, 38), and together had investments of about US$ 100 billion in 1999, or about 93 percent of all investment by the Spanish financial sector in the region (Rozas Balbontin 2001, 28). For all of these primarily European firms, the external commercial policy of the European Union offered a potential recourse to regain a competitive position. 5.1.1 The Institutional Background: Trade Policy in the European Union As European firms faced problems of declining competitiveness, they turned to-wards the E U institutions responsible for conducting the common trade policy of the Community.2 The institutional setting primarily determined the pattern of lob-bying and the access points for private sector interest, but it also had a minor effect on the outcome: since the E U Commission did not have a mandate to negotiate a comprehensive investment chapter along the lines of N A F T A Chapter 11, the final text of the EU-Mexico FTA only referred to the bilateral investment treaties sev-eral member states had signed with Mexico. While the E U Commission has the 2Since the common external commercial policy stems from the Treaty of Rome, the correct legal term refers to the European Economic Communities (EEC). For reasons of simplicity, the following only refers to the European Union (EU). 94 sole right to initiative and conduct of trade negotiations, it has much more limited power over domestic legislation concerning services and investment. As a result, most lobbying at the E U level is directed at the Commission, especially regard-ing technical aspects of agreements and offers made in negotiations. However, industrial associations in the member states also put pressure on national govern-ments. The Council of Ministers of the member states has the power of decision and authorizes the Commission to conduct negotiations in accordance with Arti-cle 113 of the Treaty of Rome (Article 133 under the Treaties of Maastricht and Amsterdam) by issuing a mandate. This mandate defines the room for manoeuvre of the Commission in negotiations. Trade interests of the member states, both offensive and defensive in policy parlance, therefore also play out in the setting of the mandate. The details of the Council decisions are prepared by the Committee of Per-manent Representatives (COREPER), consisting of the national ambassadors to the E U . Below this level, the Article 133 Committee of officials from the member states and its sectoral working groups consult with the Commission and interest groups. The members of this committee, often with detailed knowledge of the is-sues, form an important link in conveying member states' aims to the Commission. The External Economic Relations Committee of the European Parliament (EP) provides information to MEPs, whose assent is required for Association Agree-ments (as in the case of Mexico and Chile), but not for other trade agreements. Association Agreements were originally only intended to prepare countries for EC accession, but have now come to signify strong economic and political ties. The Economic and Social Committee, a legacy from the European Community of Coal and Steel, has an advisory role in raising issues to the Commission. Reflecting an early concern about the N A F T A negotiations, the EP External Economic Relations Committee published a report in 1992 that warned of com-petitive disadvantages for European firms in NAFTA. The report pointed out those industries in which the N A F T A negotiations converged on very restrictive rules of origin of concern to European firms: automobile parts and components produc-ers in all of NAFTA, and producers of fabrics who would potentially lose market share in Canada under NAFTA's triple transformation rule. A 1993 information note published by the European Commission echoed these concerns, although it argued that broad tariff reductions negotiated in the Uruguay round promised to limit the preferential benefits of N A F T A for US and Canadian companies. This was much less clear in the services sector, given that the GATS negotiations pro-ceeded slowly. The Commission preferred the multilateralization of the N A F T A service commitments under the GATS, but acknowledged that: \"the absence of 95 such multilateralism could lead to considerable trade diversion in the services sector, to the detriment of Community trade in services and particularly to its potential trade in services with Mexico\" (15). At the time, these issues remained abstract. Within a few years, though, problems of European firms were channelled into political action. 5.1.2 Moving Toward Free Trade with Mexico Shortly after N A F T A came into effect, European firms began to voice concerns. A 1995 survey by the Delegation of the European Commission in Mexico found that European companies in fields such as telecommunications, automobiles, chem-ical, pharmaceuticals and footwear felt severely disadvantaged by N A F T A (11-13). Reacting to this information, the Council adopted a \"basic paper\" that ex-pressed the readiness to start talks with Latin American countries aiming at \"more far-reaching agreements\" (Commission of the European Communities 1995). A Commission communication to the Council and the EP, published on 8 February 1995, urged action in relations with Mexico: \"If the E U fails to take appropri-ate steps, its relations with Mexico run the risk of being eroded by N A F T A (...). The available figures on direct investment show that the risk of European opera-tors being marginalized is a very real one.\" It warned that, \"without a new, more advantageous contractual framework for trade, Mexico has considerable scope for protecting its market while increasing its customs tariffs within GATT limits,\" and requested a mandate to negotiate a new framework agreement with Mexico (13, 17). A draft version of a negotiating brief to the Council for an agreement with Mexico was submitted on 23 October 1995. However, the offer by the Mexican President to sign an FTA, conveyed for the first time in 1995, had to be rejected by the Commission because of French worries, expressed in the 133 Committee, over agricultural imports of citrus fruits that could displace those from French overseas territories (Inter Press Service, 8 February 1995). In May 1995, the European Commission, the Council and the Mexican government signed a Joint Solemn declaration to establish the founda-tions of a future \"framework\" agreement. Still, E U member states differed in their commitment to negotiating an actual FTA with Mexico. At the Council meeting in February 1996, Spain and the United Kingdom supported the negotiation of a free trade agreement in a \"single phase\" undertaking (Europe Daily Bulletin, 10 May 1996), but were blocked by France (European Report, 28 February 1996; European Information Bulletin, 27 February 1996, 1; A P Press Business News, 26 February 1996). 96 The Council finally reached a compromise in May 1996, deciding on a man-date of gradual liberalization and negotiations on a sectoral basis (AP Press Busi-ness News, 13 May 1996). The process was slowed down even further because services fell under member state jurisdiction and required unanimity in the Coun-cil , while trade in goods could be negotiated by the Commission with a qualified majority backing the mandate. Finally, on 11 June 1997, the E U and Mexico arrived at a framework agreement that established a joint council of Commis-sion, member states and Mexico that would negotiate on bilateral trade and ser-vices liberalization, with the eventual aim of a fully-fledged FTA. This agreement was followed by acrimonious arguments in the council of permanent representa-tives of the member states (COREPER), in which France threatened to block the framework agreement, probably to prevent agricultural liberalization in an FTA (Sanahuja 2000, 53). In the meantime, between 1995 and 1997, Spain, Belgium, France, Italy, the U K and Germany negotiated bilateral investment treaties with Mexico to establish a minimum of legal guarantees equal to N A F T A Chapter 11, but without equivalent liberalization of Mexico's investment regime. Talks to-wards a free trade agreement began on 14 July 1998 (Financial Times, 15 July 1998). After nine rounds of negotiations, the agreement was finalized in Novem-ber 1999 and initialled in December of the same year. 5.1.3 Coalitions and Aims of European Lobbies During the protracted process that led to the EU-Mexico FTA, both member states and the E U Commission came under increasing pressure from lobbyists to counter the competitive disadvantages created by NAFTA. The motto of lobbyists was to achieve \" N A F T A parity,\" an expression coined by Commission officials to describe the same access to the Mexican market as guaranteed for the US and Canada. Specific requests, though, reflected the different concerns of exporters who sold to Mexican customers, exporters that were mostly suppliers to European firms, and investors both in services and manufacturing. The coalitions that emerged closely match those predicted in Chapter 2. Those manufacturing firms that had already invested or planned to invest to use Mexico as an export platform lobbied most strongly for an FTA. Since they were also the most affected by the discriminatory provisions of NAFTA, they specifically sought to counter these parts of the agreement. Services firms sought parity in market access, which required liberalized investment in the sector. Of firms with export interests, those in sectors with close links as suppliers to investors actively supported an FTA. By contrast, conventional exporters lobbied less. Furthermore, 97 the negotiating process and outcome corroborate a key claim made in Chapter 2: Interests convergence primarily on FDI. Conventional exporters were not only less active, but also less successful in achieving their aims, since neither the E U Commission nor the Mexican accorded them high priority. In the light of Volkswagen's particular interests, the company was clearly the most vocal individual supporter of an FTA. The company had decided to produce the new Beetle and Jetta models exclusively in Mexico for both the US and Euro-pean markets, as well as all Golfs for the N A F T A market. As such, it had the dual interest of negotiating tariff-free access to the E U and dismantling tariff barriers on parts. Furthermore, Volkswagen sought a larger quota to export its luxury cars Audi and V W Passat to Mexico. Volkswagen therefore combined the interests of mass producers and exporters. Just like the US manufacturers who started to use Mexico as an export platform, the company wanted to produce entry-level cars in specialized factories that achieved high volumes of production. Volkswagen lobbied the Commission from 1996 on and became closely in-volved in 1997 and 1998, with regular correspondence and five personal visits at the Commission by its representatives.3 One of the key issues for Volkswagen was to lower the Mexican tariffs on vehicle parts. According to a Volkswagen repre-sentative, \"even 3 percent tariffs [i.e. US tariffs on cars] are difficult with narrow margins, while 15 percent [i.e. Mexican tariffs on parts] are outright impractica-ble.\"4 The Mexican side, under pressure from local parts producers, resisted this issue \"in a dramatic battle\" almost until the end, as most European parts producers preferred to export to Mexico. 5 Since the Mexican association of car producers Mexicar also sided with Volkswagen, the Mexican government eventually gave in. Compared to NAFTA, the deal was easier to accept for the Mexican side because Volkswagen was already achieving a higher overall content local quota than US manufacturers. Furthermore, growing exports from Mexico to Europe promised to compensate Mexican parts producers. However, Volkswagen also managed to press for a special clause that stretched the limits of the agreement: some new Beetle models would be outfitted with an engine declared as originating in the Community but produced in Hungary, which did not join the E U until May 2004.6 C L E P A , the European association of automobile parts suppliers, also lobbied for interviews with interest group representative and EU officials, Brussels, June 2004. interview, Brussels, June 2004. 5Just like in the NAFTA negotiations, industry interest representatives were based in a nearby hotel room, referred to as \"el cuarto al lado,\" and frequently consulted during the negotiations. Interview with EU official, Brussels, June 2004. interview with interest group representative, Brussels, June 2004. 98 lower tariff on parts and components, but was rather ineffective. While partly due to Volkswagen's pressure for specific tariff offers that matched its suppliers' exports, Mexican negotiators also resisted CLEPA's requests strongly.7 To serve the upper market segment, Volkswagen pushed for a bigger, firm-specific import quota on the basis of their status as a long-term incumbent in Mexico. Other European car manufacturers, in particular Mercedes Benz and B M W , had different aims. Since they only planned to serve the small market seg-ment for luxury cars by exporting, they were mainly interested in an export quota for all firms and eventual tariff-free access to the Mexican market. The European Association of Automobile Manufacturers (ACEA) lobbied on their behalf, trying to balance the interests of Volkswagen and the exporters.8 A C E A frequently met with Commission officials to press its position, although the interests of Volkswa-gen clearly predominated, since they were partly represented at both sides of the table. While the Commission tried to get E U automobile producers to compro-mise, arguing that it \"could not well discriminate against some of our own [EU] companies,\"9 the Mexican side preferred Volkswagen's proposal that the Mexican government would allocate the quota. One E U official stated that the negotiators \"almost got a sectoral deal in cars for Volkswagen.\"1 0 European automobile manufacturers also sought to counter the interest of US manufacturers by using rules of origin, recreating the scenario that had evolved during the N A F T A negotiations. The Commission insisted on using the standard formula of all E U agreements on the grounds of simplicity. This coincided with the manufacturers' interests, but proved problematic for Chrysler. The company, which at the time planned to export its PT Cruiser model from Mexico to the E U , only achieved a 30 percent Mexican content quota and lobbied the Commission to count all N A F T A parts as Mexican. While this would have matched Volkswa-gen's success in pressing its point regarding the Hungarian beetle engines, the Commission refused and the Mexican side appeared disinterested.11 Exporter interests further up in the production chain also lobbied in favour of the agreement. European chemical manufacturers' exports to Mexico primarily supply European investments. The umbrella organization of the chemical industry, CEFIC, became a staunch supporter of an agreement with Mexico to level the playing field with the US competition. Between 1996 and 1999, the organization interview with interest group representative, Brussels, June 2004. interview with interest group representative, Brussels, June 2004. 'interviews with EU officials, Brussels, June and July 2004. 10Interview with EU official, Brussels, June 2004. \"interview with interest group representative, Brussels, June 2004. 99 met on a regular basis with the responsible desk officers in the D G Trade and emphasized the concerns of its member companies.12 Early letters from CEFIC to the Directorate General stressed the need for an FTA to counter the negative effects in NAFTA. In an April 1996 letter to D G Trade, CEFIC stated that: (...) Considering (...) the particular situation created by Mex-ico's inclusion in the NAFTA, establishing a bilateral free-trade area should not only revive E U relations with Mexico but also bring about some compensation towards E U industry's 'third-country' position vs. US and Canadian competitors on the Mexican market.\" When the Commission took up the Mexican offer to negotiate an agreement, CEFIC was among the most active supporters. CEFIC specifically asked that any tariff reductions by the Mexican side had to match those negotiated under N A F T A . 1 3 A March 1999 letter to the Commission emphasized that: In earlier correspondence regarding this Free Trade Agreement CEFIC has already pointed at the disadvantages for E U chemical in-dustry resulting from the N A F T A agreement. While US industry ben-efits mainly from zero or reduced import duties (...), E U industry is still faced with the negative impact of Mexico's non-participation in the [WTO] Chemical Tariff Harmonisation Agreement. In the services sector, the key lobbying association was the European Services Forum (ESF). While not originally created to focus on bilateral agreements,14 the association followed the negotiations closely. In line with the general pattern of service sector lobbying that also became visible during the N A F T A negotiations, the position papers supplied by ESF requested the same liberalization measures as those sought in the WTO. In monthly meetings with the responsible desk offi-cer at the Commission, the ESF pressed for an opening of the Mexican services markets.15 Particularly strong interests targeted the financial sector, where Span-ish banks sought to achieve access equal to N A F T A by investing directly or by taking over Mexican banks. B B V and B C H pushed for an agreement primar-ily by lobbying the Spanish representatives in the Council and the Committee of I^nterview with interest group representative; Interview with EU official, Brussels, June 2004. 'interview with interest group representative, Brussels, June 2004. 'interview with interest group representative, Brussels, June 2004. 'interview with EU official, Brussels, June 2004. 100 Permanent Representatives (COREPER), since services commitments had to be listed and requested by member states. However, the two banks also sought to in-fluence Commission negotiators directly.1 6 Requests from the Commission were addressed at both D G Trade and the Directorate General for Financial Services. Although H S B C had comparable interests in Mexico, the bank did not lobby ac-tively. One E U official speculated that the push by the two Spanish banks was already so strong that H S B C opted to \"get a free ride.\" 1 7 Evidence from the negotiating aims and tactics of E U officials offers further support for the hypothesis of levelling the playing field for investors and upstream producers. In the tariff negotiations, the E U refused to negotiate on a product-by-product basis and insisted on a deal across the board. The timetable for the tariff phasing was to match NAFTA, with products in three lists: An A list with immediate liberalization, a B list with liberalization by 1 January 2003 to coincide with NAFTA, and a C list with liberalization by 2007. A deal then had to be struck on which product groups fell in which categories, with an allocation again based on the speed by which N A F T A had liberalized them.1 8 Despite the EU's entrenched protectionist agricultural policy, relatively little lobbying revolved around market access in agricultural goods, where export in-terests in very different products predominated. Mexico suggested a slower tariff phasing than the E U and insisted on the exclusion of any product benefiting from E U export subsidies, since these measures had been eliminated under N A F T A (Europe Daily Bulletin, 12 March 1999). Only one lobby group was involved until the very end of the negotiations: the European spirits and wine producers, who pressed for Mexico's acceptance of E U geographic denominations, but also appeared willing to accept matching Mexican requests regarding Mexican spir-its. E U officials judged that the agricultural component was neither in violation of WTO Art. X X I V , nor threatening to any particular interest group in the E U , since Mexican exports were only competing in a handful of product categories, and there only with third-country interests.19 E U services negotiators opted for a deal that locked in Mexico's commitments. Since Mexico had to adapt its domestic regulatory framework to fulfill the N A F T A schedule for services liberalization, the E U could achieve parity by securing iden-tical commitments for its industries. Its negotiators therefore pressed for a simple standstill clause. The Mexican side perfunctorily mounted a brief resistance, but I^nterview with EU official, Brussels, July 2004. See also El Pais, 5 May 1999. 'interview, Brussels, June 2004. I^nterview with EU official, Brussels, July 2004. ''interview with EU official, Brussels, June 2004. 101 then agreed. Their main aims in services were to attract more capital into the banking sector, still weak from the 1994 financial crisis, and to balance US invest-ment with inflows from the E U . European negotiators also pressed for clear terms on equal access to maritime transport and its facilities for E U providers. This measure complemented the trade liberalization by clearing the path for European freight companies and was largely undisputed.20 However, given the contested mandate for services liberalization, the negotiators postponed the negotiation of actual commitments by either side. The European service industry appeared con-tent with a level playing field. E U negotiators, however, failed to achieve a key goal in the related field of standards and regulations. This conforms to the expectation of a first-mover ad-vantage bestowed by a PTA in services. Mexican negotiators argued that four years after NAFTA, so much of its trade was with the US that they could not legislate the adoption of any European standards. To the dissatisfaction of E U negotiators, Mexico reserved the right to use N A F T A government procurement rules, although the E U had pressed for its standard clauses.21 The argument so far underscores that firms with interests linked to FDI were the most active in their support of an EU-Mexico FTA. By contrast, supporters of finished goods were less concerned about market access. In services, firms sought to close the gap with their US and Canadian competitors as much as possible, al-though the latter's first-mover advantages could not be fully compensated. As the next section shows, the final outcome of the negotiations offers further evidence for the interests of these firms. 5.1.4 Outcomes: Beyond N A F T A Parity The treaty establishing a free trade area in goods, regulating government procure-ment and cooperation and consultation on other matters, was signed in March 2000. 2 2 The E U Commission cited as its achievements similar or better condi-tions for automotive exports to Mexico than from the Canada and the US by re-ducing tariffs from 20 to 3.3 percent immediately and the complete elimination by 2003. In addition, Mexico committed itself to a quota for the import of vehicles: 14 percent of the Mexican sales of the previous year in units until December 2003, then 15 percent until December 2006. 2 3 Paragraphs 3.1 and 3.2 grant Mexico the 20Interview with EU official, Brussels, June 2004. 21Interview with EU official, Brussels, June 2004 22Formally Decision 2\/2000 of the EC-Mexico Joint Council of 23 March 2000. 2 3 Annex II to Decision Nr. 2\/2000 of the EC-Mexico Joint Council of 23 March 2000. 102 temporary right to allocate up to 11 percent of this quota to incumbent firms in compliance with the 1989 Mexican auto decree and to prefer exporters based on traditional trade patterns, which in practice strongly favours Volkswagen. Car parts and components that fall under the Mexican Auto decree are tariff free by 2007. Likewise, this deal favoured the incumbent manufacturer Volkswagen and its parts suppliers over exporters. Tariffs on agricultural goods levied by either party are only phased out over long periods of time of up to ten years. The agreement also maintained import quotas by both sides for certain products, but the trade in these products appears too limited to warrant a challenge in the WTO by competitors. Regarding investment and services, the Joint Council agreed that specific com-mitments would have to be negotiated within three years. In investment, the agree-ment only refers to the commitments under the O E C D and bilateral investment treaties, and points to a future review. In fact, the Commission had negotiated what one source called a \"sexy investment chapter better than NAFTA,\" but dropped it to seal the agreement in services.24 However, despite the conflict over the man-date, lobbying pressures from the financial sector were apparently strong enough to motivate a special agreement on financial services. Chapter III on financial services contains NT and M F N clauses, a standstill clause and provisions for lib-eralization within three years. Notably, Article 23 establishes a special committee that reviews the agreement to allow additional, matching commitments, should ei-ther party liberalize its financial services sector under a regional agreement. This clause offers a qualified commitment to maintain parity even if future preferential liberalization takes place in other deals. Within a few months of the conclusion of the FTA, European car manufactur-ers announced major investments in Mexico. Renault returned to Mexico after a 15-year absence with a joint production plant with Nissan, following the merger of the two car manufacturers in April 1999. Volkswagen earmarked US$ 1 billion for the expansion of Mexican operations, which by then provided a third of the parts imports for its German factories and exported 300 New Beetle models daily to the E U . Peugeot executives visited Mexico in January 2000 to survey sites for future investments (Financial Times, 5 January 2000). DaimlerChrysler unveiled plans to invest US$ 2 billion in Mexico over the next five years (Financial Times, 24 May 2000). While EU-Mexican trade is still dwarfed by Mexican trade with the US, Euro-pean firms regained market share in the first years after the entry into force of the Interviews with E U officials, Brussels, June and July 2 0 0 4 . 103 agreement. European merchandise exports to Mexico expanded by 28.1 percent between July 1999 and June 2003. A considerable share consists of intermediate and capital goods: Electrical and power generating machinery, transport equip-ment and parts, and chemical products make up 58 percent of exports (EC Com-mission 2004). The FTA appeared to have levelled the playing field and restored the competitiveness of European firms in Mexico. The previous discussion has shown that trade diversion and the effects of N A F T A on European FDI in Mexico triggered a countermove by the E U , lead-ing to the first expansion of its free trade network across the Atlantic. Yet the E U strategy had an unexpected consequence in itself: it led trade policymakers in Japan to reconsider the country's exclusive focus on the multilateral trade regime. 5.2 The Japan-Mexico Free Trade Agreement While the Japan-Mexico FTA is of limited importance in terms of the actual vol-ume of trade between the two countries, it signifies an important step for Japanese trade policy: The FTA represents the first substantive bilateral agreement that cov-ers agricultural products. N A F T A thus triggered a major shift, highly contested domestically, in Japan's foreign economic policy, underscoring the competitive pressures exerted by the proliferation of PTAs. Moreover, as described in the following chapter, it led Japan to the active pursuit of bilateral agreements with neighbouring countries and beyond, with trade bureaucrats leading policy in reac-tion to demands by multinational firms. The coalitions of Japanese firms that supported the FTA corresponded to those lobbying for the EU-Mexico FTA. Japanese investment in Mexico, however, is smaller and narrowly concentrated in a few industries. While automotive firms and their suppliers acted similar to their E U counterparts, the service sector was much less engaged due to its limited interests in Mexico. By contrast, electronics firms were particularly hurt by N A F T A and lobbied actively. 