WPS7304 Policy Research Working Paper 7304 Is the WTO Passé? Kyle Bagwell Chad P. Bown Robert W. Staiger Development Research Group Trade and International Integration Team June 2015 Policy Research Working Paper 7304 Abstract The WTO has delivered policy outcomes that are very dif- surveys a growing economics literature on international ferent from those likely to emerge out of the recent wave of trade agreements and argues on this basis that the WTO is preferential trade agreements (PTAs). Should economists see not passé. Rather, and subject to some caveats, this survey of this as an efficient institutional hand-off, where the WTO research to date suggests that the WTO warrants strong sup- has carried trade liberalization as far as it can manage, and is port while a more cautious view of PTAs seems appropriate. now passing the baton to PTAs to finish the job? This paper This paper is a product of the Trade and International Integration Team, Development Research Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at cbown@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Is the WTO passé? Kyle Bagwell Chad P. Bown Robert W. Staiger Stanford and NBER The World Bank and CEPR Dartmouth and NBER JEL: F13, F14 Keywords: trade agreements, terms of trade, WTO, preferential trade agreements, dispute settlement Bagwell: Department of Economics, Stanford University; 579 Serra Mall, Stanford, CA 94305-6072; tel: +1.650.723.3251, email: kbagwell@stanford.edu, web: http://web.stanford.edu/˜ kbagwell/. Bown: Development Research Group (DECTI); The World Bank, 1818 H Street NW, MSN MC3-303, Washington, DC 20433 USA; tel: +1.202.473.9588, email: cbown@worldbank.org, web: https://sites.google.com/site/chadpbown/. Staiger: Department of Economics, Dartmouth College; Hanover, NH 03755 USA; tel: +1.603.646.9345, email: rstaiger@dartmouth.edu, web: http://www.dartmouth.edu/˜ rstaiger/. Bagwell thanks the Center for Advanced Studies in the Behavioral Sciences for support and hospitality. Bown acknowledges …nancial support from the World Bank’ s Multi-Donor Trust Fund for Trade and Development. We thank Steven Durlauf, Douglas Irwin, Nuno Limão, Daniel Tre‡ er and six anonymous referees for very useful and detailed comments on an earlier draft. Semira Ahdiyyih provided outstanding research assistance. 1 Introduction Together with its predecessor the General Agreement on Tari¤s and Trade (GATT), the World Trade Organization (WTO) has delivered policy outcomes for its member governments that are very di¤erent from those likely to emerge out of the recent wave of preferential trade agreements (PTAs). Over nearly 70 years, the GATT/WTO concluded 8 rounds of multilateral trade negotiations, reducing the average ad valorem tari¤ on industrial goods to below 4 percent and expanding the s membership from 23 to 161 economies. But the GATT/WTO liberalization multilateral system’ process has ground to a halt with the 9th and seemingly moribund Doha Development Round. Furthermore, the scope of GATT/WTO liberalization, with its focus on border measures, has mainly been shallow.1 By contrast, PTAs have emerged as the vehicle by which countries reduce their tari¤s from current WTO levels down to zero, albeit on a discriminatory basis: the number of PTAs has expanded from roughly 100 in 1990 to nearly 400 today. And the intended scope of PTA liberalization, which reaches further and further behind the border, is increasingly deep. Should economists see the current state of a¤airs as an e¢ cient institutional hand-o¤, with the GATT/WTO having carried trade liberalization as far as it could manage, and now passing the baton to PTAs to …nish the job and help governments arrive at their international e¢ ciency frontier?2 And if so, can PTAs rely on their own systems of dispute resolution to ensure that governments remain at the frontier? If these questions can be answered in the a¢ rmative, then economists could view PTAs as a legitimate successor to the GATT/WTO and reasonably conclude that “the WTO is passé.” But there are alternative interpretations of these developments. One possibility is that PTAs are indeed needed to complete or complement the liberalization process and move governments to the international e¢ ciency frontier, but that a central role for dispute resolution would continue to reside at the WTO. Under this view PTAs and the WTO are complementary to an e¢ cient multilateral trading system, and both deserve support. More ominously, the current state of a¤airs might be seen as ultimate proof that PTAs are stumbling blocks to the multilateral system. According to this interpretation, the WTO still has important liberalization work to do, but it has stalled out short of its goal because of the existence and ready availability of PTAs. From this perspective, liberalization under the GATT/WTO may have ground to a halt short of the international e¢ ciency frontier, but PTAs should be seen as a 1 Here and throughout our survey we focus on international agreements to liberalize market access for traded goods and services, which for short we refer to as “trade agreements.” The GATT focused on liberalizing market access for goods and, as we explain further below, took a shallow integration approach. The WTO’ s General Agreement on Trade in Services (GATS) extended the market access focus of GATT to trade in services, but GATS has yet to produce meaningful liberalization (Francois and Hoekman, 2010). The WTO Agreements also include the Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS), the principal concern of which is the protection of intellectual property rights rather than issues of market access. While distinct from our trade agreements focus, in the conclusion we brie‡ y discuss a literature related to the TRIPS agreement in the context of our discussion of linkage across issue areas in trade agreements. 2 By “international e¢ ciency frontier” we mean policy choices that could not be adjusted to generate Pareto gains across countries when each country’ s welfare is judged by the preferences of its government. This focus on government preferences generally follows the literature, though a distinction is made in some of the literature between ex-ante and ex-post government preferences. We discuss these points further in our survey. Our survey does not focus on research that assesses the value of trade agreements from the perspective of a speci…c country or subset of countries. 1 root cause of the WTO’s current woes rather than its legitimate successor. A third possibility is that governments may have already achieved the international e¢ ciency frontier under the GATT/WTO liberalization process – or if not yet, they could achieve it with selective …xes – so that the WTO is in better shape than it appears; and rather it is the kind of additional liberalization associated with the recent rise of PTAs that represents a failure of e¢ cient international trade policy cooperation. With this interpretation PTAs are liberalization run amok. In this paper we make use of a growing economics literature on international trade agreements to sort through these interpretations and suggest answers to the questions posed above. To facilitate our discussion, we adopt a simple organizing principle: we group papers in the literature by their stance on what makes a trade agreement valuable to its member governments, that is, by the nature of the “problem”that a trade agreement is supposed to “solve”for its member governments. According to this organizing principle, there are four strands of the literature. The oldest and most established strand of the literature is the “terms-of-trade”theory of trade agreements. This theory posits that governments use trade agreements to undo the policy ine¢ - ciencies that are associated with unilateral policy choices when those choices can shift the costs of intervention onto trading partners through movements in foreign exporter prices (terms of trade). In this theory, addressing an international externality (which travels through the terms of trade) is the central purpose of a trade agreement. The “commitment”theory also has a well-established his- tory in the literature, but here the central role for an international externality is absent. Instead, governments value trade agreements as a way to tie their hands (make commitments) against their own lobbies and citizens. The two remaining strands of the literature, what we call the “delocation/pro…t-shifting” and the “o¤shoring” theories of trade agreements, are more recent ar- rivals. They can be viewed as attempts to identify new international externalities that go beyond the terms-of-trade externality to include the local prices in each country, and that can give rise to and shape international trade agreements. The delocation/pro…t-shifting theory argues that such non-terms-of-trade externalities have been important for understanding real-world trade agreements all along, while the o¤shoring theory suggests that non-terms-of-trade externalities may only have become prominent with the recent rise of o¤shoring and international supply chains. As might be anticipated, the strength of the literature’s support for the various interpretations of recent developments depends on which purposes are central to real-world trade agreements. While we discuss below evidence that lends support to all four theories, a growing body of evidence points to the terms-of-trade theory as central for understanding the actual trade agreements that we see. We therefore …rst evaluate these developments from the perspective of the terms-of-trade theory, surveying both the theoretical and empirical literature to assess the various interpretations and establish some initial answers. We then survey the commitment, delocation/pro…t-shifting and o¤shoring theories, describing where they yield di¤erent assessments of these interpretations, and we utilize this description in combination with a survey of the relevant empirical literature to suggest quali…cations to the answers provided by the terms-of-trade theory. To preview, the literature we survey does not support the view that the WTO is passé. On 2 the contrary, from the perspective of the terms-of-trade-theory strand of the literature, economists should reserve their strongest backing for the WTO and adopt a more cautious view of PTAs. The commitment, delocation/pro…t-shifting and o¤shoring theories do raise important caveats to unquali…ed support for the WTO, and there are features of PTAs that these theories support. But until more empirical evidence suggests otherwise, these other strands of the literature do not establish that PTAs rather than the WTO should be entrusted with the rules of globalization. To set the stage, we next provide a brief overview of the main institutional features of the world trading system, focusing on the multilateral framework provided by the GATT/WTO and the current state of PTAs. Section 3 reviews the terms-of-trade theory of trade agreements and surveys the empirical literature that relates to its essential tenets. In Sections 4 and 5 we use the terms-of-trade strand of the literature as a lens through which to evaluate the GATT/WTO and PTA approaches to trade liberalization, and from this perspective we interpret recent developments in the world trading system. Section 6 surveys the literature on the commitment, delocation/pro…t- shifting, and o¤shoring theories of trade agreements and identi…es insights from each that suggests quali…cations to the answers provided by the terms of trade theory. Section 7 turns to an evaluation of dispute settlement in the world trading system. Finally, Section 8 concludes, identi…es directions for future research, and discusses the possible role of critical mass and plurilateral agreements in strengthening and revitalizing the GATT/WTO approach. 2 The World Trading System: a brief overview Individuals and …rms ultimately drive globalization, but governments set the rules of the game, and the rules can be very important to the outcome. Here we brie‡y summarize the two main sets of rules for the world trading system: the policy commitments and their enforcement under the GATT/WTO, and the sets of rules associated with the web of PTAs currently in force. We describe how the recent wave of PTAs is changing the rules of globalization along a number of important dimensions relative to the rules established by the GATT/WTO and even previous PTAs, and we suggest that there are important choices embedded in these two institutional forms. 2.1 The GATT/WTO Multilateral System We begin with some background on the GATT/WTO multilateral System. A short history of tari¤ liberalization under the GATT and WTO From the backdrop of the Smoot-Hawley tari¤s imposed by the US in 1930 and the international retaliatory response that followed, the GATT was created in 1947 with 23 countries and grew in membership over the next …ve decades before being consolidated into the WTO in 1995. As of 2015, the WTO counts 161 member economies –including both the EU and each of the 28 EU member states individually.3 3 The EU is a member of the WTO; for legal reasons it was o¢ cially known until 2009 as the European Communities. The 28 individual countries of the EU are also WTO members in their own right. The EU is a single customs union 3 The trans-Atlantic economies of the US, Canada, and a number of European countries were not only a driving force behind the creation of the GATT, but they provide perhaps the most familiar story line for how the GATT facilitated gradual, multilateral trade liberalization and allowed coun- tries to sustain an extensive period of low most-favored-nation (MFN) tari¤s. These countries used the GATT forum to reduce MFN tari¤s reciprocally through periodic negotiating rounds (WTO, 2007), and they then locked in those low tari¤s through legally binding commitments. Table 1 illustrates the multilateral trade liberalization process of negotiations (percentage tari¤ cuts) cov- ering 1947-1994, and Table 2 documents the resulting average applied ad valorem tari¤ rates for a number of these countries in 1952 and again in 2005 after eight rounds of GATT negotiations.4 The trans-Atlantic GATT experience is not, however, how countries have universally liberalized their MFN tari¤s or even entered into the GATT/WTO system. There are two prominent classes of exceptions that are best illustrated by the GATT/WTO experience of other member countries. First, some countries did not enter the system at its inception; indeed, many “latecomers” did not seek or were not admitted entrance into the agreement until well after the initial set of GATT Contracting Parties had already substantially negotiated MFN tari¤ liberalization. As such, the GATT/WTO has had the ‡exibility to accommodate accessions by major economies, including West Germany in 1951, Japan in 1955, China in 2001, and Russia in 2012.5 Second, many developing countries chose not to participate in the reciprocal tari¤ liberalization negotiations that took place under successive GATT rounds. Instead, countries including GATT founders India and Brazil requested and utilized “special and di¤erential treatment” exemptions from reciprocity in order to pursue import substitution policies. While such countries may currently apply relatively low (in historical terms) MFN tari¤s, their liberalization episodes frequently were not undertaken reciprocally, but instead unilaterally (e.g., India) or in concert with a period of preferential liberalization (e.g., Brazil). These and other countries also did not follow the trans- Atlantic approach of gradually lowering their MFN tari¤s over decades; instead, their period of low and sustained multilateral tari¤s began suddenly and not until the 1990s. And unlike the US and the EU, the relatively low MFN tari¤ rates that countries like Brazil and India apply have not been legally bound under the WTO at similarly low levels. s 161 members notwithstanding, there are at least three dozen countries that Finally, the WTO’ are not yet members. While most are developing countries and some liberalized their trade regimes with a single trade policy and tari¤, and the European Commission “speaks” on behalf of the EU member states in most WTO matters. Nevertheless, most other customs unions are not represented in the WTO in this manner, with individual countries retaining WTO membership rights and obligations. 4 By 1952, average import tari¤s expressed in ad valorem terms had already fallen substantially from peak levels in the 1930s and 1940s due to a combination of in‡ ation, as many were imposed as speci…c duties, and the negotiated liberalization of the …rst three GATT rounds. Irwin (1995, Table 5.2) reports average tari¤ rates in 1931 (after the US imposition of its Smoot-Hawley tari¤) for France, Germany, and Italy of 38, 40, and 48 percent, respectively. Irwin (2011, 2012) describes the political-economy forces behind the import protection that increased sharply during the Great Depression, and Irwin, Mavroidis, and Sykes (2008) describe the negotiations that ultimately led to establishment of the GATT in the late 1940s. WTO (2007) also provides an extensive analysis tracking the multilateral trade liberalization that took place over the 60 year period following the GATT 1947 inception. 5 China was an original Contracting Party to the GATT but withdrew in 1950. The other two original Contracting Parties to subsequently withdraw from the GATT were Lebanon and Syria. 4 independently of the WTO (through either preferential or unilateral tari¤ liberalization), there remain roughly 500 million people that reside in countries entirely outside of the WTO system. Seven percent of the global population has not taken on WTO obligations and does not enjoy the WTO legal bene…ts that we describe in more detail below. Contemporary tari¤ commitments under the WTO Table 3 summarizes many of the salient features resulting from the GATT/WTO’s “shallow” integration approach to trade lib- eralization, including information on contemporary multilateral tari¤s across and within the major economies. The table splits countries into three groups –the high income members of the Group of 20 (G20), the emerging economy members of the G20 (which includes the BRICS, namely, Brazil, Russia, India, China and South Africa), and a selected sample of other major developing countries with 2012 populations of over 50 million - some of which are not (yet) WTO members, as they are currently only WTO “observers.”6 Overall, the tari¤ data indicate substantial heterogeneity across countries and industries and include many examples of applied MFN import tari¤s, as well as the bindings that have been legally negotiated to constrain them, that are not close to free trade. Consider …rst the US. The simple average MFN tari¤ that the US applies to imports from any other WTO member is 3.4 percent. One hundred percent of the US tari¤ lines are bound at some level, and the simple average binding rate is 3.5 percent. The rate above which a country promises not to raise its applied MFN tari¤ is referred to as the binding, and the binding serves as the formal legal commitment that a country submits to the WTO membership. The WTO permits countries to apply tari¤s below their bound rates, provided that such o¤erings are made to all other members on a nondiscriminatory (MFN) basis. That the US applied MFN import tari¤ is pushing up against the binding rate is an indicator that it has very little scope to unilaterally increase its applied MFN import tari¤s without running afoul of WTO rules. While average US applied and bound MFN tari¤s are quite low, there is considerable hetero- geneity both across and within even high-income countries. Most major industrialized economies have almost universal binding coverage and applied rates that are relatively close to their tari¤ bindings. Nevertheless, while applied MFN tari¤s may be low in historical terms, they range from an average of 2.7 percent (Australia) to 13.3 percent (South Korea). There also remain important examples of outliers or tari¤ “peaks” in high-income economies; for example, 2.7 percent of US tari¤s have applied MFN rates higher than 15 percent, with the highest rate being 350 percent. Canada, the EU, and South Korea each have more than 5 percent of MFN tari¤ lines with rates higher than 15 percent, and maximum applied rates in these economies are greater than 500 percent. Tari¤s exhibit even more heterogeneity across emerging and developing economies. While av- erage applied MFN tari¤s are also relatively low for these countries in historical terms, the rates applied by even the relatively advanced (G20) emerging economies are typically substantially higher than their high-income country counterparts. Furthermore, some countries (e.g., India) have not committed to legally binding a signi…cant share of their tari¤ lines at any level. Finally, within 6 Governments with WTO observer status are non-members that are granted limited rights (e.g., access to certain WTO meetings) and are expected to uphold certain obligations (e.g., minimal contributions to the WTO’s budget). 5 the set of products that countries have committed to legally bind, there can be signi…cant di¤eren- tials between applied rates and the binding commitment. This last point holds for all of the G20 emerging economies (including Argentina, Brazil, India, and Mexico) with the exception of the relatively new WTO accession countries of China (2001) and Russia (2012), for which the existing membership demands included relatively low levels of MFN tari¤ bindings.7 Heterogeneity across the tari¤ data can be even more extreme for other major (but poorer) developing countries. Some WTO members (e.g., Bangladesh, Burma, Nigeria) have committed to upper limits for tari¤ bindings on fewer than 20 percent of their import tari¤ lines. Even on products for which these WTO members bind their tari¤s, the average binding rates may be more than 100 percentage points higher than applied rates. There are also important di¤erences in applied MFN tari¤ heterogeneity within countries across sectors. As one important example, Table 3 shows many instances of sharp di¤erences between average applied tari¤s in agricultural products relative to overall rates of protection. Within the G20, a few countries such as Argentina, Australia, and Brazil o¤er lower average import tari¤s for agriculture than they do for other products. For most others, however, the rates in agriculture are substantially higher (Anderson, Rausser, and Swinnen, 2013). Finally, applied MFN tari¤s are not the only important trade policy instrument within the multilateral WTO system. An increasing number of countries since the early 1990s have begun to invoke GATT/WTO exceptions to their negotiated tari¤ bindings and use the temporary trade barrier (TTB) policies of antidumping, safeguards, and countervailing duties. Before 1990, industrialized economies such as Australia, Canada, the EU and the US dominated overall use of TTBs, and especially the most predominant antidumping policy (Blonigen and Prusa, 2003). Since the early 1990s, a number of emerging economies have subsequently become major users of TTBs (Bown, 2011a) as they reduced their applied import tari¤s. The last column of Table 3 provides data on the import coverage of the TTBs cumulatively applied in 2012. As examples, eleven di¤erent G20 economies had more than 1 percent of their tari¤ lines also subject to an imposed TTB in 2012; some of these countries did not even have an antidumping law in place 25 years earlier. Many also had one or more episodes over this 25 year period during which the cumulative TTB import coverage rose to as high as 4-6 percent. Table 3 reveals two other features of TTB use. First, not all WTO members use these policies. Indeed most of the poorest WTO members have never implemented a formal antidumping or safeguard proceeding, a feature that can be partially explained by the fact that the tari¤ bindings of these countries are su¢ ciently above their applied rates that they can adjust tari¤s upward unilaterally in response to shocks. Second, even members of a customs union –i.e., countries that eliminate tari¤s on internal trade with each other and share a common applied MFN tari¤ toward non-members, examples of which we describe in more detail below – do not necessarily apply a common set of TTB policies. In 2012, for example, customs union partners Argentina and Brazil 7 As of 2012, Russia had not yet fully phased in its MFN applied tari¤ cuts under its WTO accession terms and thus its average applied rate was still above its average binding commitment. 6 had di¤erent shares of product lines covered by TTBs, as did the partners Turkey and the EU. GATT/WTO commitments relating to behind-the-border measures The GATT tradi- tionally eschewed e¤orts to negotiate restrictions on the use of behind-the-border measures of its member governments. As Hudec (1990) describes in his depiction of the genesis of GATT’s shallow- integration approach, while governments understood that behind-the-border measures could have trade e¤ects, the GATT never had its heart in deep integration: “...The standard trade policy rules could deal with the common types of trade policy measure governments usually employ to control trade. But trade can also be a¤ected domestic’measures, such as product safety standards, having nothing to do by other ‘ with trade policy.”[When GATT was created in 1947,]...governments would never have agreed to circumscribe their freedom in all these other areas for the sake of a mere trade agreement.” Hudec (1990) The WTO emphasizes a shallow-integration approach as well but has attempted to venture into the realm of “deeper” integration, most substantively with the aborted Doha Round attempts to negotiate directly over the “Singapore issues” of foreign investment and competition policy.8 It is also important to point out that many of the deep-integration issues of apparent focal interest to recent PTAs are not entirely absent from consideration by the WTO Agreements (WTO, 2012). It is simply that the GATT/WTO shallow-integration approach has addressed such issues di¤erently. For example, it is true that WTO member governments maintain considerable freedom to implement unilaterally a variety of public policy interventions, including those that adversely a¤ect trade ‡ows. Examples include allowances for the protection of plant, animal and human health as well as the establishment of product standards. But the WTO provides governments with guidance so as to discourage such interventions from becoming non-tari¤ barriers applied without a legitimate public policy motive. For trade in goods, these exceptions and guidelines are outlined in the GATT’s basic rules on national treatment found in Article III and are further elaborated under the GATT’s original Article XX, and the WTO’s Agreements on Technical Barriers to Trade (TBT), and Sanitary and Phytosanitary (SPS) Measures. Furthermore, countries can bring non- tari¤ issues to light by also …ling “speci…c trade concerns”with relevant WTO standing committees. Finally, the GATT/WTO provides member governments with the right to initiate, under formal dispute settlement proceedings, “violation” complaints against behind-the-border measures that violate these guidelines, and even “non-violation” complaints against behind-the-border measures that do not violate the guidelines but still erode negotiated market access commitments. 8 In addition to foreign investment and competition policy, the Singapore issues included trade facilitation and government procurement. Trade facilitation focused on the removal of non-tari¤ barriers “at the border” (e.g., procedures for clearing customs), and the Doha Round has produced a Trade Facilitation Agreement (TFA). The attempt in the context of the Doha Round to negotiate multilateral rules for government procurement failed along with the attempts to negotiate a multilateral agreement covering foreign investment and competition policy, and instead a revised version of the plurilateral Government Procurement Agreement (GPA) …rst signed in 1979 was negotiated among 43 WTO members and entered into force on April 6 2014. The WTO TRIPS Agreement is a deep integration agreement, but it is not considered a market access agreement (see also note 1). 7 Dispute Settlement under the WTO and GATT Access to formal dispute settlement pro- cedures has always been part of the GATT/WTO system. The 1947 GATT’s Article XXIII estab- lished the basic provisions whereby government-to-government dispute resolution would take place, and over the subsequent 50 years, Contracting Parties initiated more than 250 disputes under the GATT fora in attempts to formally resolve a variety of trading frictions that arose.9 Legal schol- ars generally characterize the GATT-provided mediation that took place during this period as a “diplomacy-based” approach to dispute resolution. Many elements of the system changed dramatically in 1995 with the WTO inception, as the current system is much more “legalistic”than its GATT predecessor. Members initiated nearly 500 formal WTO disputes against one another between 1995 and 2014, or nearly twice as many as during the GATT period of 1947-1994. The literature identi…es a number of contributing explanations, including that more countries are now actively involved in the trading system, there is substantially more trade, and countries have taken on more legally binding commitments. Over time, more and more WTO members have found themselves involved in formal disputes. To date, nearly 50 out of the 161 WTO members have initiated a case as a “complainant”(i.e., the plainti¤) and more than 50 members have faced a dispute as a “respondent” (i.e., the defendant). More than half of the membership has been formally involved in at least one dispute via the legal status as an “interested third party.” This can be an important role even for countries without trade stakes in a particular dispute, given that jurisprudence arising from a dispute between any two countries –e.g., a policy dispute pitting Colombia versus Panama or Moldova versus Ukraine– could have policy implications for the entire WTO membership, including the US and EU. The US and the EU are the two most frequent WTO litigants; combined they have initiated roughly 40 percent of all disputes, roughly 50 percent of all disputes involve one or the other as a respondent, and a signi…cant share involves one challenging the other. Nevertheless, many other industrialized countries have also been frequent WTO litigants, including Australia, Canada, Japan, New Zealand, and South Korea. Finally, the share of WTO disputes involving developing country members, with the exception of least developed countries, has risen over time, and includes a large number of developing country versus developing country disputes.10 Developing countries that are frequent WTO litigants include Argentina, Brazil, Chile, China, Colombia, Guatemala, Honduras, India, Indonesia, Mexico, Panama, Peru, Philippines, and Thailand. 2.2 PTAs While a central pillar of the GATT/WTO system is the MFN principle, GATT Article XXIV provides an exception to MFN that allows GATT/WTO members to form PTAs that satisfy certain features. The key stipulations are that the PTA must eliminate tari¤s on “substantially all” trade among the member countries, and that the external MFN tari¤s that member countries continue 9 See Bown (2002; Table 1). 10 The poorest and least developed country members of the WTO system –of which there are dozens –are almost entirely absent from participation in formal WTO dispute settlement. Bown and Hoekman (2008) provide a discussion of the political-economic hurdles faced by these countries in the WTO that can help account for this fact. 8 to apply to imports from outside the PTA not increase as a result of PTA formation. For decades during the post-World War II period, much of the analysis of PTAs centered primarily on one successful experience of regional integration - i.e., the continuing and ongoing evolution of western Europe. The 1951 Treaty of Paris established the European Coal and Steel Community (ECSC) which was expanded with the 1957 Treaty of Rome to create the six-country s EU is the result of continued integration over the European Economic Community (EEC). Today’ subsequent …ve decades, including numerous country accessions (28 member countries, as of 2015) as well as substantial “deepening” of negotiations and agreements beyond trade preferences and toward factor market, economic, monetary, and even political integration. Beginning in the late 1980s, a number of other potentially economically meaningful PTAs arose that have subsequently been sustained. These include the 1987 CUSFTA (Canada-US Free Trade Agreement) that was subsequently expanded into the NAFTA (North American Free Trade Agree- ment) through the addition of Mexico in 1994. There are also increasingly important developing country PTAs, including the MERCOSUR (Mercado Común del Sur) customs union involving Argentina, Brazil, Paraguay and Uruguay in the early 1990s, ASEAN (Association of Southeast Asian Nations) Free Trade Area involving Brunei, Indonesia, Malaysia, Philippines, Singapore and Thailand in the early 1990s, and CAFTA-DR (Central American Free Trade Area – Dominican Republic) involving the US and …ve Central American economies in the mid-2000s. As of 2014, the WTO reports that it has been noti…ed of nearly 600 reciprocal trade agreements in existence, and nearly 400 agreements are currently in force (WTO, 2014a).11 Tari¤s and behind-the-border measures in PTAs There are two key areas in which PTAs push beyond the multilateral, WTO commitments. The …rst is by reducing import tari¤s even lower than WTO levels, albeit on a discriminatory basis. The second is by negotiating beyond tari¤s directly over new, behind-the-border policy instruments. WTO (2011) provides a recent and relatively comprehensive characterization of the patterns of tari¤s and trade taking place under PTAs. With the sharp increase in PTAs since 1990, the value of trade between PTA members has grown faster than the world average; not surprisingly the share of intra-PTA trade in world trade has nearly doubled from 18 percent in 1990 to 35 percent in 2008.12 And when intra-EU trade ‡ows are included in these statistics, intra-PTA trade as a share of world trade increased from 28 percent to 51 percent over this period. However, the WTO (2011) data analysis, based on a matching of product-level trade ‡ows to tari¤s and preferential tari¤s to MFN tari¤s, reveals a number of other stylized facts, some of which challenge the conventional wisdom regarding the degree to which PTAs serve as a force for discriminatory tari¤ liberalization. First, while many theoretical models typically assume PTAs result in zero applied tari¤s be- 11 The number of noti…cations and trade agreements in force di¤er for several reasons. One is because noti…cations include not only new agreements, but also the accession of new countries to existing agreements - e.g., Croatia’ s accession to the EU in 2013. Second, some agreements noti…ed to the GATT/WTO later become “inactive” (or no longer in force), when they become superseded by a subsequent agreement that was later noti…ed and which is currently in force - e.g., CUSFTA is no longer in force as it was superseded by NAFTA. 