5.2.1 Japanese Foreign Direct Investment in Mexico Japanese investment in Mexico is small compared to that from the US and the E U . It is, however, heavily concentrated in the maquiladoras: Sectorally in con-sumer electronics and auto parts, in which Japanese manufacturers have a tradi-tional competitive advantage; geographically in Baja California and the northeast of Mexico. Japanese firms used their US subsidiaries to set up production facilities 104 in Mexico, thus benefiting from the Mexican duty drawbacks and the limited-duty imports under the US tariff headings 806.30 and 807.00. This particular own-ership structure prevails today beyond the maquiladora sector: of 205 Japanese firms in Mexico in 2002, 135 are wholly or majority-owned by the US subsidiary of the Japanese company (Toyo Keizai Shinbunsha 2003). In 1990, Japanese in-vestment in the maquiladoras represented 6 percent of the total number of plants, with about 25,000 employees, of which 55 percent produced electronics parts, while most automotive parts firms produced wire harnesses (Koido 1991). A l -though mainly aimed at the US market, maquiladora plants also began to export to other countries in the region (Nikkei Sangyo Shinbun, 30 July 1991). Since the late 1980s, major Japanese companies such as Sony, Sanyo, Sharp, Hitachi and Matsushita (Panasonic) produced consumer electronics goods in Mex-ico, in particular T V sets. Until the mid-1990s, most exports of assembled goods went to the US market. As a result of US, Japanese and European investment in the industry, the share of Mexican exports of the US colour television market jumped from 2.6 percent in 1985 to 40 percent in 1989 (Koido 1991, 64). In many cases, labour-intensive production in Mexico was integrated with Japanese-owned factories in the twin city across the border in the US (Ueda 2001, 310-313). How-ever, until recently, core components such as cathode ray tubes were sourced from Japan and other Asian countries (Nikkei Sangyo Shinbun, 30 July 1991). Prior to NAFTA, the maquiladora industry was characterized by minimal local input. Between 1980 and 1989, the percentage of Mexican inputs even fell from 10 per-cent in some plants located in Mexico's interior region to less than 5, while most border-region maquiladoras never used more than two percent domestic parts and raw materials (Wilson 1992, 48^19). Non-NAFTA firms were far from attaining anything close to the level of regional content required by the agreement. Except for Nissan, and a small assembly operation of Honda motorcycles in Guadalajara established in 1985, no Japanese vehicle producers were present in Mexico. Nissan established its subsidiary in Mexico in 1961 and took up pro-duction five years later. Pursuing a strategy similar to Volkswagen in Mexico, the company initially aimed at the domestic market only, with its plant in Cuer-navaca producing low-priced small cars. Immediately prior to NAFTA, Nissan had a domestic market share of 24 percent and exported the required minimum of 30 percent of its Mexican production (Moreno Brid 1996, 19), primarily to Cen-tral American markets. Following the conclusion of NAFTA, Nissan began to spe-cialize its North American operations by transferring the production of the Sentra model from its US to its Mexican operations, to be exported to US and Canadian markets (United States Department of Commerce 1998, 8). The company invested 105 over US$ 1 billion in its second, recently built plant in Aguascalientes to achieve greater economies of scale (Moreno Brid 1996, 29) and targeted new exports mar-kets, including the sales of 20,000 cars to Chile in reaction to the Mexico-Chile FTA (El Financiero, 17 October 1995). Not operating any production facilities in the maquiladora sector, Nissan Mexico did not use duty-drawback schemes and was not integrated much with the company's overseas operations. Consequently, it adopted a passive stance towards the N A F T A negotiations at first (Cameron and Tomlin 2000, 134-135). 5.2.2 N A F T A affects Japanese Investment in Mexico NAFTA's high rules of origin and the phasing out of duty-drawback schemes pre-sented Japanese investors with a severe problem. Instead of paying zero tariff for imported intermediate goods destined to be re-exported, Japanese firms would face a 10 to 15 percent M F N tariff from 2001 on (United States Trade Represen-tative 2000, 284). Cathode tubes for T V sets mentioned before would now carry a 16 percent duty (Nihon Keizai Shinbun, 12 December 2002). Electronics firms such as Hitachi and Kenwood therefore shifted production and sourcing from Asia to Mexico in response to NAFTA. For many firms, this entailed higher costs, since their production strategy was based on a high percentage of imported intermediate and capital goods from Japan, as well as parts from Southeast Asia, often about 10 percent cheaper than comparable Mexican components (Nihon Keizai Shinbun, 15 December 1998; 24 May 2000; Edgington and Fruin 1994, 256). Represen-tatives of Kenwood, producing car audio components in Ciudad Juarez, noted to the Japanese press that they had built the plant under the assumption of global sourcing opportunities. Hitachi shifted the control over the production of parts from Malaysia to Tijuana, while Sanyo began to domestically procure parts for its compressor production in central Mexico (Nihon Keizai Shinbun, 5 January 1996). Mitsubishi moved all of its production of circuits for large-screen TVs from Asia to Mexico, a measure followed by Korean and Taiwanese competitors, while Sony switched towards importing cathode-ray tubes to Tijuana from the US (Mendiola 1999, 33). However, in many instances it proved difficult to establish an adequate Mexican supply base within the short timeframe of five years until the abolishment of the maquiladora benefits (Nihon Keizai Shinbun, 5 January 1996). NAFTA's rules of origin not only affected the maquiladora sector. Despite Nis-san's long presence in Mexico and its domestic market orientation, the company was forced to adapt its strategy to changing circumstances. When N A F T A entered into force, Nissan sourced 50 percent of its parts from within Mexico. However, 106 to raise this content to 62.5 percent, it would be difficult to import more parts from the US while still fulfilling the foreign exchange balancing requirement mandated by the 1989 auto decree (Nikkei Sangyo Shinbun, 30 July 1991). Consequently, Nissan expressed concern about the state of the Mexican parts industry that did not meet its requirements in terms of quality and efficiency (JETRO 1994, 20). A second, idiosyncratic factor forced Nissan abandon its principal orientation towards the Mexican domestic market: the Mexican currency crisis of 1994 and the resulting devaluation of the peso. In the wake of the crisis, domestic demand in Mexico bottomed out. For the US automakers in Mexico, this was no problem: not only were their operations oriented towards export, but their shipments also became more competitive with a depreciating Mexican currency. Excess produc-tion in Mexico could be exported to the US and Canada, thus balancing out costs within the North American market. This option did not exist for Nissan, which did not produce any model in Mexico that could be sold in the home market Japan. Collapsing sales forced the company to lay off ten percent of its workforce in Mex-ico. Nissan therefore began to pursue previously considered plans to use Mexico as an export base to the US (Walzer 1995). As a result, benefits N A F T A bestowed on incumbents in the form of a dif-ferential phase-out of rules of origin lost their lustre. This moved the company's position closer to that of other Japanese firms. After the merger with Renault, Nissan changed its strategy towards the global integration of operations. Whereas the company had previously relied on its network of Japanese-owned suppliers or joint ventures with Mexican manufacturers, it shifted to low-cost global sourcing, including Renault's European suppliers. Nissan planned to increase vehicle pro-duction from 200,000 to 330,000 units by 2001. In addition to the Sentra models for the US market, the smaller Renault models Megane and Clio began rolling off production lines in Nissan's factories. However, the merger with Renault also implied a specialization and concentration of sourcing efforts that conflicted with the 70 percent N A F T A content Nissan achieved so far in its Mexican production (Nikkan Kogyo Shinbun, 1 March 2000; 17 June 2001). Although N A F T A raised the costs for newcomers, several Japanese firms de-cided to expand their operations in Mexico. Honda announced in 1994 that it would start assembling cars in Mexico. Toyota began talks with Mexican officials about the same time (Moreno Brid 1996, 26), but did not announce the set-up of production facilities until 2002, when it opened a plant in Tijuana to produce SUVs for the whole North American market, aiming for the production on a scale of 30,000 units by 2004 (Ward's Auto World, 1 September 2003). Major parts producers such as Denso and Toyota Tsusho followed their most important buyer, 107 investing US$ 6.6 million in a supplier park (Nihon Keizai Shinbun, 23 September 2002; 19 August 2003). Mexico became a key production platform for Japanese firms, making it turn more important to receive the necessary political backing from Japan to remain competitive. This required a major change in Japanese trade policy. 5.2.3 A Regional Trade Policy Option for Japan Japanese trade policy reflects the \"fragmented character of state authority\" (Calder 1988, 528) vis-a-vis domestic actors in Japan. External economic policy is con-ducted by several ministries at the behest of their respective constituencies, with only the Ministry of Foreign Affairs (MOFA) balancing the various interests. A n analysis of Japanese policy outcomes based on interest groups therefore faces par-ticular problems. Mulgan's assessment of agricultural lobbying forces (George Mulgan 1999, xviii) holds true for other actors as well: Many bodies cannot be clearly delineated as either public or private, operating as auxiliary agencies of government and interest group representatives at the same time. Agricultural interests organized in the Japan Agricultural Cooperatives Group (JA Group, formerly known as Nokyo), acting through its central union Zenchu link protectionist forces with the ruling Liberal-Democratic Party. They have fre-quently prevailed in setting policy, often by directly lobbying parlamentarians and the Ministry of Agriculture, Forests and Fishery (MAFF) (George Mulgan 1999, 565-570). Other lobby groups do not enjoy comparable access to members of the Japanese Diet. Their activities therefore focus on formal representations to advisory councils (shingikai) and the Vice-Ministers of the respective ministries, personal appeals to high-level bureaucrats, the P M and Cabinet ministers, publi-cations and media contacts. In the case of trade policy, governmental organizations such as the Japan Ex-ternal Trade Organization (JETRO), quasi-non-governmental organizations like the Japan Institute for Overseas Investment (JOI) as well as industry associations frequently survey Japanese companies to channel concerns to M E T I . 2 5 Nippon Keidanren, the peak industry association, publishes formal requests (yosei). How-ever, many lobbying efforts leave no paper trail and can only be reconstructed on the basis of the information provided by the actors involved. This fact, com-bined with a preference of firms to let bureaucrats take the lead in public, has 25In 2000, the former Ministry of International Trade and Industry (MITI) was reorganized and renamed Ministry of Economy, Trade and Industry (METI). 108 furthered the view of powerful METI bureaucrats.26 METI 's capacity to initi-ate policy changes, though, has long been questioned in other domains, because information asymmetries abound between the small numbers of bureaucrats com-pared to numerous, sometimes multinational firms (Okimoto 1989). In terms of trade policy, this implies that bureaucrats often know very little about the specific interests of exporters and investors in foreign markets, unless firms provide the information themselves. Moreover, Japanese elite bureaucrats can still hope for lucrative positions in the private sector upon retirement, known as amakudari or descent from heaven (Curtis 1999,233-234), a practice that is likely to make them receptive to industry demands. A few key bureaucrats in both M E T I and M O F A must be credited with translating firm demands into the pursuit of preferential trade agreements. However, this policy only emerged after other options had been exhausted and the European Union demonstrated the viability of extra-regional, defensive PTAs. During most of the 1990s, Japan's regional trade policy initiatives were limited to the Asia Pacific Economic Cooperation (APEC), an intergovernmental forum intended to liberalize trade and investment around the Pacific Rim. Originating in a MITI initiative, Australia put forth the official proposal (Berger 1999). While the forum received considerable political and scholarly attention (Beeson and Jaya-suriya 1998; English 1999; Gallant and Stubbs 1997; Ravenhill 2000; Terada 1998), its lack of credible commitment mechanisms made trade and investment liberalization difficult to achieve. Although the 1994 A P E C summit in Bogor de-clared the goals of free trade in the region by 2010 for developed and 2020 for developing countries, the proposed unilateral liberalization skirted domestically sensitive sectors. In consequence, the 1994 Individual Action Plans and the 1997 initiative for an early voluntary sector liberalization (EVSL) foundered on a diver-gence of interests: under domestic pressure, the Japanese government was unable to commit to an opening of its agricultural market, while the US insisted on re-ciprocal concessions (Ravenhill 2000). Given this background of resistance to regional initiatives and the limited Japanese FDI in Mexico, few expected an FTA between the two countries to materialize. Motivating the Japanese trade bureau-cracy into action required lobbying by firms over a period of several years. 26This view goes back to the classic work by Johnson (640, 1982) It is extended to trade policy in Ogita (2003) 109 5.2.4 Lobbying Efforts in Mexico and Japan Concurrent with their investments, Japanese firms voiced their concerns over N A F -TA to quasi-non-governmental organizations. As in the case of the EU-Mexico FTA, the most vocal supporters were firms with investment interests in Mexico and exporters whose sales supplied these investments. Exporters who sold to Mexican buyers showed much less interest, since the Mexican market in itself is too small to warrant separate lobbying efforts. However, since Japanese farmers feared Mexican imports and had no counterbalancing export interests, protection-ist agricultural interests proved to be much stronger than in the EU-Mexico case. A 1997 JOI hearing found that 35 percent of Japanese firms in Mexico sourced most of their inputs from Japan, citing a lack of local suppliers and high technolog-ical and quality requirements as reasons. Of these firms, a \"considerable number\" worried about the abolishment of the maquiladora benefits under NAFTA. Au-tomotive and machinery manufacturers were mostly concerned about NAFTA's rules of origin and the low technological standards of Mexican parts producers. At the same time, though, many firms cited advantages that N A F T A had brought for their Mexican investment, mostly in the form of lower taxes for parts imports from the US (Japan Institute for Overseas Investment 1997, 13). The immediate solution to the higher cost burden of Japanese firms would have been a unilateral tariff reduction by Mexico. To this end, both MITI and associa-tions of Japanese firms in Mexico addressed their concerns directly to the Mexican government during the negotiations of NAFTA. The Japan Automobile Manufac-turers Association (JAMA) issued a statement that criticized the proposed rules of origin for automobiles in N A F T A (Jiji Nyusu, 13 August 1992). In Septem-ber 1992, Japanese trade minister Watanabe asked the Mexican undersecretary for trade at SECOFI, Pedro Noyola, to offer tariff relief on parts to compensate Japanese firms for the loss of maquiladora benefits (Japan Economic Newswire, 29 September 1992). From 1994 on, Japanese firms organized in the Japan Maquiladora Associa-tion lobbied the Mexican government directly for tariff reductions, again arguing that N A F T A manufacturers would be unable to supplant their Japanese counter-parts within a time frame of ten years (Nihon Keizai Shinbun, 5 January 1996). The Japanese Chamber of Commerce and Industry, representing both maquiladora firms as well as smaller suppliers to Nissan's operations in Mexico, likewise en-gaged in lobbying efforts. While it obtained zero-tariff access for semiconductors and integrated circuits by convincing SECOFI bureaucrats that these parts could not be sourced in Mexico (Byrne 1994), it failed to achieve the same for the broad 110 range of electronic parts and machinery that its member companies hoped to im-port tariff-free. At the same time, the development of a broad base of Mexican suppliers to the maquiladora failed to materialize (United States General Account-ing Office 2003, 35). In response to these efforts, SECOFI promised a sectoral promotion program-me that offered selective tariff reductions (Programa de Promotion Sectoral, PRO-SEC), to be launched in 2000. PROSEC granted reduced tariffs of 5-10 percent on parts imports for the maquiladora sector upon request by individual firms. How-ever, it was criticized by the maquiladora industry as too limited in the coverage of products (United States General Accounting Office 2003, 39), too burdensome for individual firms and unpredictable in its application (Japan Machinery Center for Trade and Investment 2001). Lobbying in Japan also put the government under pressure. Already in a 1991 interview, the regional managers of several Japanese firms in Mexico referred to N A F T A as an \"assault on Japanese firms\" (Nikkei Sangyo Shinbun, 30 July 1991). A Japanese government panel at the Economic Planning Agency (EPA) criticized NAFTA's rules of origin in August 1993 (Japan Economic Newswire, 4 August 1993). Shortly after NAFTA's passage by the US Congress in November, executives of Hitachi and Nissan expressed worry about its discriminatory nature (Japan Economic Newswire, 18 November 1993). The Japanese Federation of Industries (Keidanren) commented negatively on the GATT review of N A F T A in 1994. In 1997, the Japanese Institute for Overseas Investment cited the high cost of importing machinery into Mexico from Japan as problem for Mexico's investment climate (Japan Institute for Overseas Investment 1997). However, the link between investor demands and a possible bilateral trade deal had yet to be made explicit. This responsibility fell on the Mexican side: during a lunch meeting at the Imperial Hotel in Tokyo in summer of 1998, Blanco informed the JETRO Chairman and former MITI Vice-Minister Hatakeyama Noboru that he had concluded a framework agreement with the E U that would lead to an FTA. He invited Hatakeyama to come to Mexico City to discuss the possibility of a Japan-Mexico FTA (Hatakeyama 2003b, 24). This conversation reportedly convinced Hatakeyama that Japan had to pursue bilateral options to maintain its competitive-ness, and motivated him to raise the issue with Prime Minister Obuchi and MITI Minister Kaoru Yosano. Still, signing an FTA, especially with a Latin Ameri-can country, represented a major change in Japanese foreign economic policy. 2 7 Yosano therefore first asked the International Trade Policy Bureau of MITI under Interview with former senior MITI official, Tokyo, November 2002. I l l Director General Konno Hidehiro to prepare a policy study on a Japanese FTA op-tion that would help establish a consensus in the Ministry. In addition, the Japan External Trade Organization (JETRO) undertook a study, based on seven con-sultative meetings with business representatives between February and December 1999, that emphasized the same factors identified by industry lobbyists (2000, 4-5): Mexico's high tariffs are a problem. The unadjusted average tariff in Mexico stands at a high 13.2% (1998). (...) As a result, Japanese exports are levied higher tariffs than FTA signatories. (...) This means that the bond system would not be used for imports of parts and materials that would then be exported to N A F T A regions, starting from the end of 2000. (...) Yet, the alternative measure [PROSEC] does not offer the same benefits as the previous scheme. (...) Fur-thermore, supporting industries such as local auto parts and electronic and electric parts are nonexistent, and transport-related infrastructure is underdeveloped. These factors are raising the cost of local produc-tion activities. In 1998 and 1999, President Zedillo and Trade Minister Blanco spoke on three occasions about the impending changes in the maquiladora system before Japanese business audiences at Keidanren. Using the opportunity, President Ze-dillo specifically floated the idea of a bilateral FTA between Japan and Mexico (Nihon Keizai Shinbun, 15 September 1998; 7 October 1998; Nihon Sangyo Shin-bun, 2 March 1999). Keidanren's Japan-Mexico Economic Committee supported the proposal in a study published in April 1999 (Keidanren 1999). Parallel efforts were undertaken at the Economic Affairs Bureau at the Min-istry of Foreign Affairs (MOFA) under Director General Tanaka. However, unlike MITI with its close contacts to Japanese multinational firms with considerable overseas investment, M O F A was more concerned about balancing the interests of more protectionist ministries, as well as the implications for Japanese foreign pol-icy in general.28 MITI's White Papers on Foreign Trade began over time to reflect the changes in attitudes towards FTAs, from carefully positive to fully supportive. In comparison, the M O F A Diplomatic Blue Books are more cautious and trail the METI Papers by a year in their positive assessments (Ogita 2002, 5-9). In order to establish the parameters of a possible agreement, M E T I Minister Hiranuma Takeo and Mexican Economic Minister Derbez launched a bilateral Interview with MOFA official, Tokyo, March 2003. 112 study group (Nichi Boku kyudo kenkyukai) of academics, trade bureaucrats and business leaders, the latter representatives of firms with investments in Mexico, such as Sharp and Mitsui-Bussan, a general trading company {sogo shosha). The study group met seven times between September 2001 and July 2002. In the final report, the Japanese side brought up the aforementioned concerns: an increase in Mexican tariffs for parts and machinery, especially regarding the electronics industry in the maquiladora sector, a loss of competitiveness vis-a-vis firms from N A F T A countries and the E U , the ineffective nature of the PROSEC scheme, and the remaining restrictions on investment in Mexico. The report judged that \"since an FTA can be concluded between\/among a small number of countries, it may become an effective measure to promptly solve the present tariff problems under the situation in which Japanese enterprises are subject to serious disadvantages due to Mexico's conclusion of N A F T A and the effectuation of the EU-Mexico FTA\" (Japan-Mexico Joint Study Group 2002, 18). 5.2.5 Negotiating Free Trade between Japan and Mexico Actual negotiations between Japan and Mexico did not take place until 2002. For \"tactical\" reasons, MITI and M O F A bureaucrats first sought an FTA with Singa-pore:2 9 The city-state did not export agricultural products in significant amounts, allowing a dry run for negotiators and establishing the precedent of a bilateral agreement. However, the negotiations with Singapore already foreshadowed the problem of liberalizing agricultural trade. Due to vehement lobbying by agricul-tural groups, the talks remained stalled for weeks because of Singapore's exports of goldfish and orchids, creating friction in the Japanese negotiating team and leading to the specific exclusion of these two products in the final agreement (New York Times, 27 January 2002). Nevertheless, the M O F A Treaty Bureau managed to hold its line that an FTA had to cover 90 percent of all trade to fulfill the GATT Art. X X I V requirement of liberalizing substantially all trade, a position at that point still considered inviolable. 3 0 Formal negotiations between Japan and Mexico began on 18 November 2002 in Tokyo. Unlike Singapore, Mexico is a major competitive exporter of agricul-tural products, in particular pork and citrus fruits. This prompted an immediate reaction from L D P constituents in rural areas, including two visits by the chair-man of the Japanese association of agricultural cooperatives (JA group) to the L D P Interview with former senior MITI official, Tokyo, December 2002. 'Interviews with senior MOFA official, Tokyo, March and December 2003. 113 headquarters (Nihon Keizai Shinbun, 12 December 2002; Nikkei Weekly, 19 May 2003). M A F F officials repeatedly stated that they could not compromise on the issue, even at the risk of failure of the negotiations with Mexico (Nihon Keizai Shinbun, 9 May 2003). To counter these moves, Keidanren, the Japan Chamber of Commerce and Industry, the Japan Association of Corporate Executives (keizai doyukai) and the Japan Foreign Trade Council issued a joint plea, timed to co-incide with a working level meeting of both sides in Tokyo on August 11-15, to conclude the FTA with Mexico as soon as possible (Nihon Keizai Shinbun, 6 August 2003). Separately, Keidanren issued several policy papers calling for the rapid con-clusion of a free trade agreement to counter NAFTA's effects. The association stressed the need to be able to import components, machinery, and chemical prod-ucts from Japan to shorten lead times in starting up production in Mexico. Its joint business council with Mexican executives led efforts on both sides of the Pa-cific (Keidanren 1999).31 Following the conclusions of the Japan-Singapore FTA, business leaders intensified their lobbying efforts at the Ministerial level. Nissan and Toyota demanded strong efforts from M E T I in several meetings with trade bureaucrats during 2002 and 2003. In particular Nissan took a similar position as Volkswagen had done during the EU-Mexico negotiations. Unlike Volkswa-gen, though, the company emphasized that it did not want to export to Japan from Mexico, but that it needed an FTA to preserve its relations with Japanese suppliers. Toyota stressed its investment plans in Mexico that could become unprofitable if Mexican tariffs on car parts were not lowered.3 2 Similar to ACEA's efforts, J A M A showed particular interest in the negotia-tions, but mostly argued on behalf of the export interests. Its president and chair-man of Honda Motor Corporation, Munekuni Yoshihide, lobbied L D P Diet mem-bers directly in a meeting on October 3. Munekuni cited the Mexican threat to raise tariffs on imported cars to 50 percent (Nihon Keizai Shinbun, 17 October 2003). M E T I officials, however, gave Nissan's role much greater weight, \"since the 20,000 or so cars that Japan exports to Mexico pale in comparison to the hun-dred thousands Nissan can produce there and sell in the US,\" which in turn would generate business for Japanese suppliers and machine tools producers.33 This view was supported by lobbying efforts by the association of the machine tools industry, the Japan Machinery Center for Trade and Investment (JMCTI, 3 1 See also http:\/\/www.keidanren.or.jp\/japanese\/policy\/2000\/016.html 32Interviews with METI officials, January 2003; Interview with MOFA official, December 2003. 33Interviews with METI officials, January 2003. 114 Nihon kikai yushutsu kumiai). In comparison to European upstream producers, the Japanese machine tools industry took longer to define its interests. Although broadly supportive of a bilateral deal, JMCTI contracted a detailed study in 2001 of the benefits of a bilateral trade deal to identify which demands ought to be put forward to METI . The study came to the conclusion that an FTA with Mex-ico could resolve the specific problems created by the N A F T A rules of origin in conjunction with high Mexican M F N tariffs. JMCTI representatives pressed this point in consultations with M E T I bureaucrats.34 In addition, a related association, the Japan Business Council for Trade and Investment Facilitation (bdeki toshi enkatsuka bijinesu kyogikai), published a sur-vey of member companies in October 2002 to identify problems of investors in Mexico that was submitted to M E T I . 3 5 Car parts manufacturers noted \"a 25 per-cent tariff that its competitors from the US, E U and other countries Mexico has an FTA with do not have to pay. (...) We request that an FTA be signed as soon as possible.\" Other firms referred to problems to fulfill the trade balancing require-ments of the auto decree and the failure of the PROSEC programme to offer the same benefits as pre-NAFTA duty drawback schemes. Machinery firms specifi-cally called for the conclusion of an FTA because they felt they could not compete given existing Mexican tariffs (Hideya 2002, 16). The lobbying in support of an FTA confirms the expectations. Like in the case of the EU-Mexico FTA, the interests of firms with investment were given a higher priority by negotiators. Crucial support for a bilateral deal came primarily from intermediate and capital goods producers, who were hurt the most by a loss of competitiveness. Despite these lobbying efforts, the negotiations nearly collapsed in October 2003, when the Japanese side refused to open its agricultural market for critical exports like pork and orange juice. Remaining differences could finally be ironed out during meetings in early March 2004 (Nihon Keizai Shinbun, 9 March 2004), leading to the official signing of the FTA during a visit by Prime Minister Koizumi to Mexico City in September 2004. Japan offered only moderate liberalization of its agricultural market, retaining quotas for oranges and pork. The tariff formula for pork offers access at a rate of 4.3 percent only if is sold at the equivalent of the price demanded by Japanese farmers, the remainder still faces a 49 percent tariff (Toyo Keizai Weekly, 29 November 2003). Overall, the agreement only liberalizes Interviews with interest group representatives, Tokyo, February 2003. Interview with interest group representatives, Tokyo, February 2003. 115 86 percent of bilateral trade. A deal could thus only be struck by reneging on the earlier M O F A position that substantially all trade had to be covered. 5.3 Alternative Accounts Did the E U or Japan reach out to Mexico because multilateral liberalization did not make progress? In the case of the E U , at least two arguments stand against this thesis. First, the E U made a turnaround in its policy toward Mexico. Dur-ing the early 1990s, European investors showed little interest in Mexico. Firms with investment in Mexico were still oriented toward the domestic market. In this situation, even the limited potential of Mexican agricultural exports proved enough of a stumbling block to prevent bilateral negotiations. European firms start to lobby only after the effects of N A F T A on outsiders became evident, at which point their interests outweighed domestic forces. Second, the E U Commis-sion itself perceived the multilateral negotiations as much more important than any bilateral agreements. According to E U officials, Trade Commissioner Lamy made it clear upon taking up his post that under his direction, no new bilateral negotiations would be initiated before the conclusion of the Doha round. 3 6 With-out pressure by European firms in services and manufacturing, the E U might have negotiated bilateral deals in its immediate region, but would hardly have reached out to Latin America. Japanese decision-makers first considered the Japan-Mexico FTA in summer of 1998, before the failed attempt at starting a WTO Millennium round. This suggests that problems at the multilateral level were of lesser importance. More-over, two Japanese ministries remained strongly committed to the WTO, albeit for different reasons. M A F F preferred the WTO precisely because it was slower in its liberalization of agriculture and because protectionist forces had allies in the E U agricultural sector.37 Several senior M O F A officials held the view that the WTO was more important than ever, since it could help Japan to monitor and enforce China's commitments to trade liberalization.3 8 Likewise, M E T I officials attributed Japan's push for rules on FDI in the Doha round to the need to monitor China better.39 This corresponds closely to findings by Searight (1999), who ar-gues that multilateralism became more important in Japanese trade relations with 36Interviews with EU officials, Brussels, June 2004. 37Interview with MAFF official, Tokyo, March 2003. 38Interviews with senior MOFA officials, Tokyo, February 2003. 39Interview with senior METI official, Tokyo, March 2003. 116 the US. While Japan pursues FTAs with emerging markets, it remains committed to the multilateral trade regime to guide its relations with major economic powers. The Japanese move toward an FTA with Mexico invites the second alternative interpretation of policy leadership by its bureaucracy rather than the private sector (Ogita 2003, 243-244). M E T I itself claims in its 2001 White Paper that besides avoiding the exclusion from other agreements, Japan's FTAs would advance new trade rules, maintain momentum for multilateral free trade, and promote domestic reform (Ministry of Economy, Trade and Industry 2001). Although sound at first glance, these arguments rest on weak evidence. While no business organization specifically requested an FTA until 1998, neither did M E T I trade bureaucrats. The unexpected proposal for an FTA came from the Mexican side, just like in the case of NAFTA. A look at the process shows that the Mexico FTA was the first to be considered, with the Singapore Agreement only a tactical move of bureaucratic politics unsupported by business interests. M E T I bureaucrats primarily sought an FTA on behalf of their constituents, Japan's multinational firms. Japanese demands in the negotiations matched closely the problems previously cited by firms. Yet as an export market, Mexico ranks only low in Japanese priorities, as METI officials readily admit. One of the most senior former MITI officials referred to the Japan-Mexico FTA as \"purely defensive.\" 4 0 Nor were Japanese FTA initiatives launched to promote domestic reform in Japan. When agricultural liberalization stalled the Japan-Mexico negotiations, chief FTA proponent Hatakeyama reversed the logic: Japan had to reform itself to be able to pursue bilateral deals (Hatakeyama 2003a). This indicates that business demands and support were crucial for the FTA policy to become viable. 5.4 Conclusions Both case studies underscore the centrality of foreign direct investment for the proliferation of preferential trade agreements. Export interests matter most when directly linked to such investment, as in the example of European chemical indus-try interests in Mexico. Investors who rely on an export platform such as Mexico to serve a third-country market are highly sensitive to cost. In consequence, in-vestor firms and their upstream chain of producers are the most vulnerable to exclusion from a key export platform, and lobby the most for countering agree-'Interview with former MITI official, Tokyo, November 2002. 117 merits. The lobbying behaviour of investors such as Japanese electronics firms, Volkswagen and Nissan supports this account. For conventional exporters of manufactured or agricultural goods in developed countries, emerging markets are small and less important. For these firms, PTAs appear to offer no specific benefit over multilateral liberalization\u2014in fact, neither for the developed or the developing country. If they are shut out of these markets because of high M F N tariffs, bilateral agreements may be easier to achieve, but offer no intrinsic benefits over multilateral liberalization. By contrast, for industries such financial services, in which market structures penalize a late entry, the discriminatory effects of PTAs with services liberaliza-tion are more pronounced. The case of the European financial services industry in Mexico shows that these industries seek parity in market access because they face a higher cost of exclusion. As elaborated in the next chapter, neither N A F T A nor the counter-agreements are idiosyncratic events. Although the investor interests depend on the specific features of the emerging market country, a competitive dynamic unfolds in a grow-ing number of cases in the Americas and Asia. 118 Chapter 6 The Unfolding Competition Mexico attracted investment in services and manufacturing across a range of busi-ness activities, much of it drawn to a huge pool of low-cost labour in proximity of the US market. As the previous chapters have shown, not only US firms, but also their competitors from non-member countries invested massively in Mexico. The competitive disadvantages created by N A F T A thus affected a broad coalition of firms in Europe and Japan, creating a powerful constituency in favour of defen-sive agreements. Importantly, these interests focused on foreign direct investment rather than exports of finished goods. But are investment interests crucial in moti-vating FTAs? Was Mexico an outlier, given its high M F N barriers and proximity to the US, whereas under normal circumstances, FTAs secure export markets? As the cases in this chapter demonstrate, Mexico was by no means excep-tional. Within few years it became evident that N A F T A was only the first in a wave of FTAs between developed and developing countries. Once policy makers recognized this fact, they became increasingly concerned about the competitive effects of FTAs. The previous chapter has argued that bureaucrats in the E U and Japan explicitly considered these effects, relayed to them by firms and lobbyists. This chapter shows that as they began pursuing FTAs with developing countries, US policy makers in turn considered bilateral options. To offer further tests the argument of a competitive dynamic driven by foreign direct investment and accel-erated by the strategic interaction of home states, this chapter surveys two kinds of cases in which either the services or the manufacturing sector featured promi-nently. The agreements with Chile show that even very open markets with low M F N barriers can trigger a competitive dynamic, as long as firms engage in foreign direct investment. Chile thus serves to counter an alternative explanation: with its low tariffs and small domestic market, we would not expect an FTA if access for finished goods exports were the main concern. If even the limited investment interests of those firms active in Chile motivated FTAs despite political obstacles, then this offers further support for the central hypothesis of this study. Service 119 investment in Chile, a comparatively small market, generated business support for an FTA in the US and the E U . In the services sector, imperfectly competitive market structures help drive bilateral agreements, even if a host country has comparatively low multilateral barriers to trade and investment. During the second half of the 1990s, this situa-tion arose when Chile attracted significant amounts of services FDI from Spanish companies through concessions from the Chilean government. By acquiring most desirable takeover targets among Chilean firms, these investors excluded com-petitors. For European firms, an FTA with Chile therefore primarily promised a locking-in of commitments. These developments spurred their US competitors into action. While earlier proposals from the Chilean side to accede to N A F T A had foundered because of the failure of the Clinton administration to obtain Con-gressional authority to negotiate, defined more precisely below, firms now made the case to US lawmakers that they needed an FTA to become competitive when entering the Chilean market. Although the debate around the authority to nego-tiate was clearly a domestic political issue in the US, the arguments put forth by lobbyists suggest that competitive considerations became increasingly important. To maximize variation on the dependent variable, the analysis includes a neg-ative case: Chile sought an FTA with Japan in 2001. The proposal was favoured by several actors close to the Japanese government and easier to achieve than the FTA with Mexico, since Chilean products threatened fewer import-competing industries in Japan. Support from business interests, however, was insufficient to generate political momentum. No Japanese firms stood in direct competition with US and European counterparts or were put at a disadvantage by the FTAs these two actors sought with Chile, undermining the case for an agreement be-tween Japan and the Southern Cone country. Notwithstanding the \"icebreaker\" agreement between Japan and Singapore, a political move to overcome bureau-cratic opposition in M E T I and M O F A , interests by investing firms are the primary driving force of North-South-FTAs. As the first North-South-FTA, N A F T A set an example of how to facilitate the use of a developing country as export platform while raising costs for com-petitors. The previous chapter argued that Japanese policy makers heeded the lessons of N A F T A and transferred them to boost the competitiveness of Japanese firms. In the case of Thailand, Japanese automotive firms strongly supported an FTA because they were reorienting production towards exports, once the Asian Financial Crisis changed the environment for Japanese M N C s in Southeast Asia. In an unprecedented development, Japanese firms began to re-export parts and complete vehicles from Thailand back to the home market. Reflecting these de-120 velopments, the FTA initiatives advanced by METI and M O F A provide backing for firm strategies similar to those of US manufacturers in Mexico. Whereas in the past, Japanese firms primarily manufactured vehicles for protected local markets in the A S E A N countries, all major firms have started to expand and retool their operations in Thailand for overseas sales. Although Japanese firms do not primar-ily seek an FTA to prevent US firms from threatening the Japanese home market, they voice concerns about the better treatment accorded to foreign investors from other countries. Unsurprisingly, Japanese automotive firms and their suppliers have emerged as the key supporters of a Japan-Thailand FTA, drawing them into a battle with protectionist forces in Japan. The chapter first presents an overview of the evolution of Chile as FDI host. It then analyzes the failed bid for N A F T A accession, the negotiations between the E U and Chile, and moves by the US to attain parity for its investors through a bilateral FTA. After comparing the actual EU-Chile and US-Chile agreements, it briefly surveys the failed bid for a Japan-Chile FTA. Finally, the chapter offers an account of the initiative for a Japan-Thailand FTA, and draws conclusions from the case studies. 6.1 The Far Side of the World: FTAs with Chile Just like Mexico, Chile has sought to actively promote itself as an investment location since the early 1990s, and to expand its exports by signing free trade deals with developed countries. Unlike most Latin American neighbours, how-ever, Chile carried with it hardly any legacy of import-substitution industrializa-tion. Under the military regime from 1973 to 1989, the country's economy was transformed and fully oriented towards international markets. The liberalization of imports proceeded apace, following the prescriptions of a group of tecnicos, economists trained at the University of Chicago (Teichman 2001). Import tariffs were reduced to a flat rate of eleven percent, with the exception of a brief period between 1982-84, when Chile struggled with the Debt Crisis that engulfed most of the Third World. As a result of these policies, Chile diversified its exports away from an al-most exclusive dependence on copper, reducing its share in exports from close to 80 percent in 1973 to around 45 percent during the 1990s (Meller 1996, 275, cited in Grugel and Hout 1999, 62). Rather than growing manufacturing exports, how-ever, diversification meant an increase in overseas sales of other primary goods such as wine, fruit, vegetables and fish. Chile lacks any a significant manufac-121 turing base, whether foreign-owned or domestic. At the same time, the relatively small market of 15 million people with a per capita income of only US$ 5,000 (in purchasing power terms) in 1995 shows the limited importance of Chile as an export market. In the second half of the 1990s, however, the country became an important location for foreign direct investment, heavily concentrated in the service sector, from Europe and the US. 6.1.1 Foreign Direct Investment in Chile In comparison with other emerging markets, Chile has been enormously success-ful in attracting FDI. In relation to its GDP, Chile hosts more FDI than the other three big Latin American economies, as shown in figure 6.1. 70 0-1 , , , , , , , , , , , , , , , , , , , , , 1980 1985 1990 1995 2000 Year Figure 6.1: FDI stock to GDP ratio, Chile, 1980-2003 Source: World Development Indicators (2004) The military regime liberalized foreign investment already in 1974 by enacting Decree Law 600 (DL 600), enshrining non-discrimination and access to virtually all productive activities. D L 600 reserves the right of the governmental Foreign Investment Committee to screen investment. Investing firms sign a contract with 122 the Chilean government that guarantees access to the necessary foreign exchange and optionally locks in an effective tax rate of 42 percent (UNCTAD 2004). In practice, Chile's economy has been open to foreign investment since the mid-1970s. Most FDI simply reacted to natural resource endowments and comparative advantage. Chile is blessed with the world's biggest deposits of copper ore, most of which can be mined close to the surface, as well as precious metals and natural gas resources. Consequently, mining attracted the lion's share of FDI, most of which came from the US and Canada. Yet from the mid-1990s on, investors from the European Union began acquiring Chilean services companies on a broad scale. The vast majority of these acquisitions were undertaken by Spanish companies: of the cumulative FDI from the European Union, about 80 percent are from Spain. Since the late 1990s, Spanish investment represented the second-biggest stock of FDI in the country, closing the gap to the US, as shown in Figure 6.2 ( E C L A C 2001, 106). 1994 2003 Figure 6.2: Shares of individual countries in Chile's FDI stock. Provisional figures for 2003. Source: Comite de Inversiones Extranjeras de Chile. http:\/\/www.foreigninvestment.cl. Spanish investors focused on three industries: financial services, energy, and telecommunications. In all three, their massive entry resulted in commanding shares in market and ownership that put competitors from other countries, in par-ticular US firms that attempted to enter markets later, at a disadvantage. The most striking wave of acquisitions took place in the banking sector. The same two Span-ish banks that invested broadly in Mexico, B B V A and B S C H , also bought major stakes in Chilean banks. Both banks faced a dilemma in the E U market, where a lack of candidates for a takeover limited the opportunities for growth. Santander, one of the banks that later merged into B S C H , began its expansion with a small 123 share in Banco Santiago, the biggest Chilean retail bank, that it gradually built up over the following years. In 1996, it acquired a controlling share in Banco Santander Chile, the second-biggest bank ( E C L A C 2002, 125). In 1998, Span-ish competitor B B V A bought Banco Hipotecaria de Fomento, now B B V A Banco BHIF (Calderon and Casilda 1999, 24). The rapid expansion contributed to the importance of Latin America in these banks' portfolios: by 1999, B S C H had 29 percent of its assets in the region, B B V A 21.1 percent (Calderon and Casilda 2000, 77). Both banks used their subsidiaries to expand into Mercosur, testimony to Chile's role as entry point to the rest of the Southern Cone (Calderon and Casilda 1999, 32). In addition, both BSCH's and BBVA's subsidiaries are active in the pension fund market. Chile's registered pension plans, called AFPs, are limited to 20 percent foreign investments, and can only be managed by subsidiaries of foreign banks, not by branch operations. This regulatory element and the dominant posi-tion of the two Spanish banks facilitated their access to this market: B B V A holds 40 percent of the biggest Chilean fund A F P Provida, B S C H 100 percent of the fourth-biggest fund A F P Summa Bansander. The concentration of market control in the hands of these two banks also worried Chilean regulators, who first contem-plated legislative proposals that would have limited market shares to 20 percent per bank, then proposed a veto clause for future takeovers ( E C L A C 2000, 128\u2014 29). Eventually, the Chilean central bank and B S C H came to an agreement that the Spanish firm would reduce its stakes in one of the two local banks it controlled. Nevertheless, the position of these two banks prevented viable competition. In ad-dition, both banks hold minority stakes in other Spanish enterprises that are active in the region (Calderon and Casilda 2000, 76). Such investment focused mostly on the telecommunications and energy industries. Just like other service industries, energy companies had been privatised much earlier in Chile than in other Latin American countries. With the first steps at liber-alization in neighbouring countries, Chilean firms successfully branched out into these markets. By acquiring a Chilean energy provider, foreign investors therefore not only gained access to local, but also regional markets, and to important know-how about the business opportunities in these countries ( E C L A C 2001). Most important among these investors is Endesa, a Spanish electricity provider, Chile's biggest foreign firm on the basis of sales. Outbidding the US firm Duke Energy in May 1999, Endesa acquired a 60 percent stake in Endesa Chile. Through its Chilean subsidiary, Endesa also holds shares in energy companies in Argentina, Brasil and Peru (Rozas Balbontin 2001, 77-84). Other foreign firms that made major acquisitions in the Chilean energy sector were TotalFinalElf (France), AES 124 and Pennsylvania Power & Light (USA), and HydroQuebec (Canada). Almost 80 percent of the accumulated investment in the sector occurred between 1996 and 1999 (72-73). In the telecommunications sector, two firms came to dominate the market. In 1996, Telefonica de Espana bought local telephone operator C T C Chile, gaining almost 85 percent market share ( E C L A C 2000, 121-22), while Telecom Italia ac-quired the smaller Entel Chile. Telefonica expanded into Chile for similar reasons as the Spanish banks: only fully privatised in 1997, it had limited opportuni-ties in European markets dominated by the global giants Vodafone and Deutsche Telekom, but had a competitive advantage when entering Latin American markets based on familiar language, legal and cultural backgrounds. Competing US firms chose not to enter the market at the time, with the exception of Bell South, which built up a mobile phone network in Chile, but later withdrew from the market and sold its subsidiary to Telefonica. Similar to the regulatory process in most developed countries, Chilean au-thorities limited the number of licences for mobile telephony. This had the unin-tended effect of benefiting established, European operators. E U firms preferred the European G S M standard and are making plans for the introduction of their own \"3rd Generation\" technology (W-CDMA), while US firms hope to establish the CDMA2000 standard that has already been successfully introduced in Japan. Entering the market with a different standard would raise costs, because existing networks would have to be refitted, and worked against competition because con-sumers could not use the same telephones when switching providers. Regulatory authorities were therefore inclined to prescribe the adoption of common standards as precondition for licences ( E C L A C 2000, 181). This issue would later reappear in the FTA negotiations between the US and Chile. In all three industries, European firms established dominant positions through the acquisition of Chilean firms. No trade or investment agreement was neces-sary, since the Chilean government freely handed out the necessary concessions in accordance with D L 600. Concurrently, it made less far-reaching commitments under the GATS: Foreign Investors could obtain a concession and did so in prac-tice, but the government was under no legal obligation. European firms used the expansion room provided by regulators to the fullest. Their dominant market share and the remaining regulatory barriers impeded the entry of competitors. US firms paid relatively little attention to the Chilean investment location at the time. This lack of business interest contributed to the delay of a project that would have opened the Chilean service sector further, possibly enabling new firms to enter the 125 market: The negotiation of Chile's accession to NAFTA, or a bilateral agreement with comparable commitments. 