12 The data reported here and below derives speci…cally from section II of WTO (2011, pp. 47-86). 9 tween partners, real-world PTAs do not always lead to zero tari¤s on all intra-PTA goods trade. Empirically, the many negotiated exceptions within PTAs have resulted in a signi…cant number of PTA tari¤s remaining at levels above zero, including eight percent of tari¤s for the major PTAs of the US, Canada, EU, and Japan (Damuri, 2012). Indeed, in an analysis of the PTAs involving 85 countries and 90 percent of world trade in 2007, the WTO (2011, pp. 124-125) …nds that roughly 66 percent of tari¤ lines with MFN tari¤ “peaks” (MFN rates de…ned as greater than 15 percent) have not been reduced at all through PTAs. Hence, while existing PTAs should be viewed as a signi…cant force in eliminating (roughly one third of, and on a discriminatory basis) the tari¤ peaks that remain among WTO members, a majority of these tari¤ peaks are nevertheless still in place. Second, while a large and increasing share of world trade takes place between PTA members, this share substantially overstates the amount of preferential trade between members. In many instances there is no preference margin because the MFN tari¤s are also zero. Furthermore, even where positive preference margins exist, exporters may not utilize available preferences because of both the resource costs (to sourcing inputs from less e¢ cient suppliers in PTA markets) and bureaucratic costs (to proving legal compliance) due to rules of origin and local value-added requirements needed to gain access to the lower preferential rates.13 How much trade really takes place under preferential tari¤s? First, between 49 percent (includ- ing intra-EU trade) and 65 percent (excluding intra-EU trade) of world trade takes place between countries that are not part of a common PTA.14 Second, excluding intra-EU trade, the WTO es- timates that only 16 percent of global trade is eligible for any preferential tari¤s and less than 2 percent is eligible to receive preferences with margins above 10 percentage points. Including intra- EU trade in these statistics implies that 30 percent of global trade is eligible for any preferential tari¤s and 4 percent is eligible for margins over 10 percentage points. Despite the explosive increase in PTA adoption, the WTO estimates that overall, excluding (including) intra-EU trade, 84 percent (70 percent) of world merchandise trade still takes place on an MFN basis. These numbers can help put the impact of existing PTAs in perspective. PTAs have served as the primary conduit for tari¤ discrimination in the WTO system. And they have led to discriminatory tari¤ reductions below MFN levels that are far from insigni…cant. But it would appear that to date PTAs have not delivered discriminatory tari¤ liberalization on a wide enough scale to cause widespread “trade diversion” (the reduction in imports from third countries; see Viner, 1950).15 Finally, an increasingly important characteristic of many of the current PTA negotiations is 13 See, however, Keck and Lendle (2014) for a recent challenge to the position that preferences often go unutilized. 14 Considering these …gures with and without intra-EU trade ‡ ows may be important depending on the context, given that the EU is a unique PTA in that it is not only a customs union but has undertaken deeper integration along many dimensions - including factor markets and monetary integration for a substantial subset of member countries - and also steps toward political integration. 15 A potentially important caveat to this last observation, however, is suggested by the results of Handley (2014) and Handley and Limão (forthcoming) which we discuss further below: in the presence of policy uncertainty there can be large di¤erences between the trade e¤ects of an applied MFN tari¤ of zero that is bound at a much higher level in the WTO and a PTA tari¤ that is both applied and bound at zero. Such di¤erences are missed by a focus on preference margins relating to applied tari¤s alone, and inferences about the degree of trade diversion caused by existing PTAs which adopt this focus could be signi…cantly understated as a result. See also the discussion in Bhagwati (2008) and the survey in Panagariya (2000). 10 that they are no longer primarily about tari¤ liberalization but instead are pushing toward deeper integration that addresses non-tari¤ and behind-the-border policies.16 The nascent literature on deeper integration currently splits new PTA issue areas into two categories. The …rst are “WTO- plus” PTA provisions - i.e., those that also exist under the WTO, but where PTA members use their agreement to take on commitments to go further. Tari¤s are the clearest example; e.g., WTO members make legally binding MFN tari¤ commitments, and PTAs involve partners lowering at least some of those tari¤s toward each other even further. Other examples include services, intellectual property rights, and product standards - each of which has at least some basic WTO coverage. The second category for PTA provisions are “WTO-extra”areas, and these involve issues that are not yet explicitly addressed by the WTO. Examples of WTO-extra areas include labor standards, environmental standards, foreign direct investment provisions, movement of capital, competition policy, data protection, and even potential cooperation over other domestic regulations in order to help achieve improved levels of “regulatory coherence” across PTA member countries. Horn, Mavroidis, and Sapir (2010) characterize the “depth”of PTA provisions by applying this categorization to the many US and EU PTAs in existence as of 2008.17 The initial evidence was that EU PTAs tend to have many more WTO-extra provisions but that the pattern is reversed when the analysis conditions on the legal enforceability (under dispute settlement) of the provisions, as US PTAs contain more legally enforceable WTO-extra provisions. In follow-up work, the WTO (2011, Section D) extended this approach in order to characterize 14 di¤erent “WTO-plus”provisions and 38 di¤erent “WTO-extra” provisions for a wider sample of PTAs, including a number involving only developing countries. Their work has established a new and rich set of databases for future research to explore the heterogeneity in application of these provisions across di¤erent PTAs. Dispute settlement under PTAs In contrast to the WTO, there is very little empirical record of sustained and e¤ective dispute resolution taking place under the major PTAs. With the exception of the EU, dispute settlement provisions in most PTAs have rarely been used, and when actually triggered, their record of resolving disputes is mixed at best.18 It is also not uncommon for the use 16 Some of these non-tari¤ policies under negotiation are applied at the border. For example, policies like an- tidumping and safeguards are applied at the border but frequently as quotas or price undertakings. Other examples of non-tari¤ barriers that arise at the border may include customs regulations, import valuation, etc. 17 Other recent contributions characterizing and assessing such PTA provisions include work by WTO Secretariat legal sta¤ (Chase et al, 2013) and political scientists (Elsig and Allee, 2015). Note that the latter assess a larger cov- erage of dispute settlement provisions in preferential agreements in a publicly available “design of trade agreements’ ‘ (DESTA) database (Dür, Baccini, and Elsig, 2014). See also WTO (2011). 18 The EU has a di¤erent institutional design, including a supra-national framework that initiates disputes against member states from within and thus does not rely exclusively on the “state-to-state”framework of dispute resolution found in the WTO and many other PTAs. One result is that the EU’ s dispute settlement provisions have led to thousands of disputes. Tallberg and Smith (2014, p. 126) report that the supranational European Commission initiated more than 30,000 cases over 1978-2009 against its member states. Furthermore, the Commission only referred 11.5 percent of these initiated disputes to the European Court of Justice for a legal decision. On the other hand, EU member states have initiated only a handful of disputes against one another. Finally, the EU’ s free trade agreement with Iceland, Liechtenstein, and Norway under EFTA contains a supranational Surveillance Authority (SA) modeled similarly to the European Commission; Tallberg and Smith (2014, p. 138) report that the SA initiated roughly 400 disputes against the three EFTA member states over 1994-2008 and that the member states …led zero EFTA disputes against one another during this period. 11 of PTA dispute settlement procedures to generate third-country spillovers and thus wider disputes that are left for the WTO system to resolve, or for PTA members to simply ignore the existence of their PTA’s dispute settlement provisions in order to take frictions directly to the WTO for resolution. We illustrate with examples from two di¤erent PTAs. Consider …rst the MERCOSUR customs union and its dispute settlement procedures. While Tallberg and Smith (2014) report that very few (roughly 20) disputes were initiated under MER- COSUR between 1993 and 2005, one particularly high pro…le MERCOSUR dispute ended with Brazil imposing a new import restriction on retreaded tires from non-MERCOSUR partners but not on its MERCOSUR partners. This policy discrimination arose after a MERCOSUR legal rul- ing in 2002 that PTA partners must be exempted from application of such import restrictions. Citing a similar MERCOSUR rule, in 1997 Argentina had imposed a new import restriction on footwear from non-MERCOSUR partners but not on its MERCOSUR partners. Because MER- COSUR rules apparently required that imports from MERCOSUR partners be exempted from the s and Argentina’ policies, Brazil’ s newly imposed import-restricting policies provided an additional implicit preference to PTA partners relative to non-partners. In both instances, non-MERCOSUR countries, including the EU and Indonesia, challenged the discriminatory treatment under formal WTO dispute settlement procedures.19 Like MERCOSUR, NAFTA also has its own dispute settlement provisions, and they have also rarely been triggered; e.g., fewer than 15 disputes were initiated under NAFTA between 1994 and 2010, and NAFTA dispute settlement largely fell into disuse after 2001 (Tallberg and Smith, 2014). Nevertheless, the small number of NAFTA disputes should not necessarily be interpreted as evidence that its PTA partners are not experiencing bilateral trading frictions that require third party mediation. The three NAFTA partners (US, Canada, and Mexico) have taken more than twice as many disputes against one another to formal WTO dispute settlement since NAFTA’s inception than they have taken to the NAFTA forum. And some of these bilateral frictions - e.g., over US-Canada trade in softwood lumber; over US-Mexico trade in the related products of sugar, corn, high-fructose corn syrup, and ultimately soft drinks - actually started as formal NAFTA disputes but could not be resolved under the NAFTA forum. The disputes escalated and ultimately spilled over to require resolution through formal WTO litigation.20 2.3 Di¤erent Paths Forward The WTO and PTAs are on di¤erent trajectories. The extent of their divergence to date may still be modest, but the “mega-regional” PTAs currently under negotiation, such as the Transatlantic Trade and Investment Partnership (TTIP) between the US and EU, or the Trans Paci…c Partnership (TPP) between the US, Japan, and other Paci…c trading partners, could change this dramatically. 19 These disputes are described in greater detail in Section 7.3 below. Bown and Trachtman (2009) provide a discussion of the WTO dispute over Brazil - Retreaded Tyres. 20 Davey and Sapir (2009) discuss the evolution of the US-Mexico disputes over sweetners that ultimately culminated in the WTO’ s Mexico - Taxes on Soft Drinks case, and Bown and Sykes (2008) describe the …fth WTO dispute brought by Canada over US - Softwood Lumber. 12 In short, globalization is looking increasingly di¤erent under these two sets of rules. Perhaps nowhere is this better illustrated in current a¤airs than by the potential implications depending on whether the US and the EU throw their weight behind the WTO and a re-energized Doha Round, or rather put their e¤orts into negotiating new PTAs. Consider the likely di¤erences in globalization’s outcomes depending on which of these strategies is pursued. If the US and the EU were to put their full support behind the Doha Round, even the most ambitious conclusion of the round would by all accounts entail relatively small cuts in average tari¤s and more substantial but still modest reductions in the remaining tari¤ peaks (and agricultural ex- port subsidies). And any nod to deep integration would likely be modest. Contrast this description of a successful WTO Doha Round with what has been leaked about the TTIP and TPP initiatives. Consider …rst the TTIP negotiations. The TTIP has adopted as its main focus the streamlining of domestic standards across the Atlantic. The BBC puts it this way: “Direct tari¤s on goods and services between the two are already low, but there are other barriers such as regulatory and safety standards, inspection procedures, and preferences for domestic business. Removing these could signi…cantly reduce the costs for companies doing transatlantic business.” BBC News (7/8/2013) Consider next the TPP negotiations. Here again the focus is on harmonizing domestic standards. As Marketplace Morning Report (1/28/2014) put it, “The Trans-Paci…c Partnership has been called NAFTA on steroids.” The New York Times continues: “If successful, the TPP agreement would eliminate most remaining tari¤s on nearly $2 trillion in goods and services exchanged between the United States, Australia, Brunei, Canada, Chile, Japan, Malaysia, Mexico, New Zealand, Peru, Singapore and Vietnam. [But the TPP negotiations] “...would go far beyond lowering tari¤s, with provisions requiring countries to maintain compatible regulatory regimes, facilitate corporate …- nancial transactions, establish copyright and patent protections to govern intellectual property rights and to safeguard foreign investors.” NYTimes (2/5/2014) Evidently, the likely form of liberalization under mega-regional initiatives di¤ers substantially from that which might be expected through re-energized multilateral negotiations in the WTO. It is thus important to have a reasoned and informed general perspective about the relative merits of regional and multilateral liberalization initiatives. Toward this goal, in the next …ve sections, we consider the implications of the theoretical and empirical literature on trade agreements. It is also important in this context to consider possible means through which the WTO might be further revitalized. We postpone discussion of this issue until the concluding section. 3 The Terms-of-Trade Theory of Trade Agreements In this section we o¤er a brief review and empirical assessment of the essential predictions of the terms-of-trade theory, and we use the theory as a lens through which to view the broad contours 13 of existing WTO policy commitments, o¤ering an initial consideration of the possibility that the WTO membership might have already arrived at the international e¢ ciency frontier. 3.1 Theory To present the essential predictions of the terms-of-trade theory, we develop a benchmark two- country two-good perfectly competitive general equilibrium trade model. We …rst develop the model under the assumption that governments have only tari¤s as instruments of policy intervention, and consider the purpose of a trade agreement. We then extend the benchmark model to allow that governments also have behind-the-border policy instruments at their disposal in order to consider whether the purpose of a trade agreement is changed in this richer policy setting. Trade model We assume that readers are familiar with the two-country two-good perfectly competitive general equilibrium trade model, and we focus here only on the essential notation and equations.21 The two countries, home (no ) and foreign ( ), produce, consume and trade two goods, x and y , under conditions of perfect competition, with home the natural importer of x and foreign the natural importer of y . The relative price facing home producers and consumers in their local market is p px =py , while the relative price facing foreign producers and consumers in their local market is p px =py . With non-prohibitive home and foreign ad valorem tari¤s denoted respectively by t and t , and with (1+ t) and (1+ t ), international goods-market arbitrage implies p = pw p( ; pw ) and p = pw = p ( ; pw ), where pw px =py is the “world” (i.e., untaxed) relative price. The foreign terms of trade is pw , and the home terms of trade is 1=pw . Production (as well as the distribution and level of factor incomes) is fully determined in each country by the local relative price in that country, while each country’s consumption depends on both the local relative price in that country and the terms of trade (with the latter, together with the local price, determining the tari¤ revenue collected by the country and distributed to its consumers). Each country’s trade is simply the di¤erence between its consumption and production. Hence, for any local and world prices, home imports of x can be written as M (p; pw ) and home exports of y can be written as E (p; pw ). The analogous functions for foreign are M (p ; pw ) and E (p ; pw ). For any prices, the home and foreign national budget constraints are then given by the respective trade balance equations pw M (p; pw ) = E (p; pw ); and (1) M (p ; pw ) = pw E (p ; pw ); (2) ew ( ; with the equilibrium world price p ) then determined by the market clearing condition for x M (p( ; pw ); pw ) = E (p ( ; pw ); pw ); (3) 21 A more complete description of the model appears in any undergraduate International Economics textbook (or see Bagwell and Staiger, 2010a, for a recent development of the model in the context of the trade agreements literature). 14 and with Walras’ Law ensuring that the y -market clears as well. Finally, with market-clearing ew ) and p ew ), p = p ( ; p local and world prices written as p = p( ; p ew ( ; ew = p ), we impose the following standard price assumptions to rule out the Metzler and Lerner Paradoxes: ew ( ; )) dp( ; p dp ( ; pew ( ; )) > 0> (4) d d @pew ( ; ) ew ( ; ) @p < 0< : @ @ According to (4), each country’s tari¤ is “protective” of its import-competing sector (i.e., the imposition of a tari¤ raises the local price of the import good), and each country is “large”in world markets and can improve its terms of trade with an increase in its tari¤. Government preferences When it comes to the goals of trade policy, real-world governments have diverse sets of preferences, in some cases adopting trade policies that would seem to promote aggregate national income while in other cases adopting trade policies with a clear distributional goal in mind. This diversity is re‡ected in the trade policy literature, where assumed government preferences range from national income maximization (see Dixit, 1987, Johnson, 1953-54, Kennan and Riezman, 1988 and Mayer, 1981 for important formalizations, and Kowalczyk and Riezman, 2011 for a recent survey) to those of a representative democracy as re‡ected in the preferences of the median voter (see Mayer, 1984 for the initial formulation, and Dutt and Mitra, 2002 and Dhingra, forthcoming for important follow-up work) to those of a government in‡uenced by lobbies (for early formalizations, see Olson, 1965, Caves, 1976, Brock and Magee, 1978, Feenstra and Bhagwati, 1982, Findlay and Wellisz, 1982, Hillman, 1982 and Baldwin, 1987; and see Grossman and Helpman, 1994 and 1995a, for the canonical treatment in the more recent literature). The diversity of government preferences, both in the real world and in the formal trade policy literature, raises the question whether these preference di¤erences across governments might trans- late into di¤erent purposes across the trade agreements that governments negotiate. To ensure that our answers regarding the purpose of a trade agreement are not dependent on adopting a particular formulation of government preferences from this diverse set, we follow Bagwell and Staiger (1999a, 2002) and adopt a “reduced form”approach to modeling government preferences, representing the ew ), ew ) and W (p ; p objectives of the home and foreign governments with the general functions W (p; p respectively. We thus represent welfare in terms of the prices that the tari¤s induce rather than directly in terms of the tari¤s themselves. We place no restrictions on a government’s preferences over its local prices. As local prices determine the level and distribution of factor incomes, this allows us to incorporate all of the formal models of trade policy motives mentioned above. We do impose one assumption on the government welfare functions, namely, that holding its local price …xed, the welfare of a government increases when its terms of trade improve: Wp ew ) < 0 < Wp ew (p; p ew ): ew (p ; p (5) 15 This assumption, which amounts to a statement that each government would like more tari¤ revenue if it could achieve this extra revenue without experiencing any change in its local price, is met by each of the formal models of trade policy determination that we mentioned above.22 Nash tari¤s In the absence of a trade agreement, the two governments choose their tari¤s unilaterally and non-cooperatively, and we assume that these choices are characterized by an interior Nash equilibrium. The …rst-order conditions that de…ne the Nash tari¤ s are ew ) dW (p; p dp @pew = Wp + Wp ew = 0, and (6) d d @ ew ) dW (p ; p dp @p ew = Wp + Wpew = 0: d d @ s best-response tari¤, while the bottom The top equation of (6) de…nes the home government’ s best-response tari¤, with the Nash tari¤s de…ned where equation de…nes the foreign government’ both governments are on their respective reaction curves. Notice that, as the top equation of (6) highlights, in the Nash equilibrium the home government strikes a balance between the e¤ects on its welfare of the local-price and world-price movements induced by its tari¤ choice. The welfare implications of the local-price movement are domestic in nature, re‡ecting the trade-o¤ for the home government between the bene…ts of any distributional changes (e.g., induced political support) and the costs of the induced economic distortions. By contrast, the welfare implications of the world-price movement are international in nature, as they re‡ect the bene…ts to the home government of shifting some of the costs of its policy choice onto the foreign country. The cost shifting occurs because an improvement in the home country’s s terms of trade. An analogous terms of trade is necessarily a deterioration in the foreign country’ interpretation holds for the foreign government, as the bottom equation of (6) highlights. In the special case where governments maximize national income with their unilateral tari¤ choices, (6) de…nes the standard (Johnson, 1953-54) “optimal tari¤” for each country, which is simply the inverse of the trading partner’s export supply elasticity. As Johnson demonstrated, when governments seek to maximize national income, setting a tari¤ at this level is the optimal way for a country to exploit its monopoly power on world markets. For this case, on the margin the tari¤ creates costly distortions in the local market (the local-price movements in the …rst terms in (6)) but some of these costs are borne by the trading partner (via the world-price movements in the second terms in (6)). For the more general cases of government preferences included in (6), the local-price movements carry additional welfare implications for the governments and this leads to Nash tari¤s that will in general di¤er from the Johnson “optimal tari¤” formula, but the trade-o¤ faced by each government in setting its unilaterally optimal tari¤ is otherwise the same. 22 We also assume su¢ cient concavity of W and W so that the second-order conditions for the optimization problems that we consider below are satis…ed. 16 E¢ ciency frontier If a trade agreement is to be useful to governments, there must be an ine¢ - ciency associated with the Nash tari¤ choices of the governments when evaluated with reference to their objectives. A trade agreement can then provide value to both governments by correcting this ine¢ ciency. Absent such an ine¢ ciency, it would not be possible for a trade agreement to yield Pareto bene…ts for the governments involved. We therefore next characterize the e¢ ciency frontier. The international e¢ ciency frontier is de…ned by the set of tari¤s that satisfy the familiar d d 23 Making use of the home and foreign tangency condition d jdW =0 = d jdW =0 . government welfare functions ew ) W (p; p and W (p ew ), this tangency condition can be written as ;p w dp @pw e @pe Wp + Wp ew ] @ [ W p + Wp d ew @ dp @pe w = 1 @pe w : (7) Wp d + Wp ew @ [ ew ] @ Wp + Wp The characterization of the e¢ ciency frontier provided by (7) is a generalization of the more familiar Mayer (1981) locus of e¢ cient tari¤s for the case of national-income maximizing governments. As is well-known, when governments maximize national income, reciprocal free trade ( = 1 = ) is e¢ cient; yet as Mayer pointed out, this is but one point on the e¢ ciency frontier de…ned by the locus of points = 1= which ensure equality of the home and foreign local prices p and p . Under the assumption that governments maximize national income and with the particular forms ew ) that this implies, (7) reduces to the Mayer locus ew ) and W (p ; p for W (p; p = 1= . With the e¢ ciency frontier de…ned, it is now a simple matter to use (6) and (7) to con…rm that Nash policies are indeed ine¢ cient in this model. Hence, according to the terms-of-trade theory, an ine¢ ciency exists when governments set their tari¤s non-cooperatively, and therefore there is a role for a trade agreement to address this ine¢ ciency and improve the welfare of each government. Interpreting the purpose of a trade agreement The ine¢ ciency of Nash policies is not surprising. After all, the cost-shifting motives that are embodied in the second term of each reaction curve in (6) impose a negative externality on the trading partner, pointing to an obvious source of ine¢ ciency. And in fact it can be con…rmed that Nash tari¤s are higher than e¢ cient tari¤s.24 But recalling that our reduced form government preferences are speci…ed in a way that is su¢ ciently general to capture all of the leading models of trade policy determination, it seems reasonable to expect that additional sources of ine¢ ciency might also arise depending on which model of trade policy determination is relevant; and in this light it would be surprising if the cost- shifting externality were the only source of ine¢ ciency that a trade agreement can correct in our model. Yet this is what the terms-of-trade theory implies. To see this, let us suppose that the home and foreign governments were replaced by hypothetical 23 When international lump sum transfers are available at the negotiation stage, as is typically assumed in the trade agreements literature when partial equilibrium models of trade are employed, internationally e¢ cient policies maximize the joint government surplus, with the division of surplus then allocated across participating governments via lump sum transfers. In later sections when we refer to e¢ cient policies in the context of partial equilibrium models we will adopt this joint-government-surplus-maximizing perspective unless otherwise noted. 24 See Bagwell and Staiger (1999a, 2002). 17 governments that were not motivated by the terms-of-trade implications of their unilateral trade- policy choices; that is, let us consider a hypothetical home government for which Wp ew 0 and a hypothetical foreign government for which Wp ew 0. If these hypothetical governments were to select their tari¤s non-cooperatively, then according to (6) their tari¤ choices would satisfy ew ): ew ) = 0 = Wp (p ; p Wp (p; p (8) Following Bagwell and Staiger (1999a, 2002), we refer to tari¤s that satisfy (8) as politically optimal tari¤ s. If politically optimal tari¤s are e¢ cient, where as in (7) e¢ ciency is evaluated relative to actual home and foreign government preferences, then we may conclude that the terms-of-trade externality is the sole rationale for a trade agreement in this model. But it is immediate that (7) is satis…ed when evaluated at tari¤s that satisfy (8), and hence politically optimal tari¤s are e¢ cient. Of course, as (7) suggests, there will in general be an entire locus of tari¤ combinations that satisfy the condition for e¢ ciency, and the politically optimal tari¤s represent only one point on this locus. The politically optimal tari¤s are arguably focal, however, as they remedy the terms-of- trade ine¢ ciency in a direct way. For example, in the case where governments maximize national welfare, we have already observed that e¢ cient tari¤s lie on the locus de…ned by = 1= as Mayer (1981) showed. The politically optimal tari¤s in this case correspond to the reciprocal free trade point = = 1, which seems focal because it conforms to the trade policies that national income maximizing governments would have adopted in the …rst place if they were not motivated by terms-of-trade/“optimal tari¤” considerations.25 Having determined that the cost-shifting externality is the only source of ine¢ ciency of the Nash policies according to the terms-of-trade theory, it is worth emphasizing now the role that the large- country assumption plays in establishing a purpose for a trade agreement in this theory. In a world of small countries where no country can impact its terms of trade with its tari¤ choices, no country can engage in international cost-shifting and Nash policies are therefore e¢ cient. According to the terms-of-trade theory, then, the ine¢ ciencies that a trade agreement can address are associated with the policies of governments that exercise market power on world markets. This is of course the same conclusion that Johnson (1953-54) drew, but that is in fact the striking point: the introduction of government preferences that can capture the wide diversity of government motives that we see in the real world does not qualify, complicate or change this conclusion. When it comes to the purpose of a trade agreement, therefore, the terms-of-trade theory embod- ies a very simple idea. A trade agreement can be valuable to governments, but only if in the absence of an agreement governments would attempt to shift costs onto one another and as a consequence adopt ine¢ cient unilateral policies. The terms-of-trade externality is simply the mechanism by which this cost-shifting occurs. But while simple and intuitive at one level, the practical relevance 25 Our claim that politically optimal tari¤s are focal does not translate into a claim that a trade agreement necessarily delivers the politically optimal tari¤s. That depends, in part, on whether politically optimal tari¤s lie on the contract curve, which in turn requires that countries not be too asymmetric. For example, as is well known, in the case of national income maximizing governments, a su¢ ciently big country can “win the tari¤ war,”meaning that it is better o¤ in the Nash equilibrium than at reciprocal free trade (see Johnson, 1953-54, and Kennan and Reizman, 1988). 18 of this idea has traditionally met with deep skepticism among many economists. Some of this skepticism re‡ects a lack of empirical evidence relating to the central tenets of the theory, such as the degree and prevalence of market power that real-world governments can wield in international markets and whether their unilateral tari¤ choices re‡ect this market power when they possess it, and in Section 3.2 we survey the recent empirical work that is beginning to …ll this gap. But some of the skepticism re‡ects a more visceral objection to the plausibility of the theory. Here we brie‡y consider three of the main objections of this kind. A …rst objection is that the terms-of-trade theory unrealistically posits that governments seek to maximize national income with their tari¤ choices. But as we have just illustrated, while the terms- of-trade theory was originally posed by Johnson (1953-54) under this assumption, it holds equally well when governments have political motivations. A second objection is that there is a disconnect between the theory and the way real-world governments think. This objection is often expressed by the view that real-world governments almost never mention the “terms of trade”in their policy discussions, and instead the language of real trade agreements and the negotiators who craft them emphasizes “market access.” But this disconnect may be more apparent than real, because the insights of the terms-of-trade theory can be easily translated into the language of market access. Speci…cally, when a government raises its import tari¤, it shifts in its import demand curve, and the resulting “price e¤ect” under which the home country enjoys a terms-of-trade improvement is accompanied by a “volume e¤ect”under which the foreign country experiences a reduction in access to the home market. Using this link between price and volume e¤ects, the terms-of-trade theory can then be recast using the market-access language that trade-policy negotiators favor.26 A third objection relates to the role played by tari¤ revenue in the theory’s account of the terms-of-trade motives of governments, and the apparent unimportance of tari¤ revenue to real-world governments. There are two main responses to this objection. First, as we describe in the next section, the cost- shifting motives at the center of the terms-of-trade theory do not hinge on government pursuit of tari¤ revenue.27 Second, it is not clear that tari¤ revenue should be seen as unimportant to real-world governments. For example, as Kim (2013) points out, the US collected $31 billion in tari¤ revenue for FY 2012, an amount comparable to what the US spent in FY 2012 on foreign aid ($23 billion) and foreign military assistance ($14 billion) combined. Behind-the-border measures We now brie‡y discuss an extension of the benchmark model in which governments also have behind-the-border policy instruments at their disposal, and consider whether the purpose of a trade agreement is changed in this richer policy setting. Our extended model broadly mirrors that of Bagwell and Staiger (2001b). Speci…cally, we introduce into the trade model of the previous section a home standard, , and a foreign standard, . To …x ideas we interpret these standards as a labor regulation (e.g., maximum s production possibilities: legal work-hours per week) in each country that impacts that country’ 26 Bagwell and Staiger (2002) provide a formal de…nition of market accees and further develop the relationship between the terms-of-trade theory and the language of market access. 