6.1.2 A Road Not Taken: Chile's Accession to N A F T A Demonstrating its commitment to hemispheric liberalization, Chile originally plan-ned to join N A F T A under the treaty's accession clause shortly after the agreement would come into effect. At the Summit of the Americas in Miami in 1994, the three N A F T A members officially invited Chile to join the North American trade agreement. Negotiating accession, however, proved impossible, since the Clinton administration and Congress failed to come to an agreement over the conditions under which the President would be granted fast-track authority. Since trade agreements change tariffs and thus affect revenue, the US con-stitution grants Congress the right to pass the laws implementing an agreement. Given this division of powers, countries negotiating with the US cannot be sure that an agreement will be binding, unless Congress ties its own hands and agrees beforehand to either accept or reject what the President has negotiated. Since 1934, Congress has ceded this power almost without interruption, modifying the Trade Act in 1962, 1979 and 1988 to take into account the negotiation of non-tariff barriers in the Kennedy and Tokyo rounds. This fast-track procedure restricts the role of Congress, specifically of the House Ways and Means and Senate Finance Committees, to a simple up or down vote during a limited time frame. The 1988 Omnibus Trade Act had granted Presidents Bush Sr. and Clinton the power to negotiate the Uruguay round and NAFTA. The expiration of the clause coincided with the negotiations for Chile's N A F T A accession. Building on the N A F T A precedent, USTR Kantor announced in February 1995 that the Clinton administration would negotiate agreements on labour and the environment in future preferential trade agreements, including the use of trade remedy laws to ensure compliance (Inside US Trade, 9 February 1995). This proposal met with immediate resistance from Republicans in the House. Representative Crane, Chairman of the Committee on Ways and Means, stated that fast-track would not be granted i f trade agreements were to become a vehicle for environmental goals, and that Chile's accession to N A F T A would be delayed until at least after the 1996 US Presidential election (Inside US Trade, 12 May 1995). Despite these difficulties, the Chilean government signalled its willingness to begin negotiations. Working level talks kicked off in July 1995 in Mexico City. However, it shortly became apparent that the US would be unable to commit to what it negotiated. Although the House Ways and Means Committee offered the 126 compromise of granting fast-track to negotiate labour and environmental clauses with Chile only, Senate Republicans under Bob Dole resisted this effort, possibly in light of the upcoming Presidential campaign. Unable to come to a resolution, the House of Representatives abandoned the effort to grant fast-track authority to President Clinton in January 1996 (Inside US Trade, 13 December 1995, 5 January 1996). In reaction, both Mexico and Canada decided to negotiate separate FTAs with Chile. Canada signed an agreement in November 1996; Mexico and Chile up-graded their 1990 FTA to include chapters on services and investment with a new treaty, signed in October 1998. The Chilean government considered it highly un-likely that fast-track would be granted to the re-elected Clinton administration: In March 1997, it announced that it would negotiate with the US as part of the FTA A process if and when the US President would be in the position to do so (Inside US Trade, 21 March 1997). For the remainder of Clinton's time in office, he failed to obtain fast-track authority to negotiate bilateral trade agreements. Throughout the debate, most US industries remained uninvolved in the debate, offering little support for a push to achieve fast-track. Whether because of an ac-curate reading of the political situation or limited interest in the Chilean market, an invitation for the submission of statements to the USTR on the proposed ac-cession to N A F T A elicited few responses. The partial exception was the service sector. A T & T urged US negotiators to press for a further opening of the Chilean telecommunications market and lower tariffs on related products. In this case of \"rent chain,\" the link between between service firms as investors and suppliers of telecommunications products as followers reversed the scenario of FDI in Mexico described in Chapter 3. The company's letter (1995) to the USTR referred directly to competitive considerations: Concessions under the N A F T A would have the advantage of being limited to the closed circle of N A F T A member countries. (...) Chile-an concessions may be easier to achieve in the accession negotiations and they could also provide a competitive advantage to firms in this hemisphere to the extent that the GATS negotiations do not achieve similarly favorably results. A T & T also asked that Chile be pressed to drop its investment review mecha-nism under D L 600, and that N A F T A Chapter 11 be harnessed to secure an open investment environment in Chile (Inside US Trade, 21 April 1995). During an hearing before the International Trade Commission, the American Chamber of 127 Commerce in Chile supported the further liberalization of the telecommunica-tions sector and demanded the opening of the financial services industry via ne-gotiations for N A F T A accession, since the de-facto moratorium on new banking licences had effectively shut US banks out of the Chilean market (Inside US Trade, 21 July 1995). Such competitive considerations would later reappear in the negotiations of the US-Chile FTA. For the time being, though, they were too limited to motivate a concerted lobbying effort by US service industries. This position would soon change when the E U began to negotiate an FTA with Chile. 6.1.3 Iberian Ties: The EU-Chi le FTA While the negotiations between the US and Chile stalled, the E U moved quickly to establish ties with the Southern Cone. Due to their strong investment interests, Spanish firms became the principal supporters of free trade agreements with both Chile and the Mercosur countries. Unlike the Clinton administration, the Euro-pean Commission could thus rely on domestic support in negotiating the EU-Chile FTA. At the same time, the absence of a history of GATT disputes between the E U and Chile indicated that an FTA would be easier to negotiate than an agree-ment with the US. Already in 1995, 82 percent of Chilean exports entered the E U tariff-free, while the remainder carried an average tariff of only 2.4 percent. The Commission proceeded along similar lines as during the negotiation of the EU-Mexico FTA: Building on a 1990 political agreement between Chile and the E U , principally signed to express support for Chile's democratic transition, the E U concluded a \"Framework Agreement for Cooperation\" in 1996. The agree-ment followed a European Commission recommendation (1995) in expressing a commitment to negotiate a free trade agreement. In July 1998, the Commission submitted its draft mandate to the Council. The formal start of negotiations was announced at the June 1999 EU-Latin America summit in Rio de Janeiro. As Chile had begun association talks with Mercosur, the Commission mandate orig-inally called for parallel negotiations with both Chile and the Mercosur countries (Bridges Weekly, 8 December 1999). The initiation of negotiations with Mercosur was itself reportedly triggered by concerns in European capitals about the F T A A process and its implications for E U firms (Bridges Weekly, 5 July 1999). 128 Negotiating Issues Formal negotiations of the three working groups (trade in goods and related mat-ters, services and investment, and government procurement and competition rules) began in November 1999. Unlike in the EU-Mexico-FTA, the most important E U interests lay in the service sector. Given this emphasis and the precedent of the EU-Mexico FTA, E U member states were willing to grant the Commission the mandate to negotiate service commitments on their behalf. For the same rea-son, however, the Commission was tied to a positive list approach, rather than a NAFTA-style schedule for liberalization with excluded sectors.1 More problematic was the mandate to move in lockstep with both Chile and Mercosur. Under French pressure, E U member states limited possible concessions to Mercosur by stating categorically that agricultural liberalization could only be discussed when linked to progress in the multilateral trade round (Bridges Weekly, 18 April 2000). As a result, the negotiations quickly slowed down to the same pace as the WTO round. The Commission therefore proposed to the Council to separate the talks, with the aim of concluding an FTA with Chile before the WTO round (Latin America Weekly Report No. 46, 545; Inside US Trade, 21 November 2000). Following this decision, the FTA was negotiated during ten separate rounds between April 2000 and June 2002. The coalitions emerging in the E U resem-bled those of the negotiations with Mexico: France initially opposed an FTA with Chile because of its agricultural export potential. Chile not only exports fruit that would displace imports from A C P countries, but also wines and spirits that compete directly with French products. Chile was, however, willing to guaran-tee full compliance with the geographical indications for wines and spirits used in the E U . Moreover, since the growing seasons of Europe and Chile are com-plementary, French resistance in the Council was quickly overcome by the push of Spain and, to a much lesser extent, Germany. Still, the negotiations almost collapsed after the third round, when the Chilean delegation was given the im-pression that the E U would settle for a simple trade facilitation agreement that stopped short of an FTA in accordance with Art. 24 of the GATT. This prompted two visits by President Lagos to France in April and to Spain and Germany in late May of 2001 to drum up support in the respective business communities (Global Economic Newswire\/Financial Times, 31 May 2001; E l Pais 4 June 2001). In addition, service sector firms lobbied to overcome the resistance of protectionist groups. 'interview with EU official, Brussels, June 2004. 129 Demands of European Firms European negotiating goals focused primarily on the locking-in of Chilean com-mitments to keep open its services markets. In part, this reflected the relatively liberal Chilean regulatory framework. At the same time, several European firms were concerned that a further opening of the Chilean market could do little to im-prove their competitive position, but might invite others to try to enter the market. The principal goal in financial services was to get the Chilean government to lift the 20 percent ceiling on foreign content in registered pension plans. Since almost all banks were already in foreign hands, the Chilean resistance to financial market opening originated more in a concern about the loss of regulatory compe-tence than protectionism. From the Chilean point of view, the concession would have counteracted a central aim of the FTA by potentially exacerbating capital outflows rather than attracting investment (Latin America Weekly Report No. 46, 545). Still, neither of the two Spanish banks with direct interests lobbied hard for a change in the way the Chilean central bank granted concessions to foreign firms. This led to a situation in which Commission negotiators, who wanted to use the Chile FTA to set an example of far-reaching commitments for future Mercosur negotiations, at times went beyond private sector concerns.2 Among telecommunications firms, Telefonica had come into conflict with reg-ulators because of its control of 85 percent of the local telephony market via its share in Telefonica CTC Chile. In 1999, the Chilean government passed de-cree 187 to regulate rates in \"non-competitive activities\" for a five-year period. Telefonica reacted by threatening legal action in Chile and announcing a freeze on investment, forcing regulatory authorities to revise the rate decree to meet its demands ( E C L A C 2000, 121-22). Following this experience, Telefonica individ-ually urged the Commission to \"approach a N A F T A standard\" in its investment clauses to prevent a reoccurrence.3 Further opening, however, was of little inter-est to Telefonica. In the words of one source: \"Telefonica had little to gain by inviting US operators, because its main business is not in long-distance opera-tions. [They] mainly tried to prevent undue regulatory activism by the Chileans.\" The firm therefore encouraged high-level clauses that would match BIT treaties with regard to expropriation, regulatory takings, and transparency that would be advanced through lobbying by the European Services Forum and the European Telecommunications Network Operators' Association. This contrasted with the firm's close involvement in the Mercosur-European Union Business Forum, since interviews with E U official, interest group representative, Brussels, June 2004. interview with interest group representative, Brussels, June 2004. 130 Mercosur markets offered much more room for expansion.4 Given their dominant market position, European firms were less interested in a further liberalization be-yond the GATS. The manufacturing sector supported the negotiating endeavour while making few specific demands,5 especially since Chile had unilaterally re-duced its M F N tariff to a uniform rate of six percent. Volkswagen was the only firm that intervened individually to ensure that Chile lifted its ban on diesel en-gines in passenger cars.6 The compromise formula dovetailed with European firm interests. According to Chile's commitments, registered pension funds could still only be offered by banks established in Chile, which in effect strongly favoured the two main Span-ish firms. These no longer need a special authorization: from March 2005 on, only the individual plans have to be approved by the regulatory authorities. Chile also committed to the opening of postal services and marine transportation, the only important service industries which had remained closed. Since the service and investment chapters of the FTA guarantee national and most-favoured-nation treatment, European operators would be granted at the least the same conditions as any preferential agreement signed by Chile. For the dominant European firms, this was sufficient. As argued in Chapter 2, both regulatory frameworks and the structure of service markets can penalize late entry. US competitors, in compar-ison, still faced difficulties when entering the market. The opportunity to gain better access presented itself when Chile reconsidered its position on an FTA with the US. 6.1.4 Catching Up: The US-Chile F T A Although negotiations between the US and Chile began only a few months af-ter the first rounds of EU-Chile talks, the US-Chile FTA proved more difficult to conclude. In August 1999, the Chilean foreign minister proposed to explore the possibility of a Chile-US bilateral agreement, leaving the question of fast-track authority for later stages of the negotiation. Following discussions between of-ficials of both parties at the sidelines of the A P E C meeting in Santiago de Chile in September 1999, the two sides did not reach agreement until November 2000, when Presidents Clinton and Lagos announced the start of negotiations (Inside US Trade, 29 November 2000). The agreement was negotiated during 14 rounds 4Interview with interest group representative, Brussels, June 2004. 'interview with interest group representative, Brussels, June 2004. interview with interest group representative, Brussels, June 2004. 131 between December 2000 and December 2002. Like the E U , US interests primar-ily focused on the service sector, often targeting the same industries, although specific demands reflected the market position of US firms. US Firms ' Demands In both the telecommunications and the financial services industry, US firms sought to use the FTA with Chile to belatedly gain access to markets. Firm demands cen-tred on the two issues identified in the analytical framework of this study: market structures that excluded competitors, and regulatory barriers such as standards and licencing procedures. In March 2001, the Coalition of Service Industries and the American Chamber of Commerce in Chile called upon US negotiators to enable US telecommunica-tions providers to compete in the Chilean market. As part of a deal, the Chilean authorities would have to be banned from using regulatory means to exclude firms from markets, for example by limiting the number of licences if the number of op-erators was deemed sufficient by regulators. Likewise, competitors in the fixed-line business, i.e. US telephony providers, pressed for a legal right to use and lease lines of Telefonica CTC. This would have required a commitment beyond what Chile had offered in the GATS, where telephony service remained unbound, and became one of the sticking points of the negotiations (Inside US Trade, 26 March 2001; 1 June 2001; 11 January 2002). Commitments beyond the GATS also became the focus of financial services negotiations, again reflecting the relative position of US firms vis-a-vis their Euro-pean competitors. Since Chile had not offered additional liberalization measures in the 1997 Financial Services Agreement, entering the Chilean retail market in banking required either subsidiaries or joint ventures, but not mere branches of foreign-based banks. Since the acquisitions of European banks and the Canadian Scotiabank had left no desirable takeover targets in Chile, entering the market would have required considerable investment to set up a subsidiary. Consequently, the Coalition of Service Industries asked for a commitment to allow branches in a letter sent to USTR Zoellick in December 2001 (Inside US Trade, 4 December 2001). Matching an E U request, the Investment Company Institute as representative of the mutual fund industry demanded that the 20 percent ceiling on foreign con-tent in registered pension plans be lifted, and that US brokers be allowed to offers such plans in Chile. In a 1 March 2002 letter to the Treasury Secretary, the associ-ation sought to ensure that capital could be withdrawn if necessary. Between 1991 132 and 1998, Chile had applied a rule requiring investments to stay in the country for a minium of one year (the policy of encaje), with a reserve requirement between 10 and 30 percent to be deposited with a bank. Rightly or wrongly credited with helping Chile to avoid the contagion of various financial crises,7 these capital con-trols were no longer applied. Yet because US firms had fewer direct investments in the service sector and would likely be limited to minority stakes, they would have been particularly affected by a reimposition, motivating them to seek strong commitments (Hornbeck 2003). Important demands by services industries were therefore clearly motivated by the benefits the FTA would entail vis-a-vis their European competitors. Given the divisions emerging in the US Congress over trade policy, President Bush would have to obtain fast-track authority. In the lobbying process to support this goal, arguments about the competitive effects of FTAs would assume prominence. The Trade Promotion Authority Debate The fact that US firms' competitiveness declined without FTAs was not lost on lobbyists and political decision makers, leading to a March 2001 hearing of the Subcommittee on Trade of the House Ways and Means Committee entitled: \"Free Trade Deals: Is the United States Losing Ground as its Trading Partners Move Ahead?\" Lobbyists pressed Committee members to grant the president fast-track, now referred to as trade promotion authority (TPA), to allow the US to catch up. Business Roundtable President Sam Maury argued that: (...) because these FTAs increasingly cover trade and services, they often place our service industries at a competitive disadvantage against their foreign rivals [and] establish product standards that favor our foreign competitors. Their product becomes the standard while the U.S. product becomes non-standard. [T]hese FTAs grant our foreign competitors investment opportunities that U.S. investors lack (...) Without Trade Promotion Authority, our trading partners wil l be re-luctant to engage in comprehensive and time-intensive negotiations with the United States and will turn to other nations to negotiate deals that exclude the United States. Vice Chairman of the Council of the Americas and G E Latin America C E O John T. McCarter argued for the need to conclude an FTA with Chile to compete suc-cessfully with German and Japanese companies in the field of power generation. 7 See Edwards (1999) for an overview over the policy and an empirical evaluation of its success. 133 With the exception of small number of import-competing agricultural groups who opposed free trade in general (and one lone academic, Daniel Tarullo of George-town University), all parties present at the hearing stated their support for granting TPA to enable the US President to pursue more FTAs (U.S. House 2001). Similar arguments were made during the debate about the ratification of the US-Chile FTA. The National Association of Manufacturers supported the FTA and pushed for more bilateral agreements in a statement submitted to the Subcom-mittee on Commerce, Trade, and Consumer Protection of the House Committee on Energy and Commerce: \"Competition is very keen between U.S. and European firms, and every day that they have duty-free access to Chile while we don't is just one more day when we are simply giving American business to European firms.\" (Inside US Trade, 8 May 2003). During a hearing before the Subcommittee on Trade, the president of Qual-comm, one of the leading providers of mobile telecommunications technology, argued that the US-Chile FTA contained important pro-competitive and trans-parency clauses including (nonbinding) provisions calling on the Chilean govern-ment to ensure that operators had the flexibility to use a technology of their own choice. The \"technology neutrality\" question represented the competition over the establishment of standards as part of an FTA. Again, the lobbying coalition showed the existence of a clear rent chain: Centred on service firms as principal investors, it was broadened by their suppliers, in this case telecommunications equipment. More broadly, the American Chamber of Commerce in Chile argued that an FTA with Chile was crucial because in sectors such as power generation, financial services, telecommunications and transport, the country served as a testing ground for regional investments (U.S. House 2003). Finally, in his response to the Senate approval of TPA H.R. 3009, USTR Zoellick issued a statement on 1 August 2002 noting \"that the United States had fallen behind other countries that had been aggressively negotiating trade agreements (...) however, that passage of the trade legislation would offer (...) a boost to the U.S. and global market.\" The fast-track debate was as much a partisan battle between Congress and President as a question of trade policy per se. Its resolution should therefore be attributed to the election victory of a Republican presidential candidate rather than to the pressures exerted by free trade agreements signed by other countries. Still, the fact that most lobbyists cited the need to move US trade policy ahead by promoting FTAs underlines the competitive concerns raised by US firms. These issues also become apparent when considering the matching demands US and E U 134 trade policy makers made in the negotiations with Chile, and the outcomes they achieved. 6.1.5 Outcomes: Parity Between the US and the EU Despite differences in legal cultures, treaty \"templates\" and domestic institutional characteristics, the FTAs with Chile negotiated by the US and the EU are strik-ingly similar. In part, this reflects the legal expression of the \"nesting\" of interna-tional regimes (Aggarwal 1998), i.e. the need to make international agreements compatible. Accordingly, much of the language in service agreements draws on the GATS. Nevertheless, the US obtained several commitments by Chile that re-flected the competitive position and demands by its service sector outlined above. Importantly, the US-Chile FTA eliminates the right of Chilean authorities to limit the number of financial services providers through licences, quotas or an eco-nomic needs test.8 Likewise, it allows cross-border supply of AFPs from 2005 on, and provides for mutual fund management by US institutions, although they still have to be incorporated in Chile.9 Among the few reservations Chile preserved, Article 12.10 retains Chile's right to reinstate the encaje policy. Along similar lines, the US obtained commitments that levelled the playing field for its telecom-munications firms. The agreement contains an obligation to ensure that \"major providers\" (which in practice only applies to Telefonica de Chile) do not exclude competitors from their network.10 Likewise, it guarantees the \"technology neu-trality\" demanded by US telecommunications providers, calling on parties \"not prevent suppliers of public telecommunications services from having the flexi-bility to choose the technologies that they use to supply their services, including commercial mobile wireless services.\"11 By contrast, the commitments in the EU-Chile FTA still allow the Chilean gov-ernment to regulate market access, but subject to both national and most-favoured-nation treatment. In practice this implies that EU firms are guaranteed the same access as US firms. However, in specific areas the EU pressed for an obtained commitments that matched the US-Chile FTA: Chile offered the same rights to provide AFPs through cross-border service from the same point in time on as US firms.12 According to the laudatory official assessment by the Commission, 8 Article 12.4 9 Annex 12.9, Sections B and C '\"Article 13.4, Section 3 \"Article 13.14 '2Annex VIII 135 (...) coverage of financial services by both Parties is extremely sig-nificant and it represents the highest liberalisation commitments ever achieved in a bilateral trade agreement, both in terms of market ac-cess commitments and of the framework of rules to be applied by the Parties. (...) This is the most open investment regime ever granted by Chile to any third country. A recent legal evaluation notes that the US-Chile FTA (and hence by extension the agreement with the EU) established numerous precedents for liberalization beyond the GATS (Secretaria General de la Comunidad Andina 2003). In particu-lar the efforts by the US reflected the demands from service firms that spur on the competitive logic of FTAs. 6.1.6 P o l i t e l y D e c l i n i n g the U n i o n : T h e P r o p o s a l fo r a J a p a n - C h i l e F T A In contrast to the active pursuit of agreements by the US and the E U , Japan de-clined an offer from Chile to start negotiations. Japanese investment in the South-ern Cone has almost exclusively focused on extractive industries, where an FTA would offer only limited benefits. The Japanese rejection, despite strong lobbying from the Chilean side and important advocates close to the Japanese government, shows that in the absence of concrete competitive benefits to investing industries and their suppliers, bilateral agreements often do not come off the ground. The official Chilean proposal for an FTA resulted from a visit by Foreign Min-ister Valdes to Tokyo in November 1999. Following a pattern similar to other Japanese FTAs, the Japanese government first commissioned a report by a tri-partite study group, led by JETRO officials and including representatives of the Japan Mining Industry Association and the Japan Chamber of Commerce & In-dustry. The report that acknowledged that Chilean exports to Japan consisted mainly of copper, copper products that entered the country tariff-free, and prod-ucts such as farmed salmon, and called for the rapid conclusion of an FTA (JETRO 2001) . Likewise, former MITI Vice-Minister and JETRO Chairman Hatakeyama, who had been instrumental in starting the process towards the Japan-Mexico FTA, strongly advocated the agreement in the press (Japan Times, 8 June 2001; Hisane 2002) . Chilean business representatives presented their case at various fora in Japan and pointed out the competitive advantages E U and US firms would achieve through FTAs, but found little resonance (Nakamura 2002). Without significant 136 interests in the service sector, Japanese firms did not see enough benefit in an FTA with a country with already very low M F N tariff's. Chilean President Lagos undertook a second attempt at a visit to Japan in February 2003 to propose an FTA, but earned a polite rejection. According to the official communique, Prime Minister Koizumi replied that: \"developments in FTA negotiations with other countries and negotiations by the World Trade Organization would have to be taken into consideration and that although such an FTA could not be realized soon, Japan would like to respond to this matter as a medium- and long-term issue.\"13 Only in November 2004, Japan and Chile agreed on the sidelines of A P E C meeting in Santiago to install a second, higher-ranking tripartite commission to study the mutual benefits of an FTA (Japan Times, 23 November 2004).1 4 The low priority accorded to an agreement with Chile underscores the impor-tance of private sector interests in supporting FTAs and driving their competitive dynamic. It also challenges the specific interpretation of the Japanese FTA strat-egy as entirely driven by bureaucracy, with little or no private sector involvement (see e.g. Ogita 2003): Even if trade bureaucrats initiate policies, they are depen-dent on the support of business interests for their implementation. In the case of Japanese FTAs, service sector interests will be limited by the relative lack of competitiveness of Japanese service firms abroad. Business support can thus only come from the manufacturing sector, as in the case of the initiative for a Japan-Thailand F T A . 1 5 6.2 Japan's NAFTA? Moving Toward an FTA with Thailand In Southeast Asia, the \"contagion\" of the Asian Financial Crisis forced countries to try to attract more FDI, rather than volatile portfolio investment. The crisis thus boosted a process of investment liberalization that had come off a slow start in the early 1990s (Yoshimatsu 1999), and gave further impetus to the development of the A S E A N Free Trade and A S E A N Investment areas. It rapidly altered the environment for Japanese firms with existing operations in Asia that had been set 13http:\/\/www.mofa.go.jp\/region\/latin\/chile\/pv0302\/overview.html 14Without any sense of irony, the Japanese government also stated that it was ready to cooperate on arithmetic education as part of bilateral technical cooperation. 15See Manger (forthcoming) for a detailed analysis of this question in the Japanese case. 137 up under restrictive host country policies. Considering the central role played by the automotive industry, the parallels to the origin of business support for N A F T A are striking. A l l major Japanese vehicle producers and their suppliers have set up shop in Thailand with the aim of establishing the country as an export base. The proactive pursuit of an FTA represents efforts by the Japanese trade bureaucracy to provide support to Japanese firms in this endeavour, supported by strong lobbying efforts by the concerned industries. Although the negotiations are still ongoing at the time of writing, the coalitions of interest groups offer evidence in support of the main thesis of this work. Again, competitive concerns are motivating the proactive pursuit of a free trade agreement. 6.2.1 The Changing Character of FDI in Thailand Although Thailand's recent economic development was based to a great extent on FDI, it originally imposed strict requirements on manufacturing investment. Japanese manufacturing in Southeast Asia was oriented towards regional markets, with about sixty percent targeting regional sales (JBIC 2003, 18; M E T I 2002, cited in Kitamura 2003, 22). In the automobile industry, Japanese investment began in the 1970s under tightly constrained conditions. While producing cars in joint-ventures with local companies, the characteristic structure of Japanese M N C s led them to replicate their vertical production, or keiretsu networks in the region. Consequently, as Japanese car manufacturers invested in Southeast Asia, they pressed their Japanese suppliers to follow and start up subsidiary operations. Host country governments, however, mandated ever-higher local content quota (Doner 1991, 40-63; Humphrey and Oeter 2000, 59-60), making it difficult to achieve efficient production scales and driving up prices for domestic sales as a result. Given the proximity and perceived growth potential, Japanese firms sought to protect the regional markets against entrants from Europe and the US (Doner 1991, 76, 80). Consequently, Japanese manufacturers built their regional produc-tion networks around host country policies whereby locally produced parts could be sourced tariff-free within A S E A N , provided local content quota were fulfilled (Yoshimatsu 1999, 2002). In an implicit bargain, these schemes also acted as a barrier to the entry of competitors (Doner 1997, 112), because the small market size required large, \"commanding\" shares to be efficient. As a result, as of 1992, Japanese affiliates accounted for 94 percent of production in Thailand (Hatch and Yamamura 1996, 37). In the major A S E A N countries, combined imports and local assembly by Japanese firms accounted for 70-95 percent of all auto sales (Mukai, 1997; Noble 2002; Takeuchi 1993; Yoshimatsu 1999). 138 In comparison, the electronics sector in Southeast Asia is characterized by a much stronger presence of US firms and production networks. Yet, whereas US firms initially sought local low-cost suppliers in Asia to produce for re-export to other advanced markets, Japanese investment was aimed at serving nascent re-gional markets (Borrus 1999, 220). Moreover, compared to their direct competi-tors, Japanese electronics MNCs still source much more inputs from their home country (Arimura 2002; Guerrieri 2000, 50), leading to ratios of 85-90 percent of trade within the firm or with close affiliates (Ernst 2000, 83). Until the mid-1990s, the dominant position of Japanese firms and the charac-ter of FDI in Thailand therefore provided little incentive for closer economic in-tegration. Japanese investment in the auto industry was primarily tariff-jumping, oriented towards host country markets, and put a low emphasis on exports. Yet in the wake of the Asian Financial Crisis, the conditions for Japanese investment changed. Slumping demand and trade liberalization begun to affect Japanese pro-duction networks, while Korean and Western firms began to make inroads into previously uncontested markets (Noble 2002, 124; Nikkei Weekly, 3 May 2004). In the ASEAN-F ive , 1 6 non-Japanese firms increased their share of the passen-ger car market from 5-10 percent to 25-30 percent compared to the early 1990s, prompting Ford and G M to make new investments that competed with Japanese incumbents. Ford started up production in 1998, G M followed in 2000 (Mainichi Economist, 15 July 2003). Concurrently, US firms complained about remaining restrictive policies such as the A S E A N Industrial Cooperation (AICO) scheme that favoured Japanese firms, motivating further liberalization measure by host governments (Yoshimatsu 2002, 139). By supplying local and overseas markets with same products, US firms replicated their successful use of Mexico as an ex-port platform in Southeast Asia. Achieving greater economies of scale, they began to put pressure on Japanese producers (Dunne 2001). At the same time, host country governments liberalized the investment envi-ronment in order to attract more FDI. Thailand made bold steps towards liberaliza-tion of the investment regime, partly as a requirement of the IMF support package during the Asian Financial Crisis. 1 7 Equity holdings had been limited to 49 per-cent percent for companies that exported less than 80 percent of their production, a requirement that was abolished in 1998. Likewise, the Thai government elim-inated the 30 percent export requirement necessary to qualify for an exemption I^ndonesia, Malaysia, Thailand, Philippines, Singapore. I7The respective letters of intent were sent to the IMF by the Thai government on 26 May and 25 August 1998. See http:\/\/www.imf.org\/external\/np\/loi\/052698.htm and http:\/\/www.imf.org\/external\/np\/loi\/082598.htm 139 from import duties on goods used in local manufacturing. (UNCTAD 1998, 342; Nihon Keizai Shinbun, 10 December 1998). Although Japanese operations in the region faced a stiff wind of competi-tion, the depreciation of the Thai Baht also created the prospect of a highly cost-competitive production. Japanese automotive firms reacted quickly. Already be-fore the crisis, the expansion of automotive parts production for export to Japan had accelerated. Firms began a rapid process of centralization and restructuring of their operations in Thailand to achieve high capacity utilization rates (Kagaku Kogyo Nippo, 8 September 2003; Nikkei Bijinesu Daily, 12 February 2004). In an unprecedented development, Japanese manufacturers began to ship finished cars to the home market. The striking trend becomes apparent in figure 6.3. While no separate data on the exports of Japanese auto and auto parts manufacturers from Thailand to Japan is available, it is safe to assume that virtually all are ship-ments of Japanese-owned producers, given the notorious lack of success of foreign brands in Japan. Exports of parts and complete vehicles grew exponentially within a short time, although it should be noted that the overall volume is still low. 250000 -i . 200000 100000 1988 1989 1990 1991 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Years Figure 6.3: Exports of Passenger Cars and Parts from Thailand to Japan, 1988-2003. Source: UN Comtrade Database (2005) 140 Besides taking up the production of passenger cars for the Japanese market, all major manufacturers developed and started to implement plans to use Thailand as an export platform for other markets (Nikkan Jidosha Nyusu, 22 March 2003). Concurrent with the start of FTA negotiations with Thailand, Japanese firms an-nounced plans for significant investments in the country. Toyota confirmed the production of the \" I M V International Multipurpose Vehicle,\" a light pickup truck for exports to markets in Europe as well as important developing countries (Nikkei Bijinesu Daily, 8 August 2004). Suppliers closely affiliated with the group such as Aisin Seiki (brakes), Aichi Steel (engine and transmission parts), N S K (bearings) and Denso (electronics) prepared to quickly increase production (Nikkei Bijinesu Daily, 12 February 2004; Nihon Keizai Shinbun 24 August 2004). Initial plans proved to be too conservative, motivating Toyota to increase production capacity from 140,000 to 260,000 units per year (Nihon Keizai Shinbun, 7 May 2004). Nis-san presented plans to expand production capacity in Thailand by over 50 percent (Nihon Keizai Shinbun, 3 July 2004). Honda considered building a second plant for its small, environmentally friendly car model marketed as \"Life\" in Japan, because its current factory operated at full capacity (Nihon Keizai Shinbun, 6 December 2004). Even struggling automaker Mitsubishi announced a capacity expansion of 30 percent because of the rising demand for pickup trucks in Europe (Nihon Keizai Shinbun, 25 August 2004). Japanese electronics firms, having replicated the production network strategy of the automobile sector, were affected by the same forces when host countries began to liberalize their markets, although many of the new entrants were in fact local companies in Korea, Singapore and Malaysia (Borrus 1999, 224). Sim-ilar developments affected the upstream component of Japanese investment in the chemical industry (Nikkan Kogyo Shinbun, 26 May 2000). Although many firms also produced in China, Japanese multinationals were not willing to aban-don Southeast Asia as a production base, likely because operations in the A S E A N countries tend to be more profitable than those in China (JETRO survey, cited in Mainichi Economist, 15 July 2003). In consumer electronics, Thailand's wage and currency advantages attracted investment that in the past would have gone to Malaysia or Singapore (Nikkei Weekly, 3 May 2004). The changes in the operations of Japanese firms in Thailand were quickly relayed to MITI bureaucrats, who in 1999 began a formal study on how to offer government support for firms in a changed environment (Nikkan Kogyo Shinbun, 22 March 1999). Companies interested in supplying Japanese factories abroad with machinery also expressed their interest in an FTA with Thailand to M E T I in 141 consultations.18 Following discussions in the International Trade Policy Bureau as well as the regional bureaus, MITI\/METI began to develop plans for an FTA with Thailand. 1 9 Given the strength of protectionist agricultural groups, however, the success of this policy depends on Japanese firms with investment in Thailand and their chain of suppliers, the central constituency in favour of a free trade agreement. 6.2.2 The Political Struggle over an F T A with Thailand Just like in the case of the Japan-Mexico FTA, a successful negotiating outcome depends on an accommodation of Japanese agricultural pressure groups. Unlike Mexico, however, Thailand produces rice, the most sensitive product, as well as several other agricultural exports that hurt important lobbies in Japan. The con-crete threat of FTAs with Southeast Asian countries brought about an unexpected conversion of M A F F to multilateral liberalization, where it could rely on the E U as ally to slow down negotiations.20 Together with L D P politicians of the infamous ndrin zoku, or farm and forestry tribe, M A F F managed to delay negotiations that were originally scheduled to start in June 2003 after the visit of the Thai Prime Minister to Japan (Nihon Keizai Shinbun, 9 June 2003). In addition, M A F F offi-cials publicly discussed the possibility of seeking a blanket exclusion for rice as a precondition for an FTA with Thailand (Nihon Keizai Shinbun, 9 June 2003; Nikkei Weekly, 15 December 2003). To still approach close to 90 percent cover-age of trade and comply with WTO Art. 24 requirements, the remaining agricul-tural trade would have to be completely covered.21 Boneless chicken, the second major export commodity, therefore became a stumbling block, in addition to a hidden issue of labour mobility for Thai massage therapists and other professions with unclear licensing requirements (Nikkei News, 6 August 2004). To counter the protectionist groups, several industries organized a concerted lobbying effort. Keidanren acted as vanguard of these firms' interests in the public battle, starting with a November 2003 policy statement that requested the pursuit of FTAs with the advanced A S E A N countries.22 During the following months, I^nterview with representatives of the Japan Machinery Export Association, Tokyo, March 2003. 'interview with METI officials, Tokyo, December 2002. 20Interview with official of the Ministry of Agriculture, Forestry and Fisheries, Tokyo, March 2003. 2 1 Interview with MOFA official, Tokyo, March 2003 22http:\/\/www.keidanren.or.jp\/japanese\/policy\/2003\/114.html 142 the association called for the speedy conclusion of an FTA, warning that US firms were outcompeting Japanese firms in terms of using Thailand as production base because of the better investment conditions afforded to US firms under the US-Thailand Treaty of Commerce and Navigation (Nihon Keizai Shinbun, 18 April ; 12 May 2004). Similar demands had been repeatedly raised by the Japan Busi-ness Council for Trade and Investment Facilitation in a policy paper submitted to M E T I (Hideya 2003). After mentioning the initiative for an FTA as positive step, the paper also addressed the issue of intermediate and capital goods exports to Thailand: [Regarding electronics parts, ball bearings and rubbers] it is requested that a further cut in tariff rates be undertaken (...), keeping in mind the materials that cannot be produced in Thailand. Furthermore, the tariffs on cutting tools ought to be reduced rapidly. (...) On auto parts, the duty ranges from 5 to 42 percent, necessitating a significant reduction. Concern about the effect of a foundering of FTA negotiations were voiced by representatives of the textile industry (Nihon Keizai Shinbun, 9 June 2003), a sector with significant overseas production in Southeast As ia . 2 3 Again, manu-facturers cited the importance of lowering barriers on intermediate goods (Nikkei Weekly, 15 December 2003). The Japanese Chamber of Commerce in Bangkok warned that Japan would fall behind in its international competitiveness without an FTA with Thailand (Toyo Keizai Weekly, 29 November 2003, 80). Negotiations began in February 2004 in Bangkok, with initial Japanese pro-posals focusing on improving the Thai investment environment for Japanese firms by guaranteeing national treatment (Nikkei News Service, 17 February 2004). M A F F indeed sought and obtained the blanket exclusion of rice from the nego-tiations (Nihon Keizai Shinbun, 10 June 2004), as confirmed by Prime Ministers Thaksin and Koizumi (Nihon Keizai Shinbun, 10 October 2004). Reflecting the interests of Japanese automotive manufacturers, the Japanese negotiating team began pushing for a reduction of tariffs on cars and parts (Nihon Keizai Shinbun, 31 October 2004). This demand was publicly supported by Keidanren Chair-man Okuda, also President of Toyota Motors, who toured the A S E A N countries to muster business support for free trade agreements (Nihon Keizai Shinbun, 1 November 2004). See Sols (2004), for an analysis of the role of Japanese sunset industries in outward FDI. 143 Although the outcome of the negotiations is still open, it is evident that the industry interests supporting the initiative for an FTA with Thailand focus primar-ily on FDI. The Ministries advancing the cause of bilateral free trade, M O F A and in particular METI , receive crucial backing from those industries automobiles and parts, machinery and electronics that have direct stakes in improving the investment environment in Thailand and lowering tariffs on intermediate goods. Taking up demands by Japanese firms, M E T I bureaucrats advance the FTA policy to strengthen the position of Japanese firms vis-a-vis their competitors from other countries. 6.3 Bilateralism Comes into its Own In the cases of the EU-Chile and US-Chile FTA, competitiveness concerns by in-terest groups assumed considerable importance. While some E U officials saw the negotiations with Chile as a chance to set \"an example of a really good FTA\" in legal terms prior to negotiating with Mercosur, 2 4 European firms were active in lobbying in the limited areas where the FTA could enhance their competitiveness. Most importantly, interest group demands helped move the US-Chile FTA nego-tiations forward almost a decade after N A F T A accession was first considered by government officials of both parties. The renewed focus on Chile and the con-sideration of a bilateral FTA, as opposed to requests in the GATS or the F T A A negotiations, underscore the pressure exerted on policy makers not to be left be-hind, even though a pareto-optimal solution of a multilateral or at least broader regional deal is theoretically possible. When drawing the case of the Japan-Chile FTA initiative into the comparison, the limits of competitive pressures are evident. Government policy makers worry about the trade and investment benefits FTAs bestow on firms, although these may lead them to long-term strategic considerations. These competitive pressures were much weaker in the case of Chile. Japanese service sector companies have not en-gaged in the same whole scale investment in emerging markets as manufacturing companies. Consequently, they are not affected by the competitive benefits the EU-Chile or US-Chile FTA might bestow. Manufacturing firms neither had major investments in Chile, nor appeared concerned about the diversion of Chilean im-ports away from Japan\u2014although the latter factor should not be overstated, since Chile also had much lower M F N tariff barriers than other emerging market coun-Interview with EU officials, Brussels, June 2004. 144 tries. Bureaucratic self-interest and strategic calculations may be sufficient causes for an FTA, as the example of METI's Trojan horse, the Japan-Singapore agree-ment, shows, but in the absence of lobbying by multinational firms, we would not observe a rapid increase in the number of agreements. The Japan-Thailand initiative shows that once the competitive dynamic is coming into full swing, government policy makers apply the lessons learned and advance FTAs when supported by business interests. Important aspects of the FTA initiative with Thailand mirror the situation faced by US firms in Mexico in the late 1980s and 1990s. Thailand is turning into a key export platform for Japanese automotive industry, including exports to the home market that was previously served by domestic production. In sum, the cases presented in this chapter show that the proliferation of free trade agreements between developed and developing countries is driven by foreign direct investment in services and manufacturing. The competition driving these agreements is likely to accelerate, although not every preferential agreement will trigger the same defensive reactions. Chapter 7 145 Conclusion Since the days of the Phoenician merchants in the Levant, trade liberalization has been a powerful force in increasing the wealth of nations. As an avenue to devel-opment, the recent reintegration of developing countries into the global economy heralds profound changes in the international economy. North-South FTAs are an increasingly important aspect of this reintegration. This study began by asking why North-South FTAs have proliferated so rapid-ly since the early 1990s. The increasing popularity of these agreements is one of the most striking developments in the international trade regime, especially since most-favoured-nation tariffs for manufactured goods are at a historical low. Yet so far, most studies have assumed that the proliferation of North-South FTAs resem-bles a beauty contest, in which developing countries with the most liberal trade policies are rewarded with access to developed country export markets. Counter-ing this interpretation, the main conclusion of this study is that North-South FTAs are not driven by the search for export markets, but by foreign direct investment. Moreover, \"free trade agreements\" is a misnomer for the selective liberalization of trade, in which new barriers are erected as others are torn down. Even in services, cross-border supply is less of a concern than the opening of developing country markets through direct investment, or a \"commercial presence\" in GATS parlance. The econometric analysis in Chapter 4 suggested that one of the main benefits of N A F T A for Mexico indeed was a strong increase in FDI from third countries. Other developing countries have followed this example. But an increase in FDI also implies that home countries have interests in the host country that promote further bilateral agreements. These interests and the resulting FTAs were the focus of this study. In key aspects, the conclusions reached in this work differ from previous anal-yses. Several studies have sought the causes of economic regionalism at the in-ternational level. According to these authors, PTAs become more popular when the state system is no longer dominated by a hegemon, when other states start to sign PTAs, and when market access is threatened by trade disputes. Implicitly, these explanations assume that countries use PTAs to secure export markets. But 146 the small markets of most developing countries should not warrant these efforts, while the biggest, most promising export destinations among emerging markets\u2014 China, India, and Brazil\u2014are still relatively closed and not at the forefront of North-South FTA negotiations. Although the OECD countries hope to open these markets for their products, most of their efforts are currently concentrated on the WTO, despite the slow process of its negotiations. Other authors have claimed that deadlock in the WTO round has created the impetus to pursue PTAs. This interpretation, however, puts the cart before the horse. Most of the FTAs analysed in this study were well underway before the failed Seattle Ministerial, and were negotiated after the GATT had been elevated to the status of a full-blown international organization with the conclusion of the Uruguay Round. Considering the timing of many North-South FTAs, the slow progress in the WTO appears to be a post-hoc justification for a strategy previously decided upon. Once FTAs are proliferating, the slowing of WTO rounds may become a self-fulfilling prophecy, as countries hold on to their M F N barriers as bargaining chips for FTA negotiations. The strategic interaction between countries as they sign FTAs so far has been only analysed between countries as they negotiate, but not between potential com-petitors when seeking FTAs with a third country. The few analyses that have looked at the interests of developed countries have only done so in isolation, but not comparatively or with a view of the broader implications. Domestic politics explanations provide partial explanations for why firms from developed countries would support North-South FTAs, but they overlook impor-tant intended and unintended consequences of preferential liberalization. So far, these explanations have failed to systematically integrate rules of origin into their analytical framework. These rules, however, are often the main interest of multi-national firms. The principal conclusion of this work is that developed countries sign FTAs with emerging market countries because of the benefits the agreements bestow on multinational firms. These benefits fall into three categories. First, FTAs facilitate the vertical integration of production between developed and developing countries. At the same time, they can be used as protectionist devices against outsiders. Firms secure tariff-free access to the host country for capital and intermediate goods, and tariff-free imports back into the home market. FTAs are not ultimately necessary for this vertical integration to occur, but as preferential agreements, FTAs can be used to raise the barriers for non-members. Rules of origin are the primary means to do so. 147 Second, in the service sector, FTAs function as investment agreements. In many service industries, market structures create first-mover advantages. At times, these advantages are reinforced by the regulatory practices of host governments that limit market access to a few firms. At the behest of service sector firms, home country governments use FTAs to secure such first-mover advantages, or to lock in commitments made by the host country. Third, bilateral agreements between developed and developing countries cre-ate an endogenous dynamic. Frequently, firms cannot afford to eschew desirable export platforms and promising service sector markets, especially if their com-petitors invest there. If, by means of an FTA, these competitors gain preferential access to such markets or raise the cost to use these countries as export, then firms have to respond. They therefore press their own governments to sign \"defensive agreements.\" These three benefits, this study has argued, drive the recent pro-liferation of North-South FTAs. This chapter summarizes and extends the most important findings. 