27 See also the discussion of local-content requirements in Section 4.5. 19 for a given local price in a country, we assume that changes in its standard shifts its production of x and y , and hence its import demand and export supply functions and therefore the market clearing world price. Proceeding to derive the market clearing world price as before, we therefore ew ( ; ew = p now have p ; ; ew to ). In addition to the assumed responses of p and contained in (4) above, we now assume that each country can also improve its terms of trade with an increase ew =@ ew =@ < 0 and @ p in its standard (i.e., we assume @ p > 0). We assume that each government cares directly about its standard but does not care directly about the standard imposed in the other country. Hence, the choice of standards can impose inter- ew ) but by assumption does not impose non-pecuniary national pecuniary externalities (through p externalities at the international level. With this we can now express the government welfare func- ew ) and W ( tions in this extended model by W ( ; p; p ew ). We continue to impose the struc- ;p ;p ture in (5) on these extended government welfare functions, and we continue to leave unrestricted how governments care about changes in their own local prices; and we also leave unrestricted how governments care about changes in their own standards. There are now four …rst-order conditions that de…ne the Nash tari¤ and standards choices: ew ) dW ( ; p; p dp @pew = Wp + Wp ew = 0, (9) d d @ ew ) dW ( ; p; p ew @p = W + [ W p + Wp ew ] = 0, d @ dW ( ; p ; pew ) dp @p ew = Wp + Wpew = 0, and d d @ dW ( ; p ; pew ) 1 @pew = W + [ Wp + Wp ew ] = 0: d @ In addition to the two conditions de…ning the home and foreign best-response tari¤s, which are unchanged from our earlier discussion and which therefore have the same interpretation, there is an additional condition for each government in (9) that de…nes its best-response standard. Here each government weighs the direct impact on its welfare of its standards choice against the impact of this choice on its welfare that runs through the induced local- and world-price e¤ects. Turning to the e¢ cient policies in this extended setting, there are now three tangency conditions that must be satis…ed for e¢ ciency: w dp @pw e @pe Wp + Wp ew ] @ [ W p + Wp d ew @ dp @pe w = 1 @pe w , (10) Wp d + Wp ew @ [ ew ] @ Wp + Wp W Wp ew = p , and ew =@ @p ew =@ @p W p Wp = : ew =@ @p ew =@ @p The top condition in (10) is the same tangency condition as in (7) that de…ned the e¢ ciency frontier in our benchmark model. This condition can be interpreted as ensuring that the home and foreign 20 tari¤s are set so that the volume of trade between the two countries is at an e¢ cient level.28 The s mix of border remaining two conditions can then be interpreted as ensuring that each country’ and behind-the-border policies delivers in an e¢ cient way this e¢ cient level of trade volume. Finally, we may again de…ne politically optimal policies and make use of these hypothetical constructs to investigate the purpose of a trade agreement in this extended setting. We therefore consider a hypothetical home government for which Wp ew 0 and a hypothetical foreign government for which Wp ew 0. If these hypothetical governments were to select their tari¤s and standards non-cooperatively, then according to (9) their tari¤ and standards choices would satisfy ew ), and ew ) = 0 = W ( ; p; p Wp ( ; p; p (11) Wp ( ew ) = 0 = W ( ;p ;p ew ): ;p ;p As before, if these politically optimal policies are e¢ cient, where as in (10) e¢ ciency is evaluated with reference to actual home and foreign government preferences, then we may conclude that the terms-of-trade externality remains the sole rationale for a trade agreement in this extended model. But it is immediate that (10) is satis…ed when evaluated at tari¤s and standards that satisfy (11), and hence politically optimal tari¤s and standards are indeed e¢ cient. In fact, as is suggested by this result, regardless of the extent of behind-the-border measures that governments may have at their disposal, according to the terms-of-trade theory the purpose of a trade agreement remains the same. Governments use trade agreements to undo the policy ine¢ ciencies that arise with unilateral policy choices when those choices can shift the costs of intervention onto trading partners through movements in foreign exporter prices (terms of trade). We may now emphasize a further insight o¤ered by the terms-of-trade theory that follows from its stance on the purpose of a trade agreement. Speci…cally, as the only “problem” for a trade agreement to “solve” is terms-of-trade manipulation, and as the tari¤ is the …rst-best policy for terms-of-trade manipulation, the terms-of-trade theory implies that the tari¤ is the only policy that is distorted in the Nash equilibrium: behind-the-border measures are set e¢ ciently under Nash choices. This insight is lurking in the conditions presented above, and it can be con…rmed by noting that the top two conditions of (9) imply the middle condition of (10) while the bottom two conditions of (9) imply the bottom condition of (10). Hence, the only e¢ ciency condition that is not met in the Nash equilibrium according to the terms-of-trade theory is the top condition of (10) that determines the e¢ cient level of tari¤s (which are lower than their Nash levels) and the e¢ cient level of trade volume (which is higher than the Nash level). In this sense, the terms-of-trade theory provides a foundation for trade agreements that adopt a shallow approach to integration.29 Our formal discussion here has focused on a particular form of behind-the-border measure that is best thought of as a production standard, such as a workplace regulation or a regulation on the use 28 See Bagwell and Staiger (2001b) for elaboration on this interpretation and the remaining interpretations of the e¢ ciency conditions that we describe in the text. 29 More speci…cally and as we describe in more detail below, the terms-of-trade theory supports an approach to behind-the-border measures that revolves around reliance on a “market access preservation” rule. 21 of an open access resource in the production process. So it is important to note that the points we have emphasized apply more broadly to behind-the-border measures of various kinds, including tax and subsidy policies and various forms of standards beyond production standards.30 Of particular relevance to the world trading system are product standards, such as minimum burn-through rates for doors or prohibitions on lead additives to paint, that can raise the costs of supplying a market but, unlike import tari¤s, do not raise revenue. In light of the prominent role played by tari¤ revenue in our account of the terms-of-trade motives of governments described above, it might be thought that those motives do not apply to such standards. But our discussion applies equally well to these kinds of behind-the-border measures, once it is understood that international cost-shifting occurs when such product standards are imposed as long as foreign exporters do not pass the full cost of meeting the product standards on to consumers in the country where the standard applies.31 3.2 Evidence We now survey the evidence related to the central tenets of the terms-of-trade theory. We focus on three basic questions. First, how signi…cant and widespread is the market power that countries possess in world markets? Second, do the unilateral tari¤ choices of countries re‡ect the market power that they possess? And third, does the pattern of negotiated tari¤ liberalization that we observe correlate with the pattern of observed market power in the way that the theory suggests it should? Answers to these questions seem central to all of the theory’s predictions, and so we focus on them here, postponing until later points in our survey a discussion of the empirical work relating to various other predictions of the terms-of-trade theory. Market power Do countries routinely possess a degree of market power on world markets that would allow them to manipulate their terms of trade? To answer this …rst question, we begin by observing that there is a large body of indirect evidence that suggests an a¢ rmative answer. We are referring here to the literature on exchange rate pass-through. To establish the connection between the exchange rate pass-through literature and an answer to our …rst question, we proceed in two steps. First, we note that Feenstra (1989) shows theoretically that the pass-through to domestic prices associated with exchange rate shocks can be thought of as comparable in magnitude to the pass-through associated with tari¤ changes, and he o¤ers econo- metric evidence supporting this hypothesis of symmetric pass-through between tari¤s and exchange rates in the data. Second, we note that, while we have adopted above a general equilibrium set- ting to present the terms-of-trade theory, the theory can also be developed in a partial-equilibrium model, where cost shifting then occurs through changes in the terms of trade provided that foreign exporters bear some of the incidence of the import tari¤ so that it is not fully passed through to domestic prices.32 That is, incomplete pass-through of the tari¤ to domestic prices is synonymous 30 For an extension of these points to the case of domestic tax/subsidy policies, see Bagwell and Staiger (2006). 31 See Staiger and Sykes (2011) for a treatment of product standards in this context. 32 See Bagwell and Staiger (2001a) for a development of the terms-of-trade theory of trade agreements in the context of a partial equilibrium model. 22 with cost-shifting and the exercise of importer market power. Hence, by the second step, the terms- of-trade e¤ects of a tari¤ arise whenever the incidence of the tari¤ is not fully passed through to domestic prices; and by the …rst step, we may look to the vast empirical literature on exchange rate pass-through for indirect evidence about the degree of tari¤ pass-through. What does the exchange rate pass-through literature imply for the answer to our …rst question? First, in light of the conclusion from Goldberg and Knetter’s (1997) survey of this literature that pass-through rates average about 60 percent, it would appear that the existence of countries that possess substantial market power on world markets is routine. Second, after surveying this literature in his Handbook chapter, Feenstra (1995, p. 1569) concludes that, while there is strong evidence that pass-through is less than complete, the magnitude of pass-through di¤ers substantially across industries and therefore “...we should not have any presumption about the extent of terms of trade gain due to tari¤s, but must treat each industry on a case-by-case basis.” Turning to the direct evidence on this …rst question, in a provocative paper Magee and Magee (2008) construct measures of world market concentration and trade elasticities to argue that even a “large” country like the US has little market power to exert on world markets, suggesting that market power can safely be ignored when considering the e¤ects of tari¤s. But for the most part, the literature has produced results which are consistent with the conclusions from the indirect evidence emphasized above. For example, the studies of Kreinin (1961), Winters and Chang (2000), Chang and Winters (2002), Anderson and Van Wincoop (2002), and Bown and Crowley (2006), among others, all o¤er evidence that unilateral tari¤ changes can signi…cantly a¤ect a country’s terms of trade. On the other hand, in their study of New Zealand’s unilateral trade liberalization of the s, Winkelmann and Winkelmann (1998) …nd only weak evidence of incomplete pass-through 1980’ of New Zealand tari¤s to prices in the New Zealand economy, and only in a few industries, and conclude that New Zealand truly is an example of a small country in most products. Finally, using high-frequency data on the price of raw sugar in New York City from 1890 to 1930, Irwin (2015) …nds that about 60 percent of any US tari¤ increase was borne by foreign exporters with only 40 percent passed through to domestic consumer prices.33 These studies seem to con…rm the basic conclusion of Feenstra (1995), that there is abundant evidence of substantial market power but its presence varies greatly from industry to industry and country to country. This conclusion is further reinforced by the most comprehensive study to date on the issue, that of Broda, Limão and Weinstein (2008). Focusing on the unilateral tari¤ choices of 15 countries prior to their membership in the GATT/WTO, Broda et al. estimate the foreign export supply elasticities faced by each of these countries, which as they note provides an inverse measure of the market power that each was able to exert on the foreign export (world) prices. With these estimates, they con…rm that most countries, even apparently “small” countries, have signi…cant ability to alter their terms of trade on many imported products with their tari¤ choices. 33 Irwin (2015) also …nds a striking asymmetry between pass-through rates for tari¤ increases and tari¤ reductions, with the latter passed through fully to consumers. He attributes this asymmetry to asymmetric demand responses. 23 Unilateral tari¤s We next turn to the second question raised above: Do the unilateral tari¤ choices of countries re‡ect the market power that they possess? An early paper that provides evidence on this question is Olarreaga, Soloaga and Winters (1999). Examining the determinants of the common external tari¤ adopted by MERCOSUR in 1996 (a period when the common external tari¤ of MERCOSUR was essentially unconstrained by WTO commitments), they conclude that terms-of-trade e¤ects account for a substantial part (between 6 and 28 percent) of the explained variation in the structure of MERCOSUR tari¤s, despite the fact that during the period of their analysis MERCOSUR had only a 1 percent market share of world trade. This question is also addressed by the Broda, Limão and Weinstein (2008) paper discussed above. Speci…cally, after estimating the foreign export supply elasticities faced by each of the 15 non-GATT/WTO member countries in their sample, Broda et al relate this measure of the power to a¤ect world prices to the unilateral (i.e., prior to WTO accession) tari¤ choices that each country made. They …nd that, prior to joining the WTO, these countries set tari¤s an average of 9 percentage points higher on imports for which they could exert large e¤ects on world prices as compared to the tari¤s they set on imports where their ability to a¤ect world prices was limited, an impact whose magnitude is roughly comparable to the size of the average tari¤s in these countries. They also …nd that this terms-of-trade motive explains more of the cross-industry variation in tari¤s than is explained by commonly used political-economy variables. Hence, according to the Olarreaga et al. (1999) and Broda et al. (2008) …ndings, governments who set their trade policies unilaterally and noncooperatively respond to terms-of-trade motives and the market power that they possess strongly and in the way that the theory predicts. These …ndings are reinforced by the recent paper of Dhingra (forthcoming). Dhingra shows that the median-voter model of non-cooperative tari¤ determination is strongly rejected by the data in its traditional small-country formulation, but that its central predictions receive strong cross-country empirical support once it is cast in a large-country setting and the existence of market power and terms-of-trade motives are taken into account when predicting the cross-country pattern of tari¤s. Negotiated tari¤s Finally, we consider the third of our three questions: Does the pattern of negotiated tari¤ liberalization that we observe correlate with the pattern of observed market power in the way that the theory suggests it should? As with the literature that takes up this question, we focus below on the pattern of negotiated tari¤ liberalization in the GATT/WTO rather than PTAs. We consider the pattern of liberalization that has taken place within PTAs and its compatibility with the terms-of-trade theory in the context of later discussions. In light of the literature’s …nding surveyed just above that market power is a strong determinant of unilateral tari¤ choices, one way to pose this third question is to ask whether the same can be said for negotiated tari¤s. If market power is an equally strong determinant of negotiated tari¤s, then this would indicate a lack of support for the view that trade agreements serve to limit the exercise of market power by their member governments; on the other hand, if negotiated tari¤s 24 are unrelated to market power, then together with the …ndings reported in the literature that unilateral tari¤s are strongly related to market power, this would suggest that trade agreements do indeed limit the exercise of market power. This is the approach that Broda et al. (2008) take. Focusing on the US, they …nd that US non-tari¤ barriers and so-called “statutory” tari¤ rates – neither of which has been the subject of negotiations within the GATT/WTO – are signi…cantly and positively related to the degree of market power which the US exerts on the world prices of its import products, while the US MFN tari¤s –which have been subjected to the many rounds of GATT/WTO negotiations –exhibit no such relationship. A related approach is taken by Nicita, Olarreaga and Silva (2014). They focus on the nature of the tari¤ commitments made by WTO member countries –commitments that as we have described above take the form of bindings de…ning the maximum allowable level for the tari¤ –and exploit the fact that countries di¤er in the degree to which their negotiated WTO tari¤ commitments constrain their applied tari¤s (i.e., the tari¤ levels that they actually set). Nicita et al. observe that tari¤s that are unconstrained by WTO bindings should exhibit a positive correlation with market power, for the reasons associated with unilateral tari¤s that we describe above. But Nicita et al. also derive a new prediction: they argue that tari¤s constrained by WTO bindings (the “cooperative tari¤s”) should exhibit a negative correlation with market power. This prediction follows under their assumptions that (a) exporters enjoy extra political economy weight in the objectives of their governments and (b) their governments lack trade instruments of their own (e.g., export subsidies) to shift surplus to these exporters. Under these assumptions, the only way for a government to help its exporters is to negotiate a tari¤ cut in the foreign markets served by its exporters, and the political payo¤ to the government from negotiating such tari¤ cuts will be higher the greater is the importer market power in the foreign market where the negotiating e¤orts are focused (and hence the greater the exporter-price e¤ect of the tari¤ cuts). On this basis, they predict that the cooperative tari¤ levels will be negatively correlated with importer market power. Examining the tari¤s of 101 WTO members, Nicita et al. …nd that the sign of the correlation between tari¤ levels and market power indeed switches from positive to negative as the WTO tari¤ bindings vary from levels that are well above applied tari¤ rates to levels that are at the applied tari¤ rates.34 Beshkar, Bond and Rho (forthcoming) derive a related prediction in an extension of the basic terms-of-trade model that emphasizes the trade-o¤ between commitments and ‡exibility. (We discuss their paper further in Section 4.2.) Within this extended model, they show that applied tari¤s are more likely to be set at their WTO-bound levels at any point in time the greater is the market power for that product possessed by the importing government; and with a su¢ ciently high level of import market power, applied tari¤s are always set at the level of the negotiated binding. Beshkar et al. …nd strong evidence for these predicted relationships between negotiated tari¤ commitments and importer market power with data on the tari¤s of 109 WTO member countries. 34 Nicita, Olarreaga and Silva (2014) also report that the expected positive correlation between tari¤s and market power emerges in their data only once the WTO tari¤ commitments are su¢ ciently above the applied tari¤ levels. They view this as something of a puzzle, and suggest that it may re‡ ect some form of implicit cooperation among WTO member countries. 25 A di¤erent approach to this question is taken by Bagwell and Staiger (2011). They use the terms-of-trade theory to derive an expression for the component of the noncooperative tari¤ that embodies the international cost-shifting motive. They then use this expression to derive the pattern of negotiated tari¤ cuts that is implied by the terms-of-trade theory; intuitively, if the cost-shifting component is big, then the negotiated tari¤ cut implied by the terms-of-trade theory should be large. Working from this basic insight, Bagwell and Staiger show that the tari¤ cuts predicted by the terms-of-trade theory can be expressed as a function of pre-negotiation import volumes and prices, and measures of the power to a¤ect world prices. The predicted relationship is then confronted with data from the accession negotiations of 16 countries that joined the WTO subsequent to its creation in 1995, and strong and robust support for the predictions of the terms-of-trade theory are found in the observed pattern of negotiated tari¤ concessions. Ludema and Mayda (2013) advance the literature on this question by allowing that free-rider issues associated with the MFN principle (which we discuss at length further below) might im- pede governments from fully addressing terms-of-trade manipulation in their GATT/WTO tari¤ negotiations. Augmenting the basic terms-of-trade model with a model of endogenous bargaining participation, they show that negotiated tari¤s should be negatively related to the product of the importer’s market power and exporter concentration. Intuitively, the lower is exporter concentra- tion, the more severe will be the free-rider problem associated with MFN, and the less e¤ective negotiations will be in ridding the tari¤ of its cost-shifting component; and the larger the market power, the larger this cost-shifting component will be. Ludema and Mayda take this prediction to the tari¤ data for 30 WTO-member countries and …nd strong support, concluding (p. 1837) that “...the internalization of terms of trade e¤ects through WTO negotiations has lowered the average tari¤ of these countries by 22% to 27% compared to its noncooperative level.” Yet a di¤erent perspective on answering this question is provided by Bown and Crowley (2013a), who investigate empirically some of the predictions of the terms-of-trade theory when that theory is developed in a repeated-tari¤-game setting subject to stochastic trade volume shocks and where self- enforcement constraints are binding.35 A basic prediction of the terms-of-trade theory in this setting is that, where the self-enforcement constraint is binding, cooperative tari¤s must rise somewhat in the face of import surges in order to mitigate the increased temptation implied by the higher import volume to shift costs onto foreign exporters with an even higher tari¤. Among the additional predictions of the model highlighted by Bown and Crowley is that, for a given import surge, a cooperative tari¤ increase is more likely on a product if the market power possessed by the importing government on that product is higher. Utilizing data on the time-varying protective measures (anti-dumping and safeguard actions) of the US over the period 1997-2006, the authors con…rm these and other predictions of the terms-of-trade theory in this setting. Summarizing, the empirical papers surveyed above provide a growing body of evidence consis- tent with the central tenets of the terms-of-trade theory of trade agreements. While we discuss 35 Bagwell and Staiger (1990) develop this model, which we describe further along with further discussion of the …ndings of Bown and Crowley (2013a) in Section 4.2. 26 below evidence that lends some support as well to the other strands of the trade agreement lit- erature, our review here suggests that, at a minimum, the terms-of-trade theory is central for understanding actual trade agreements, and provides a reasonable basis from which to seek initial s motivating questions. answers to our survey’ 3.3 Has Globalization under the WTO Gone Far Enough? With the broad features of the terms-of-trade theory of trade agreements described, we now pose a …rst and basic question. Viewed from the lens of this theory, might it be possible to conclude that globalization under the GATT/WTO has already allowed its member governments to reach the international e¢ ciency frontier? This is a question that we must confront in the ensuing pages in the process of answering the motivating questions of our survey. But we pause here to dispel one natural misconception, namely, that an immediately plausible answer to this question might be “No, as long as trade is not truly free.” The terms-of-trade theory requires a more nuanced answer because, according to this theory, the purpose of a trade agreement is not to secure free trade, but to remove the ine¢ cient cost-shifting component from the unilateral tari¤ choices of its member governments. As we have observed, this outcome could be compatible with free trade, but only under certain strong assumptions about the preferences of governments, assumptions which seem unlikely to be met in the real world. Accordingly, the continued existence of tari¤s is not by itself evidence of further work to be done. To know whether globalization under the WTO has gone far enough requires, according to the terms-of-trade theory, a more nuanced assessment of the theoretical and empirical issues addressed by the literature we survey below. Much as the continued existence of tari¤s cannot by itself be interpreted as a sign of lin- gering ine¢ ciencies in the GATT/WTO system, neither can one interpret the lack of negotiated GATT/WTO commitments on behind-the-border measures as a sure sign of GATT/WTO failings according to the terms-of-trade theory. Indeed, as is suggested by our discussion of behind-the- border measures above and as we highlight further below, the GATT/WTO shallow-integration approach is compatible with e¢ ciency in the presence of certain kinds of accompanying rules, rules that we will suggest at a broad level …nd representation in the GATT/WTO. Again with regard to behind-the-border measures, to know whether globalization under the WTO has gone far enough requires, according to the terms-of-trade theory, a more nuanced assessment. Finally, we emphasize two important points that are implicit in our discussion here and should be kept in mind as we proceed through the survey. First, for the most part the terms-of-trade theory of trade agreements treats government preferences as …xed and sovereign, much as con- sumer preferences are taken as sovereign in standard consumer theory, and seeks to understand trade agreements as agreements that lead to Pareto improvements for the member governments when gauged from the perspective of their own preferences. This is clearly not the only possible assumption, and below we also discuss some papers in this strand of the literature that entertain alternative assumptions. But the assumption does resonate well with the “member driven”nature of real-world trade agreements, especially the GATT/WTO. As we later discuss, the commitment 27 theory strand of the trade agreements literature can be interpreted as adopting a major departure from this assumption, and so we consider these alternatives in more depth when we survey that strand of the literature. A second point is that the terms-of-trade theory (and the other strands of the trade agreements literature as well) takes the policy instruments that a government has at its disposal as …xed. Hence, while it is well-known (Bhagwati and Ramaswami, 1963) that a tari¤ is a second-best instrument for virtually every domestic policy goal that could be imagined, the trade agreements literature starts from the view that there are often constraints (usually unmodeled) that prevent governments from using the …rst-best instruments for their policy goals and lead them to use tari¤s to achieve these goals instead. And given that these governments are constrained to use tari¤s for, e.g., distributional reasons, the trade agreements literature then explores the role of trade agreements in eliminating ine¢ ciencies from their tari¤ choices.36 This feature is important to keep in mind, as without it the notion that the use of tari¤s could ever be compatible with a position on the e¢ ciency frontier (beyond tari¤s that reside on the Mayer locus) would seem strange. 4 Evaluating the GATT/WTO approach to trade liberalization We now turn to an evaluation of the GATT/WTO approach to trade liberalization from the per- spective of the terms-of-trade theory of trade agreements literature. This strand of the literature seeks to evaluate the design and performance of the GATT/WTO as an institution that could plausibly help governments in their attempts to solve the terms-of-trade externality problem that we described in Section 3. We use our survey of this literature to ask: Is the GATT/WTO well- designed at a fundamental level to help governments address the terms-of-trade problem? And can the successes and failures of the GATT/WTO record be broadly understood to re‡ect the strengths and weaknesses of these design features when interpreted within the context of the terms-of-trade problem? If the answers to these questions are a¢ rmative, then we tentatively conclude that the GATT/WTO is an institution worthy of the continued support of economists. This is not to say that PTAs might not also be deserving of support - we take up that question in the next section - but rather to say that the GATT/WTO appears fundamentally well-designed and thus worthy of support when viewed from the perspective of the terms-of-trade theory.37 We organize our discussion into six parts. Section 4.1 considers the GATT pillars of reciprocity and the most-favored nation (MFN) clause, and the general self-enforcing nature of GATT/WTO commitments. Sections 4.2 through 4.5 survey the literature on more speci…c features of the GATT/WTO approach that are central to identifying current challenges. Throughout we highlight 36 We discuss some partial exceptions in the literature at later points in our survey, such as Limão and Tovar (2011), but even there the availability of …rst-best instruments are ruled out by assumption. 37 Our discussion here presupposes that institutions matter for solving problems of international trade policy co- operation. As Maggi (2014) emphasizes, this assumption indicates the existence of underlying frictions, even though these frictions are not often explicitly modeled. That instutitions matter in this context seems clear from the histori- cal experience with international trade policy cooperation, and the many failed attempts at international cooperation that preceded the creation of GATT. See Bagwell and Staiger (2010a) and the literature cited therein for further discussion of the historical antecedents of GATT. 28 a number of signi…cant challenges facing the GATT/WTO approach. Section 4.6 summarizes by o¤ering an initial terms-of-trade-theory-based perspective on how the GATT/WTO has worked, its current woes and possible …xes. 4.1 Reciprocity, MFN and enforcement In this section, we review research based on the terms-of-trade approach that examines the three pillars of the GATT/WTO architecture: reciprocity, MFN and enforcement. Reciprocity Governments reach agreements to lower tari¤s through GATT/WTO negotiation rounds. It is often observed that these negotiations re‡ect a norm of reciprocity and entail a “balance of concessions,” whereby each government makes the “concession” of lowering its import tari¤s in exchange for receiving the bene…t of a similar concession from a trading partner. Such a norm would be di¢ cult to interpret in a model with small countries. The reciprocity norm, however, admits a straightforward interpretation when countries are large. To develop this point, we return to the terms-of-trade model described in Section 3. Let us suppose that the two governments start at the Nash tari¤s. We know from our earlier discussion that these tari¤s are ine¢ ciently high, due to the terms-of-trade externality. Since each government selects its best-response tari¤ at a Nash equilibrium, no government can gain from a unilateral tari¤ cut. This simple observation provides an immediate interpretation for why a government regards its own tari¤ cut as a concession. Governments may enjoy mutual gains, however, if they jointly reduce tari¤s. Indeed, as Bagwell and Staiger (1999a, 2002) show, when governments start at the Nash tari¤s, ( N; N ), as de…ned by (6), they can mutually gain from moving to a new pair of tari¤s ( ; ) only if they both o¤er strictly lower tari¤s, so that < N and < N. A general form of reciprocity is thus necessary for mutual gains. We may also interpret reciprocity at a more speci…c level. Following Bagwell and Staiger (1999a, 2002), let us associate the principle of reciprocity with mutual changes in trade policy that bring about changes in the volume of each country’s imports that are of equal value to changes in the volume of its exports. Formally, for given initial and subsequent tari¤ pairs, ( 0 ; 0) and ( 1 ; 1 ), respectively, a set of tari¤ changes 1 0 and 1 0 satis…es the principle of reciprocity if ew1 ) ew0 [M (p1 ; p p ew0 )] = [E (p1 ; p M (p0 ; p ew1 ) ew0 )]; E (p0 ; p (12) ew0 where p ew ( 0 ; p ew1 0 ); p ew ( 1 ; p 1 ); p0 ew0 ); p1 = p( 1 ; p = p( 0 ; p ew1 ) and trade-volume ew0 . In this two-good model, as Bagwell and Staiger changes are valued at the initial world price, p establish, a set of tari¤ changes satis…es the principle of reciprocity if and only if the world price is ew1 : This …nding may be easily con…rmed by applying (1) to (12), where ew0 = p unchanged so that p (1) is applied at both the initial and subsequent tari¤ pairs. Is liberalization according to the principle of reciprocity su¢ cient for mutual gains? To address this question, we start at the Nash tari¤s as characterized in (6) and consider for simplicity the preferences of the home government. Given (4) and (5), it is straightforward to see that Wp < 0 29 when tari¤s are at their Nash levels. Thus, at Nash tari¤s, the home government would prefer more trade (a lower relative price of imports) if it could achieve a greater trade volume without s terms of trade. A unilateral tari¤ cut delivers greater trade inducing a loss in the home country’ volume for the home country, but it does so at the cost of a deterioration in the home country’s terms of trade. By contrast, mutual changes in trade policy that satisfy the principle of reciprocity enable the home country (and similarly the foreign country) to enjoy greater trade volume without su¤ering a terms-of-trade loss. Thus, and as Bagwell and Staiger (1999a, 2002) show, starting at the Nash equilibrium, trade liberalization that satis…es the principle of reciprocity is sure to raise each government’s welfare, at least initially.38 This …nding o¤ers an interpretation for the reciprocity norm that guides trade-liberalization negotiations in GATT/WTO. While more evidence is needed, recent empirical work o¤ers some support for the important role played by reciprocity in GATT/WTO tari¤ negotiations. In an early e¤ort, Shirono (2004) considers the Uruguay Round and …nds that the negotiated tari¤ reductions conform well with the reciprocity norm. Limão (2006, 2007) considers US tari¤ cuts in the Uruguay Round and also reports evidence consistent with the application of reciprocity. After constructing a measure of market-access concessions and identifying instruments that address possible endogeneity concerns, he presents evidence of reciprocity for US products that were not subject to non-tari¤ barriers. Among such goods, US import tari¤ reductions embody a general form of reciprocity, being greater for goods exported from countries whose own tari¤ cuts provided greater market access to US exports. Karacaovali and Limão (2008) provide similar support for reciprocity in a related exercise for the EU. Finally, we also note that evidence of reciprocity may be stronger in some product groups than in others. Gulotty (2014) focuses on tari¤ liberalization by the US in the Uruguay Round and reports evidence that sectors with high contract-intensive products, characteristic of global production, do not exhibit reciprocity. The principle of reciprocity is a key pillar of the GATT/WTO approach to trade liberalization. Our discussion to this point considers the principle of reciprocity as a bargaining norm that char- acterizes the manner in which tari¤s are reduced in GATT/WTO negotiations. The principle of reciprocity, however, also explicitly arises in GATT/WTO rules that govern the manner in which tari¤s may be raised as part of renegotiation or dispute resolution procedures. Most-Favored Nation (MFN) treatment A second pillar of the GATT/WTO architecture is the principle of non-discrimination. For member countries, this principle requires that a country apply the same tari¤ on a given import good, regardless of which country exports the good. This principle is embodied in the most-favored nation (MFN) rule, under which no exporter of a given good can be treated any less favorably than any other exporter. To interpret and evaluate the principle of non-discrimination, we extend the model in a simple 38 In a symmetric setting, if governments were to liberalize according to the principle of reciprocity and move thereby from the Nash equilibrium to the politically optimal tari¤s, then they would enjoy mutual gains all along this path. In an asymmetric setting, if governments were to liberalize according to the principle of reciprocity, then mutual bene…ts may disappear before the e¢ ciency frontier is reached. In both settings, mutual gains are enjoyed provided that the liberalization does not go too far. 30 fashion to allow for three countries. As before, the home country imports good x and exports good y . We now assume, however, that there are two foreign countries. Foreign countries 1 and 2 each trade with the home country, importing good y and exporting good x, and we assume for simplicity that the two foreign countries do not trade with each other. The local price in the home country is again represented as p px =py , while the local price in foreign country i, where i = 1; 2, is given by p i i i px =py . The world price for trade between the home country and foreign country i is denoted as pwi pxi =py and represents the terms of trade for foreign country i: Let ti represent the home-country ad valorem tari¤ applied to imports from foreign country i and let t i denote the ad valorem tari¤ of foreign country i applied to imports from the home country. De…ning i 1 + ti and i 1+ t i , we then have from arbitrage conditions that p i = pwi = i and p = 1 pw 1 = 2 pw 2 . From here, we see that if the home country adopts discriminatory tari¤s, de…ned by 1 6= 2, then di¤erent world prices, pw 1 6= pw 2 , obtain for its two foreign trading partners. The foreign country that receives the lower import tari¤ enjoys a better terms of trade. If instead the home country satis…es the principle of non-discrimination (i.e., the MFN rule), de…ned by 1 = 2 , then pw 1 = pw 2 pw follows, and so the two foreign countries enjoy the same terms of trade, pw . Under the MFN rule, the home country’s terms of trade are given as 1=pw . Our next steps are to determine market-clearing prices and represent each government’s welfare function. We begin with the simplest case, where the home country’s tari¤s satisfy the MFN rule. ew ( ; In that case, as noted, a single world price obtains. The market-clearing world price, p 1; 2 ); may then be determined similarly to how we do so above for the two-country model. With the ew and p i ( ew ) = p market-clearing local prices then given as p( ; p ew ) i; p ew = =p i; we may rep- ew ) and W 2 (p 2 ; p ew ); W 1 (p 1 ; p resent the respective government welfare functions as W (p; p ew ). Following Bagwell and Staiger (1999a, 2002), we may now again impose two key assumptions: each ew ( ; country can improve its terms of trade by raising its import tari¤ (i.e., @ p 1; 2 )=@ i >0> ew ( @p ; 1; 2 )=@ ); and each government prefers an improvement in its terms of trade, holding …xed its local price (i.e., Wp ew ) < 0 < Wp ew (p; p i i ew )). ew (p ; p Analogous steps apply as well for the case in which the home country utilizes discriminatory tari¤s, although in that case the home country’s terms of trade is more complex to de…ne. As Bagwell and Staiger (1999a, 2002) discuss, when the home country’s tari¤s are discriminatory, the home country’s multilateral terms of trade is de…ned as a weighted average of its bilateral terms of trade, 1=pw1 and 1=pw2 , where the weights are export shares and thus depend on foreign local prices. It follows that international externalities associated with foreign import tari¤s then travel to home’s multilateral terms of trade via foreign local-price channels as well as bilateral terms-of- trade channels. Intuitively, when the home government employs discriminatory tari¤s, it cares not just about the bilateral terms of trade but also about the share of exports that comes from each partner, as it enjoys greater tari¤ revenue when a higher share of a given import volume comes from the partner on which the import tari¤ is highest. The formal counterpart to this intuition is that the home country enjoys an improved multilateral terms of trade when a greater share of its imports comes from the partner on which it places the highest import tari¤. 31 With the three-country model described, we now consider the value of the principle of non- discrimination.39 A simple observation is already at hand: when the MFN rule is used, international externalities associated with tari¤ choices are channeled through the world price alone, just as in the two-country model. A key implication is that the main …ndings presented above for the two-country model carry over as well to the multi-country model when the MFN rule is imposed. Speci…cally, as Bagwell and Staiger (1999a, 2002) show, if home-country tari¤s are non-discriminatory, then Nash tari¤s are ine¢ cient, politically optimal tari¤s are e¢ cient, and a small liberalization between the home country and a foreign trading partner that starts at the Nash tari¤s and that satis…es the principle of reciprocity is sure to generate mutual gains for the two negotiating governments.40 The latter …nding suggests a broad complementarity between the principles of reciprocity and non- discrimination: the latter principle ensures that international externalities travel through the world price, which is a channel that the former principle is well-equipped to manage. The multi-country model also allows consideration of sequential negotiations between trading partners. The possibility of sequential negotiations is easily motivated in the trade-agreement context. First, within a given negotiation round, some bilateral negotiations may precede others, suggesting that strategic considerations associated with sequential negotiations may come into play. Second, liberalization e¤orts in the GATT/WTO system have occurred over more than 65 years in the context of eight di¤erent negotiation rounds, and so negotiations among a given set of countries naturally occur through time across di¤erent negotiation rounds. Finally, accession dates di¤er across GATT/WTO members, so that some countries participating in a given negotiation round may not have participated in earlier rounds. To capture new insights associated with the sequential nature of negotiations, we assume that the home government initially negotiates with the government of foreign country 1 and subsequently negotiates with the government of foreign country 2. A key question is whether the principle of non-discrimination may be interpreted as having e¢ ciency-enhancing implications in this sequential context. We begin by putting this question in broader context and highlighting potential ine¢ ciencies that the MFN rule may introduce. As Caplin and Krishna (1988) emphasize, the MFN rule is a restriction on the set of instruments and thus shifts in the bargaining frontier; hence, an e¢ ciency- enhancing role for non-discrimination is available only if the bargaining process in the absence of the MFN rule delivers ine¢ cient outcomes. Furthermore, in the context of sequential negotiations, it is commonly argued that the MFN rule may lead to a “free-rider” problem, whereby a country refrains from o¤ering signi…cant concessions since it expects anyhow to enjoy MFN tari¤ cuts from trading partners undertaking their own negotiations. As Caplin and Krishna argue, the MFN free- rider concern suggests that bargaining under the MFN rule may fail to deliver e¢ cient outcomes even relative to the MFN-constrained e¢ ciency frontier.41 39 For analyses of the MFN rule in other modeling frameworks, see Bagwell and Staiger (2001b), Choi (1995), McCalman (2002) and Saggi (2004), for example. Bagwell and Staiger (2010a) and Horn and Mavroidis (2001) o¤er further discussion of research on the legal and economic aspects of the non-discrimination principle. 40 By contrast, due to the presence of local-price externalities, politically optimal tari¤s are not e¢ cient when discriminatory tari¤s are used. 41 Ludema (1991) shows, however, that an MFN-e¢ cient bargaining outcome is possible, if participating countries 32 We now consider sequential bargaining in the three-country model, where the …rst-stage bargain between the governments of the home country and foreign country 1 determines ( 1 , 1) and the second-stage bargain between the governments of the home country and foreign country 2 deter- mines ( 2 , 2 ). Our …rst step is to identify a bargaining ine¢ ciency that arises in the absence of the MFN rule. As Bagwell and Staiger (2005a) show, under a slight strengthening of the assumptions given above, any proposed e¢ cient tari¤ vector is vulnerable to bilateral opportunism : by appro- priately lowering the second-stage tari¤s that they apply to one another, the governments of the home country and foreign country 2 can enjoy mutual welfare gains that come about at the expense of the government of foreign country 1. Intuitively, such second-stage tari¤ reductions lead to a terms-of-trade loss for foreign country 1; and for two reasons. First, the discriminatory tari¤ cut that the home government o¤ers stimulates export supply in foreign country 2 and thus generates downward pressure on the world price of foreign country 1’s export good. Second, the tari¤ cut o¤ered by the government of foreign country 2 induces greater demand for the home export good and thus generates upward pressure on the world price of foreign country 1’s import good. In the absence of the MFN rule, bilateral opportunism in the second-stage bargain thus ensures an ine¢ cient outcome. The prospect of second-stage bilateral opportunism may feed back and limit the scope of the …rst-stage bargain as well. If the government of foreign country 1 were to foresee that the value of any …rst-stage concession obtained from the home government might be eroded by an even greater concession extended to foreign country 2, then the government of foreign country 1 might be cautious in extending its own …rst-stage concession. In light of the inevitable bargaining ine¢ ciencies that arise in the absence of the MFN rule, we move now to the second step of our discussion and examine whether the MFN rule addresses the bilateral opportunism problem and thereby promotes e¢ ciency. As Schwartz and Sykes (1997) argue, it is natural to expect that the MFN rule could be helpful in this regard. After all, the concession-erosion concern is addressed under the MFN rule, since foreign country 1 is then assured of receiving any home tari¤ cut that is subsequently o¤ered to foreign country 2. In terms of our discussion in the preceding paragraph, the MFN rule addresses the …rst reason for the terms-of-trade loss of foreign country 1. Unfortunately, however, the MFN rule does not address the second reason. Even if the home tari¤ satis…es the MFN rule, foreign country 1 may su¤er a terms-of-trade loss due to the tari¤ cut extended by foreign country 2. Building on this point, Bagwell and Staiger (2005a) show that a subset of the tari¤s that are e¢ cient within the MFN class are also vulnerable to bilateral opportunism, even when the MFN rule is imposed. The principle of non-discrimination is thus helpful but not completely e¤ective in eliminating the scope for bilateral opportunism. Is there a simple rule which, in combination with the MFN rule, fully eliminates the scope for bilateral opportunism? In fact, as Bagwell and Staiger (2005a) show, bilateral opportunism have the option to reject an agreement and continue bargaining when another country free rides and refuses to cut its own tari¤s. For bargaining within a given round, Ludema’ s …nding suggests that the threat of delayed agreement may be an important defense against free riding. His model, however, may be less well suited for bargaining that occurs over time and across rounds, as then negotiations in one round may be undertaken with a view toward endogenously a¤ecting the outside options for acceding countries in future rounds. 33 is impossible if any bilateral negotiation must satisfy both the principle of non-discrimination and the principle of reciprocity. As in the two-country model, under the principle of reciprocity, a negotiation between the governments of the home country and foreign country 2 maintains the world price between these countries. In addition, as argued above, the principle of non-discrimination ew . It follows that a bilateral implies that foreign countries 1 and 2 have a common terms of trade, p negotiation between the governments of the home country and foreign country 2 that satis…es the principles of non-discrimination and reciprocity leaves unaltered foreign country 1’s terms of trade. Since foreign country 1’s tari¤ is not altered, foreign country 1’s local price, p 1 ew = = p 1; also remains unchanged. We thus arrive at the following welfare-preservation result : if the governments of the home country and foreign country 2 engage in a bilateral negotiation that satis…es the principles of non-discrimination and reciprocity, then government welfare in foreign country 1, ew ), is preserved. This result implies as well that the MFN rule does not generate a W 1 (p 1 ; p free-riding incentive, when bilateral negotiations also abide by the principle of reciprocity.42 To see the intuition behind the welfare-preservation result, let us suppose that the governments of the home country and foreign country 2 enter into a bilateral negotiation in which they lower their respective tari¤s. Under the MFN rule, exporters from foreign country 1 then face a lower home import tari¤, which in isolation provides improved access to the home market. But the impact of foreign country 2’s tari¤ cut also must be considered. In foreign country 2, this tari¤ cut lowers the local price of the import good relative to the export good, and thus both stimulates consumption of the import good and production of the export good. For both of these reasons, foreign country 2’s export volume increases. Hence, and as the Lerner symmetry theorem would suggest, foreign country 2’s import tari¤ cut has the same e¤ect as would an export subsidy increase. Thus, while exporters in foreign country 1 enjoy a lower home import tari¤, they also face, in e¤ect, subsidized competing exporters from foreign country 2. Under the principles of reciprocity and non- discrimination, these e¤ects exactly balance out, with the increase in home import demand exactly ful…lled by the expanded export volume from foreign country 2. Exporters in foreign country 1 thus ultimately do not gain additional access to the home market, which is to say that the bilateral negotiation has no impact on foreign country 1’s terms of trade. Bagwell and Staiger (2010b) provide further analysis of sequential bargaining in the three- country model. They assume that the home government negotiates sequentially with the two foreign governments, where the home government makes take-it-or-leave-it o¤ers and negotiations are over MFN tari¤s as well as lump-sum international transfers. The assumption that such transfers are feasible is extreme but ensures tractability. The MFN rule alone is then completely ine¤ective in addressing bilateral opportunism: for any proposed MFN-e¢ cient outcome, the governments of the home country and foreign country 2 can adjust the tari¤s and transfers under their control so as to enjoy mutual gains that come at the expense of the government of foreign country 1. In addition to this “backward-stealing problem,” they identify a “forward-manipulation prob- 42 In recent work, Ossa (2014) argues that the presence of di¤erentiated products can interfere with the ability of reciprocity and MFN to neutralized third-party e¤ects. An interesting direction for future research is to assess the welfare-preservation result in other modeling frameworks. 34 lem”: the home government may keep its MFN tari¤ ine¢ ciently high in its initial negotiation, and thus engage in “foot-dragging,” in order to endogenously generate a less attractive outside option (i.e., disagreement point) for its subsequent negotiation partner. Hence, governments gen- erally would be unable to achieve an e¢ cient outcome, even if the backward-stealing problem were addressed.43 Bagwell and Staiger (2010b) argue, however, that e¢ cient outcomes may be reached when the MFN rule is combined with other bargaining rules. The forward-manipulation problem, in particular, may be addressed if the opportunities for renegotiation are so “sweeping” as to dis- connect the initial negotiation outcome from the outside option in the subsequent negotiation. At the same time, we note that other problems may arise if the renegotiation option is too readily available, as then the signi…cance of any GATT/WTO tari¤ commitment would be put in question. We turn now to the empirical evidence that concerns the non-discrimination principle. A …rst form of evidence considers the trade-volume impacts associated with GATT/WTO membership and relates these impacts to the theoretical implications developed above. Subramanian and Wei (2007) …nd that GATT/WTO membership is associated with large and signi…cant trade-volume impacts for developed countries but that the trade-volume impacts of membership are small or nonexistent for most developing countries.44 Since developed countries have been the most active participants in GATT/WTO rounds, one interpretation of this …nding draws from the welfare- preservation result presented above. In particular, if developed countries negotiate tari¤ reductions that broadly adhere to the principles of reciprocity and non-discrimination, then the trade-volume impacts on third-party countries should be limited. As Bagwell and Staiger (2014) discuss, an implication of this interpretation of relevance for the Doha Round is that substantial trade-volume gains for developing countries from negotiated tari¤ reductions may be achieved most e¤ectively if, in markets where they are large, developing countries negotiate reciprocally with each other and with developed countries. This implication runs counter to the non-reciprocal approach for developing countries in the Doha Round, as codi…ed under “special and di¤erential treatment” clauses. A second interpretation of the Subramanian-Wei …nding is that developed countries have managed to circumvent the MFN rule and discriminate against non-participating GATT/WTO members. Further empirical analysis of this interpretation is certainly warranted. We note, however, that in the speci…c context of GATT/WTO bilateral dispute settlement negotiations, Bown (2004a) …nds evidence that countries comply with the MFN rule.45 43 Limão (2007) explores a related forward-manipulation strategy. In his model, a government engages in foot dragging in order to in‡ uence a future bargain involving a preferential trade agreement with non-trade objectives. We provide further discussion of preferential trade agreements in Section 5. 44 The uneven trade e¤ects of GATT/WTO membership found by Subramanian and Wei overturn the …ndings of an earlier paper by Rose (2004) that constrained the trade e¤ects of GATT/WTO membership to be the same for all countries and found no membership e¤ect at all. Evidence that the trade e¤ects of GATT/WTO membership are restricted primarily to developed countries has subsequently been con…rmed by a number of papers (see, for example, Chang and Lee, 2011, Eicher and Henn, 2011, and Dutt, Mihov and Van Zandt, 2013, though Eicher and Henn attribute to PTAs most of the trade e¤ects that Subramanian and Wei attributed to WTO membership). 45 Subramanian and Wei (2007) report small and insigni…cant impacts of a developing country’ s WTO membership on its overall imports, which under balanced trade implies comparably small and insigni…cant impacts on its overall exports. They also report that developing countries – whether or not they are WTO members – export more to developed countries that are WTO members, and in their conclusion emphasize this as a possible source of gain 35 A second form of evidence relates GATT/WTO negotiated import tari¤ reductions to industry and country characteristics. Ludema and Mayda (2009, 2013) relate negotiated bilateral tari¤ reductions to measures that capture foreign exporter concentration and importer market power. In s tari¤ resulting particular, Ludema and Mayda (2013) …nd evidence that the level of the importer’ from negotiations is negatively related to the product of two terms: exporter concentration (as s market power (as measured by measured by the Her…ndahl-Hirschman index) and the importer’ the inverse elasticity of export supply, on a product-by-product basis). The positive e¤ect of market power on a country’s MFN tari¤ thus diminishes as exporter concentration increases, which is consistent with the view that negotiations are especially e¤ective in “undoing” terms-of-trade- driven ine¢ ciencies when exporter concentration is large. The important role played by exporter concentration supports the existence of an MFN free-rider e¤ect. They also provide evidence that the free-rider e¤ect is quantitatively important: they estimate that between 12 and 25 percent of potential liberalization on average goes unrealized, with most of this gap falling on the exports of developing countries due to their low-concentration product mix. Their …ndings thus also suggest a novel concentration-based interpretation of the Subramanian-Wei (2007) …nding. Enforcement and repeated-game models Our preceding discussion assumes that a negotiated trade agreement can be enforced. While this abstraction is helpful for interpreting and evaluating the principles of reciprocity and non-discrimination, the manner in which a trade agreement is en- forced is also essential to understand. According to the terms-of-trade theory of trade agreements, the strategic environment confronting governments setting trade policy corresponds to a Prisoners’ Dilemma setting. In a symmetric, two-country and two-good model, for example, if governments were to attempt to cooperate with a common tari¤ strictly below the Nash level, then each gov- ernment would have an incentive to “cheat” by raising its tari¤ to the best-response level. What deters governments from cheating? Since there is no GATT/WTO jail or other external enforce- ment device, a government cooperates in a trade agreement if and only if the government perceives that such behavior is in its self interest. In other words, and as argued by McMillan (1986, 1989), Dixit (1987) and Bagwell and Staiger (1990), the theory of repeated games suggests that a trade agreement must be self-enforcing so that for each government the short-term gain from cheating is smaller than the long-term discounted cost of any consequent breakdown in cooperation. We highlight here three implications of this repeated-game perspective. First, this perspective suggests an interpretation of the decision by initial GATT Contracting Parties to concentrate protective measures, with certain exceptions, into tari¤s. As Bagwell and Sykes (2004) argue, by “tari¤ying” quantitative restrictions such as quotas, governments facilitate mutually bene…cial and reciprocal trade liberalization. Speci…cally, by imposing tari¤s rather than allocating quotas across foreign exporters, governments make market-access gains easier to assess and thereby reduce negotiation transactions costs, reduce the uncertainty facing perhaps risk-averse exporters and thus enhance the value of market-access concessions, and increase the transparency of trade-policy for developing countries associated with the WTO. Their …ndings on overall trade are the relevant …ndings for our purposes, which is why we emphasize these …ndings in the text. 36 conduct so that cheating is less tempting. The latter point corresponds to the familiar notion from repeated-game theory that cooperation is typically easier to achieve when players’actions are publicly observable or transparent.46 As Maggi (1999) shows, a second and related implication of the repeated-game perspective is that a trade agreement may facilitate self-enforcing cooperation by enhancing third-party trans- parency. Maggi distinguishes between bilateral and multilateral enforcement mechanisms. Under a bilateral enforcement mechanism, if country A deviates with respect to the tari¤ that it applies to country B, then countries A and B revert to a Nash trade war. Countries A and B continue to cooperate with country C, however. By contrast, under a multilateral enforcement mechanism, if country A deviates with respect to the tari¤ that it applies to country B, then country A reverts to a Nash trade war with both countries B and C. Maggi shows that a multilateral enforcement mechanism can achieve greater cooperation than is possible under bilateral enforcement mecha- nism. Correspondingly, when a trade institution ensures that any deviation would be observed by all member countries, the future cost of cheating could be quite severe indeed, which in turn implies that more e¢ cient tari¤s can be achieved without violating the self-enforcement constraint. A third implication of the repeated-game perspective is that novel predictions may be generated when the self-enforcement constraint binds and political or economic shocks occur.47 Intuitively, when a change in the environment upsets the balance between the short-term incentive to cheat and the long-term discounted value of cooperation, an adjustment in trade policy may be required to bring the self-enforcement constraint back into balance. We develop this point further in Section 4.2 when we interpret GATT/WTO safeguard rules. A …nal point is that our discussion here concerns retaliation that occurs o¤ the equilibrium path in the repeated game. The role of such retaliation is to induce compliance; however, o¤-equilibrium- path retaliation is, by de…nition, not predicted by the model and thus should be distinguished from retaliation that actually appears in WTO rules and practice. At various points below, we note that on-equilibrium-path retaliation consistent with WTO rules and practice emerges naturally in the model once the repeated-game model is extended to include privately observed shocks. Implications of basic WTO principles and rules We summarize above research on reci- procity, non-discrimination and enforcement, all from the perspective of the terms-of-trade theory. Our discussion suggests that the principles of reciprocity and non-discrimination are well-designed for addressing the ine¢ ciencies associated with the terms-of-trade externality. The theory also pro- vides a natural means for interpreting the self-enforcing nature of trade agreements. At the same time, our review directs attention to signi…cant challenges that confront the GATT/WTO approach. The possibility of third-party externalities warrants particular attention, whether such externalities are positive (and raise free-riding concerns) or negative (and raise bilateral-opportunism concerns). 46 An additional and important advantage that tari¤s have over quantitative restrictions on trade is that the latter may be more di¢ cult to apply and enforce on a non-discriminatory basis. See, e.g., Curzon (1965, p. 130) for an early discussion of this issue. 47 As Furusawa (1999) shows, the repeated-game approach also suggests that greater patience is not always bene…cial for a country in trade-agreement negotiations, when the negotiation outcome also must be self-enforcing. 37 The theory reviewed above suggests that third-party externalities are eliminated when tari¤ poli- cies adhere to the principles of reciprocity and non-discrimination. The extent to which negotiated tari¤ cuts satisfy the principle of reciprocity may di¤er somewhat across market settings, however, and GATT/WTO rules allow for exceptions to the MFN rule, with the most important exception being the provisions for preferential trading agreements. Addressing the potential for third-party externalities thus remains an on-going challenge for the GATT/WTO. Our review also encourages consideration of the interaction between potential free-rider bene…ts and the bene…ts of GATT/WTO membership for developing countries. Recent empirical work indicates that many developing countries have not experienced signi…cant trade-volume bene…ts from GATT/WTO membership. The research reviewed above delivers two possible interpretations. First, if free-rider bene…ts are signi…cant, then countries may perceive modest gains from pursuing trade-liberalization negotiations in general but perhaps especially with developing countries whose exporters often operate in unconcentrated industries. Second, if free-rider bene…ts are largely eliminated, then developing countries may bene…t little from pursuing a non-reciprocal approach to trade liberalization. Together, these interpretations suggest that the bene…ts of GATT/WTO membership for developing countries may be enhanced if negotiated tari¤ reductions adhere closely to the principles of reciprocity and non-discrimination and if, in markets where they are large, developing countries negotiate reciprocally with each other and with developed countries. 