7.1 Findings: Proactive Agreements The case of the N A F T A negotiations showed that rules of origin assume promi-nence in North-South FTAs. US manufacturing firms lobbied strongly for rules of origin if they had production facilities in Mexico, or if they were part of the North American rent chain of US multinationals. For many US firms, N A F T A was not a precondition for the continental integration of their operations. Rather, the agree-ment allowed them to go the last mile by using Mexico as an export platform for the home market, while raising the costs for European and Japanese competitors. Most prominent among the supporters of N A F T A was the automotive industry. Automotive products made up the biggest share of US firms' exports back to the home market. The operations of US automotive firms are textbook examples of vertical integration: high-volume, low-margin parts and entry level cars are pro-duced in Mexico, while the more capital-intensive development and production of luxury cars remains in the US. Such investment created relatively high-paying jobs in Mexico, assuring US auto firms influence on both sides of the negotiating table beyond what finished goods exporters from the US and Mexico could hope to attain. The degree of restrictiveness of the rules of origin that firms demanded de-pended on their supply networks. Firms that sourced globally rather than region-ally, for example in the US computer industry, favoured an open regime. By con-148 trast, automotive firms used primarily North American components and pressed for strict rules of origin. Since these rules extended protection up the production chain, their suppliers supported these demands. Even among automotive firms, differences emerged depending on the degree of regional sourcing, with G M de-manding less restrictive rules because of a higher foreign content in some of its operations. Unsurprisingly, import-competing firms in the US as well as labour demanded the strictest rules of origin. For these actors, however, any liberalization would have been problematic, whether bilateral or multilateral. In determining its position toward these rules of origin, the Mexican govern-ment was sandwiched between two competing goals. On the one hand, stricter rules of origin would act as a deterrent to outside investment that would have used Mexico as a beachhead. While the Mexican government sought this investment, the US side wanted to exclude it as much as possible. On the other hand, Mexican domestic suppliers would benefit from strict rules of origin because they would induce a switch towards intermediate goods from within NAFTA. In the end, the latter position prevailed, as the positions of the US automotive firms and Mexican parts producers converged. Furthermore, this outcome may have contributed to Mexico's raising of M F N barriers after N A F T A came into force. By themselves, rules of origin are an administrative burden on firms, but they exert their full force when they interact with host country tariffs. Accordingly, US firms sought to prevent Mexico from rebating its most-favoured-nation tariffs for firms from non-NAFTA countries: The maquiladora system of export-processing zones had to be phased out. N A F T A enabled US firms to produce anywhere in Mexico with the same benefits the maquiladoras created, while severely restricting the ability of non-NAFTA firms to do so. Besides multinational manufacturing firms that integrate their production ver-tically, service sector firms become key supporters of North-South FTAs. Many of these firms follow manufacturing firms and offer services such as export fi-nancing. The case study of N A F T A showed that in particular banks and insurance companies from the southern US supported the agreement as a business oppor-tunity. These firms probably would not have lobbied by themselves for an FTA without the manufacturing multinationals taking the lead. By contrast, service firms that wanted to invest in the developing country become key supporters. In the case of NAFTA, US financial services firms used the agreement to gain access to a new market. The weakly capitalized Mexican banks were attractive takeover targets, allowing US banks to acquire a retail branch network at a low cost. For these service sector firms, a bilateral agreement was an opportunity for expansion. If US service firms were aware of the potential benefits of preferential 149 liberalization, they did not state these as openly as they did in the case of the US-Chile FTA. N A F T A created limited first-mover advantages for US firms. Mexico negotiated the opening of important service industries under N A F T A and subse-quently adapted its domestic regulation in many aspects, granting non-NAFTA firms similar conditions with a delay of a few years. First-mover advantages emerged most clearly in Chile, when the government granted European service firms concessions that allowed them to establish commanding positions in the market. In the case of N A F T A as well as the FTAs with Chile, exports of finished goods mattered less than the interests of firms with investment in the partner country. Finished goods exporters lobbied less energetically than investing firms, but also achieved less in terms of market access. The interests of developed and developing countries primarily converge on foreign direct investment, according multinational firms and their suppliers much greater influence. In sum, in both the manufacturing and the service sector, the interplay of reg-ulatory frameworks and foreign direct investment can be used by firms to their advantage. While rules of origin are are obvious barriers to outsiders in the man-ufacturing sector, preferential liberalization can create comparable barriers in the service sector. These increased costs of entry trigger defensive agreements. 7.2 Findings: Defensive Agreements If developing countries attract beachhead FDI after they sign a first FTA with a developed country, then non-member firms that invest have a strong incentive to lobby for tariff reduction. Lower tariffs take away much of the force of strict rules of origin, because even if the required regional content quota is not met, only reduced host country and low developed country tariffs have to be paid. Non-member firms thus primarily look for parity with the firms of the first partner. In a slow, piecemeal fashion, proactive and defensive FTA wil l therefore whittle away the remaining tariff barriers of the developing country\u2014although only for those goods of interest to these firms. Service liberalization as part of an FTA levels the playing field by lifting restrictions on foreign participation and limiting the regulatory competence of host states. This study has evaluated these arguments in two case studies of defensive agreements. The reactions to NAFTA, an agreement laden with discriminatory clauses against outsiders, provided the starting point. Following NAFTA's entry into force, direct investment by European firms in Mexico began to grow rapidly, creating a strong incentive for these companies to 150 support an EU-Mexico FTA. Volkswagen was the most important individual firm to lobby for a defensive agreement. The company's position resulted from a strate-gic shift from domestic market orientation toward the use of Mexico as an export platform. In adopting this strategy, Volkswagen was trailing the US auto firms by several years. Once the decision was made, however, the company's operation would have a faced an increased tariff burden because of NAFTA: Firms from third countries that wanted to use Mexico as beachhead were the direct targets of rules of origin and bans on M F N rebates. Although V W did not own maquiladoras due to its domestic market orientation, it was particularly hurt by the combination of NAFTA's rules of origin and Mexican M F N tariffs. Since newer automobile models produced for the US and Canadian markets used higher non-NAFTA con-tent, the differential phase-in of rules of origin requirements for incumbent firms did not benefit Volkswagen as much as the US Big Three automakers. Volkswagen therefore supported an EU-Mexico FTA to lower the costs N A F T A imposed on its operations. The company mobilized its rent chain of suppliers, many of which had followed its main customer and invested in Mexico, to sup-port a bilateral trade agreement. Just like in the case of NAFTA, the interests of home and host governments converged on FDI. Given the prospect of increased investment by V W in Mexico in order to export cars to the US and back to Euro-pean markets, the Mexican government responded positively to these initiatives, in particular when supported by the Mexican car parts manufacturers. The con-trast to the liberalization of finished goods is striking: Manufacturers that sought market access for exports from Europe fared badly in the negotiations, obtaining only a slow expansion of quotas over several years. Reflecting both smaller investment interests and greater domestic resistance, the Japanese \"defensive agreement\" came much later than in the E U case. Aboli-tion of the maquiladora system was bound to affect Japanese firms in the electron-ics industry the most. These firms used Mexico as a backdoor for the US market, producing T V sets and other consumer electronics goods in customs-bond facto-ries. Once the Mexican government was barred from rebating its M F N tariff for the parts imports of these firms, they had to pay 15 to 30 percent import duty. Notably, Japanese firms exhausted all other means to avoid the impact of N A F T A on their Mexican operations, unsuccessfully lobbying the Mexican government to grant tariff reductions. These efforts contrast with Nissan's relative indifference to N A F T A in its early years. After the merger with Renault, however, the company followed Volkswagen's footsteps and used Mexico as production site to supply the US market. Just like in the case of Volkswagen, the sequence of decisions sup-ported the hypotheses put forth in this study. Firms that established beachheads in 151 an FTA lobbied for defensive agreements, with those firms most affected by rules of origin and bans on tariff rebates reacting the fastest. Neither the E U nor Japan would have accorded an FTA with Mexico high priority in the absence of the discriminatory effect of N A F T A on their invest-ment. The E U first rejected a Mexican offer, then wavered for several years in its commitment to a bilateral initiative. Japan had to overcome ideological and ma-terial resistance in the domestic arena before it could sign any preferential trade agreements at all. Clearly, weaker investment links suffice to trigger defensive agreements, and competitive pressures can motivate FTAs that would otherwise not have been feasible. The Japan-Thailand FTA case study focused on an additional dimension of the competitive dynamic. Following the Asian Financial Crisis, Thailand be-came a highly attractive export platform for the automobile industry. Japanese firms began to emulate their US counterparts by producing in cars in an emerging market country for both home and third country markets. This new strategy de-parts from traditional practice of directly investing in the host market, in particular the US, to circumvent restrictions on exports. Its active support by the Japanese trade bureaucracy suggests that the competitive dynamic of North-South FTAs has reached a second stage, accelerating as states begin to preemptively sign bilateral agreements. M E T I and M O F A bureaucrats, a few years before still committed to multilateralism, applied the lessons of N A F T A to Japanese trade policy. Once bilateralism becomes popular, it creates an endogenous dynamic. This dynamic is also evident in the service sector cases analysed in this study. Japanese firms were not active in the lobbying for these agreements, given their limited overseas operations, all of which fall into the category of intermediate service providers. European and US firms, however, actively supported FTAs to bolster their competitive positions. European financial services firms, in particular the two Spanish banks B B V A and B S C H , supported an EU-Mexico FTA to remain on par with their US com-petitors. Just like in manufacturing, achieving N A F T A parity in Mexico's service commitments was the goal. European banks did not enjoy the same liberaliza-tion and guarantees that N A F T A offered. While US banks established themselves quickly in the Mexican market, European competitors had to use their US sub-sidiaries to enter Mexico. European banks and financial intermediaries also saw indirect benefits as shareholders in companies active in the region, leading them to support the EU-Mexico FTA. The cases of the telecommunications and banking markets in Chile have shown that as long as host country governments retain reservations or limit market access, 152 firms can gain advantages over competitors. In the agreements with Chile, manu-facturing firms played only a minor role as part of the rent-chain of service firms. This reversed the pattern observed in the FTAs with Mexico. European service sector firms used government concessions to establish themselves in service mar-kets, effectively shutting out US firms because of remaining regulatory barriers and limited number of government licences. As a result, Washington lobbyists used the US-Chile FTA to make highly specific demands that would allow them access to telecommunications networks and banking systems dominated by Euro-pean providers. While lobbyists had earlier stressed that the advantages of N A F T A accession would be limited to member country firms, they later emphasized that US firms needed an FTA to regain markets and to avoid being left behind. By contrast, E U firms mainly sought to lock in the gains made through Chile's open policy towards FDI without inviting further competition. In all these cases, mar-ket structures and regulatory frameworks in the service sector raised entry barriers and triggered a competitive dynamic. Contrasting the agreements with Mexico and Chile has also shown the first-mover advantage FTAs create in the field of standard-setting. While Mexico adopted mostly US standards after NAFTA, the dominant position of E U firms in Chile reversed this situation. Accordingly, the E U pressed unsuccessfully for common standards in the negotiations with Mexico, while the US tried to attain technology neutrality in Chilean regulatory policy by means of the US-Chile FTA. Manufacturing and services FDI in emerging markets differ in their objec-tives. While the former primarily seeks efficiency by turning developing coun-tries into export platforms, the latter is mostly market-seeking. Such different circumstances, however, lead to similar outcomes, since preferential liberaliza-tion of the investment environment and tariffs on intermediate goods discriminate against firms from third countries. The proliferation of FTAs that first and fore-most facilitate investment therefore becomes endogenous. 7.3 Implications What are the limits of the proliferation of PTAs in general and North-South FTAs in particular? Most importantly, with every new agreement that creates rules of origin for manufactured goods, transaction costs increase. Complex tracing meth-ods and origin certificates burden firms immensely. As a result, as Anson et al. (2003) have shown, these rules depress trade and undermine the liberalizing po-tential of FTAs. Moreover, since they often reflect the lobbying efforts of differ-153 ent interest groups, they inevitably differ between FTAs. So far, only the E U has started to harmonize its rules of origin across all its bilateral agreements, but the impact is still not uniform because of varying host country tariffs. Seen in a favourable light, rules of origin are just a snapshot of manufacturing industry interests at the time of negotiation. For example, the strict N A F T A rules of origin for T V sets will eventually become redundant as cathode ray tubes are replaced by liquid crystal and plasma displays manufactured in Korea and China. Alternatively, rules of origin might eventually become such a burden on trade that their cost outweighs their benefits. The N A F T A countries have undertaken the first tentative steps towards simplifying the existing rules of origin that guide their trade relations. Likewise, the WTO has begun to investigate the subject. However, the relatively technical nature of rules of origin suggests that such efforts will take a long time and are by no means guaranteed to succeed. As the case studies in this work have shown, North-South FTA negotiations are frequently dominated by the interests of a few industries or even a handful of firms. Liberalization often focuses on the goods of interest to these firms. FTAs do not perform much better than multilateral rounds when it comes to reducing barriers on exports of finished goods, and usually fail to make inroads into pro-tected agricultural markets. The weak disciplines of GATT Art. X X I V permit \"dirty\" FTAs such as the Japan-Mexico agreement, in which less than 90 percent of trade between the partners are liberalized. Such violations of Art. X X I V are likely to become more common in the future, since neither the US nor the E U are in a position to cast the first stone and demand a WTO examination of FTAs. Yet because North-South FTAs primarily focus on investment and the attendant trade, they come dangerously close to sectoral liberalization. Sectoral reduction of trade barriers threatens to undermine the multilateral trade regime. Negotiators have long valued package deals to facilitate the liberal-ization of the most stubbornly protected sectors. Preserving the potential for such \"issue linkage\" was one of original rationales for disciplines on regional trade agreements. As Davis (2003, 364) warns, Pursuit of single-sector liberalization where liberalization may be easier to achieve wil l bring closer the time when cross-sector linkages are no longer available. The economic gains from rapid liberalization of select sectors may outweigh the need for liberalization of highly protected sectors in the short term. However, over time greater losses may accumulate if persistent pockets of protection remain as sources of trade distortion and diplomatic tensions. (...) To avoid getting 154 left without any source of leverage, developing countries may need to demand a greater proportion of liberalization of agricultural and textile markets by developed nations relative to measures they agree to undertake for service sector or investment liberalization. This passage aptly characterizes the current state of WTO negotiations. While many observers blame the stalled WTO for the popularity of preferential agree-ments, the findings of this study suggest that the causal arrow may (perhaps con-currently) be reversed. As North-South FTAs advance liberalization of trade and investment where the interests of developing and developed countries converge, the WTO may be left with the most intractable problems. Failure of multilateral rounds becomes a self-fulfilling prophecy if no package deals can be made. By implication, bilateral deals undermine multilateral liberalization by satisfying the only constituencies that could balance protectionist agricultural interest groups. Liberalization through bilateral deals is likely to remain partial and incom-plete. Although the negotiation of an individual agreement may be easier than a multilateral round, the sum of bilateral PTAs increases transaction costs. More-over, as the case studies have shown, the succession of proactive and defensive FTAs means that developed countries end up with parity. The proliferation of bilateral FTAs could therefore be seen as a prisoner's dilemma-type game: once countries turn towards bilateralism (with the best intentions), multilateralism un-ravels. Eventually, the burden of compliance with numerous agreements may lead countries to reconsider the advantages of multilateral liberalization. Students of international cooperation have long recognized the power of international institu-tions to reduce transactions costs (Keohane 1982). Moreover, by lowering M F N tariffs, multilateral liberalization can take away the protectionist force of rules of origin: If the US did not impose any tariff on imported vehicles, it would not matter whether they were produced in Mexico, Slovakia or Korea. Such consider-ations are likely to become more important when the most attractive host countries have become FDI \"hubs,\" the centres of a dense network of \"spokes\" of preferen-tial trade agreements. At that time, it may also be possible for these host countries to drop their remaining M F N tariffs. For example, once the Mexican FTAs with the E U and Japan are in full effect, most of Mexican trade will be covered by pref-erential agreements. This will spur on the necessary adjustment of the domestic parts industry. Mexico will then be in a position to reduce tariffs on these goods. At this point, the country could follow Chile's example, matching preferential lib-eralization with equivalent unilateral or multilateral commitments. If this holds 155 for enough countries, the current wave of preferential agreements may turn out to be beneficial after all. These considerations lead to several avenues for future research. 7.4 Avenues for Further Inquiry At the outset, this study made a crucial assumption: Developing countries sign FTAs with bigger partners primarily to attract foreign direct investment. The econometric analysis in Chapter 4 has offered evidence that at least in the case of Mexico, this goal has been attained by drawing in FDI from third countries. This also implies that developing countries are locked into their own competition, advertising themselves as export platforms. Political scientists and economists have so far mostly assumed that FTAs matter as commitment devices that lock in a liberal economic policy. But as the case study of N A F T A in Chapter 3 suggests, vertical FDI, the kind of investment these countries are likely to attract, appears to respond primarily to differences in factor proportions. Credible commitments ex ante seems less important. For export-oriented manufacturing FDI, it follows that countries are likely to compete with similarly endowed developing nations. Alter-natively, developing countries may sign FTAs with neighbours to attract FDI into an expanded market, as in the case of Chile's association with Mercosur. Explor-ing this dimension of preferential trade agreements as an alternative explanation to the \"credible commitments\" hypothesis would shed much light on the choices developing countries face. A second aspect that warrants further exploration is the role of international agreements in liberalizing investment in services. This study has shown that ser-vices FDI plays an important, sometimes even crucial role in North-South eco-nomic relations. Because of a paucity of data and the particular, not yet well understood characteristics of service markets, little research on service sector lib-eralization has been generated in political economy beyond the contributions by Aggarwal (1992) and Bhagwati (1991). Empirically, the analysis of international agreements on services investment could span different regions, as Asian coun-tries begin to open these economic sectors several years after Latin American markets have done so. Since the regulation of services has so far been mostly in the domestic domain, but is increasingly constrained by international agreements, it offers a promising field for the study of negotiations across different levels. Finally, for several decades after World War II, the \"generic institutional form\" (Ruggie 1993) of multilateralism was seen as in many ways superior to bilateral-156 ism. Yet studies of the foundations of multilateral norms are few and far between. The cases analysed in this work show that for developed countries, the departure from multilateral norms is by no means easy. Often, such norms have become deeply embedded in diplomatic practice and national laws. With the exception of European integration, the US did not support preferential agreements until the late 1980s. Despite of the European experience with regionalism, several E U members perceived bilateral deals with countries beyond the immediate neighbourhood as problematic. Even when other countries had begun to negotiate numerous FTAs, Japanese trade bureaucrats hesitated to abandon the country's \"pure\" multilateral-ist trade policy orientation. Preferential trade agreements are deviations from the most-favoured-nation rule, a norm that is closely tied to the principle of sovereign equality. Yet while there are numerous studies that explore the welfare effects of a most-favoured-nation rule, we know very little about the foundations of this norm, its evolution and economic and political dynamics. Such an inquiry would contribute to a better understanding of the theoretical foundations of the global trade regime. 7.5 Outlook This study has analysed the interests of developed countries in signing agreements with much smaller and poorer partner countries that are hardly important as mar-kets. It has stressed the importance of foreign direct investment in North-South relations, and offered a perspective to integrate FDI in the study of preferential trade agreements. The explanation advanced in this work can be usefully extended to many of these agreements, for example FTAs between emerging market coun-tries with different factor endowments in which investment flows primarily in one direction. The FTA between Korea and Chile is among the first such agreements. Such agreements are likely to become even more common in the future. The E U is negotiating with Mercosur. Despite a stalled F T A A process, the western hemisphere will see an even further expansion of its dense network of preferential trade agreements. Finally, the Asia Pacific region experiences a similar prolifera-tion of FTAs. Preferential agreements will remain an important field of study in political economy. Bibliography Newspapers and Magazines Bridges Weekly (Geneva) Business Mexico (Mexico City) Chemical Week E l Financiero (Mexico City) E l Pais Europe Daily Bulletin (Brussels) European Information Bulletin (Brussels) European Report (Brussels) Financial Times (London edition) Inside US Trade Japan Times Kagaku Kogyo Nippo Latin America Weekly Report Mainichi Economist New York Times Nihon Keizai Shinbun Nikkan Jidosha Nyusu Nikkan Kogyo Shinbun Nikkei Bijinesu Nikkei Sangyo Shinbun Nikkei Weekly Toyo Keizai Weekly Wall Street Journal Electronic News Services (accessed through Lexis-Nexis) A F X News A P Press Business News Inter Press Service Global Economic Newswire Japan Economic Newswire Electronic Databases U N Comtrade Database (2005) World Bank World Development Indicators (2004-2005) U N C T A D World Investment Directory (2004) 159 Literature Cited Adams, Richard, et al. 2003. The trade and investment effects of preferential trading arrangements\u2014Old and new evidence. Canberra: Australian Pro-ductivity Commission. Adcock, Robert and David Collier. 2001. \"Measurement validity: A shared stan-dard for qualitative and quantitative research.\" American Political Science Review 95, no. 3:529-46. Aggarwal, Vinod K . 1992. \"The political economy of service sector negotiations in the Uruguay Round.\" Fletcher Forum 16, no. 1:34-54. \u2014. 1998. Institutional designs for a complex world: Bargaining, linkages, and nesting. Ithaca, NY: Cornell University Press. Aggarwal, Vinod K. , Robert O. Keohane and David B. Yoffie. 1987. \"Dynamics of negotiated protectionism.\" American Political Science Review 81, no. 2:345-66. Aitken, Norman D. 1973. \"The effect of the E E C and EFTA on European trade: A temporal cross-section analysis.\" American Economic Review 63, no. 5:881-92. Alt, James E., et al. 1996. \"The political economy of international trade: Endur-ing puzzles and an agenda for inquiry.\" Comparative Political Studies 29, no. 6:689-717. Amin, Ruzita M . , Zarinah Hamid and Norma M . Saad. 2004. Economic inte-gration among the League of Arab States: Empirical evidence. Paper pre-sented at the Conference on \"The Prospects of Arab Economic Cooperation to Boost Savings and Investment\", 16-18 June, at Alexandria, Egypt. Anderson, James E. and Eric van Wincoop. 2003. \"Gravity with gravitas: A solution to the border puzzle.\" American Economic Review 93, no. 1:170-92. Anderson, Michael and Stephen Smith. 1999. \"Do national borders really mat-ter? Canada-US regional trade reconsidered.\" Review of International Eco-nomics 7, no. 2:219-27. 160 Anson, Jose, et al. 2003. Rules of origin in North-South preferential trading arrangements with an application to NAFTA. CEPR Discussion Paper No. 4166, London: Centre for Economic Policy Research. Aoki , Masahiko, Hyung-Ki K i m and Masahiro Okuno-Fujiwara, eds. 1996. The role of government in East Asian economic development. Oxford: Claren-don Press. Arimura, Sadanori. 2002. \"Ajia kiki go, zaitoajia nikkei kigyo seizo to kenkyu kanpatsu ni kansuru chosa hokoku (Research report on the manufacturing and r&d of Japanese companies in East Asia after the Asian crisis).\" Toa Keizai Kenkyu (East Asian Economic Research) 61, no. 3:293-313. Asociacion Mexicana de la Industria Automotriz. 1993. Balance de la industria automotriz. Mexico City: Asociacion Mexicana de la Industria Automotriz. \u2014. 2004. Ubicacion de las empresas: Estado de Mexico. http:\/\/www.amia.com.mx\/edomex.htm. Baier, Scott L . and Jeffrey H . Bergstrand. 2001. \"The growth of world trade: Tariffs, transport costs, and income similarity.\" Journal of International Economics 53, no. 1:1-27. \u2014. 2004. \"Economic determinants of free trade agreements.\" Journal of Inter-national Economics 64, no. 1:29-63. Baier, Scott L . , Gerald P. Dwyer and Robert Tamura. 2005. Factor returns, institutions, and geography: A view from trade. Unpublished manuscript. Atlanta, G A : Federal Reserve Bank of Atlanta. Balasubramanyam, V. N . , D. Sapsford and D. Griffiths. 2002. \"Regional inte-gration agreements and foreign direct investment: Theory and preliminary evidence.\" Manchester School 70, no. 3:460-82. Baldwin, Richard E. 1996. \"A domino theory of regionalism.\" In Expanding membership of the European Union, edited by Richard E. Baldwin, Pertti Haaparanta and Jaakko Kiander, 25-48. Cambridge: Cambridge University Press. Banco Bilbao Vizcaya. 1998. Informe anual. Bilbao. 161 Baron, David P. 1995. \"The nonmarket strategy system.\" Sloan Management Review 37, no. 1:73-85. \u2014. 1997. \"Integrated strategy and international trade disputes: The Kodak-Fujihim case.\" Journal of Economics & Management Strategy 6, no. 2:291-346. \u2014. 1999. \"Integrated market and nonmarket strategies in client and interest group politics.\" Business and Politics 1, no. 1:7-34. Bayoumi, Tamim and Barry Eichengreen. 1997. \"Is regionalism simply a di-version?\" In Regionalism versus multilateral trade arrangements, edited by Takatoshi Ito and Anne O. Krueger, 141-65. Chicago: University of Chicago Press. Beach, Derek. 1997. The negotiation of the Europe Agreements agreements with Poland, Hungary, and Czechoslovakia\u2014a three-level game. Paper pre-sented at the Workshop on East-West Regimes after the Cold War, 15-16 May, at Thorkil Kristensen Institute, Centre for East-West Research, South Jutland University Centre. Beeson, Mark and Kanishka Jayasuriya. 1998. \"The political rationalities of regionalism: A P E C and the E U in comparative perspective.\" Pacific Review 11, no. 3:311-36. Berger, Allen N . , Asl i Demirgiic-Kunt, Ross Levine and Joseph G. Haubrich. 2004. \"Bank concentration and competition: An evolution in the making.\" Journal of Money, Credit, and Banking 36, no. 3:433-51. Berger, Mark T. 1999. \"APEC and its enemies: The failure of the new regional-ism in the Asia-Pacific.\" Third World Quarterly 20, no. 5:1013-30. Bhagwati, Jagdish. 1991a. \"Trade in services and the multilateral trade ne-gotiations.\" In Political economy and international economics, edited by Douglas A . Irwin, 282-306. Cambridge, M A : MIT Press. \u2014. 1991b. The world trading system at risk. Princeton, NJ: Princeton University Press. Bhagwati, Jagdish and Arvind Panagariya. 1996. The economics of preferential trade agreements. Washington, DC, College Park: AEI Press. 162 Blank, Stephen and Jerry Haar. 1998. Making NAFTA work: U.S. Firms and the new North American business environment. Miami, FL : North-South Center Press. Blomstrom, Magnus and Ar i Kokko. 1997. Regional integration and foreign direct investment, N B E R Working Paper 6019, Cambridge, M A : National Bureau for Economic Research. B M W de Colombia. 2004. Camino exito. http:\/\/www.bmw.com.co\/cog\/fascinacion\/ Tradicion\/story\/caminoexito.htm. Boisso, Dale and Michael Ferrantino. 1997. \"Economic distance, cultural dis-tance, and openness in international trade: Emprical puzzles.\" Journal of Economic Integration 12, no. 4:456-84. Borrego, Rene. 1991. \"One-stop service.\" Business Mexico, 1 October, 10-12. Borrus, Michael G. 1999. \"Exploiting Asia to beat Japan: Production networks and the comeback of U.S. Electronics.\" In Japanese multinationals in Asia: Regional operations in comparative perspective, edited by Dennis J. Encar-nation, 213-37. Oxford: Oxford University Press. Brainard, S. Lael. 1997. \"An empirical assessment of the proximity-concentration trade-off between multinational sales and trade.\" The American Economic Review 87, no. 4:520-44. Brenton, Paul. 2003. \"Integrating the least developed countries into the world trading system: The current impact of European Union preferences under 'everything but arms'.\" Journal of World Trade 37, no. 3:623-46. Busch, Marc L . and Helen V. Milner. 1994. \"The future of the international trading system: International firms, regionalism, and domestic politics.\" In Political economy and the changing global order, edited by Richard Stubbs and Geoffrey R. D. Underbill, 259-76. Toronto: McClelland & Stewart. Business Mexico Special Edition. 1992. \"Foreign investment in NAFTA: A U.S. Perspective.\" Business Mexico 2, no. 1:24-26. Byrne, Eileen. 1994. \"A Yen for Mexico.\" Business Mexico 4, no. 3: 46-47. 163 Biithe, Tim. 2002. \"Taking temporality seriously: Modeling history and the use of narratives as evidence.\" American Political Science Review 96, no. 3:481-93. Buthe, Tim and Helen V. Milner. 2004. The politics of foreign direct investment into developing countries: Increasing FDI through policy commitment via trade agreements and investment treaties. Paper presented at the Annual Meeting of the American Political Science Association, 2-5 September, at Chicago, IL. Calder, Kent E. 1988. \"Japanese foreign economic policy formation: Explaining the reactive state.\" World Politics 40, no. 4:517-41. Calderon, Alvaro and Ramon Casilda. 1999. Grupos financieros Espanoles en America latina: Una estrategia audaz en un dificil y cambiante entorno Europeo. Santiago de Chile: Economic Commission for Latin America and the Caribbean. \u2014. 2000. \"La estrategia de los bancos Espanoles en America latina.\" Revista de laCEPAL 70:71-89. Calderon, Alvaro, Michael Mortimore, and Wilson Peres. 1995. Mexico's incor-poration into the new industrial order: Foreign investment as a source of international competitiveness. Santiago de Chile: Economic Commission for Latin America and the Caribbean. Cameron, Maxwell A . 1997. \"North American trade negotiations: Liberalization games between asymmetric players.\" European Journal of International Relations 3, no. 1:105-39. Cameron, Maxwell A . and Brian W. Tomlin. 2000. The making of NAFTA: How the deal was done. Ithaca, NY: Cornell University Press. Carillo, Jorge. 2000. \"The integration of the Mexican automobile industry to the USA: Between policies and corporate strategies.\" In Actes du GER-PISA, edited by GERPISA, 55-77. Val d'Essonne: Groupe d' Etude de Recherche Permanent sur 1' Industrie et les Salaries de 1' Automobile, Uni-versite d'Evry. Carlsen, Laura and Donald McCarthy. 1991. \"Trade talks to shape auto industry future.\" Business Mexico 1, no. 7:20-22. 164 Carr, David L . , James R. Markusen and Keith E. Maskus. 2001. \"Estimating the knowledge-capital model of the multinational enterprise.\" American Economic Review 91, no. 3:693-708. Caves, Richard E. 1996. Multinational enterprise and economic analysis. Cam-bridge: Cambridge University Press. Cerny, Philip G. 1990. The changing architecture of politics: Structure, agency, and the future of the state. London: Sage. Chant, John F. 1993. \"The financial sector in NAFTA: Two plus one equals re-structuring.\" In Assessing NAFTA: A trinational analysis, edited by Steven Globerman and Michael Walker. Vancouver: The Fraser Institute. Chase, Kerry A . 2003. \"Economic interests and regional trading arrangements: The case of NAFTA.\" International Organization 57, no. 1:137-74. \u2014. 2004a. Foreign direct investment and commitment problems in regional trad-ing arrangements. Paper presented at the Annual Meeting of the American Political Science Association, 2-5 September, at Chicago. \u2014. 2004b. \"From protection to regionalism: Multinational firms and trade-related investment measures.\" Business and Politics 6, no. 2:1-36. Chrysler Corporation, Ford Motor Company, and General Motors Corporation. 1991. Proposed policy positions for the automotive provisions of a North American Free Trade Agreement. Unpublished Manuscript. Detroit. Claessens, Stijn, Asl i Demirguc-Kunt and Harry Huizinga. 2001. \"How does foreign entry affect domestic banking markets?\" Journal of Banking & Fi-nance 25, no. 5:891-911. Commission of the European Communities (EC Commission). 1993. Infor-mation note: North American Free Trade Agreement (NAFTA). Brussels: European Commission. \u2014. 1995. Towards closer relations between the European Union and Mexico: Commission communication to the Council and Parliament, COM\/95\/0003, Brussels: European Commission. 165 Commission of the European Communities (EC Commission). 2004. Bilateral trade relations with Mexico.http:\/\/europa.eu.int\/comm.\/trade\/issues \/bilateral\/countries\/Mexico\/index_en.htm. Curtis, Gerald L . 1999. The logic of Japanese politics: Leaders, institutions, and the limits of change. New York: Columbia University Press. Davis, Christina L . 2003. Food fights over free trade: How international insti-tutions promote agricultural trade liberalization. Princeton, NJ: Princeton University Press. Deardorff, Alan V. 1995. Determinants of bilateral trade: Does gravity work in a neoclassical world? N B E R Work Paper 5377, Cambridge, M A : National Bureau of Economic Research. Delegation of the European Commission in Mexico. 1995. Informe de evolution de las relaciones UE-Mexico. Mexico City. Depken, Craig A . II. and Jon M . Ford. 1999. \" N A F T A as a means of raising rivals' costs.\" Review of Industrial Organization 15, no. 2:103. Destler, I. M . 1986. American trade politics: System under stress. Washington, DC: Institute for International Economics. Destler, I. M . 1989. American trade politics. 2nd edition. Washington, DC: Institute for International Economics. Destler, I. M . , John S. Odell and Kimberly Ann Elliott. 1987. Anti-protection: Changing forces in united states trade politics. Washington, DC: Institute for International Economics. D i Mauro, Francesca. 2000. The impact of economic integration on FDI and ex-ports: A gravity approach. CEPS Working Document 156, Brussels: Centre for European Policy Studies. Diebold, William. 1952. The end of the ITO. Princeton, NJ: International F i -nance Section, Dept. of Economics and Social Institutions, Princeton Uni-versity. Doner, Richard F. 1991. Driving a bargain: Automobile industrialization and Japanese firms in Southeast Asia. Berkeley and Los Angeles: University of California Press. 166 \u2014. 1997. \"Japan in East Asia: Institutions and regional leadership.\" In Net-work power: Japan and Asia, edited by Peter J. Katzenstein and Takashi Shiraishi, 197-233. Ithaca, NY: Cornell University Press. Downs, Anthony. 1957. An economic theory of democracy. New York: Harper. \u2014. 1967. Inside bureaucracy. Boston: Little, Brown and Co. Dunne, Michael J. 2001. \"Free trade: Who wants it?\" Automotive News Inter-national, February, 32-33. Dunning, John H . 2002. Theories and paradigms of international business ac-tivity. Cheltenham, U K : Edward Elgar. Duttagupta, Rupa and Arvind Panagariya. 2003. Free trade areas and rules of origin: Economics and politics. IMF Working Paper 03\/229, Washington, DC: International Monetary Fund. Economic Commission for Latin America and the Caribbean (ECLAC) . 2000. Foreign investment in Latin America and the Caribbean. Santiago de Chile. \u2014. 2001. Foreign investment in Latin America and the Caribbean. Santiago de Chile: E C L A C . \u2014. 2002. Foreign investment in Latin America and the Caribbean. Santiago de Chile: E C L A C . Eden, Lorraine and Maureen Appel Molot. 1992. Fortress or free market? N A F T A and its implications for the Pacific Rim. Ottawa: Centre for Trade Policy and Law, Carleton University, Asian Pacific Research and Resource Centre. \u2014. 1993. The NAFTA's automotive provisions: The next stage of managed trade. Toronto: C D . Howe Institute. Edgington, David W. and W. Mark Fruin. 1994. \" N A F T A and Japanese in-vestment.\" In Foreign investment and NAFTA, edited by Alan M . Rugman, 253-75. Columbia, SC: University of South Carolina Press. Edwards, Sebastian. 1999. \"How effective are capital controls?\" Journal of Economic Perspectives 13, no. 4:65-84. 167 Eichengreen, Barry and Douglas A . Irwin. 1995. \"Trade blocs, currency blocs and the reorientation of world trade in the 1930s.\" Journal of International Economics 38, no. 1-2:1-24. Ekholm, Karolina, Rikard Forslid and James R. Markusen. 2003. Export-plat-form foreign direct investment, N B E R Working Paper 9517, Cambridge, M A : National Bureau of Economic Research. Elster, Jon. 1989. Nuts and bolts for the social sciences. Cambridge: Cambridge University Press. Elster, Jon, and Aanund Hylland. 1986. Foundations of social choice theory. Cambridge: Cambridge University Press. English, H . Edward. 1999. \"Asia-Pacific crossroads: Regime creation and the future of A P E C . \" Pacific Affairs 72, no. 4:572-74. Ernst, Dieter. 2000. \"Evolutionary aspects: The Asian production networks of Japanese electronics firms.\" In International production networks in Asia: Rivalry or riches? edited by Michael G. Borrus, Dieter Ernst and Stephan Haggard, 80-109. London: Routledge. Estevadeordal, Antoni and Kati Suominen. 2004. Rules of origin in FTAs in Eu-rope and in the Americas: Issues and implications for the EU-Mercosur inter-regional association agreement. INTAL-ITD 15, Washington, DC: IADB-INTAL. Ethier, Wilfred J. 1998a. \"The new regionalism.\" Economic Journal 108, no. 449:1149-61. \u2014. 1998b. \"Regionalism in a multilateral world.\" Journal of Political Economy 106, no. 6:1214-45. \u2014. 2001. \"The new regionalism in the Americas: A theoretical framework.\" North American Journal of Economics and Finance 12, no. 2:159-72. Evans, Peter B . 1979. Dependent development: The alliance of multinational, state, and local capital in Brazil. Princeton, NJ: Princeton University Press. Evans, Peter B. , Harold K . Jacobson and Robert D. Putnam. 1993. Double-edged diplomacy: International bargaining and domestic politics. Berkeley and Los Angeles: University of California Press. 168 Falvey, Rodney E. 1981. \"Commercial policy and intra-industry trade.\" Journal of International Economics 11, no.:495-511. Falvey, Rodney E. and Geoff Reed. 2000. Rules of origin as commercial policy instruments. Research paper 2000\/18, Nottingham: Centre for Research on Globalisation and Labour Markets, University of Nottingham. Feenstra, Robert C. and Gordon H . Hanson. 1996. \"Foreign investment, out-sourcing, and relative wages.\" In The political economy of trade policy: Papers in honor of Jagdish Bhagwati, edited by Robert C. Feenstra, Gene M . Grossman and Douglas A . Irwin. Cambridge, M A : MIT Press. Feenstra, Robert C , Gordon H . Hanson and Deborah L . Swenson. 2000. \"Off-shore assembly from the united states: Production characteristics of the 9802 program.\" In The impact of international trade on wages, edited by Robert C. Feenstra, 85-122. Chicago: University of Chicago Press. Feenstra, Robert C , James R. Markusen and Andrew K . Rose. 2001. \"Using the gravity equation to differentiate among alternative theories of trade.\" Canadian Journal of Economics 34, no. 2:430-47. Fernandez de Castro, Rafael. 2000. \"Las instituciones del T L C A N : Una eval-uation a los cinco afios.\" In Socios naturales? Cinco anos del tratado de libre comercio de America del norte, edited by Beatriz Leycegui and Rafael Fernandez de Castro. Mexico D.F.: Instituto Tecnologico Autonomo de Mexico. Fernandez, Raquel and Jonathan Portes. 1998. \"Returns to regionalism: An analysis of nontraditional gains from regional trade agreements.\" World Bank Economic Review 12, no. 2:197-220. Fink, Carsten and Carlos A . Primo Braga. 1999. How stronger protection of intellectual property rights affects international trade flows. Working Paper 2051, Washington, DC: World Bank. Frankel, Jeffrey A . 1997. Regional trading blocs in the world economic system. Washington, DC: Institute for International Economics. Frankel, Jeffrey A. , Ernesto Stein and Shang-jin Wei. 1995. \"Trading blocs and the Americas: The natural, the unnatural, and the super-natural.\" Journal of Development Economics Al, no. 1:61-95. 169 Freeman, Harry L . 1998. \"The role of constituents in U.S. Policy development towards trade in financial services.\" In Constituent interests and U.S. Trade policy, edited by Alan V. Deardorff and Robert M . Stern, 183-94. Ann Arbor, MI : University of Michigan Press. Freund, Caroline. 2000. \"Different paths to free trade: The gains from regional-ism.\" Quarterly Journal of Economics 115, no. 4:1317-41. Frieden, Jeffry A . 1991. Debt, development and democracy: Modern political economy and Latin America, 1965-1985. Princeton, NJ: Princeton Univer-sity Press. Frischkorn, Allen R., Jr. 1993. \"Why N A F T A will benefit us telecommunication equipment manufacturers.\" Telecommunications, Feb 1993, 17. Gallant, Nicole and Richard Stubbs. 1997. \"APEC's dilemmas: Institution-building around the Pacific Rim.\" Pacific Affairs 70, no. 2:203-18. Garay, Luis Jorge and Rafael Cornejo. 2001. Metodologia para el analisis de regimenes de origen: Aplicacion en el caso de las Americas. INTAL Work-ing Paper No. 8, Washington, DC: IADB. George, Alexander L . and Timothy J. McKeown. 1985. \"Case studies and theo-ries of organizational decision making.\" Advances in Information Process-ing in Organizations 2:21-58. George Mulgan, Aurelia. 1999. The politics of agriculture in Japan. New York: Routledge. Gowa, Joanne and Edward D. Mansfield. 1993. \"Power-politics and interna-tional trade.\" American Political Science Review 87, no. 2:408-20. Gowa, Joanne S. 1995. Allies, adversaries, and international trade. Princeton, NJ: Princeton University Press. Graham, Edward M . 1994. \"NAFTA, foreign direct investment, and the United States.\" In Foreign investment and NAFTA, edited by Alan M . Rugman, 105-23. Columbia, SC: University of South Carolina Press. \u2014. 1996. Global corporations and national governments. Washington, DC: Institute for International Economics. 170 \u2014. 2000. Fighting the wrong enemy: Antiglobal activities and multinational enterprises. Washington, DC: Institute for International Economics. Graham, Edward M . and Erika Wada. 2000. \"Domestic reform, trade and in-vestment liberalisation, financial crisis, and foreign direct investment into Mexico.\" World Economy 23, no. 6:777-97. Greenaway, David. 1992. \"Trade related investment measures and development strategy.\" Kyklos 45, no. 2:139-59. Greenaway, David, Robert Hine and Chris Milner. 1994. \"Country-specific fac-tors and the pattern of horizontal and vertical intra-industry trade in the U K . \" Weltwirtschaftliches Archiv 130, no. 1:77-100. Greene, William H . 2002. The bias of the fixed effects estimator in nonlinear models. Unpublished Manuscript. New York: Stern School of Business, New York University. \u2014. 2003. Econometric analysis. Upper Saddle River, NJ: Prentice Hall. Grossman, Gene M . and Elhanan Helpman. 1994. \"Protection for sale.\" Ameri-can Economic Review 84, no. 4:833-50. \u2014. 1995. \"The politics of free-trade agreements.\" American Economic Review 85, no. 4:667-90. Gruben, William C. 2001. \"Did N A F T A really cause Mexicos high maquiladora growth?\" Federal Reserve Bank of Dallas Economic and Financial Review, third quarter, 11-21. Gruben, William C , John H. Welch and Jeffery W. Gunther. 1993. \"U.S. Banks, competition, and the Mexican banking system: How much wil l N A F T A matter?\" Financial Industry Studies, October, 11-22. Grugel, Jean and Wi l Hout. 1999. Regionalism across the North-South divide: State strategies and globalization. London: Routledge. Guerrieri, Paolo. 2000. \"International competitiveness, regional integration, and corporate strategies in the East Asian electronics industry.\" In International production networks in Asia: Rivalry or riches? edited by Michael G. Bor-rus, Dieter Ernst and Stephan Haggard, 32-56. London: Routledge. 171 Haggard, Stephan. 1995. Developing nations and the politics of global integra-tion. Washington, DC: Brookings Institution. \u2014. 1997. \"Regionalism in Asia and the Americas.\" In The political economy of regionalism, edited by Edward D. Mansfield and Helen V. Milner, 20-49. New York: Columbia University Press. \u2014. 1998. \"The political economy of regionalism.\" In The post-NAFTA political economy: Mexico and the Western Hemisphere, edited by Carol Wise, 302-38. University Park, PA: University of Pennsylvania Press. Hall, Peter. 1986. Governing the economy. New York: Oxford University Press. Hallward-Driemeyer, Mary. 2003. Do bilateral investment treaties attract FDI? Only a bit. And they could bite. Unpublished Manuscript. Washington, DC: World Bank. Hanson, Gordon and Helen Shapiro. 1994. Volkswagen de Mexico's North American strategy. HBS Case No. 9-794-104, Boston, M A : Harvard Busi-ness School Publishing. Hatakeyama, Noboru. 2003a. \"Nichiboku jiyu boeki koshS ketsuretsu ga shime-su keizai senryaku fuzai seiji ga kawaranai ka giri nihon botsuraku suru (the collapse of Japan-Mexico free trade negotiations shows: If the policy lacking an economic strategy does not change, it will be Japan's downfall).\" ChudKoron 118, no. 12:50-55. \u2014. 2003b. \"Short history of Japan's movement to FTAs (Part 1).\" Journal of Japanese Trade and Industry 21, no. 6:24-25. Hatch, Walter and Kozo Yamamura. 1996. Asia in Japan's embrace: Building a regional production alliance. Cambridge: Cambridge University Press. Hausman, Jerry A . 1978. \"Specification tests in econometrics.\" Econometrica 46, no. 6:1251-71. Hayes, John P. 1993. Making trade policy in the European community. New York: St. Martin's Press. Head, Keith and John Ries. 2001. \"Overseas investment and firm exports.\" Review of International Economics 9, no. 1:108-22. 172 Head, Keith, John Ries and Barbara J. Spencer. 2004. \"Vertical networks and US auto parts exports: Is Japan different?\" Journal of Economics and Man-agement Strategy 13, no. 1:37-67. Helpman, Elhanan. 1984. \"A simple theory of international trade with multina-tional corporations.\" Journal of Political Economy 92, no. 3:451-71. Helpman, Elhanan and Paul R. Krugman. 1985. Market structure and for-eign trade: Increasing returns, imperfect competition, and the international economy. Cambridge, M A : MIT Press. Hideya, Taida, ed. 2002. Issues and requests relating to trade and investment in 2002. Tokyo: Japan Business Council for Trade and Investment Facilitation. \u2014. ed. 2003. Issues and requests relating to trade and investment in 2003. Tokyo: Japan Business Council for Trade and Investment Facilitation. Hirsch, Moshe. 2002. \"International trade law, political economy, and rules of origin.\" Journal of World Trade 36, no. 2. Hirschman, Albert O. 1981. Essays in trespassing: Economics to politics and beyond. Cambridge: Cambridge University Press. Hisane, Masaki. 2002. FTA with Chile just isn't in the cards. Japan Times, 12 July. Hoekman, Bernard M . and Michael P. Leidy. 1993.\"Holes and loopholes in in-tegration agreements: History and prospects.\" In Regional integration and the global trading system, edited by Kym Anderson and Richard Blackhurst, 218-45. Hemel Hempstead, Hertfordshire: Harvester Wheatsheaf. Hornbeck, J.F. 2003. The U.S.-Chile free trade agreement: Economic and trade policy issues. CRS Report for Congress RL31144, Washington, DC: Con-gressional Research Service. Humphrey, John and Antje Oeter. 2000. \"Motor industry policy in emerging markets: Globalisation and the promotion of domestic industry.\" In Global strategies and local realities: The auto industry in emerging markets, edited by John Humphrey, Yveline Lecler and Mario Sergio Salerno, 42-71. New York: St. Martin's Press. 173 Husan, Rumy. 1997. \"The continuing importance of economies of scale in the automotive industry.\" European Business Review 97, no. 1:38. Ioannou, Lori. 1994. \"Better banking with NAFTA.\" International Business 7, no. 1:40-43. Japan Bank for International Cooperation (JBIC). 2003. \"Wagakuni seizSgyo kigyo no kaigai jigyo hakkai ni kansuru chosa hokoku: 2002 nen do kaigai chokusetsu toshi anketo chosa kekka (report on the inquiry on the expansion of the overseas activities of our country's manufacturing industry compa-nies: Result from the 2002 enquete on overseas direct investment).\" Kaigai Toyushi (Overseas Finance and Investment) 12, no. 1:12-19. Japan External Trade Organization (JETRO). 1994. \" N A F T A hirogaru takaku-teki na eikyo (NAFTA expansion: The multilateral effect).\" JETRO Sensor 44, no. 12:10-40. \u2014. 2000. Report by the committee for closer economic relations closer eco-nomic relations between Japan and Mexico, Tokyo: JETRO. \u2014. 2001. Study report on the Japan-Chile free trade agreement. Tokyo: JETRO. Japan Institute for Overseas Investment (JOI). 1997. \"Tokushu mekishiko no toshi jokyo: JOI chosa kekka hoko: Nikkei sangyo shinshutsu jokyo oyobi toshi kankyo hoka (special on the Mexican investment environment: Report on the findings of Japanese institute for overseas investment study: The cur-rent situation and investment environment of Japanese entreprises).\" Kaigai Yushi 6, no. 4:8-18. Japan Machinery Center for Trade and Investment (JMCTI). 2001. \"PROSEC ni kan suru yobosho wo mekishiko keizai daiji ni tei (Proposal for the requests regarding PROSEC to the Mexican minister of the economy).\" JMC Journal 49, no. 4:5-6. Japan-Mexico Joint Study Group on the Strengthening of Bilateral Economic Relations. 2002. Final report, Tokyo and Mexico City: Ministry of Econ-omy Trade and Industry (METI) and Secretaria de Comercio y Fomento Industrial (SECOFI). Jensen-Moran, Jeri. 1996. \"Choice at the crossroads: Regionalism and rules of origin.\" Law and Policy in International Business 27, no. 4:981-89. 174 Johnson, Chalmers A . 1982. MITI and the Japanese miracle: The growth of industrial policy, 1925-1975. Stanford, C A : Stanford University Press. Johnson, Jon R. 1993a. \" N A F T A and the trade in automotive goods.