4.2 Bindings, Binding Overhang and Safeguards/Contingent Protection A fundamental design feature of the GATT/WTO is that governments negotiate “tari¤ bindings” or “bound tari¤ levels” rather than exact tari¤ levels. For a given country and good, a bound tari¤, which is also referred to as a “tari¤ cap,” identi…es the maximal import tari¤ that can be applied. A tari¤ cap thus permits “downward ‡exibility,” in that a government can apply a tari¤ that is strictly below the tari¤ cap. In that event, “binding overhang” is said to occur.48 But a tari¤ cap constrains “upward ‡exibility,” since under normal circumstances the applied tari¤ cannot exceed the cap. Some potential for upward ‡exibility does exist within the GATT/WTO agreement, however, when certain contingencies arise. In this section, we review economic research that interprets tari¤ caps, binding overhang, and contingent protection or “safeguards.” We develop our discussion in the context of a standard partial-equilibrium model with two symmetric countries and three goods.49 One of the goods is the import good for the home country, while another good is the import good for the foreign country. The third good is a standard nu- meraire good, which is produced in both countries under constant returns to scale where labor is the only factor in the model. For the two non-numeraire goods, production occurs in each country under conditions of perfect competition and with diminishing marginal productivity. The resulting supply functions are upward sloping. We assume further that the consumers in both countries 48 The phrase “water” or “water in the binding” is sometimes used interchangeably with “binding overhang.” 49 Symmetry here means that the two countries are “mirror images” of one another. This model is frequently used in trade-policy research. See, for example, Bagwell and Staiger (2001a). 38 have a common utility function, which is additively separable and takes a quasi-linear form. The consumption of the numeraire good exhibits constant marginal utility while the consumption of each of the non-numeraire goods exhibits diminishing marginal utility. The latter property deliv- ers downward sloping demand functions for the non-numeraire goods. Finally, each country has available a tari¤ for its import good, and we assume that trade in the numeraire good is untaxed. Tari¤ bindings and tari¤ caps With this standard model in mind, we now consider the in- terpretation of tari¤ caps. At the time of negotiation, we imagine that governments are in an “ex ante” stage, with each government being uncertain about the importance that it will place in the future on pro…ts in the import-competing sector relative to consumer welfare. After government preference shocks are realized, tari¤s are applied in a manner consistent with the negotiated trade agreement. In this “ex post” stage, a government’s optimal unilateral tari¤ and also the e¢ cient tari¤ are higher when the government’s “type” (i.e., the welfare weight that the government at- taches to import-competing pro…ts) is larger. Due to the terms-of-trade externality, the optimal unilateral tari¤ is higher than the e¢ cient tari¤ for any government type, provided only that the e¢ cient tari¤ does not prohibit all trade. In this context, the challenge is to design a trade agree- ment that permits some ‡exibility, so that applied tari¤s may respond to preference shocks and thereby facilitate greater e¢ ciency, without opening the door to opportunistic tari¤ hikes. To …x ideas, we begin with a couple of extreme scenarios in which governments can design a trade agreement in the presence of preference shocks that achieves full e¢ ciency (i.e., an e¢ cient tari¤ in each state). First, if governments’realized types were publicly observed and veri…able, and if there were no contracting costs, then governments could write a “state-contingent”contract that delivers full e¢ ciency. Second, even if each government’s type were privately observed, standard arguments establish that governments could again achieve full e¢ ciency if a lump-sum contingent transfer instrument were available.50 These scenarios are extreme, however. Governments are likely to have some private (or at least unveri…able) information about their preferences, and explicit monetary transfers are not required in GATT/WTO rules and are rarely used in WTO dispute resolution.51 Motivated by these considerations, we therefore turn now to a scenario in which governments negotiate a trade agreement under uncertainty, have private information about their respective preferences at the time that tari¤s are applied, and do not have available an instrument with which to e¤ect contingent transfers. Allowing for a continuum of possible government types that are distributed uniformly, Bagwell and Staiger (2005b) consider this scenario in a linear-quadratic model and compare two possible trade agreements.52 In the …rst agreement, governments adopt a rigid tari¤ rule, under which each government commits to an exact tari¤ level for all types. Since a government’s type enters its welfare function in a linear fashion, the most e¢ cient agreement of this 50 See Bagwell and Staiger (2005b) for details in the trade-agreement context. 51 Further, even monetary transfers may entail ine¢ ciencies, due to the distortions associated with raising funds through taxation. 52 In a linear-quadratic model, economic agents have quadratic payo¤s and the corresponding demand and supply functions are linear. 39 kind places the rigid tari¤ at the level that is e¢ cient for the average type.53 They then compare this agreement with one in which each government adopts a weak-binding tari¤ rule, consistent with GATT/WTO rules, under which it commits to a tari¤ cap. They …nd that the most e¢ cient weak- binding yields strictly higher expected joint welfare than does the most e¢ cient rigid tari¤, and that the most e¢ cient weak binding is strictly higher than the most e¢ cient rigid tari¤. Intuitively, under a weak binding, the applied tari¤ exhibits binding overhang when a government draws a low s type is high and type, and so the binding only constrains the applied tari¤ when the government’ the e¢ cient tari¤ is thus also high. Amador and Bagwell (2013) generalize this analysis in several directions. Allowing for more general payo¤ and distribution functions, they derive conditions under which a trade agreement with a tari¤ cap (i.e., a weak-binding tari¤ rule) maximizes expected joint welfare among all incentive- compatible trade agreements. They thus provide a …rst theoretical explanation for the use of tari¤ caps in an optimal trade agreement. Their approach is to represent the problem of …nding an optimal trade agreement for the import good of the home country as a delegation problem, in s expected welfare corresponds to the associated expected joint welfare of the which the principal’ two governments and the agent is the government of the home country.54 In this context, a trade agreement identi…es a set of permissible tari¤s. In the ex ante stage of trade-agreement design, the two governments thus choose the set of permissible tari¤s for the home-country import good that maximizes ex ante joint welfare while satisfying the incentive-compatibility constraint that the home government will choose its preferred tari¤ from this set after observing its type. The tari¤s induced by rigid-tari¤ rules and weak-binding tari¤ rules are of course candidate solutions, but so, too, are many discontinuous tari¤ functions. Amador and Bagwell further allow that the trade agreement may specify that tari¤ choices are bundled with money-burning expenditures perhaps corresponding to administrative procedures. They characterize a rich set of environments, which includes the linear-quadratic model with a uniform distribution as a special case, in which the optimal trade agreement takes the simple form of a tari¤ cap. The optimal trade agreement then also exhibits binding overhang and does not employ money burning.55 Beshkar, Bond and Rho (forthcoming) extend the linear-quadratic model to a setting with asymmetric countries. They restrict attention to tari¤ caps and provide theoretical and empirical 53 As discussed in note 23, under the assumption that governments have available at the negotiation stage an instrument with which to make non-contingent, lump-sum transfers, e¢ cient policies for this partial equilibrium model maximize expected joint government welfare. 54 A delegation game is a principal-agent game in which the agent has private information and transfers are infeasible. The delegation game was …rst introduced and analyzed by Holmstrom (1977). 55 Related results arise in other settings as well. Amador and Bagwell (2013) move beyond the partial-equilibrium model with perfect competition and use their main propositions to establish conditions under which a tari¤ cap is also optimal for a monopolistic-competition model of trade with a …xed number of …rms. (Ossa, 2012, explores a similar model while focusing on other issues.) Amador and Bagwell (2012) similarly employ these propositions to provide conditions for the optimality of a tari¤ cap in a linear-quadratic model with a uniform distribution when private information concerns the weight that tari¤ revenue receives in the government welfare function. Finally, the models discussed here assume a continuum of possible types. As Bagwell (2009) con…rms, the optimal trade agreement does not take the form of a tari¤ cap in the linear-quadratic model when government preferences concerning the relative importance of import-competing …rms are drawn from two possible types. 40 support for the prediction that the optimal tari¤ cap is higher, and thus the likelihood of binding overhang is greater, when importer market power is lower. To gain some intuition for their theo- retical …ndings, it is helpful to consider the extreme case of a small country. Since the tari¤ policy of a small country imposes no terms-of-trade externality on its trading partner, the optimal trade agreement would permit such a country to impose its unilaterally optimal trade policy for what- ever preference shock it experiences. A high (i.e., unrestrictive) tari¤ cap achieves this goal. More generally, when a country has more market power in a sector, its tari¤ policy imposes a greater externality on its trading partner, and so a lower tari¤ cap, with an associated reduced frequency of binding overhang, is optimal. Using data on applied and bound tari¤s for WTO member countries, they then provide strong empirical support for this prediction.56 Tari¤ caps and binding overhang have also received attention in other modeling frameworks. In a model with contracting costs, Horn, Maggi and Staiger (2010) compare a weak binding rule and a rigid tari¤ rule. The framework is di¤erent from that considered by Bagwell and Staiger (2005b) but interestingly points to a related set of insights as regards bindings and overhang. Horn, Maggi and Staiger (2010) show that a weak binding rule is preferred to a rigid tari¤ rule, since the former permits e¢ ciency-enhancing downward ‡exibility, and they also note that the weak binding rule is characterized by binding overhang. A di¤erent approach is pursued by Maggi and Rodriguez-Clare (2007). As we also discuss in Section 6.1, they analyze a model in which trade agreements address both commitment and terms-of-trade problems. In their model, binding overhang does not occur in equilibrium; however, the potential to apply a tari¤ below the bound level induces ex post lobbying that has the bene…cial e¤ect of diminishing an ex ante problem of excess investment. The tari¤-cap theory described above establishes a rationale for tari¤ caps when governments have private information and contingent transfers are infeasible. As we argue next, in addition to providing an interpretation of tari¤ caps and binding overhang, this theory provides a foundation from which to understand contingent protection. A key idea is that contingent protection potentially may enhance e¢ ciency by “linking” tari¤ choices through time for a given government or across governments, and creating thereby some scope for imperfect contingent transfers. More generally, contingent protection provides a form of upward ‡exibility that in some cases may enhance e¢ ciency when certain shocks occur. At the same time, it also must be emphasized that tari¤ caps are a valuable means of “stabilizing” tari¤ commitments and diminishing the potential for “unwinding” tari¤ commitments.57 The optimal rules for contingent protection thus re‡ect a delicate balance between maintaining reduced tari¤s in response to the terms-of-trade externality and providing some upward ‡exibility in light of various shocks that may confront governments. Contingent protection such as safeguards and antidumping Bagwell and Staiger (2005b) explore a role for contingent protection when governments experience preference shocks that are 56 See Bacchetta and Piermartini (2011) for additional empirical evidence regarding tari¤ caps and binding overhang. 57 For further discussion of the importance of tari¤ stabilization, see Curzon (1965, Chapter 4). Focusing on India, Bown and Tovar (2011) provide empirical evidence that countries use antidumping and safeguard exceptions to unwind commitments to lower tari¤s in the presence of domestic political-economic pressure. 41 privately observed. Formally, they consider a repeated-game model in which government’s privately observed types are iid through time, and they show that expected joint government welfare may be improved when the tari¤ choices of any given government are appropriately linked through time.58 Their analysis is motivated by the “escape clause” de…ned in the WTO Agreement on Safeguards. This agreement describes contingencies under which a country can set a tari¤ above the tari¤ cap and thus achieve some degree of upward ‡exibility.59 As Bagwell and Staiger note, an interest- ing feature of the WTO safeguards agreement is that it embodies a dynamic-use constraint : if a government imposes escape-clause protection in an industry for X years, then it is not allowed to reimpose escape-clause protection in that industry for the next X years. This constraint introduces an opportunity cost to a government from selecting an escape-clause tari¤, so that incentive com- patibility is achieved when only a government with a su¢ ciently high type in the current period imposes an escape-clause tari¤. The dynamic-use constraint may thus promote further e¢ ciency by facilitating more e¢ cient tari¤s when a government draws a su¢ ciently high type. The prospect of improved e¢ ciency emerges because a government that goes above the cap today e¤ectively makes a transfer to the other government in the form of an improved continuation value. Martin and Vergote (2008) develop a related set of insights for a situation in which the tari¤ choices of one government are linked to those of the other government. They consider anti-dumping duties as allowed under certain contingencies in the WTO Anti-dumping Agreement. Motivated by empirical work by Blonigen and Bown (2003), Feinberg and Reynolds (2006) and Prusa and Skeath (2002) that provides evidence of a retaliatory role for anti-dumping duties, Martin and Vergote argue that such on-equilibrium-path retaliation may facilitate e¢ ciency gains by ensuring that the home government protects its import-competing industry with an anti-dumping duty only when the current importance of that industry to the home government is su¢ ciently great. In this case, when the home government imposes an anti-dumping duty, a link is forged to a future retaliatory anti-dumping duty of the foreign country; and upon raising its applied tari¤ via an anti-dumping duty the home government thereby again makes a transfer to the foreign government in the form of an improved continuation value. Our discussion of contingent trade-policies above focuses on the idea that the potential for upward ‡exibility might improve e¢ ciency when governments are privately informed about their preferences, if contingent transfers achieved through continuation values ensure that only govern- ments with high types exercise this potential. The WTO Safeguards Agreement and also the WTO Anti-dumping Agreement, however, explicitly indicate contingencies under which upward ‡exibility can be exercised. For example, the WTO Agreement on Safeguards permits the application of a safeguard tari¤ when the domestic industry is seriously injured as a result of increased imports. If such contingencies are interpreted as de…ning veri…able market conditions, then an alterna- tive modeling approach is suggested under which a government that seeks to impose a contingent trade policy must incur the necessary costs to publicly verify that the relevant contingent state is 58 Their work builds on a literature in game theory that associates continutation values with transfers. See Athey and Bagwell (2001) and Fudenberg, Levine and Maskin (1994). 59 For an overview of the the WTO safeguards agreement, see Wauters (2010). 42 present.60 Bond and Beshkar (2012) o¤er a …rst example of this modeling approach. They analyze a partial-equilibrium model with asymmetric country sizes that features costly state veri…cation, in the speci…c sense that at a cost the importing government can publicly verify the welfare weight that determines its type. A novel feature of their model is that both caps and escape clauses are endogenously determined as part of an optimal trade agreement. Interestingly, they …nd that circumstances under which the possibilities of overhang (downward ‡exibility) and escape (upward ‡exibility) co-exist are relatively rare. Intuitively, when higher types use the escape clause, the cap can be set at a lower level; and this implies in turn that the likelihood of overhang is small. An alternative theory of escape clauses dispenses with the possibility of private information and focuses instead on the self-enforcement constraints that underlie any cooperative trade agreement. Bagwell and Staiger (1990) o¤er a …rst paper of this kind. They consider a partial-equilibrium model with two countries, in which publicly observed trade-volume shocks occur in an iid fashion over time. When a country imports a large volume, it has an increased incentive to cheat on a cooperative agreement and select its unilaterally optimal tari¤. The cost of such behavior is that it may induce governments to abandon cooperation and revert to Nash trade policies in the future. If governments maximize national income and are su¢ ciently patient, then they can enforce free-trade policies for all possible trade-volume shocks. If governments have moderate patience, however, then they can enforce free-trade policies only in periods with low trade volumes. When trade volumes take higher values, the incentive to cheat is acute at free-trade policies and the gains from defection overwhelm the discounted future cost of initiating a Nash trade war. Governments with moderate patience can still cooperate in the presence of large trade volume shocks, but they do so by setting positive (and below-Nash) tari¤s. Intuitively, by cooperating with a positive tari¤, governments reduce the incentive to cheat and bring it back in line with the future discounted cost of Nash reversion. The positive tari¤s that accompany high trade-volume shocks can be interpreted in terms of an escape clause, and Bagwell and Staiger show that an escape clause of this kind is a feature of an optimal self-enforcing trade agreement for governments with moderate patience.61 Bown and Crowley (2013a) provide a …rst empirical investigation of the cross-sectional and intertemporal predictions of the Bagwell-Staiger (1990) model. As discussed, that model predicts that an import tari¤ increase is more likely when the import volume increases. In addition, con- ditioning on a positive import surge, the model predicts that the probability of a tari¤ increase is positively associated with less elastic import demand and export supply functions and also with less variable import demand volumes. Since the escape clause featured in the Bagwell-Staiger model could correspond to any instrument of special protection, such as a safeguard tari¤ or an anti-dumping duty, Bown and Crowley analyze the model’s predictions using data on US import tari¤ increases arising under the US antidumping and safeguard laws and …nd strong support for the predictions of the Bagwell-Staiger model. A special feature of the Bagwell-Staiger (1990) model is that trade-volume shocks are iid through 60 As Sykes (2003) discusses, however, the extent to which the WTO Agreement on Safeguard s and subsequent legal cases serve to articulate a clear set of contingencies may be questioned. 61 Their work builds on methods developed by Rotemberg and Saloner (1986) for the study of collusion. 43 time. A trade-volume shock thus a¤ects the incentive to cheat in the current period but has no direct impact on the discounted future value of cooperation. Bagwell and Staiger (2003) extend the model to allow for both iid shocks to the trade-volume level and stochastic but persistent trends in the growth rate for trade volume.62 At any given point in time, governments observe the current shock to the trade-volume level and observe as well whether the trading relationship is in a fast- or slow-growth phase, where the relationship moves between the two phases according to a Markov process. They …nd that the most-cooperative trade agreement for governments with moderate patience is countercyclical: all else equal, the most-cooperative tari¤s are lower in a fast-growth phase. The key intuition is that governments have more to lose from initiating a trade war in a (persistent) fast-growth phase; as a consequence, they can then withstand that heightened incentive to cheat that a lower cooperative tari¤ implies.63 Bown and Crowley (2013b) provide an empirical investigation of the macroeconomic determinants of time-varying trade policy. Using quarterly data for the US, EU, Australia, Canada and South Korea, they …nd evidence of a countercyclical trade policy response in the pre-Great Recession period (namely, the …rst quarter of 1988 through the third quarter of 2008).64 Their …ndings are broadly consistent with the theoretical predictions in Bagwell and Staiger (2003) as well as those in Crowley’s (2011) reciprocal dumping model, which predicts an increase in import restrictions in response to macroeconomic weakness abroad. A di¤erent perspective on safeguards is o¤ered by Sykes (1991, 2003). He observes that in- struments of special protection are often applied to assist declining domestic industries. Firms in such an industry retain a large share of the bene…ts of price-increasing protection, since temporary protection is unlikely to encourage entry in a declining industry. Thus, …rms in a declining industry s wel- may lobby hard for protection and, therefore, …gure prominently in the domestic government’ fare function. Foreign exporters are naturally harmed by the domestic tari¤; however, if the foreign export industry is growing, then they may complain little about facing protection, since their pro…ts eventually would be lost to entry anyhow. These …rms therefore may not …gure prominently in the s welfare function. Thus, governments in an ex ante state may be attracted to foreign government’ a safeguard rule for declining industries, since the welfare bene…t to the government that uses the safeguard may exceed the welfare cost to the government whose exporters face the safeguard. Implications and new directions Our survey highlights that tari¤ caps and binding overhang occur in an optimal trade agreement when governments are privately informed about their pref- erences and contingent transfers are infeasible. With this foundation in place, we then consider three complementary theories for the use of special instruments of protection. First, when govern- ments experience privately observed preference shocks, upward ‡exibility might promote further 62 Hochman and Segev (2010) extend the Bagwell-Staiger (1990) model in a di¤erent direction by allowing that governments may imperfectly observe the trade-volume shock before applying tari¤s. Another interesting extension is considered by Tabakis (2010), who examines the use of special protection when countries are transitioning into preferential trading agreements. 63 This work builds on methods used by Bagwell and Staiger (1997c) to analyze collusion over the business cycle. 64 Bown and Crowley (2014) provide additional empirical support from a sample of 13 emerging economies and use of annual data covering the period 1995-2010. 44 e¢ ciency, provided that higher tari¤s are only used by governments facing high political pressures. This incentive compatibility constraint, in turn, can be met, when current tari¤s are linked to future tari¤s so that contingent transfers can be achieved, at least to some degree, through changes in continuation values. The incentive-compatibility constraint also can be met when costly state veri…cation is feasible. Second, optimal cooperation when trade volumes are volatile entails a low baseline tari¤ coupled with an escape clause that allows for higher tari¤s when trade-volume shocks are high. Safeguards emerge in an optimal trade agreement in this setting and indeed complement tari¤ liberalization, as the baseline tari¤ would be higher were safeguards not allowed. Finally, safeguards may enable governments to achieve greater welfare by rewarding (penalizing) industries that …gure more (less) prominently in governments’welfare functions. This theory associates the use of safeguards with declining industries. While more empirical work is needed, we also identify studies that provide empirical support for themes emerging from the theoretical analyses. We conclude this section by mentioning a new literature that assesses the value of tari¤ bindings, as distinct from reductions in applied tari¤s, in the presence of policy uncertainty.65 Handley (2014) and Handley and Limão (forthcoming) focus on the implications of policy uncertainty when exporters face sunk market entry costs. They observe that tari¤ bindings reduce policy uncertainty by constraining the range of possible tari¤s, limiting losses in the worst-case scenario and thereby stimulating entry into export markets. Formalizing this insight, they use their models to empirically quantify the trade e¤ects of WTO bindings (for Australia, in Handley, 2014) and of PTA bindings (for Portugal joining the EU, in Handley and Limão, forthcoming), reporting large trade e¤ects for each case. The study of tari¤ caps and policy uncertainty represents an especially promising direction for new research. 4.3 Subsidies The appropriate treatment of subsidies in a trade agreement is subtle. On the one hand, a domestic production subsidy can be a “…rst-best” instrument with which to address a market failure that results in an ine¢ ciently low level of output. A production subsidy may also be an attractive instrument for a government with political-economic objectives that wishes to redistribute surplus to producers in a given industry. On the other hand, some restrictions on the use of domestic production subsidies are necessary, since otherwise a government could always use a domestic production subsidy to an import-competing industry so as to undermine the bene…ts o¤ered to other countries through negotiated tari¤ cuts. The appropriate treatment of export subsidies is also subtle. An export subsidy lowers the world price for the export good and thus generates a terms-of-trade gain for importing countries, but it may also displace exports from other countries and alter entry and exit patterns across countries. In this section, we review the GATT/WTO legal treatment of subsidies and discuss economic research that interprets and evaluates this treatment.66 65 In addition to the Handley (2014) and Handley and Limão (forthcoming) papers we discuss in the text, other recent papers in this emerging literature include Limão and Maggi (forthcoming) and Pierce and Schott (2015). 66 For closely related discussions, see also Bagwell (2008) and Bagwell, Staiger and Sykes (2013). 45 Subsidy rules under the GATT versus WTO The treatment of subsidies in GATT was relatively tolerant: a foreign trading partner could respond to the subsidies of the domestic country through two possible means.67 First, the foreign government could unilaterally impose a counter- vailing duty (CVD) if its import-competing industry experienced material injury as a consequence of an export subsidy given to producers in the domestic country. Second, the foreign government could …le a non-violation complaint if it had previously negotiated a tari¤ reduction from the do- mestic government on a given product and the domestic government later o¤ered a subsidy to its import-competing producers of that product. To succeed, the foreign government would have to show that a new or increased domestic subsidy program emerged that had the e¤ect of nullifying or impairing the market-access bene…ts that the foreign government had reasonably expected at the time of the tari¤ negotiation.68 In this case, the domestic government would not be required to remove the subsidy; however, it would be expected to make policy adjustments that restored the foreign country’s negotiated market access. In addition, as part of GATT negotiations, several countries agreed to restrict the use of export subsidies, especially for non-agricultural goods. s Agreement on Subsidies and Countervailing Measures The treatment of subsidies in the WTO’ (the SCM Agreement) is much more restrictive. First, except as allowed for in the Agreement on Agriculture, export subsidies (and also local-content subsidies) are prohibited. Second, “speci…c” subsidies that have “adverse e¤ects” on other members are actionable, where adverse e¤ects could take the form of an injury to an industry in another member country, the nulli…cation or impairment of bene…ts expected by another WTO Member, and “serious prejudice”to the interests of another WTO Member. The …rst two forms of adverse e¤ects are broadly re‡ected in the GATT treatment of subsidies and may be associated with the use of CVDs and non-violation complaints by the adversely impacted member country. The more novel ingredient is serious prejudice, which may occur if the e¤ect of a subsidy o¤ered by the domestic country is to cause a loss of exports by a WTO member into the domestic market or a third-country market. Importantly, a domestic production subsidy can be actionable under the SCM Agreement independently of whether the subsidy nulli…es or impairs the market-access expectations associated with an earlier tari¤ negotiation.69 WTO rules thus treat subsidies in a fairly severe manner. In contrast to import tari¤s, for which caps are negotiated, exports subsidies are banned. As well, under WTO rules, a country that uses a domestic production subsidy must withdraw it, or remove its adverse e¤ects, even if the subsidy itself does not upset any negotiated market-access expectation. We consider next research that interprets and evaluates the WTO’s treatment of subsidies. 67 Sykes (2005a) o¤ers a detailed discussion of the evolution of subsidy rules under GATT and the WTO. 68 See Petersmann (1997, pp. 151-4) for discussion of the conditions under which a subsidy could be determined to upset market-access expectations. 69 As regards countermeasures, if the domestic government refuses to remove an export subsidy, then the com- plaining Member government may take “appropriate countermeasures.” For an actionable subsidy, in the absence of an agreement on compensation, and if steps to remove the adverse e¤ects are not undertaken or if the subsidy itself is not withdrawn, then the complaining Member may be granted authority to impose countermeasures that are “commensurate” with the adverse e¤ects attributable to the subsidy. See Lawrence (2003, pp. 54-60), Mavroidis (2000), Spamann (2006), and Bown and Ruta (2010) for further discussion. 46 Production subsidies We begin with the treatment of domestic production subsidies. On the one hand, and as mentioned above, a domestic production subsidy is a …rst-best instrument with which to address a market failure that leads to an ine¢ ciently low level of production.70 A domestic production subsidy also may be an attractive instrument for a government with political-economic preferences that seeks to redistribute surplus to the import-competing industry. On the other hand, if the use of domestic production subsidies were completely unregulated, then governments would be unable to achieve e¢ cient outcomes through reciprocal tari¤ negotiations alone. Intuitively, in the absence of any restrictions on the use of such subsidies, a government that exchanged reciprocal tari¤ cuts with a trading partner could subsequently “undo”the market-access consequences of its own tari¤ cut by providing a production subsidy to its domestic import-competing industry. This discussion suggests that domestic production subsidies have a potential e¢ ciency-enhancing role to play but that their use must be regulated in some manner. Given these considerations, we may ask: how should domestic production subsidies be treated in a trade agreement? The described trade-o¤s suggest a potential answer to this question: grant each government ‡exibility over its domestic policies, provided that the overall e¤ect of its chosen domestic policies does not erode the market-access commitments made through its preceding tari¤ negotiations. Under this approach, after tari¤s are negotiated, a government would be allowed to adjust its domestic policies in any way, so long as the overall e¤ect does not result in a terms-of-trade loss for its trading partner. To analyze this approach, Bagwell and Staiger (2006) augment the two-country, general-equilibrium model of trade considered above to include domestic policies. Assuming that the set of domestic instruments available to governments is su¢ ciently rich to create a degree of policy redundancy, they …nd that GATT rules enable governments to achieve an e¢ cient outcome using tari¤ negotiations.71 A key feature of GATT rules in this context is that a government can …le a non-violation complaint if it su¤ers a terms-of-trade loss as a consequence of a change in the domestic policies of its trading partner. The WTO’s SCM Agreement places further restrictions on the use of domestic production subsidies. These restrictions limit policy redundancy and may thereby prevent governments from achieving e¢ cient outcomes through tari¤ negotiations. Bagwell and Staiger (2006) also consider a setting with limited instruments. In this setting, GATT rules are no longer su¢ cient for achieving e¢ ciency through tari¤ negotiations. Intuitively, in a limited-instrument setting, a government may be unable to reposition its subsidy to an ef- …cient level without imposing a terms-of-trade loss on its trading partner. Indeed, when the set of instruments is limited, the SCM Agreement could represent an improvement over GATT rules. For example, if governments respectively maximize national income and no market failure exists that creates a corrective role for domestic production subsidies, then the use of subsidies would be ine¢ cient. More broadly, though, market failures and/or redistributive goals suggest a potential 70 An import tari¤ is another instrument that might be used to expand domestic production levels; however, an import tari¤ is a second-best instrument. An import tari¤ is equivalent to a consumption tax and a production subsidy, and thus a¤ects both consumer and producer margins. 