\" In As-sessing NAFTA: A trinational analysis, edited by Steven Globerman and Michael Walker. Vancouver, B C : The Fraser Institute. \u2014. 1993b. What is a North American good: The NAFTA rules of origin. Toronto: C D . Howe Institute. Keidanren. Working Group on Japan-Mexico Bilateral Treaties. Japan-Mexico Economic Committee. 1999. Report on the possible effects of a Japan-Mexico free trade agreement on Japanese industry. http:\/\/www.keidanren.or.jp\/english\/policy\/pol099.htm. Kemp, Murray and Henry Wan. 1976. \"An elementary proposition concerning the formation of customs unions.\" Journal of International Economics 6, no. 1:95-97. Keohane, Robert O. 1982. \"The demand for international regimes.\" Interna-tional Organization 36, no. 2:325-55. King, Gary, Robert O. Keohane and Sidney Verba. 1994. Designing social in-quiry: Scientific inference in qualitative research. Princeton, NJ: Princeton University Press. Koido, Akihiro. 1991. \"The color television industry: Japanese-U.S. Compe-tition and Mexico's maquiladoras.\" In Manufacturing across borders and oceans: Japan, the United States, and Mexico, edited by Gabriel Szekely, 51-75. La Jolla, C A : Center for U.S.-Mexican Studies, University of Cali-fornia San Diego. Krishna, Kala and Anne O. Krueger. 1995. \"Implementing free trade areas: Rules of origin and hidden protection.\" In New directions in trade theory, edited by Alan Deardorff, James Levinsohn and Robert Stern, 149-87. Ann Arbor: University of Michigan Press. Krishna, Pravin. 1998. \"Regionalism and multilateralism: A political economy approach.\" Quarterly Journal of Economics 113, no. 1:227-51. 175 Krueger, Anne O. 1995. Trade policies and developing nations. Washington, DC: Brookings Institution. Krueger, Anne O. 1993. Free trade agreements as protectionist devices: Rules of origin. N B E R Working Paper 4352, Cambridge, M A : National Bureau of Economic Research. \u2014. 1995. American trade policy: A tragedy in the making. Washington, D . C : AEI Press. Kunimoto, Robert and Gary Sawchuk. 2004. \"Moving toward a customs union: A review of the evidence.\" Horizons 7, no. 1:23-31. Kurian, George Thomas, ed. 1989. Geo-data: The world geographical encyclo-pedia. Detroit, MI: Gale Research. Lawrence, Robert Z. 1996. Regionalism, multilateralism, and deeper integra-tion. Washington, DC: Brookings Institution. Levy, Philip I. 1997. \"A political-economic analysis of free-trade agreements.\" American Economic Review 87, no. 4:506-19. Levy Yeyati, Eduardo, Ernesto H . Stein and Christian Daude. 2003. Regional integration and the location of FDI, Working Paper, Washington, DC: Inter-American Development Bank. Lewer, Joshua and Neil Terry. 2002. \"The trade effects of financial flow liberal-ization: A focus on Africa.\" Journal of International Business Research 2, no. 1:1-12. Lieberman, Marvin B . and David Montgomery, B . 1988. \"First-mover advan-tages.\" Strategic Management Journal 9, no. Special Issue:41-58. \u2014. 1998. \"First-mover (dis)advantages: Retrospective and link with the resource-based view.\" Strategic Management Journal 19, no. 12:1111-25. Lijphart, Arendt. 1975. \"The comparable cases strategy in comparative re-search.\" Comparative Political Studies 8, no. 2:158-77. Lipsey, R.G. 1957. \"The theory of customs unions: Trade diversion and welfare.\" Economica 24, no. 93:40-46. 176 Lipson, Charles. 1985. Standing guard: Protecting foreign capital in the nine-teenth and twentieth centuries. Berkeley and Los Angeles: University of California Press. Loungani, Prakash, Ashoka Mody and Assaf Razin. 2002. \"The global dis-connect: The role of transactional distance and scale economies in gravity equations.\" Scottish Journal of Political Economy 49, no. 5:526-43. Lustig, Nora. 1998. Mexico: The remaking of an economy. Washington, DC: Brookings Institution Press. Manger, Mark S. Forthcoming. \"Competition and Bilateralism in Trade Policy: the Case of Japans Free Trade Agreements.\" Review of International Politi-cal Economy. Mansfield, Edward D. and Marc L . Busch. 1995. \"The political economy of nontariff barriers\u2014a cross-national analysis.\" International Organization 49, no. 4:723-49. Mansfield, Edward D. and Eric Reinhardt. 2003. \"Multilateral determinants of regionalism: The effects of GATT\/WTO on the formation of preferential trading arrangements.\" International Organization 57, no. 4:829-62. Markusen, James R. 1995. \"The boundaries of multinational enterprises and the theory of international trade.\" Journal of Economic Perspectives 9, no. 2:169-89. Markusen, James R. and Anthony J. Venables. 1998. \"Multinational firms and the new trade theory.\" Journal of International Economics 46, no. 2:183-203. Markusen, James R., Anthony J. Venables, Denise Eby Konan and Kevin H . Zhang. 1996. A unified treatment of horizontal direct investment, vertical direct investment, and the pattern of trade in goods and services. N B E R Working Paper 5696, Cambridge, M A : National Bureau of Economic Re-search. Maxfield, Sylvia and Adam Shapiro. 1998. \"Assessing the N A F T A negotiations: U.S.-Mexican debate and compromise on tariff and nontariff issues.\" In The post-NAFTA political economy: Mexico and the Western Hemisphere, 177 edited by Carol Wise, 82-118. University Park, PA: Pennsylvania State University Press. Mayer, Frederick. 1998. Interpreting NAFTA: The science and art of political analysis. New York: Columbia University Press. McAlinden, Sean B. 1997. \"What's the right size for an assembly plant?\" Auto-motive Manufacturing & Production 109, no. 10:14-15. Mclntyre, John R., Rajneesh Narula and Len J. Trevino. 1996. \"The role of export processing zones for host countries and multinationals: A mutually beneficial relationship?\" International Trade Journal 10, no. 4:435-66. McKeown, Timothy J. 1999. \"Case studies and the statistical worldview: Review of King, Keohane, and Verba's designing social inquiry.\" International Or-ganization 53, no. 1:161-90. McLaren, John. 2002. \"A theory of insidious regionalism.\" Quarterly Journal of Economics 117, no. 2:571-608. Meade, James E. 1955. The theory of customs unions. Amsterdam: North-Holland. Meller, Patricio. 1996: Un siglo de economia politica Chilena. Santiago de Chile: Editorial Andres Bello. Mendiola, Gerardo. 1999. Mexico: Empresas maquiladoras de exportacion en los noventa. Santiago de Chile: E C L A C . Messmer, Ellen. 1992. \"Mexico may relax networking policies.\" Network World, 7 December, 33-34. Milner, Helen V. 1988. Resisting protectionism: Global industries and the poli-tics of international trade. Princeton, NJ: Princeton University Press. \u2014. 1997. \"Industries, governments, and the creation of regional trade blocs.\" In The political economy of regionalism, edited by Edward D. Mansfield and Helen V. Milner, 77-106. New York: Columbia University Press. Ministry of Economy Trade and Industry (METI). 2001. Tsusho hakusho (White paper on international trade). Tokyo: METI . 178 Ministry of Economy Trade and Industry (METI). 2002. Tsusho toshi hakusho (White paper on international trade and investment). Tokyo: METI . Ministry of International Trade and Industry (MITI). 1999. Tsusho hakusho (White paper on international trade). Tokyo: MITI. Molot, Maureen Appel. 1993. Driving continentally: National policies and the North American auto industry. Ottawa: Carleton University Press. Molot, Maureen Appel and Lorraine Eden. 1991. From silent integration to strategic alliance: The political economy of North American free trade. Oc-casional Papers no. 17, Ottawa, ON: Centre for Trade Policy and Law, The Norman Paterson School of International Affairs. Moravcsik, Andrew. 1993. \"Preferences and power in the European commu-nity: A liberal intergovernmentalist approach.\" Journal of Common Market Studies 31, no. 4:437-524. \u2014. 1997. \"Taking preferences seriously: A liberal theory of international poli-tics.\" International Organization 51, no. 4:513-53. \u2014. 1998. The choice for Europe: Social purpose and state power from Messina to Maastricht. Ithaca, NY: Cornell University Press. Moreno Brid, Juan C. 1994. La competitividad de la industria automotriz en Mexico. Mexico City: E C L A C . \u2014. 1996. Mexico's auto industry after NAFTA: A successful experience in restructuring? Working Paper No. 232, Notre Dame, IN: Kellogg Institute of International Studies, University of Notre Dame. Motor and Equipment Manufacturers Association. 1991. Discussion paper: Rule of origin for the US-Mexico-Canada free trade agreement. Unpublished Manuscript. Washington, DC: M E M A . Mukai, Juichi. 1997. \"Dai kyoso jidai no shosha wa dare ka (Who will be the winner in the age of mega-competition?).\" Sekai 631, 223-30. Matyas, Lazlo. 1997. \"Proper econometric specification of the gravity model.\" World Economy 20, no. 3:363-68. 179 \u2014. 1998. \"The gravity model: Some econometric considerations.\" World Econ-omy 20, no. 3:397-401. Nakamura, Akemi. 2002. European competition cited: Japan urged to push for Chile FTA. Japan Times, 30 May. Nelson, Douglas. 1988. \"Endogenous tariff theory: A critical survey.\" American Journal of Political Science 88, no. 3:796-838. Neyman, J. and Elizabeth E. Scott. 1948. \"Consistent estimates based on par-tially consistent observations.\" Econometrica 16, no. 1:1-32. Noble, Gregory W. 2002. \"On the road to Asia: Japanese automakers in A S E A N . \" In Winning in Asia, Japanese style: Market and nonmarket strategies for success, edited by Vinod K . Aggarwal and Shujiro Urata, 123-56. New York: Palgrave MacMillan. Ogita, Tatsushi. 2002. An approach towards Japan's FTA policy. Working Paper 01\/02 No. 4, Chiba: A P E C Study Center, JETRO Institute of Developing Economies. \u2014. 2003. \"Japan as a late-coming FTA holder: Trade policy change for Asian orientation?\" In Wither free trade agreements? Proliferation, evaluation and multilateralization, edited by Jiro Okamoto, 216-51. Chiba: Institute of Developing Economies, Japan External Trade Organization. Okimoto, Daniel I. 1989. Between MITI and the market: Japanese industrial policy for high technology. Stanford, C A : Stanford University Press. Olson, Mancur. 1971. The logic of collective action: Public goods and the theory of groups.. Cambridge, Mass: Harvard University Press. Oman, Charles. 1994. Globalisation and regionalisation: The challenge for developing countries. Paris: Development Centre of the Organisation for Economic Co-operation and Development. Ostry, Sylvia. 1990. Governments & corporations in a shrinking world: Trade & innovation policies in the United States, Europe & Japan. New York: Council on Foreign Relations Press. Panagariya, Arvind. 1999. \"The regionalism debate: An overview.\" World Econ-omy 22, no. 4:477-511. 180 \u2014. 2000. \"Preferential trade liberalization: The traditional theory and new de-velopments.\" Journal of Economic Literature 38, no. 2:287-331. Panagariya, Arvind and Ronald Findlay. 1996. \"A political economy analysis of free trade areas and customs unions.\" In The political economy of trade reform: Essay in honour ofJagdish Bhagwati, edited by Robert C. Feenstra, Douglas A . Irwin and Gene M . Grossman, 265-87. Cambridge, M A : MIT Press. Pastor, Manuel and Carol Wise. 1994. \"The origins and sustainability of Mexico free-trade policy.\" International Organization 48, no. 3:459-89. Peres Nunez, Wilson. 1990. Foreign direct investment and industrial develop-ment in Mexico. Paris: Development Centre of the Organisation for Eco-nomic Co-operation and Development. PreuBe, Heinz Gert. 2000. Sechs Jahre Nordamerikanisches Freihandelsabkommen eine Bestandsaufnahme, Tubinger Arbeitspapiere Nr. 183, Tubingen. Przeworski, Adam and Henry Teune. 1970. The logic of comparative social inquiry. New York: Wiley-Interscience. Putnam, Robert D. 1988. \"Diplomacy and domestic politics\u2014the logic of two-level games.\" International Organization 42, no. 3:427-60. Ramamurti, Ravi. 2000. \"Risks and rewards in the globalization of telecom-munications in emerging economies.\" Journal of World Business 35, no. 2:149-70. Ramirez, Miguel D. 1989. Mexico's economic crisis: Its origins and conse-quences. New York: Praeger. Ravenhill, John. 2000. \"APEC adrift: Implications for economic regionalism in Asia and the Pacific.\" Pacific Review 13, no. 2:319-33. \u2014. 2003. \"The new bilateralism in the Asia Pacific.\" Third World Quarterly 24, no. 2:299-317. Richardson, J. David. 1991. U.S. Trade policy in the 1980s: Turns\u2014and roads not taken, N B E R Working Paper 3725, Cambridge, M A : National Bureau of Economic Research. 181 Robert, Maryse. 2000. Negotiating NAFTA: Explaining the outcome in culture, textiles, autos, and pharmaceuticals. Toronto: University of Toronto Press. Ros, Jaime. 1994. \"Mexico and NAFTA: Economic effects and the bargaining process.\" In Mexico and the North American Free Trade Agreement: Who will benefit? edited by Victor Bulmer-Thomas, Nikki Craske and Monica Serrano. New York: St. Martin's Press. Rose, Andrew K . 2004. \"Do we really know that the WTO increases trade?\" American Economic Review 94, no. 1:98-114. Rozas Balbontin, Patricio. 2001. La inversion Europea en la industria energetica de america latina. Santiago de Chile: Economic Commission for Latin America and the Caribbean. Ruggie, John G. 1993. Multilateralism matters: The theory and praxis of an institutional form. New York, NY: Columbia University Press. Russell, Joel. 1992. \"Continental banking: Integration of financial services across the border.\" Business Mexico 2, no. 5:44-46. Sanahuja, Jose Antonio. 2000. \"Trade, politics, and democratization: The 1997 global agreement between the European Union and Mexico.\" Journal of Interamerican Studies and World Affairs 42, no. 2:35-64. Santin Quiroz, Osvaldo. 2001. The political economy of Mexico's financial reform. Aldershot: Ashgate. Schott, Jeffrey J. 1983. \"The GATT ministerial: A postmortem.\" Challenge 26, no. 2:40-45. \u2014. 1989. \"More free trade areas?\" In Free trade areas and U.S. Trade policy, edited by Jeffrey J. Schott, 1-58. Washington, DC: Institute for International Economics. \u2014. 1992. \"Comment.\" in North American Free Trade: Assessing the Impact, edited by Nora Lustig, Barry O. Bosworth, and Robert Z. Lawrence, 238-241. Washington, DC: Brookings Institution. Searight, Amy. 1999. \"MITI and multilateralism: GATT and the evolution of Japanese trade policy\". Ph.D. dissertation, Department of Political Science, Stanford University. 182 Secretaria General de la Comunidad Andina. 2003. Analisis del tratado de libre comercio Chile-Estados Unidos. Lima: Secretaria General de la Comu-nidad Andina. Sesit, Michael R. 1992. \"U.S. Brokers seek wider access to the shielded Mexican market.\" Wall Street Journal, 12 February. Silverstein, Jeff. 1992. \"Survival of a dinosaur: After NAFTA, wil l maquiladoras face extinction or prosperity?\" Business Mexico 2, no. 11:4-6. Simmons, Beth A . and Zachary Elkins. 2004. \"The globalization of liberal-ization: Policy diffusion in the international political economy.\" American Political Science Review 98, no. 1:171-89. Slaughter, Matthew J. 2000. \"Production transfer within multinational enter-prises and American wages.\" Journal of International Economics 50: 449-72. Soloaga, Isidro and L . Alan Winters. 2001. \"Regionalism in the nineties: What effect on trade?\" North American Journal of Economics and Finance 12, no. 1:1-29. Stephenson, Sherry M . 2002. \"Regional versus multilateral liberalization of ser-vices.\" World Trade Review 1, no. 2:187-209. Szekely, Gabriel. 1991. \"In search of globalization: Japanese manufacturing in Mexico and the United States.\" In Manufacturing across borders and oceans: Japan, the United States, and Mexico, edited by Gabriel Szekely, 1-24. La Jolla, C A : Center for U.S.-Mexican Studies, University of California San Diego. Takeuchi, Junko. 1993. \"Foreign direct investment in A S E A N by small- and medium-sized Japanese companies and its effects on local supporting in-dustries.\" RIM: Pacific Business and Industries 4, no. 22:36-57. Teichman, Judith A . 2001. The politics of freeing markets in Latin America: Chile, Argentina, and Mexico. Chapel H i l l : University of North Carolina Press. Terada, Takashi. 1998. \"The origins of Japan's A P E C policy: Foreign minister Takeo Miki 's Asia-Pacific policy and current implications.\" Pacific Review 11, no. 3:337-63. 183 Thompson, Aileen J. 1994. \"Trade liberalization, comparative advantage, and scale economies\u2014stock-market evidence from Canada.\" Journal of Inter-national Economics 37, no. 1-2:1-27. Tinbergen, Jan. 1962. Shaping the world economy: Suggestions for an interna-tional economic policy. New York: Twentieth Century Fund. Tornell, Aaron and Gerardo Esquivel. 1995. The political economy of Mexico's entry to NAFTA. Cambridge, M A : National Bureau of Economic Research. Ward's Auto World. 2003. \"Toyota ups output target.\" 2003. 1 September, 12. United States Banker. 1992. \"U.S. Banks look North, South.\" United States Banker, June, 42-44. Ueda, Akira. 2001. \" N A F T A to mekishiko no makiradora kogyo: Keizai togo to takoku sekikigyo (NAFTA and Mexico's maquiladora industry: Regional integration and multinational corporations).\" Doshisha Shogaku 51, no. 3:292-320. United Nations Centre on Transnational Corporations (UNCTC). 1992. Foreign direct investment and industrial restructuring in Mexico: Government pol-icy, corporate strategies and regional integration. New York: United Na-tions. United Nations Conference on Trade and Development (UNCTAD). 1998a. Bi-lateral investment treaties in the mid-1990s. New York: United Nations. \u2014. 1998b. World investment report 1998: Trends and determinants. Geneva: U N C T A D . \u2014. 2001. Host country operational measures. New York: United Nations Con-ference on Trade and Development. \u2014. 2004. World investment report: The shift towards services, Geneva: U N C -TAD. U.S. Congress. 1992. North American Free Trade Agreement. 102nd Congress. 2nd Session. Serial 102-135. 184 U.S. House. Committee on Ways and Means, Subcommittee on Trade. 2001. Free trade deals: Is the United States losing ground as its trading partners move ahead? Hearing before the Subcommittee on Trade of the Committee on Ways and Means. 107th Congress. 1st Session. \u2014. 2003. Implementation of U.S. Bilateral free trade agreements with Chile and Singapore: Hearing before the Committee on Ways and Means, Subcom-mittee on Trade. 108th Congress. 1st Session. United States General Accounting Office. 2003. International trade: Mexico's maquiladora decline affects U.S.-Mexico border communities and trade; re-covery depends in part on Mexico's actions. Report GAO-03-891, Washing-ton, DC: GAO. United States International Trade Commission (USITC). 1990. Review of trade and investment liberalization measures by Mexico and prospects for future United States-Mexican relations. Washington, DC: U.S. International Trade Commission. \u2014. 1991. Rules of origin issues related to NAFTA and the North American au-tomotive industry: Report to the committee on ways and means, U.S. House of representatives, on investigation no. 332-314 under section 332 of the tariff act of 1930. Washington, DC: U.S. International Trade Commission. \u2014. 1993. Potential impact on the U.S. Economy and selected industries of the North American free-trade agreement: Report to the committee on ways and means of the United States house of representatives and the committee on finance of the United States senate on investigation no. 332-337 under section 332 of the tariff act of 1930. Washington, DC: U.S. International Trade Commission. United States Trade Representative (USTR). 2000. National trade report on for-eign trade barriers. Washington, DC: United States Trade Representative. Vandevelde, Kenneth J. 1993. \"U.S. Bilateral investment treaties: The second wave.\" Michigan Journal of International Law 14, no. 3:621. Vernon, Raymond. 1971. Sovereignty at bay: The multinational spread of U.S. Enterprises. New York: Basic Books. 185 Viner, Jacob. 1950. The customs union issue. New York: Carnegie Endowment for International Peace. Wade, Robert. 1990. Governing the market: Economic theory and the role of government in East Asian industrialization. Princeton, NJ: Princeton University Press. Walzer, Robert P. 1995. \"Car makers speed up export production.\" Business Mexico, April , 32-34. Warr, Peter C. 1990. \"Export processing zones.\" In Export promotion strategies: Theory and evidence from developing countries, edited by Chris Milner, 130-61. New York: New York University Press. Warr, Peter G. 1989. \"Export processing zones and trade policy.\" Finance & Development 26, no. 2:34. Williams, Barry. 2003. \"Domestic and international determinants of bank prof-its: Foreign banks in Australia.\" Journal of Banking & Finance 21, no. 6:1185-210. Welch, John H . and William C. Gruben. 1993. \"A brief modern history of the Mexican financial system.\" Financial Industry Studies, October, 1-11. Wilson, Patricia A . 1992. Exports and local development: Mexico's new maqui-ladoras. Austin: University of Texas Press. Womack, James P. 1991. \"A positive sum solution: Free trade in the North American motor vehicle sector.\" In Strategic sectors in Mexican-US free trade, edited by M . Delal Baer and Guy F. Erb, 31-65. Washington, DC: Center for Strategic and International Studies. Wonnacott, Ronald J. 1993. The NAFTA: Fortress North America?. Toronto: C D . Howe Institute. Woo-Cumings, Meredith ed. 1999. The developmental state. Ithaca: Cornell University Press. Wooldridge, Jeffrey M . 2003. Introductory econometrics: A modern approach. Mason, OH: Thomson South-Western. 186 Wunsch-Vincent, Sacha. 2003. \"The digital trade agenda of the U.S.: Par-allel tracks of bilateral, regional and multilateral liberalization.\" Aussen-wirtschaft 58, no. 1:7-46. Yang, Y. Helio. 1998. \"Post-NAFTA production and sourcing development in the maquiladora program.\" Industrial Management & Data Systems 98, no. 5-6:269-78. Yoshimatsu, Hidetaka. 1999. \"The state, MNCs , and the car industry in A S E A N . \" Journal of Contemporary Asia 29, no. 4:495-516. \u2014. 2002. \"Preferences, interests, and regional integration: The development of the A S E A N industrial cooperation arrangement.\" Review of International Political Economy 9, no. 1:123-49. 187 Appendix I Estimating Trend Changes In order to estimate breakpoints in the trend in US FDI stock in Mexico, 1966-2002, we can fit a piecewise linear function with one or more breakpoints. A change in trend can be inferred if the function with cusp at ii fits ln y(t) = rt + A, where y(t) is the stock of US FDI in Mexico in year t, 1966 < t < 2002, and r and A are constants, better than any other linear or piecewise linear function. We take the dependent variable in logarithmic form to be able to estimate an approximation of percentage changes (Wooldridge 2003, 187-189). The best fit of the function can be estimated using ordinary least squares estimation by minimizing the sum of the total sum of squares of the estimated function up to t\\ and the total sum of squares of the estimated function after t\\. Given the small number of data points, this can easily be done numerically, as shown in the table below. Breakpoint SSQ 1 SSQ 2 Total SSQ 1985 0.776 1.718 2.495 1986 0.837 1.430 2.268 1987 0.869 1.141 2.010 1988 0.914 0.885 1.799 if*-198'9j,lifi-3isSi A:.;.,\"'<> 704 i-7pt\"v:.35sl 1990 1.153 0.576 1.729 1991 1.371 0.489 1.860 1992 1.646 0.410 2.056 1993 1.945 0.340 2.286 1994 2.273 0.281 2.554 1995 2.631 0.196 2.827 1996 2.969 0.122 3.091 1997 3.353 0.074 3.427 1998 3.826 0.027 3.853 1999 4.332 0.014 4.346 2000 4.999 0.000 4.999 Table 1.1: Numerical Minimization of Sum of Squares 188 Appendix II Investment Liberalization in Mexico Sector 1993 Foreign Investment Law NAFTA WTO\/GATS 1973 Foreign Investment Law Primary Agriculture and Livestock 49 49 unbound (100)* Production Cooperatives 10 10 0 Fishing 49 49 49 Forestry 49 49 0 Mining Not specified 49(100 subject to approval until January 1999; thereafter unrestricted) 49\/34 for special concessions Coal Mining Not specified 49(100 subject to approval until January 1999; thereafter unrestricted) 34 Industry and Manufacturing Auto Parts 49 rising to 100 in January 1999 49 rising to 100 in January 1999 unbound 40 Basic Petrochemicals 0 0 0 Manufacturing of Buses\/Trucks 49 rising to 100 in January 1999 49 rising to 100 in January 1999 40 Table II.l: Comparison of Mexican Investment Regimes (percentages of foreign ownership permitted) 189 Construction 49 (100 subject to approval until January 1999; thereafter unrestricted) 49(100 subject to approval until January 1999; thereafter unrestricted) (100)* Electricity 0 0 0 Munitions Manufacturing 49 49 49 Nuclear Energy 0 0 0 Petroleum and other Hydrocarbons 0 0 0 Radioactive Materials Retail Gasoline and LPG 0 0 0 0 0 0 Secondary and Tertiary Petrochemicals 100 100 100 Financial Sector Banking 30 100** 30 30 Brokerage Houses 30 100** 30 Credit Unions 0 0 0 Financial Groups 30 100** 30 Financial Leasing 49 100** 30 Foreign Exchange Houses 49 100 30 Insurance 49 ioo** 30 Mutual Fund Companies 49 100 0 Other Services Bus Transportation 0 (up to 49 in 1995,51 in 2001, 100 in 2004) 0 (up to 49 in 1995,51 in 2001, 100 in 2004)\/ 100 for terminals 49 0 Business Real Estate 100* 100* 30-year trust in restricted areas\/ 100* otherwise ... Table II.l continued 190 Cable T V 49 49 0 Drilling Services for Oil and Gas 49 * * * 49 * * * 100* Mail 0 0 0 Maritime Services 49 49 * * * 0 Periodicals 49 49 for Mexican editions and 100 for foreign editions 100* Telegraphic and Radio Telegraphic Services 0 0 0 Television 0 0 0 Wholesale and retail trading 100 100 100 not mentioned Tourism 49 *#* 100 49 * * * All other services 49 100 49 ... Table II.l continued Unbound sectors under the GATS are equivalent to the 1993 Foreign Invest-ment Law. * Subject to government approval. In practice, the application of the policy varied as described in the text. ** With size and market share restrictions as described in the text. *** Up to 100 percent with special government authorization. 191 Appendix III Summary Statistics Variable Obs Mean Std. Dev. Min Max InFDI 13183 15.968 7.606 0 26.505 InRGDPl 13183 26.608 1.904 20.178 29.850 lnRGDP2 13183 25.917 1.736 20.003 29.850 InDGDPcap 13183 9.109 1.249 -0.685 10.956 InRDKPW 11872 14.216 1.301 5.294 16.285 lndist 13183 8.422 1.006 5.011 9.875 T R A D E 0 P E N 1 13183 40.410 18.909 5.221 167.687 T R A D E 0 P E N 2 13183 50.533 44.308 4.248 289.239 Table III.l: Descriptive Statistics ","@language":"en"}],"Genre":[{"@value":"Thesis\/Dissertation","@language":"en"}],"GraduationDate":[{"@value":"2005-11","@language":"en"}],"IsShownAt":[{"@value":"10.14288\/1.0092259","@language":"en"}],"Language":[{"@value":"eng","@language":"en"}],"Program":[{"@value":"Political Science","@language":"en"}],"Provider":[{"@value":"Vancouver : University of British Columbia Library","@language":"en"}],"Publisher":[{"@value":"University of British Columbia","@language":"en"}],"Rights":[{"@value":"For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https:\/\/open.library.ubc.ca\/terms_of_use.","@language":"en"}],"ScholarlyLevel":[{"@value":"Graduate","@language":"en"}],"Title":[{"@value":"Competitive liberalization : the proliferation of preferential trade agreements","@language":"en"}],"Type":[{"@value":"Text","@language":"en"}],"URI":[{"@value":"http:\/\/hdl.handle.net\/2429\/16896","@language":"en"}],"SortDate":[{"@value":"2005-12-31 AD","@language":"en"}],"@id":"doi:10.14288\/1.0092259"}