71 Su¢ cient policy redundancy is satis…ed if each government has available an import tax, a domestic production subsidy and a domestic consumption tax. 47 role for domestic production subsidies in an e¢ cient outcome. As Bagwell and Staiger argue, tight restrictions on subsidies could then have a “chilling e¤ect” on tari¤ negotiations.72 To the extent that the SCM Agreement imposes tight restrictions on the use of domestic production subsidies, GATT rules on subsidies are then preferred to those in the WTO. GATT non-violation rules on domestic policies identify an attractive approach in granting ‡ex- ibility to governments up to the point where an externality is imposed on trading partners. At a practical level, however, it also must be acknowledged that non-violation rules have their own lim- itations.73 As two illustrations, we mention that it may not be obvious what a government should reasonably expect at the time of negotiation, and it also may not be clear where to draw the line in terms of which sorts of domestic policy changes are appropriately disciplined using non-violation nulli…cation-and-impairment complaints. Export subsidies in two-country models We turn now to the treatment of export subsidies. We begin with a basic observation: in the simple two-country, two-good general equilibrium model described above, an increase in the domestic country’s export subsidy would have the same e¤ect on prices and thus government welfare functions as would a decrease in its import tari¤. This observation, known as the Lerner symmetry theorem, follows since either policy change would lower (raise) the relative price of the domestic import good in the domestic (world) market. One implication is that both policy changes would generate a terms-of-trade loss (gain) for the domestic (foreign) country. Since each government’s welfare is expressed as a function of the relative price in its country and its country’s terms of trade, it follows that the two policy changes a¤ect government welfare functions in the same fashion as well. Hence, our arguments above carry over immediately when governments select export instead of import policies. In particular, Nash export policies are ine¢ cient and result in too little trade, and governments can mutually gain from an agreement on export policies only if they agree to make reciprocal adjustments that lead to greater trade volume. This result is familiar when governments respectively maximize national income. In that case, s optimal unilateral policy is an export tari¤. The key free trade is e¢ cient, but each government’ intuition is that a government can use an export tari¤ to induce its competitive export industry to restrict output as would a monopolist, where monopoly rents are now retained in the form of tari¤ revenue. In the resulting Nash equilibrium, both governments impose export tari¤s, and the trade volume is ine¢ ciently low. More generally, the sign of a government’s unilaterally optimal export policy depends on the government’s speci…c preferences. The Nash export policy is an export subsidy for a government that gives su¢ cient welfare weight to the interests of its export sector.74 The fundamental point, though, is the governments’noncooperative export policies, whatever their sign, induce too little trade from the governments’joint perspective. A trade agreement can thus generate mutual gains for governments only if it facilitates reciprocal 72 The idea is that governments may be hesitant to negotiate tari¤ bindings if subsidies are disciplined heavily, since tari¤s may then be the best remaining means of providing assistance to domestic import-competing industries. 73 Another important caveat is that a market-access-preservation rule may fail to be optimal in settings with private information, as Lee (2014) argues. We discuss his and related papers in greater detail in Section 5. 74 See, for example, Bagwell and Staiger (2001a) and Grossman and Helpman (1995a). 48 increases in export subsidies (or reciprocal decreases in export tari¤s) relative to noncooperative levels. Intuitively, an increase in a country’s export subsidy generates a positive terms-of-trade externality for its trading partner, whose consumers now enjoy a lower price on their import good. While this argument makes sense within the context of the terms-of-trade approach to trade agree- ments, it runs completely counter to the treatment of export subsidies in the WTO. There are thus two possibilities: either the WTO’s prohibition on export subsidies is misguided, or the simple two- country, competitive-markets version of terms-of-trade theory is missing something important. To explore the latter possibility, we discuss next a sequence of enriched terms-of-trade models within which to further explore the treatment of export subsidies. Export subsidies in third-market models One potential cost of export subsidies not featured in the above discussion is that an export subsidy o¤ered in one country lowers the world price and thereby imposes a negative terms-of-trade externality on other exporting countries. To explore this issue, we follow the “strategic-trade” literature and construct a “third-market” model.75 Speci…- cally, we illustrate our points using a simple partial-equilibrium model with a single good, where all …rms are located in Countries 1 and 2 and all consumers are located in Country 3. Firms in Countries 1 and 2 then compete for sales to consumers in Country 3. We ask two questions. First, when would a government have a unilateral incentive to o¤er an export subsidy? Second, when if at all should an international trade agreement discipline the use of export subsidies? Following the pioneering model of Brander and Spencer (1985), we focus on a setting in which a …xed number of …rms engage in Cournot competition for sales in Country 3. The key issues involved can be illustrated in a simple third-market model with two …rms, wherein Firm 1 is located in Country 1, Firm 2 is located in Country 2 and all consumers are located in Country 3. In the absence of any subsidy, Firms 1 and 2 have the same constant marginal cost. The governments of Countries 1 and 2 respectively maximize national income. The game has two stages. The governments simultaneously select their respective speci…c (i.e., per unit) export subsidies, and after observing these selections Firms 1 and 2 simultaneously select their respective output levels. A key …nding is that, starting at free trade, a government that maximizes national income now has a unilateral incentive to o¤er an export subsidy. As is now well known, an export subsidy lowers s output reaction curve. The strategic the cost of the exporting …rm and thus shifts out this …rm’ advantage of such a shift is that the other exporting …rm responds by reducing its own output. An export subsidy thus “shifts pro…ts” to the subsidizing country. The overall e¤ect of a strategic export subsidy is to expand aggregate output and thus induce a fall in the world price. Hence, as in the competitive model, an export subsidy generates a positive terms-of-trade externality for the 75 As we discuss in Section 6.2, some recent research evaluates whether the strategic-trade and “delocation”theories of export subsidies can be interpreted as providing a rationale for trade agreements that is distinct from that provided by the terms-of-trade externality. Bagwell and Staiger (2012a, 2015) argue, however, that the problem for a trade agreement to solve in the pro…t-shifting and delocation settings can be given a terms-of-trade intepretation, provided that both import and export policies are available. In particular, they show that politically optimal policies remain e¢ cient in these settings when a full set of trade policies is available. We thus include some discussion of these theories here, as part of our discussion of the implications of the terms-of-trade approach for the treatment of export subsidies, while postponing further discussion of the rationale of trade agreements in these settings until Section 6.2. 49 importing country. A new feature of the third-market model is that an export subsidy generates as well a negative terms-of-trade externality for the other exporting country. Of course, the other exporting country has a similar incentive to subsidize exports, and the resulting subgame perfect Nash equilibrium involves export subsidies from both exporting countries. The two exporting countries end up worse o¤ when export subsidies are legal, since their strategic e¤orts o¤set and simply result in a lower world price. Consumer and global economic welfares, however, are higher when subsidies are allowed than would be the case were subsidies banned. The described model therefore provides an interpretation for why exporting countries would seek a ban on export subsidies as a means to keep the price high, but it suggests that an international trade agreement designed to maximize the combined welfare of all countries in fact should encourage even greater use of export subsidies than occurs in the noncooperative equilibrium.76 This simple third-market model has been extended in many directions to allow for alterna- tive forms of imperfect competition, multiple exporting …rms in exporting countries, consumers in exporting countries, endogenous cost technologies and other considerations.77 Such extensions provide important quali…cations about the sign of the unilaterally optimal export policy, but a robust feature of models with a …xed number of …rms is that a more expansionary export policy provides a positive terms-of-trade externality to importing consumers. Returning now to the two questions identi…ed above, we note that the simple third-market model described here provides an interpretation for the unilateral appeal of export subsidies, but it also suggests that export subsidies may be under-supplied. The model thus does not provide a foundation from which to understand a cap on export subsidies, much less the WTO’s prohibition on export subsidies. The model also fails to provide an interpretation for the fact that the WTO treats export subsidies more severely than import tari¤s.78 In short, we conclude that the simple third-market model fails to rationalize the treatment of export subsidies in the WTO.79 Export subsidies and industrial policy The models described so far do not focus on the long- run implications of export policies for industrial structure. To explore this issue, we now follow Venables (1985) and consider the “delocation” e¤ects of trade policies in a two-country partial equilibrium model with two-way trade in a homogeneous good. The game has three stages. In the 76 The market is initially distorted with too little production as a consequence of oligopolistic competition. The equilibrium with strategic export subsidies expands output closer to the competitive level, and a trade agreement could further improve global welfare by increasing subsidy levels so that the competitive level of output is produced. 77 Eaton and Grossman (1986) and Maggi (1996) provide the key studies examining the dependence of the sign of optimal export policies on the form of imperfect competition in the product market. Bagwell and Staiger (1994) argue that the sign of the optimal strategic R&D policy is less sensitive to the form of imperfect competition. See Brander (1995) for a review of the strategic-trade literature. 78 We develop these conclusions for a model with imperfect competition. As Bagwell and Staiger (2001c) show, if governments have political-economic objectives that give su¢ cient weight to export interests, then similar conclusions hold in a third-market model with perfect competition when marginal cost is increasing. 79 The simple third-market model described here neglects many potential welfare costs associated with export subsidies. Export subsidies may generate distortions in production and encourage rent-seeking behavior, for example. We do not intend to minimize the importance of such considerations; however, we also point out that similar welfare costs are associated with import tari¤s. For further discussion of these and other neglected welfare costs associated with export subsidies, see Bagwell, Staiger and Sykes (2013, pp. 186-9). 50 …rst stage, governments simultaneously select (speci…c) import and export tari¤s. Each government seeks to maximize national income in its country. In the second stage, after observing trade policies, potential entrants decide whether to locate in the domestic or foreign market, where entry entails a positive …xed cost. In any country, entry occurs until expected pro…ts (including the …xed cost) are driven to zero. Finally, in the third stage, after observing trade policies and the numbers of …rms located in each of the two countries, the entering …rms simultaneously choose Cournot output levels, where an individual …rm selects both an output level for the market in which it is located as well as a separate output level for exportation into the other market. The two markets are segmented, and a positive (per-unit) transport cost must be incurred for exported units. A key feature of this model is that it exhibits a Metzler paradox: if a government raises its import tari¤ (or raises its export subsidy), the price of the good within its country falls. To see the intuition, suppose that we start at global free trade with levels of entry in each country that generate zero pro…t for each …rm, and suppose that the domestic government then imposes a slight import tari¤ (or a slight export subsidy). Holding …xed the numbers of …rms in each country, domestic …rms would then enjoy positive pro…ts while foreign …rms would experience negative pro…ts. Some adjustment in the patterns of entry across the two countries is thus necessary to restore zero pro…ts. Due to positive transport costs, each …rm sells greater output in its local market than in its export market; hence, the only way to reduce the pro…t of a domestic …rm while increasing the pro…t of a foreign …rm is to adjust entry patterns until the domestic price falls and the foreign price rises. Consequently, the domestic policy change must induce (reduce) entry into the domestic (foreign) market to such an extent that the domestic price falls (foreign price rises). In this sense, when a government imposes a higher import tari¤ (or export subsidy), it “delocates” …rms from the other country to its own country. As Venables (1985) shows, if all policies are initially set at free trade, then the domestic gov- ernment can gain by imposing a small import tari¤. Producer surplus is una¤ected by the change, since free-entry conditions ensure that …rms earn zero pro…t. But a small import tari¤ generates positive tari¤ revenue and also leads to a lower price and higher consumer surplus in the domestic country, due to the Metzler paradox. A small export subsidy likewise leads to a lower price and higher consumer surplus in the domestic country; however, the small export subsidy imposes a cost in the form of subsidy expenses. Venables shows that, when demand and costs are linear, a small export subsidy also generates a net gain for the domestic government. Both policy changes result in a higher foreign price, lower foreign consumer surplus and lower foreign government welfare. Starting at free trade, therefore, Venables’ analysis shows that export subsidies are unilaterally attractive and impose a negative terms-of-trade externality on the trading partner. Bagwell and Staiger (2012b) generalize the analysis of the linear Cournot delocation model. They show that the Nash policies for governments in fact are characterized by the use of import tari¤s and export tari¤s. Thus, while an export subsidy is unilaterally attractive for a government when its import policy is free trade, the government prefers an export tari¤ when its import tari¤ is optimally set at a positive level. Intuitively, when a positive import tari¤ is in place, an export 51 tari¤ generates additional tari¤ revenue on imports by encouraging foreign entry and thus exports. They also …nd that free trade in import and export policies is e¢ cient. Together, these …ndings suggest a possible interpretation of the treatment of export subsidies in GATT/WTO. The linear Cournot delocation model suggests that governments would perceive a unilateral gain from using export subsidies only once import tari¤s were negotiated to levels su¢ ciently close to free trade. From this perspective, it is not surprising that GATT rules did not impose strong restrictions on the use of export subsidies. Over time, however, as import tari¤s were negotiated through GATT rounds to lower levels, governments may have perceived a unilateral gain from imposing export subsidies. Furthermore, since free trade is an e¢ cient outcome in the linear Cournot delocation model, governments could achieve mutual gains given low import tari¤s if they were to cap or even prohibit export subsidies. In this way, the model o¤ers a potential e¢ ciency- enhancing interpretation of the prohibition of export subsidies in the WTO SCM Agreement. Among the models reviewed above, the linear Cournot delocation model o¤ers the most suc- cessful interpretation of the treatment of export subsidies in the WTO. At the same time, we note that the model is not completely successful. The linear Cournot delocation model also predicts that governments would bene…t from a prohibition of import tari¤s, and so it does not deliver an interpretation for why export subsidies are treated more severely in the WTO than are import tari¤s. We also note that the model rests on a speci…c structure (Cournot competition, segmented markets) and applies only for governments that focus on the long-run implications of trade policy.80 The strategic “pro…t-shifting”models described above may be more appropriate if governments are focused on trade-policy implications that manifest over the short run. Implications for subsidy rules under the WTO In sum, the appropriate treatment of sub- sidies in a trade agreement is a subtle issue. Our review of the literature in this section focuses on models for which trade policies generate terms-of-trade externalities for trade partners. The review reinforces the subtle implications of subsidies: domestic productive subsidies can play both e¢ ciency-enhancing and opportunistic roles, export subsidies generate positive externalities to for- eign consumers and negative externalities to foreign …rms in models with …xed industrial structures, and export subsidies may generate negative externalities to foreign consumers in long-run settings with endogenous entry and exit. On the whole, our review does not provide strong support for the speci…c treatment of subsidies in the WTO. We describe work suggesting that WTO rules on domestic production subsidies may be a step backwards relative to GATT rules, and we also summarize a range of models under which export subsidies are actually under-supplied relative to 80 Bagwell and Lee (2015) examine trade policies in the alternative long-run model of Melitz and Ottaviano (2008), wherein …rms have heterogeneous costs and engage in monopolistic competition. They show that, starting at global free trade, a country gains when it introduces a small import tari¤ or a small export subsidy, provided in the latter case that transportation costs are low and productivity dispersion is high. These interventions, however, lower the welfare of the trading partner. Other work examines optimal export policy when relative wage e¤ects are induced through general equilibrium channels. For example, in Ricardian settings, Itoh and Kiyono (1987) characterize a welfare-enhancing role for targeted export subsidies, and Costinot, et al. (2015) o¤er a full characterization of optimal trade policies. Relatedly, Demidova (2015) considers a general equilibrium version of the Melitz-Ottaviano model and shows that the Metzler Paradox then no longer obtains. 52 the e¢ cient level for governments. While the linear Cournot delocation model provides a potential interpretation for an agreement to limit or even prohibit the use of export subsidies, the existing formal models that we review do not identify a reason for treating export subsidies more severely than import tari¤s. 4.4 Non-Violation Complaints, Shallow Integration and National Treatment The central implication of the terms-of-trade theory of trade agreements is that governments set unilateral tari¤s at levels that are ine¢ ciently high, since they fail to internalize the terms-of-trade implications of their tari¤ policies for each other. A trade agreement can “undo” this ine¢ ciency by facilitating mutually advantageous and reciprocal tari¤ reductions that expand the volume of trade to more e¢ cient levels. Tari¤s are not the only instruments, however, that impact the terms of trade. For large countries, domestic taxes, subsidies and standards may also a¤ect the terms of trade and lead thereby to possible ine¢ ciencies. At the same time, domestic policies may have legitimate and even …rst-best roles to play as instruments with which to address market failures or distributional concerns within a country. Attempts to regulate domestic policies through a trade agreement therefore must balance the possible opportunistic use of such policies against the e¢ ciency-enhancing roles that such policies may play. We thus arrive at the following question: how should domestic policies be treated in a trade agreement? In this section, we describe research that responds to this question while utilizing the terms-of-trade approach to trade agreements. In fact, we have already encountered this question above in the speci…c context of our discussion of the treatment of domestic production subsidies. As we note there, a natural answer to this ques- tion is that a trade agreement should grant each government ‡exibility when choosing its policies, provided that the overall e¤ect of any policy changes does not erode the market-access commitments achieved through its preceding tari¤ negotiations. Bagwell and Staiger (2001b) formally explore this answer in a two-good general equilibrium model of trade in which governments have available domestic policies (e.g., labor or environmental standards) as well as tari¤ policies. Working with a model in which all international externalities ‡ow through the terms of trade, their main …nding is that e¢ ciency can be achieved through tari¤ negotiations alone, provided that each government is free to make subsequent adjustments in its domestic and trade policies that leave its negotiated market access commitment (i.e., the terms of trade) unaltered. As discussed in Section 4.3, Bagwell and Staiger (2006) obtain a related …nding in the speci…c context of domestic production subsidies. As noted in Section 3.1, a key assumption of the Bagwell-Staiger (2001b) model is that each government only has a direct interest in the domestic policy adopted by its country, where this interest in turn may re‡ s welfare ect various national considerations that impact the government’ (e.g., a government may have a direct interest in the health and safety of its citizens, or in the s borders). The lack of any direct interest by any one environmental quality within its country’ government in the domestic policy selected by another government indicates that the model does not allow for non-pecuniary international externalities (e.g., global pollution). For this family of models, the domestic policy choices of one government therefore impact the welfare of another 53 government only indirectly, through the terms of trade. Notice, though, that “race to the bottom” concerns are about the pecuniary international externalities (trade e¤ects) associated with a choice of weak standards, and so this family of models is capable of capturing those concerns. When international externalities travel only through the terms of trade, the main …nding of Bagwell and Staiger (2001b) suggests that a “shallow integration” approach to trade agreements may su¢ ce. Governments need not negotiate directly over domestic policies; instead, they may negotiate over tari¤s alone, provided that the market-access gains so achieved are secure. The important task for an international agreement in this context is then to ensure that negotiated market-access concessions are secure against opportunistic policy adjustments. Bagwell and Staiger (2001b) and Bagwell, Mavroidis and Staiger (2002) argue that current GATT/WTO rules, which focus on market access, can with some strengthening strike the right balance, so that governments can set e¢ cient domestic policies while pursuing international negotiations over tari¤s alone. Non-violation complaints and the preservation of market access In particular, GATT rules that permit non-violation complaints are a potentially attractive means of securing market- access concessions. A government may …le a non-violation complaint when a trading partner under- takes a policy change that nulli…es or impairs the market access gains that a government reasonably expected as part of an earlier negotiation. For example, following a tari¤ negotiation, the possibility of facing a non-violation complaint might deter a government from an opportunistic (terms-of-trade improving) relaxation in the production standards that it requires for an import-competing industry. Existing rules are insu¢ cient, however, to enable a government to adjust its policy mix following a tari¤ negotiation by raising its standards in the import-competing industry while also raising its import tari¤ so as to maintain its negotiated market-access commitment. Bagwell, Mavroidis and Staiger (2002) propose that a modi…cation to GATT rules of renegotiation, whereby a government could use a higher standard in an import-competing industry as compensation for a higher import tari¤, could in principle provide the needed ‡exibility. Ederington (2001) explores related themes in a model in which all international externalities travel through the terms of trade and any agreement on tari¤s and domestic policies must be self- enforcing. In his model, each government has two instruments – an import tari¤ and a domestic production tax –and the latter instrument has a legitimate role since domestic production generates a non-pecuniary externality that resides entirely within the country in which production occurs. Both policies a¤ect the terms of trade, and the challenge is to ensure that the e¢ cient policy mix is selected. In Ederington’s model, the e¢ cient domestic policy is a Pigouvian tax that o¤sets the domestic distortion, and the e¢ cient level of the domestic tax in fact is independent of the import tari¤ and thus the level of market access. Consistent with the themes developed above, Ederington shows that the most-cooperative solution is achieved when domestic policies are set at the e¢ cient (Pigouvian) level and import tari¤s are lowered so as to expand market access to the level that is as close to e¢ cient as possible before the self-enforcement constraint of the repeated game binds. The key intuition is that an e¢ cient domestic policy raises the discounted future value of cooperation, 54 which in turn enables governments to enforce lower tari¤ levels.81 The principle of national treatment Our discussion of the terms-of-trade implications of do- mestic policies to this point emphasizes the bene…t that the domestic country may enjoy when domestic production standards in an import-competing industry are relaxed in an opportunistic manner. As Staiger and Sykes (2011) argue, however, the bulk of WTO disputes concern instead cases in which foreign suppliers complain about standards that apply to their own products. Follow- ing a similar line of reasoning, we may anticipate a potential incentive for the domestic government to set standards on foreign products in an opportunistic fashion that could undermine the security of negotiated market-access gains for the foreign exporters. A key design feature of GATT/WTO rules that guards against such opportunism is the “national treatment” principle. This princi- ple, which is embodied in GATT Article III, the WTO Agreement on Technical Barriers to Trade (TBT), and the WTO Agreement on Sanitary and Phytosanitary Measures (SPS), restricts the ability of member governments to impose regulations on foreign suppliers that are more stringent than those imposed on domestic suppliers. To formally explore the e¤ectiveness of the national treatment principle, Staiger and Sykes (2011) adapt and extend the general insights of Bagwell and Staiger (2001b) to a product-standards setting. In the Staiger-Sykes model, the domestic government chooses trade policy as well as do- mestic tax and regulatory policy with respect to a product that is domestically produced and also imported. Regulatory policy has a legitimate role to play in their model, since domestic consump- tion generates a negative and non-pecuniary consumption externality that resides entirely within the domestic country. Staiger and Sykes show that governments of large countries indeed have incentive to impose discriminatory tax and regulatory policies against foreign imported products once import tari¤s are bound. The model thus provides an interpretation of the national treatment principle as a guard against such opportunistic behavior. Moreover, when product-speci…c consumption taxes are infeasible, they further show that the domestic government has an incentive to impose ine¢ - ciently stringent non-discriminatory product standards even in the presence of a national treatment clause, since foreign exporters bear some of the cost of achieving higher product standards that bene…t domestic consumers. In light of their …ndings, they conclude that the national treatment principle can play an important role in preventing tax and regulatory discrimination but leaves a potentially important role for the non-violation clause to address nondiscriminatory regulations that are excessively stringent.82 The national treatment principle has been interpreted and evaluated in other studies as well. Horn (2006) and Horn, Maggi and Staiger (2010) examine the national treatment principle with a 81 Lee (2007) considers a related model but with the important di¤erence that each government is privately informed about the magnitude of the domestic production externality in its country. As Lee shows, in this case it may be optimal to distort the tari¤ in order to limit the potential for disguised protectionism, which occurs when a government misrepresents its information by selecting a low production tax even though the externality cost is high. 82 Staiger and Sykes (2011) also observe that one of the …ndings reported by Broda, Limao and Weinstein (2008) and that we discussed in Section 3.2 above – that US non-tari¤ barriers are positively related to US market power over world prices –may re‡ ect limitations of the GATT/WTO national treatment and non-violation clauses to police terms-of-trade manipulation through behind-the-border measures. See also Staiger and Sykes (2013, 2015). 55 focus on domestic taxes rather than regulatory standards. Grossman, Horn and Mavroidis (2013) also provide an extensive study of the GATT national treatment provision and argue that case law, economic theory and the negotiation record all suggest that the purpose of the national treatment provision is to outlaw protectionist use of domestic policies.83 Moving beyond shallow integration? The case for shallow integration described above rests on the assumption that all international externalities are pecuniary and travel through the terms of trade. This case can be weakened, however, when governments possess private information, a point we discuss in Section 5.3. And as Antràs and Staiger (2012a, 2012b) argue, more complex forms of integration may be required in the presence of o¤shoring, which alters how prices are determined and complicates the nature of international pecuniary externalities. We discuss their work in Section 6.3. Similarly, “deeper”forms of integration may be needed if the trade agreement is created with the goal of also addressing non-pecuniary international externalities (e.g., global pollution).84 Limão’s (2005) model, which we discuss next, o¤ers one illustration of this point. Limão (2005) explores a model of self-enforcing cooperation among governments, with the new feature that production in the import-competing sector generates a negative non-pecuniary exter- nality that travels (at least to some degree) across national borders. Each government selects an import tari¤ and a domestic production tax, and both policies a¤ect the terms of trade. In Limão’s model, therefore, international externalities travel through the terms of trade as well as through a non-pecuniary channel. His …ndings illustrate that a form of “deep integration” is attractive to governments in such a setting as a means of relaxing enforcement constraints, where deep in- tegration in this context refers to a policy linkage whereby the potential for retaliation in both policies deters deviations in any one policy alone.85 In particular, Limão …nds that governments can achieve higher welfare in a self-enforcing agreement when the policies are linked; however, there is no guarantee that linkage raises the level of cooperation in each policy.86 In total, our survey of research in this section provides support for the shallow integration approach of the GATT/WTO when externalities are pecuniary and travel through the terms of trade. GATT/WTO rules concerning non-violation complaints and national treatment can be interpreted from this perspective and rest on a solid economic foundation. Anticipating discussion in later sections, we also note that the case for shallow integration can be weakened when governments possess private information, and arguments for deeper integration may also emerge in settings where international externalities travel through other channels. 83 Other related studies include Battigalli and Maggi (2003), Copeland (1990) and Costinot (2008). 84 For discussion of the extent to which GATT/WTO rules can be used to address non-pecuniary international externalities, see, e.g., Bagwell, Mavroidis and Staiger (2002) and Trebilcock, Howse and Eliason (2013, Chapters 17 and 18) and the references cited therein. 85 Retaliation in Limão’ s (2005) repeated-game model occurs o¤ the equilibrium path and thus carries the interpre- tation of a breakdown in cooperation with respect to the relevant policies. 86 See also Spagnola (1999a,b), who develops related themes for a distinct class of interdependent payo¤s. 56 4.5 Investment and Services The creation of the WTO in 1995 includes new agreements related to investment and services. In this section, we consider these new agreements and discuss economic research utilizing the terms- of-trade theory that interprets and evaluates the provisions contained therein. Foreign direct investment, local content, and international cross-ownership We start with the treatment of investment in the GATT/WTO, with an initial focus on foreign direct in- vestment (FDI). The past two decades have witnessed a signi…cant growth in FDI activity.87 This growth encourages consideration of the investment measures that host governments may impose and the appropriate treatment of those measures in the GATT/WTO. Investment measures inter- act with GATT rules when they have direct e¤ects on trade; in particular, local content, export and trade-balancing requirements may distort investment decisions and generate tension with basic GATT rules concerning national treatment and quantitative restrictions.88 As Trebilcock, Howse and Eliason (2013, Chapter 15) discuss, the WTO Trade-Related Investment Measures Agreement (TRIMs) builds on GATT rules to subject some measures with direct e¤ects on trade, such as local content requirements and quantitative restrictions, to explicit scrutiny under GATT norms. The appeal of additional restrictions on investment measures, however, is controversial, with some coun- tries expressing concerns about the broader implications of extensive investor-protection provisions in trade agreements. Investment has been removed as a topic for further discussion in the WTO Doha Round. A number of investment agreements have arisen, however, via Bilateral Investment Treaties (BITs) and as part of investor-protection provisions in PTAs. The purpose of restrictions on investment measures in a trade agreement can be developed at several levels. We mention two here.89 First, in the absence of a trade agreement that imposes restrictions on investment measures, a government might be tempted to impose measures favoring local input suppliers as a means of generating advantageous price changes for its country. Second, a trade agreement that appropriately restricts investment measures may also encourage e¢ ciency- enhancing FDI when the host government is otherwise unable to commit not to expropriate foreign rents once the foreign …rm has sunk costs. We brie‡y discuss the …rst purpose next and note that 87 FDI may be attractive to …rms for a variety of reasons. For example, FDI may provide access to cheap inputs, reduce trade costs and facilitate “tari¤ jumping.” FDI also may be advantageous relative to outsourcing as a means of maintaining tighter control over technology. At the same time, …rms considering FDI confront a variety of possible costs, including the possibility of rent expropriation via government policy changes after sunk costs are incurred. For a survey of research on multinational …rms, see Antràs and Yeaple (2014). 88 Local content requirements concern measures that require foreign-owned …rms to discriminate between domestic and imported goods that are used as inputs for production in the host country; export requirements concern measures that require exportation of a certain percentage of the foreign-owned …rm’ s output; and trade-balancing requirements concern measures that impose a relationship between what the foreign-owned …rm is allowed to import into the host country and the value of its exports. In some situations, trade-balancing requirements may be defended under GATT rules if they are necessary to address a balance-of-payment crisis. See Trebilcock, Howse and Eliason (2013, Chapter 15) for further discussion of the implications of GATT rules for the treatment of investment measures. 89 Another possible consideration is that restrictions on investment measures may play an e¢ ciency-enhancing role in managing unilateral policies designed to enhance technology spillover. A full analysis of the international externalities associated with such policies requires a comparison of the market-access bene…ts that …rms may enjoy via FDI with the technological knowledge that they may provide. 57 the second purpose is more directly associated with the “commitment theory”of trade agreements developed in Section 6. Bagwell and Sykes (2005) examine conditions under which a local content requirement generates advantageous price changes for the domestic (i.e. host) country. As they emphasize, a local content requirement is logically distinct from import tari¤s and quotas, since a local content requirement does not generate government revenue. They consider a simple two-country partial-equilibrium model in which a single homogenous input supplied in both the domestic and foreign country is used by foreign-owned …rms to manufacture a …nal good for sale in the domestic market. When markets are competitive and the domestic country is small in that its local content requirement does not a¤ect the world (i.e., foreign) input price, a domestic local content requirement raises the domestic input price, and thus redistributes domestic surplus and creates deadweight loss, but does not generate an international externality.90 If instead a foreign monopolist supplies the …nal good, then a domestic local content policy may redistribute surplus from the foreign monopolist to domestic input suppliers.91 Such a policy becomes more appealing to the domestic government when the foreign monopolist does not respond by signi…cantly reducing output. In turn, a signi…cant output reduction is less likely if the domestic country has market power (i.e., is large) so that its local content policy induces a fall in the foreign input price that o¤sets to some degree the rise in the domestic input price. The end result is that, for settings in which market power is present, local content policies may be unilaterally appealing to the domestic government and harmful to the foreign trading partner. From this perspective, restrictions on the use of local content requirements rest upon a solid economic foundation when market power is present. Blanchard (2010) explores a di¤erent aspect of the relationship between foreign investment and trade agreements. She does not focus on investment measures and rules that restrict such measures; instead, she considers the broad implications of general cross-border equity holdings for optimal tari¤s and the role of the GATT/WTO. Augmenting the two-country, two-good general equilibrium model of trade described above to include exogenous international cross-ownership, she identi…es the channels through which cross-border ownership impacts the optimal tari¤. The internal e¤ect is that a country has less incentive to maintain a high tari¤ in the presence of a larger degree of foreign ownership in the local import-competing industry, and the external e¤ect is that a government likewise has less incentive to raise its tari¤ for a terms-of-trade gain when its constituents hold a greater stake in the foreign export industry.92 Finally, the compositional e¤ect 90 Bagwell and Sykes (2005) assume that the local input supply function is upward sloping and that the …nal-good production technology exhibits constant returns to scale. In the case of a small domestic country and a competitive …nal-good market, a domestic local content requirement raises the domestic input price but has no impact on the foreign input price. Pro…t in the …nal-good sector is then likewise una¤ected, being zero in any case. The absence of an international externality in this setting implies that there is no obvious role for an international agreement on local content requirements. See Corden (1971), Grossman (1981) and Vousden (1990, Chapter 2) for further analyses of local content requirements in models with competitive markets and small countries. 91 See Brander and Spencer (1981) for analysis of a related model, in which an import tari¤ is used to extract surplus from a foreign monopolist. 92 As Blanchard (2010) notes, her analysis of optimal tari¤s under exogenous international cross-ownership gener- alizes and combines themes from previous theoretical work. 58 is that industry bias in ownership patterns may encourage a government to manipulate local prices to bene…t industries with a relatively higher proportion of national ownership. As Blanchard (2010) argues, consideration of cross-border ownership leads to interesting policy implications. An implication of the external e¤ect is that a country may welcome foreign invest- ment into its export sector as a means of encouraging a unilateral tari¤ reduction from its trading partner.93 Similarly, the internal e¤ect implies that foreign investment into an import-competing sector may encourage the host country to lower its import tari¤ as a means of extracting rent from foreign investors. Perhaps the most provocative implication of her analysis is that international ownership, by encouraging governments to liberalize their tari¤s unilaterally, may substitute par- tially (or even in some cases completely) for negotiated tari¤ reductions.94 Indeed, with a su¢ cient degree of international cross-border ownership, unilateral tari¤s are lower than e¢ cient, and the role of an international agreement is then to facilitate reciprocal and e¢ ciency-enhancing restric- tions in market access.95 Finally, Blanchard argues that the principle of reciprocity continues to serve as an important guide to e¢ cient outcomes, once the de…nition of market access re‡ects own- ership positions. More generally, the implications of cross-border ownership for the optimal design of GATT/WTO rules is an important subject that warrants further attention. Services We brie‡y consider next the treatment of services in GATT/WTO. A variety of evidence con…rms that services play an increasingly important role in modern economies.96 In a survey of research on services trade and policy, Francois and Hoekman (2010) indicate that the theoretical literature on services trade highlights the complementarity between international services trade and FDI, the implications of di¤erent market structures and national regulatory policies for services trade, and the way in which international service …rms are organized. They also describe increasing evidence that services liberalization is a major potential source of economic performance gains. Government policies that a¤ect international service …rms are disciplined in the WTO General Agreement on Trade in Services (GATS). As Trebilcock, Howse and Eliason (2013, Chapter 13, p. 480) indicate, the GATS is a “highly complex accord.” This agreement recognizes that services may be supplied through various modes and provides a framework for market-access negotiations across these modes. While GATS contains a general MFN provision (subject to some exceptions), national treatment and market access commitments apply only where WTO members make speci…c commitments to such coverage in their schedules. The impact of GATS on services reform is 93 See also Blanchard (2007) and Krishna and Mitra (2005). Blanchard and Matschke (forthcoming) provide some evidence to this e¤ect through an assessment of the impact of US multinational …rms’a¢ liate o¤shoring behavior on US tari¤ preferences. 94 While international ownership may lead in this way to lower tari¤s, Gulotty (2014) argues that greater interna- tional ownership does not similarly lead to reductions in regulatory barriers. He argues that regulatory barriers raise …xed costs, and that the associated reduction in entry may lead to net gains for e¢ cient, globalized …rms. See also Ethier (1998) for a di¤erent perspective under which regional agreements and associated foreign direct investment activity arise endogenously in response to multilateral liberalization. 95 In relation to the Bagwell-Staiger (1999a, 2002) model described above, a key di¤erence here is that international ownership operates via the external e¤ect to diminish the absolute value of Wp ew and Wpew and may even reverse the sign of these terms. 96 For further discussion, see Francois and Hoekman (2010) and Jensen (2011). 59 challenging to estimate. Francois and Hoekman (2010, p. 678) review the evidence to date and conclude that “the available, limited, evidence suggests that, with the exception of the EU, most services policy reform has been unilateral. The contribution of the GATS to services reform has been negligible.”97 Given the signi…cant potential gains from services liberalization, it is natural to ask why the reciprocity mechanism that underlies trade agreements has not played a greater role in achieving policy reforms in services. Francois and Hoekman (2010) review the literature relating to this question and identify a number of potential factors. Among these factors, we mention here one that is related to our discussion above: given that FDI is a signi…cant mode for supplying non-tradable s (2007, 2010) arguments imply that unilateral liberalization initiatives may services, Blanchard’ substitute to some degree for reciprocal trade liberalization through trade-agreement negotiations. More generally, as Antràs and Staiger (2012a, 2012b) argue, the rise of o¤shoring may have changed the nature of the international externality that a trade agreement must address, which suggests in turn that trade agreements may require additional restrictions for services policies that are associated with o¤shoring. The purpose and design of trade agreements for market settings with o¤shoring is an important direction for research, which we discuss in further detail in Section 6.3. Investment and services are of increasing importance in the international economy. The WTO includes agreements that place restrictions on measures that a¤ect investment and services, but the appropriate nature of such restrictions is controversial. Drawing on research that employs the terms-of-trade approach to trade agreements, we argue that restrictions on the use of local content requirements rest upon a solid economic foundation when market power is present. We also summarize research that indicates that international ownership may substitute to some degree for negotiated tari¤ reductions. Finally, as we note, recent research suggests a signi…cant potential for gains from services policy liberalization, even though the liberalization achieved to date through multilateral negotiations appears modest. To our minds, all of this points to a valuable role for future research directed toward understanding the impact of trade-related investment measures and services policies and, correspondingly, the appropriate design of WTO restrictions in this context. 4.6 The story line thus far... As interpreted through the lens of the terms-of-trade theory, the original 1947 GATT was created to solve the central economic problem of the day: the US Smoot-Hawley tari¤s of 1930 and the inter- national retaliatory response that followed had led to a terms-of-trade driven Prisoners’Dilemma with excessively high trade barriers. The task confronting governments was to set up an institu- tion that could work well to internalize the international (terms-of-trade) externality at the root of the high-tari¤ problem, and thereby induce governments to make the tari¤ choices they would have made had they not succumbed to the temptation of international cost-shifting (terms-of-trade 97 As Francois and Hoekman (2010, p. 678) note, countries that acceded to the WTO after 1995 tended to make more GATS commitments and represent a further exception, although care is required in assuming that GATS commitments are actually implemented. See also Eschenbach and Hoekman (2006) and Hoekman (2008). 60 manipulation) in the …rst place. If successful, GATT would lead necessarily to lower tari¤s and ex- panded market access from those countries and in those industries where signi…cant market power was present. But with signi…cant market power not universal in all countries and all industries, GATT would not lead to lower tari¤s from all countries and in all industries; and with the evident desire of governments to use trade policy for goals beyond that of national income maximization, GATT would also not be expected to lead to universal free trade. Finally, while some constraints on domestic policies would be needed to ensure that subsequent adjustments in those policies did not undo market-access commitments achieved via tari¤ restraints, GATT’s lack of any deeper integration beyond such constraints would not imply a weakness of the GATT system. The literature we have surveyed thus far lends broad support to the view that, at a fundamental level, governments succeeded with the GATT/WTO in creating an institution that is well-designed to solve the terms-of-trade problem. Many of the GATT/WTO’s core features appear sensible when interpreted in the context of this problem, and many of the outcomes negotiated within the GATT/WTO are broadly consistent with what might be expected from such an institution. The literature does, however, point out some potential di¢ culties with the GATT/WTO ap- proach, and these di¢ culties may account for some of the central challenges that the WTO confronts today. The evolution in the treatment of subsidies from GATT to the WTO is especially puzzling from the perspective of the terms-of-trade theory, both with regard to domestic subsidies and even more so with regard to the relatively severe treatment of export subsidies. Beyond the puzzling GATT/WTO treatment of subsidies, the literature emphasizes the pos- sibility of a serious free-rider problem under the MFN principle, and there is some evidence that this problem is signi…cant. To the extent that the principles of MFN and reciprocity together allow countries that negotiate reciprocal tari¤ cuts to appropriate the gains from their bargains and thereby keep free-riding to a minimum, the exemption from reciprocity granted to developing countries may ironically have kept these countries from enjoying to their full potential the bene…ts of GATT/WTO membership, again something that the evidence seems to bear out. This feature may in turn be contributing to a signi…cant “latecomer”problem for the Doha Round, as the Round grapples with how to better integrate developing and emerging economies into the world trading system when the major developed countries have already negotiated low average MFN tari¤s. Moreover, where there are signi…cant deviations from the MFN principle, such as can arise with the formation of PTAs, the literature points to these deviations as complicating international externalities beyond the simple terms-of-trade problem that the GATT/WTO seems well-designed to solve, suggesting in turn that the rise of PTAs could be creating di¢ culties for the GATT/WTO approach. Finally, while services trade and international investment ‡ows are of increasing im- portance to the global economy, the literature has developed only a nascent understanding of the international externalities associated with them, and so the ability of the GATT/WTO approach to function well in their presence is still an open question. 61 5 Evaluating the PTA Approach to Trade Liberalization We have described in the previous section how the GATT/WTO approach to liberalization derives broad support from the theoretical and empirical terms-of-trade literature. Does the terms-of- trade theory also support the view that PTA-driven liberalization can be seen as contributing to a solution to the terms-of-trade problem? In this section we review the relevant theoretical and empirical literature that speaks to this question. 5.1 PTAs, External Tari¤s and Multilateral Bargaining We begin with a focus on the impact of PTAs on the external (MFN) tari¤s of PTA-member countries, and ask: Might PTAs be seen to work in tandem with the tari¤ liberalization e¤orts of the GATT/WTO, or should PTAs be seen rather to work against these e¤orts? According to the terms-of-trade theory, non-cooperative Nash tari¤s are set ine¢ ciently high on products where the countries possess market power; hence, one way to shed light on this question is to assess the impact of PTAs on the non-cooperative external MFN tari¤s of the member countries on such products. If the formation of PTAs lowers these tari¤s, then it could be argued that PTAs work in tandem with the GATT/WTO’s own e¤orts to reduce these tari¤s, and that PTAs are hence building blocks for the needed multilateral liberalization that the GATT/WTO is also orchestrating. On the other hand, if the formation of PTAs raises these tari¤s, then PTAs would appear to increase the degree of multilateral tari¤ liberalization that is needed to reach the e¢ ciency frontier, and it could be argued that PTAs pose stumbling blocks to multilateral liberalization in this case. There is a large literature that evaluates the impact of PTA formation on the external non- cooperative MFN tari¤s of member countries. This literature has identi…ed several e¤ects of PTAs on external tari¤s, where PTAs may take the form of free-trade agreements (FTAs) or customs unions (CUs), with the key di¤erence being that, in addition to eliminating tari¤s on intra-union trade, CU members adopt a common external tari¤ policy toward the trade of nonmembers. Two of the e¤ects identi…ed by this literature operate to reduce the external tari¤s of PTA members: a “tari¤ complementarity e¤ect”that can take two forms and applies to FTAs and CUs, and a “rent destruction e¤ect” that applies to FTAs. Richardson (1995) identi…es a …rst tari¤ complementarity e¤ect: when an FTA is formed between countries that are competing importers of a common product from third countries, each FTA partner has an incentive to lower its external tari¤ on this product slightly below that of its FTA partners so as to increase its share of the tari¤ revenue collected on imports from outside the FTA, and this competition for tari¤ revenue between FTA partners can lead to a downward spiral in their external tari¤s. Bagwell and Staiger (1999b) identify a second tari¤ complementarity e¤ect: in a competing-exporter world where each FTA or CU member competes with non-members for exports to other members, when FTA or CU members reduce their tari¤s to zero on imports from one another, the resulting trade diversion (i.e., the reduction in imports by members from non-members) encourages members to lower the 62 tari¤s that they apply on imports from non-members.98 The rent-destruction e¤ect is highlighted by Ornelas (2005a,b,c 2008) and is also a force for lower external tari¤s among FTA members. In a setting where special interest lobbies push for tari¤ protection, Ornelas shows that the rents from the external tari¤s of one FTA member country spill over to producers in FTA partner countries, creating a free-rider problem for national lobbies within the FTA which interferes with their ability to obtain high external tari¤s from their governments. Two further e¤ects operate to increase the external tari¤s of PTA members, but operate only for CUs. A “market power e¤ect” (see Kennan and Reizman, 1990, Krugman, 1991, Bond and Syropolous, 1996a,b, Bagwell and Staiger, 1997b and 1999b, and Cadot et al, 1999) arises when CU members are competing importers of a common product, and can collectively exert more market power on the world price of that product with their common external tari¤ than they could individually. A separate “coordination e¤ect” (see Kennan and Reizman, 1990) operates to raise the external tari¤s of CU members even when countries are competing exporters: when one CU member raises its external tari¤ on a product that it imports from third countries, other CU members that export that product to the …rst CU member gain as they receive higher prices for their exports to the …rst CU member, and this is a positive externality of higher external tari¤s that can be internalized among CU members when they set their common external tari¤ policy. Of course, while the impact of PTA formation on non-cooperative Nash external MFN tari¤s is suggestive of the nature of the relationship between PTAs and the GATT/WTO, it provides at best an incomplete picture of this relationship. First, it is not clear that the impact of PTAs on non-cooperative MFN tari¤s translates – even with the same sign – to the impact on cooperative MFN tari¤s. For example, Limão (2007) shows that an FTA that pursues non-trade objectives can result in higher cooperative MFN tari¤s, in circumstances where FTA partners agree to provide non-trade concessions to a country in exchange for preferential market access rents created and maintained by the high external tari¤s of the country. In e¤ect, Limão demonstrates that the FTA s external MFN partners can become a force pushing against negotiated reductions in the country’ tari¤s (and hence their rents). And Bagwell and Staiger (1997a, 1997b, 1999b) show that when self-enforcement constraints bind in a multilateral agreement over external tari¤s, the formation of FTAs and CUs can have impacts on the most-cooperative MFN tari¤s achievable that are the opposite of the impacts on the Nash external tari¤s, and that vary through time if the PTAs are implemented in stages.99 Second, recall that the terms-of-trade theory directs attention to the question whether PTAs reduce the degree of ine¢ cient terms-of-trade manipulation embodied in the external tari¤s of PTA partners, and the results surveyed above are not always presented with 98 Other papers featuring a tari¤ complementarity e¤ect in various settings include Bond, Riezman and Syropoulos (2004), Bond, Syropoulos and Winters (2001), Freund (2000a), Syropoulos (1999) and Yi (1996). 99 Other work that considers the e¤ects of exogenous PTAs on self-enforcing multilateral tari¤ cooperation includes Bond and Syropolous (1996b) and Saggi (2006). In additional related work, Levy (1997) and Krishna (1998) argue that FTAs can erode the political support for further agreements to reduce MFN external tari¤s and thereby act as stumbling blocks to multilateral trade liberalization. A literature also exists that examines the impact of multilateral liberalization on the formation of PTAs. Using di¤erent frameworks, Ethier (1998) and Freund (2000a) argue that PTAs may be a response to successful multilateral liberalization. 63 this question in mind.100 Finally, while a reduction in the external MFN tari¤s of PTA members triggered by the formation of the PTA might be viewed as partially solving the terms-of-trade problem and thereby making the remaining task easier for the GATT/WTO, Bagwell and Staiger (1999a, 2001a) show that the introduction of PTAs and the violation of MFN that this implies can change (does change for FTAs, can but need not change for CUs) the nature of the problem that a trade agreement must solve, from a simple terms-of-trade problem to a more complicated problem in which international externalities also travel through local prices. Bagwell and Staiger s approach to multilateral thus argue that PTAs are inherently at odds with the GATT/WTO’ trade liberalization, which seems best-suited to address simple terms-of-trade problems. In any case, with these various e¤ects identi…ed and pointing in di¤erent directions, it is clear that theory alone cannot resolve the issue of the impact of PTA formation on the external MFN tari¤s of member countries.101 We therefore turn to the empirical literature on this question. What is the impact of PTA tari¤ liberalization on subsequent e¤orts toward multilateral tari¤ liberalization?102 Limão (2006) provides a …rst product-level investigation into whether PTAs are stumbling blocks or building blocks for multilateral liberalization. His approach involves a comparison of two di¤erent types of products – those for which a country has positive imports from PTA partners and those for which it only imports from PTA non-partners. An examination of subsequent US multilateral tari¤ changes made as a result of the Uruguay Round provides evidence that the US granted smaller MFN tari¤ reductions in products with positive US imports from PTA partners. The evidence applies not only to products imported from large PTA partners, such as the countries in NAFTA, but also to imports from smaller PTA partners. Given that even those smaller US PTA partners export in nearly 15 percent of product lines, and that these products also have positive levels of imports from non-PTA partners, a further implication is that even small US PTAs were a stumbling block to the multilateral liberalization taking place under the Uruguay Round (consistent with the model of FTAs with non-trade objectives in Limão, 2007). In addition to the US, another important environment to conduct such an exercise is the EU. Karacaovali and Limão (2008) …rst con…rm the evidence found for the US by showing that the EU cut MFN tari¤s on products not imported from PTA partners by nearly twice as much as it cut tari¤s on products imported from partners during the Uruguay Round. Furthermore, the size of the EU stumbling block e¤ect is larger for the products that are exported by more PTA partners. Second, they exploit additional margins of the data on the EU PTAs by grouping PTA partners based on whether they eventually acceded to the EU between the ends of the Tokyo Round and Uruguay Round of negotiations. They …nd evidence consistent with theoretical predictions that accession countries should not trigger stumbling block e¤ects,103 and that the stumbling block e¤ect 100 For example, the rent destruction e¤ect identi…ed by Ornelas (2005a,b,c 2008) does not have a clear prediction related to this question. 101 Bagwell and Staiger (1998) and Freund and Ornelas (2010) reach a similar conclusion. 102 See Freund and Ornelas (2010) for a more extensive survey of this empirical literature. 103 According to the theoretical model of Karacaovali and Limão (2008), EU accession countries would be eligible to receive a transfer or revenue collected under the common external tari¤ which o¤sets the potential loss in intra-PTA trade that they would su¤er as a result of additional multilateral tari¤ reduction. 64 is only associated with products from countries with which the EU had PTAs in place at that time. Changes to multilateral tari¤s need not only take place in the context of GATT/WTO negoti- ating rounds. It is natural, however, to analyze tari¤ changes achieved through negotiation rounds when considering the US and EU during the recent period in which detailed tari¤ data are avail- able. This is true for two reasons. First, for most products, US and EU applied MFN tari¤s are relatively close to their legal tari¤ bindings so that the tari¤s cannot be increased without violating multilateral commitments. Second, US and EU tari¤s were relatively low to begin with during this period, and so there is also not much scope for downward variation in the form of additional uni- lateral reductions. However, these two conditions do not apply to a number of developing countries in the international trading system which had both i) su¢ ciently high applied MFN tari¤s at the time of PTA implementation to allow for the possibility of meaningful unilateral reductions, and ii) legal binding commitments su¢ ciently above their applied MFN rates to allow applied rates to legally increase as well without violating these commitments. Estevadeordal, Freund and Ornelas (2008) exploit these features of the data for 10 Latin Amer- ica countries over 1991-2000. They assess patterns of applied MFN tari¤ changes following the implementation of PTA tari¤ reductions, under agreements like MERCOSUR, the Andean Com- munity, and other intra-Latin American PTAs formed in the decade. These countries exhibit, on average, a positive relationship between changes in preferential tari¤s and subsequent changes to s preferential agreements worked as a building applied MFN tari¤s –evidence that Latin America’ block toward unilateral MFN liberalization during this period. Higher shares of intra-PTA imports are also associated with reductions in applied MFN tari¤s even for relatively small preference mar- gins. However, the results are limited to the free trade areas and do not hold for Latin America’s trade agreements that were formulated as customs unions during that period, such as MERCOSUR. In a related approach, Calvo-Pardo, Freund, and Ornelas (2011) assess multilateral tari¤ changes following the preferential tari¤ reductions associated with the creation of the ASEAN Free Trade Area in 1992. ASEAN is another setting in which preferential liberalization led to applied MFN tari¤ cuts, with MFN tari¤ cuts found to be larger in products with larger preference margins and thus a greater scope for trade diversion.104 What are the potential explanations for the di¤erences across settings? One potential contribu- tor is government policy responsiveness to the threat of trade diversion. The Estevadeordal, Freund and Ornelas (2008) and Calvo-Pardo, Freund, and Ornelas (2011) settings resulted in original PTA liberalization that led to large preference margins (relative to pre-PTA applied MFN rates). Con- sistent with the tari¤ complementarity e¤ect, economically costly trade diversion could have arisen if governments did not subsequently also lower their applied MFN tari¤s. The US and EU en- 104 Tovar (2012) also studies how developing countries make unilateral changes to applied MFN tari¤s after the formation of a PTA. She examines four countries after the formation of the CAFTA-DR in 2004. The results for CAFTA-DR are di¤erent than the earlier studies as they suggest at least an initial stumbling block e¤ect. Countries increased (or decreased by less) MFN tari¤s on products that had previously been subject to larger preferential tari¤ reductions in the …rst two years after PTA implementation, and this is somewhat o¤set in the subsequent two year period during which the countries reduced their MFN tari¤s. An interesting feature of this evolving e¤ect is that it is consistent with the pattern expected under an FTA according to the model of Bagwell and Staiger (1997a). 65 vironments, on the other hand, subject to the Limão (2006) and Karacaovali and Limão (2008) studies, resulted in preference margins that were much smaller with potentially less scope for trade diversion. Second, the US and EU preferences were more unilateral in nature. The theory in Limão (2007) emphasizes the non-reciprocal nature of US and EU preferences and that they were o¤ered as compensation for countries that took up non-trade obligations in areas such as environmental or labor standards, intellectual property rights protection, and supporting the war on drugs. An open question for research is whether building block e¤ects may be more likely to dominate in reciprocal PTAs.105 The rising importance of “WTO-extra”provisions in PTAs (Horn, Mavroidis, and Sapir, 2010) suggests that this should be a priority area for additional research. Preferential tari¤s and MFN tari¤s are certainly not the end of the line when it comes to trade policy, as there are a number of other potential non-tari¤ barriers to trade. The GATT/WTO provides several exceptions that countries can invoke to implement higher levels of protection for legitimate environmental, health, or other public safety concerns, for example under Article XX. Furthermore, most of the same major economies involved in multilateral and preferential trade lib- eralization since the late 1980s are also major users of antidumping and safeguards (Bown, 2011a), another major category of GATT/WTO exceptions allowing countries to temporarily implement higher levels of import protection under certain conditions.106 While the general relationship be- tween PTA liberalization and non-tari¤ barriers use is not yet well understood, here we highlight a recent paper that initiated the investigation of this relationship. Antidumping is the most frequently applied TTB (temporary trade barrier) policy in use across countries and over time since the 1980s. Prusa and Teh (2010) use a cross-country sample involving 80 PTAs and antidumping use dating back to 1980. While there is only a modest impact of PTA formation on the overall use of antidumping, after controlling for other aggregate-level determinants, there is evidence of important di¤erences in policy treatment between PTA partners and non- partners. PTA implementation is associated with both a reduction in antidumping actions against new PTA partners and an increase against non-partners. Prusa and Teh attribute some of this post-PTA change in behavior to PTA variation in the legal provisions a¤ecting antidumping use - i.e., the sort of “WTO-plus”provisions described in more detail in Section 2.2. In any event, these …ndings suggest that reorientation of TTBs toward the imports of non-PTA partners could be an important avenue by which PTAs lead to rising external tari¤s. Finally, in addition to the literature we survey above, an active literature models the choice countries face between PTA formation and multilateral trade bargaining.107 That literature is 105 Recent work by Mai and Stoyanov (forthcoming) provides initial evidence on this topic. They examine the e¤ect of CUSFTA on Canadian trade policy. Consistent with the tari¤ complementarity e¤ect, they …nd that the CUSFTA led to declines in Canadian external tari¤s. But they also …nd that Canadian external tari¤s declined more slowly in industries that generated more revenue for US exporters to Canada. This latter e¤ect suggests an attempt by Canada to limit preference erosion in industries of particular importance to its PTA partner (i.e., the US). 106 Bown, Karacaovali, and Tovar (2015) provide a more general survey on the relationships between PTAs and the use of temporary trade barriers such as antidumping and safeguards actions. 107 Other related research considers the endogenous network of stable PTAs when multilateral liberalization is not included as an option. See Goyal and Joshi (2006), Furusawa and Konishi (2007), Mrázová, Vines and Zissimos (2013) and Yi (1996). Another related strand of the literature considers endogenous formation of CUs using the core 66 concerned with the question whether global free trade is more or less likely to be achieved when PTAs are available as an alternative to multilateral tari¤ negotiations. Guided by the terms-of- trade theory, our focus here is on a related but distinct question, namely, whether PTAs contribute to (building blocks) or interfere with (stumbling blocks) the ability of multilateral negotiations to achieve (globally) e¢ cient policy outcomes when judged against the governments’own preferences. With regard to this statement of the stumbling block/building block question, the …ndings of this related literature can be summarized as follows. First, when there are no bargaining frictions, as is assumed by most of the literature, e¢ ciency would always be achieved under multilateral negotiations if PTAs were banned, and so PTAs can’t possibly be building blocks in the sense we are interested in here. In this case PTAs can facilitate the attainment of global free trade, but only when global free trade does not mark a Pareto improvement over the outcome that would be delivered if PTAs were banned (see, e.g., Aghion, Antràs and Helpman, 2007, and Saggi and Yildiz, 2010).108 Nevertheless, in this no-bargaining-frictions case, PTAs can be stumbling blocks to Pareto e¢ cient outcomes under certain conditions (see, e.g., Aghion et al., 2007, Seidman, 2009, and Saggi, Yildiz and Woodland, 2013). Second, when there are bargaining frictions (as in McCalman, 2002, and McLaren, 2002) so that a building block role for PTAs is possible, no such building block role has been found, but a stumbling block role has again been shown to be possible. Hence, while these papers show that PTAs can serve as a building block for the attainment of global free trade, if anything this branch of the literature reinforces the view that PTAs should be viewed with some caution from the perspective of e¢ ciency when judged against the governments’own preferences. 5.2 PTAs and Third-Country Externalities According to the terms-of-trade theory, the purpose of a trade agreement is not to secure free trade, but to allow governments to internalize the terms-of-trade externalities associated with their tari¤ choices. As we discuss in Section 4.1, the GATT/WTO appears well-equipped to help governments internalize terms-of-trade externalities, in part through its norms of reciprocity and MFN which can help to keep the terms-of-trade consequences for third countries to a minimum when subsets of coun- tries negotiate tari¤ cuts. PTAs, by de…nition, deviate from the MFN norm, raising the possibility that, rather than contributing to a solution to the terms-of-trade problem, PTAs are a surviving vehicle for imposing terms-of-trade externalities on third countries within the GATT/WTO sys- tem. Under this possibility, some PTAs may be viable, in the sense that their member governments support their formation, only because they have been able to use discriminatory tari¤ cuts be- as the solution concept. See, for example, Riezman (1995, 1999) and the survey by Kowalczyk and Riezman (2011). 108 This is a point also made by Maggi (2014). The attainment of global free trade does not lead to a Pareto improvement in Aghion et al. (2007) because governments are assumed to maximize something other than real national income in the relevant case; and in Saggi and Yildiz (2010) this is so because international lump sum transfers are assumed to be unavailable. We note, however, two quali…cations. First, as Freund (2000b) argues in an oligopoly context, the path by which global free trade is achieved may matter. In her model, world welfare is higher when global free trade is achieved through expanding preferential agreements rather than through multilateralism. Second, and as we discuss brie‡ y in the next section (see note 122), a possible commitment theory interpretation of these …ndings could suggest that PTAs enhance e¢ ciency when viewed from an ex-ante perspective. 67 tween them to impose negative terms-of-trade externalities on third countries and convert those third-country losses into their own gains.109 This possibility is emphasized by Bagwell and Staiger (2005a), and it provides one reason according to the terms-of-trade theory why the proliferation of PTAs could re‡ect a development which is ine¢ cient from a global perspective. A necessary feature for PTAs to impose negative terms-of-trade externalities on third countries is that the discriminatory market access granted to PTA partners diverts trade volumes that would otherwise have occurred between PTA member countries and third countries. This trade diversion e¤ect of a PTA is the trade volume reduction that can lead to changes in trade prices with third countries, and hence to third-country terms-of-trade impacts. A number of papers have emphasized the likelihood that PTAs rely on substantial trade diversion in order to keep them viable (see, for example, Grossman and Helpman, 1995b, and Krishna, 1998; and see Ornelas, 2005a,b,c for quali…cations to this claim). The possibility described above is a particular version of this claim, in which the third-country terms-of-trade externality associated with trade diversion is the mechanism by which an otherwise nonviable PTA is kept viable. What is the evidence regarding the importance of third-country externalities imposed by PTAs? The evidence is mixed, with some studies …nding substantial trade diversion and terms-of-trade impacts of PTA formation on third countries and other studies …nding only insigni…cant e¤ects. Chang and Winters (2002) take up the international externality question by investigating the experience of third-country export prices to Brazil in light of MERCOSUR. Relying on product- level unit values data to proxy for export prices, they …nd that intra-PTA tari¤ reductions are empirically associated with the price declines of third-country (Chile, Korea, Japan, US) exports to Brazil, relative to the prices of these third-country’s exports of the same products to the rest of the world.110 Furthermore, welfare calculations arising from the model’s estimates indicate PTA non-partner countries such as the US and Germany experienced sizeable welfare losses due to the price declines, even after taking into account the e¤ects of Brazil’s subsequent MFN tari¤ reductions in many of the same products. Winters and Chang (2000) present a similar approach by examining s 1986 EC accession on US and Japanese exports to Spain. They argue that the impact of Spain’ these earlier results are not as strong due to methodological and data issues, including the reliance on data at higher levels of aggregation. Nevertheless, results from this study are consistent with the Chang and Winters (2002) evidence from MERCOSUR. In particular they …nd that each 1 percent preferential Spanish tari¤ cut toward new PTA members is associated with roughly a 0.5 percent export price decline for Japanese and US exporters to Spain, relative to these new PTA partners’(France, Germany, Italy, UK) export prices of the same good. That important negative terms-of-trade externalities arise after PTA formation is consistent with some, though certainly not all, of the evidence arising from other studies. Romalis (2007), for example, found that the EU’s trade with the US, Canada, and Mexico was negatively impacted by 109 Note that a PTA can generate a negative terms-of-trade externality for third countries even when a (modest) tari¤ complementarity e¤ect is present, since third countries receive smaller tari¤ cuts than do member countries. 110 In related work, Schi¤ and Chang (2003) …nd that even the threat of duty-free exports from Argentina into Brazil resulted in price declines for US exports into Brazil. 68 the implementation of NAFTA, con…rming signi…cant trade diversion e¤ects, but found little in the way of third-country price impacts associated with these trade volume reductions. Both Romalis (2007) and Clausing (2001) found only insigni…cant trade diversion e¤ects from the Canada-US FTA, a result consistent with the analysis of Tre‡er (2004), who found that Canada’s trade creation associated with CUSFTA dominated the welfare e¤ects of any trade diversion. Using data on the manufacturing trade and FTAs for 64 countries over the period 1990-2002, Dai, Yotov and Zylkin (2014) report large trade diversion impacts of FTA formation, with the largest third-country impacts su¤ered by existing FTA members when a country joins a new FTA from which its other FTA partners are excluded. On the other hand, Frazer and Van Biesebroeck (2010) found no evidence that US preferences under the African Growth and Opportunity Act (AGOA) drew African exports in those products away from the EU in ways that may have a¤ected EU consumer prices.111 A potential limitation of this literature is suggested by the results of Handley (2014) and Handley and Limão (forthcoming) which we have discussed at various points in the survey. They show that in the presence of policy uncertainty there can be large di¤erences between the trade e¤ects of, on the one hand, an applied MFN tari¤ of zero that is bound at a much higher level in the WTO, and on the other hand a PTA tari¤ that is both applied and bound at zero. Such di¤erences are likely to be missed in studies such as those above that attempt to relate trade e¤ects of PTAs to preference margins on applied tari¤s alone. Also relevant is the paper by Prusa and Teh (2010) discussed previously: their …nding that PTA formation is associated with both a reduction in antidumping actions against PTA partners and an increase against non-partners suggests a form of third-country externality that may be di¢ cult for typical studies of the trade impacts of PTAs to measure. Finally, in recent work, Spearot (2015) develops a general equilibrium version of the Melitz and Ottaviano (2008) model which allows for di¤erences in …rm productivity distributions across supplying countries, and shows how the model can be structurally estimated with product level bilateral trade and tari¤ data and used for counterfactual experiments. Of particular interest to our discussion here is Spearot’s counterfactual calculations of the impacts of a country unilaterally eliminating all of its (post-2000) remaining tari¤s. As Spearot notes, if existing PTA-induced tari¤ discrimination causes su¢ cient trade diversion, even a large country such as the US with low MFN tari¤s could gain in terms of real national income by eliminating its remaining MFN tari¤s. And this is exactly what Spearot …nds for the US (and others including India, Japan and Korea). Relatedly, Caliendo and Parro (2015) and Tintelnot (2014) build on the Eaton and Kortum (2002) framework to quantify the trade impacts of NAFTA (Caliendo and Parro) and the trade and production impacts of the proposed Canada-EU trade and investment agreement (Tintelnot). Caliendo and Parro …nd small price and quantity impacts of NAFTA on the rest of the world, while Tintelnot …nds larger potential third-country impacts of the proposed Canada-EU agreement. 111 Furthermore, there is also mixed evidence on international externalities associated with the application of other discriminatory trade policy, such as antidumping. Bown and Crowley (2006, 2007) …nd international externalities associated with US antidumping imposed on Japanese exports via the trade volume (“trade de‡ ection”) and price (third-market terms-of-trade) e¤ects on its exports of those products to third markets. However, Bown and Crowley (2010) investigate similar trade restrictions imposed on a developing country exporter (China, during 1992-2001) and do not …nd evidence of trade de‡ ection to third countries in that setting. 69 Tintelnot’s …ndings are especially relevant for assessing the potential of recent PTAs to impose third- country externalities, as he focuses on the deep-integration features of the proposed agreement and their impacts on the location of multinational production facilities for export platform purposes. Taken together, we interpret the literature as indicating that the potential for PTAs to impose important third-country externalities is real and has at times probably been exercised, but that it does not appear to be a pronounced and consistent feature of existing PTAs. It is possible that PTAs have mostly not imposed signi…cant third-country e¤ects at least in part because of the role of the GATT/WTO dispute procedures in policing such e¤ects, a possibility that is given some credence by the literature we survey in Section 7.3. In any event, the literature surveyed above establishes that the potential for third-country externalities is there, and suggests that the problem could become more substantial with the increasing focus of new PTAs on deep integration. Maintaining a cautious view of PTAs in light of this potential seems warranted. 5.3 PTAs and Deep Integration We consider now the growth in deep-integration PTAs. Does the terms-of-trade theory provide support for this development? We describe above how the theory can provide support for shallow integration; this is the basic message of Bagwell and Staiger (2001b)’s claim that a well-working market-access-preservation rule can allow countries to achieve e¢ cient outcomes through tari¤ ne- gotiations without directly negotiating over domestic policies. This message survives a variety of generalizations of the original model in which it was made, but requires quali…cation when govern- ments possess private information.112 In Bajona and Ederington (2011), the private information takes a “hidden-action” form and is over the degree of a government’s intervention with domestic policies. The case for shallow integration then survives largely intact, albeit with some modi…ca- tion: an e¢ cient self-enforcing agreement generally takes the form of a minimum market-access level combined with a binding tari¤ cap. But, as Lee (2014) shows, if private information takes a s type (e.g., the magnitude of a domestic “hidden-information” form and concerns a government’ distortion associated with an externality), and if the type is uncertain at the time the agreement is written, then a simple market-access-preservation rule leads to excessive protection for some types and is not optimal. Furthermore, while a state-contingent market-access-preservation rule could implement the …rst-best allocation were the state observable, such an allocation is not incentive compatible when information is hidden. In the hidden-information setting, Lee shows that a form of deep integration is needed to construct the optimal agreement. Hence, the terms-of-trade literature suggests that the case for shallow integration can be weak- ened to the extent that private (and perhaps especially hidden) information is important in trade 112 Sauré (2014) argues that quali…cations to the case for shallow integration also arise when the agreement must be self-enforcing and there exist inter-temporal linkages that can a¤ect the deviation payo¤, so that the extent to which an agreement is self-enforcing may vary with policies, even among policies that generate the same market access. Sauré shows that the freedom over domestic policies provided under shallow integration then might induce a government to adjust its policies in a manner that ultimately changes its own gains from deviation in an unanticipated way and renders the new policies unsustainable. It is an open question whether such adjustments would be exercised in related models were governments to anticipate the consequences for their own self-enforcement constraint. 70 negotiations. While more research is clearly needed, the possibility is thus raised that governments might achieve e¢ ciency gains in trade agreements that include some deep integration rules. But granting this possibility, what form would such rules take? Could they be provided via modest adjustments to WTO rules, or are deeper forms of integration required? And, if the latter, do PTAs represent the best path forward? Or, might instead deep integration initiatives in PTAs generate third-party externalities and be ine¢ cient from a multilateral perspective? To date, the terms-of-trade literature has not gone far enough to sort through these possibilities and provide answers. But there is suggestive evidence from a small empirical literature focusing on the trade e¤ects of regional integration of TBT and SPS measures that does point to the potential for important trade-diverting e¤ects of these kinds of agreements, with particular harm to the exports of developing countries not party to the agreements (see, in particular, WTO, 2012, p. 152).113 We therefore see providing answers to these questions as an important area for future research. At this point, given the broad a¢ nity between the GATT/WTO design and solutions to the terms-of-trade problem, we view the current terms-of-trade strand of the trade-agreement literature as suggesting a presumption that any deep integration that would be required to achieve e¢ ciency (say, due to the presence of private information) is likely better provided within the GATT/WTO than by PTAs.114 But we also see signi…cant value in further research that explores whether certain forms of deep integration might exist that are achieved most easily among smaller groups of countries at similar developmental stages and that impose little or no costs on third countries. As Maggi (2014) also emphasizes, future research of this kind might explicitly include bargaining frictions, which may be higher for negotiations that involve many countries and complex issues. For such forms of deep integration, the corresponding agreement might be well placed in a PTA or, alternatively, in a plurilateral or critical-mass agreement within the WTO.115 5.4 The story line continued... As WTO liberalization e¤orts seem to have stalled, PTA liberalization has taken o¤. Has the explosion of PTAs interfered with the WTO’s ability to deliver countries to the global e¢ ciency frontier? Or are PTAs carrying countries to the global e¢ ciency frontier in ways that the WTO could not? Or, are PTAs succeeding where the WTO could not because PTAs can impose costs on third countries that WTO rules successfully internalize, in which case PTAs are likely moving the 113 See, for example, Disdier, Fontagne and Cadot (2015), who …nd evidence that regional agreements between developed and developing countries that focus on harmonizing standards tend to reduce the exports of the developing country members to third countries, and Chen and Mattoo (2008), who …nd that regional agreements to harmonize standards can reduce developing country exports from outside the region. The …ndings of these studies complement the …ndings of Tintelnot (2014) that we discuss above regarding the potential for deep integration PTAs to impose third-country externalities. 114 Even in the private-information models, the underlying problem is still the terms-of-trade problem, and so Nash domestic policies are set e¢ ciently, with the ine¢ ciency amounting simply to a level of market access that is too low. From this perspective, it is not obvious why PTAs would be better than the WTO at providing the degree of deep integration necessary to achieve e¢ cient market access levels in the presence of private information. 115 We discuss plurilateral and critical mass agreements further in the concluding section, where we consider potential approaches for strengthening the WTO. 71 world away from the global e¢ ciency frontier? Our survey of the terms-of-trade literature suggests a cautious interpretation of the bene…ts of PTAs to the world trading system. As we have emphasized, this literature provides broad support for the view that the GATT/WTO is fundamentally well-designed to minimize the in‡uence of terms-of-trade externalities on the policy choices of member governments and thereby solve the terms-of-trade problem. The literature provides a more mixed view of PTAs in this regard, with theory pointing out many avenues through which PTAs could permit terms-of-trade externalities to re-enter the calculus of trade-policy making, and empirical evidence providing only partial assurance that these avenues have not been exercised. In this sense, the terms-of-trade literature supports a cautious view of the wisdom of entrusting PTAs with the rules of globalization. The literature also provides ample reasons for caution concerning the position that PTAs are complementary to the GATT/WTO. Both the theory and evidence are mixed; hence, as a general matter, whether PTAs are stumbling blocks or building blocks for multilateral liberalization remains ambiguous. 6 Beyond the Terms-of-Trade Theory We now survey the literature on the commitment, delocation/pro…t-shifting and o¤shoring theories of trade agreements. Our purpose is to identify insights that would qualify the answers given by the terms-of-trade theory to the questions that motivate our survey. 6.1 The Commitment Theory The most established alternative to the terms-of-trade theory of trade agreements is the commit- ment theory. According to this theory, governments value trade agreements as a way to tie their hands against their own lobbies and citizens. Of course there is no reason why trade agreements couldn’t serve multiple purposes, which is to say the commitment and terms-of-trade theories need not be mutually exclusive. The question for us here is whether the commitment theory o¤ers a more supportive interpretation of PTAs relative to the GATT/WTO than does the terms-of-trade theory and, if so, whether there is su¢ cient empirical support for the commitment theory more generally to qualify or alter the initial conclusions we have drawn concerning the relative merits of PTAs and the GATT/WTO from our survey of the terms-of-trade theory literature. While expressions of the commitment theory of trade agreements can be found in a variety of early papers (see, for example, Carmichael, 1987, Staiger and Tabellini, 1987, Lapan, 1988, Matsuyama, 1990, Tornell, 1991 and Brainard, 1994), a particularly elegant treatment that has become the workhorse model of this idea is provided by Maggi and Rodriguez-Clare (1998). Their model is one of a small open economy, where the terms-of-trade argument for trade agreements is absent. The focus of the model is on the idea that an anticipated trade-policy-lobbying relationship between a government and producers in some sector is likely to distort the equilibrium allocation of resources in the economy toward that sector, and on the possibility that the lobby might then not fully compensate the government for this distortion. To formalize this possibility, Maggi and 72 Rodriguez-Clare extend the lobbying model of Grossman and Helpman (1994) to include a prior stage in which resources in the economy are allocated across sectors. They con…rm that the gov- ernment is compensated by the lobby for the ex-post distortions its trade-policy choice imposes on the economy given the sectoral allocation of the economy’s resources that are sunk at the time the trade policy choice is made; this …nding is the same as in the original Grossman and Helpman model. But Maggi and Rodriguez-Clare establish that the lobby does not compensate the govern- ment for the ex-ante distortions in the sectoral allocation of resources created by the anticipation of the government’s relationship with the lobby. This second …nding is novel, and as Maggi and Rodriguez-Clare show, it provides a reason that the government might (under certain conditions which they explore) wish to tie its hands ex ante against in‡uence by the lobby ex post. And in this way, a possible commitment role for a trade agreement is thereby identi…ed.116 The commitment theory has been used to o¤er interpretations of some of the features of the GATT/WTO which appear puzzling when viewed through the lens of the terms-of-trade theory. One example is Potipiti (2012), who uses the commitment theory to explain why, in the WTO, tari¤s are the subject of negotiated limits while export subsidies are banned outright. From the perspective of the standard terms-of-trade model, and as discussed in Section 4.3, this feature is puzzling at two levels: the standard theory suggests that, if anything, export subsidies should be encouraged, and it also does not rationalize treating export subsidies more severely than import tari¤s. Potipiti shows that these puzzles can be resolved in the context of the commitment theory, once an asymmetry between the rents earned by import-competing and exporting interests is introduced. Potipiti (2012) builds on the small-open economy model of Maggi and Rodriguez-Clare (1998). s version of this model, a government can join an agreement that bans tari¤s and/or In Potipiti’ an agreement that bans export subsidies, and doing so eliminates the anticipation of protection by the private sector and the associated ex ante investment distortion, and thereby generates a social welfare gain. Commitment to such an agreement, however, also requires the government to forfeit the political contributions it would otherwise collect for the protection it o¤ers. The government thus faces a trade-o¤, and it commits to a trade agreement covering a particular policy only if the social welfare gain from banning the use of that policy is greater than the government’s valuation of the associated loss in political contributions. Applied to export policies, the underlying Maggi and Rodriguez-Clare model can therefore account for an agreement that discourages (bans) export subsidies. But how can the asymmetry in treatment across import tari¤s and export subsidies be understood? Potipiti argues that this asymmetric treatment can arise from an underlying asymmetry in growth prospects of the two sectors that he shows occurs when trade costs are decreasing through time, and from the di¤erences in the rent-generating capacity of protection in (expanding) export and (declining) import-competing sectors that this implies. Due to the relative inability for protection to create rents in expanding as opposed to contracting sectors, he …nds that 116 See also Mitra (2002), for a similar commitment story where the avoidance of wasteful lobbying resources rather than distorted sectoral allocations is the driving factor that motivates governments to use trade agreements as a commitment device. As Bagwell and Staiger (2002) and Maggi (2014) observe, commitment arguments can also serve as reasons against joining a trade agreement, as the papers by McLaren (1997, 2002) elegantly illustrate. 73 it is sometimes optimal for a government to agree to a ban on export subsidies and thereby give up the (smaller) political rents in favor of the social welfare gain, while also not banning import tari¤s and instead opting to retain the (larger) political rents that their use generates.117 The commitment theory may also be used to interpret the evolution of rules on domestic sub- sidies from GATT to the WTO, an evolution that as discussed in Section 4.3 does not …nd easy support under the terms-of-trade theory. Here the relevant paper is Brou and Ruta (2013), who augment the Maggi and Rodriguez-Clare (1998) model by allowing the domestic government to use both a production subsidy and an import tari¤ in its relationship with the lobby in an import- competing sector.118 Taxation to raise revenue is assumed to be distortionary, so that a production subsidy does not dominate a tari¤ for redistributive purposes; rather, as Brou and Ruta show, in this setting optimal intervention will typically include a mix of tari¤s and production subsidies. In the Brou-Ruta (2013) model, the fundamental reason for signing a trade agreement that commits a government to free trade is the same as that in Maggi and Rodriguez-Clare (1998) and s anticipation of both a Potipiti (2012). But Brou and Ruta add two novel twists. First, the lobby’ tari¤ and a domestic subsidy creates ex-ante distortions that the government is not compensated for ex post, and so the government may have a direct reason to sign agreements which constrain both tari¤s and domestic subsidies. Second, if a government does sign an agreement that constrains its tari¤ only, this commitment induces the government to turn more intensively to production subsidies in its political relationship with the import-competing lobby –what Brou and Ruta term “the policy substitution problem” –and the resulting distortions are themselves welfare-reducing. What Brou and Ruta show is that in the presence of a tari¤-only commitment the new subsidies associated with the policy substitution problem can be handled with a “nulli…cation-or-impairment” rule, o¤ering support for the GATT shallow-integration approach to domestic subsidies in much the same way that the terms-of-trade theory supports GATT’s approach to domestic subsidies. But Brou and Ruta show as well that there is a remaining distortion associated with the original subsidy level that a tari¤-only agreement in combination with the GATT nulli…cation-or-impairment rule cannot address. It is with this second …nding that Brou and Ruta demonstrate that the commitment theory can provide support for the WTO’s new disciplines imposed directly on domestic subsidies, and in this sense provides support for deep integration.119 117 Potipiti’s (2012) model can explain why export subsidies might be banned while import tari¤s are not banned, but it doesn’ t explain why some limits on tari¤s might still be negotiated. However, it is not hard to see that introducing a small amount of terms-of-trade motive into the model (by relaxing slightly the small country assumption) could provide a reason for negotiating tari¤ bindings while not altering the other results of the model. 118 Limão and Tovar (2011) also study the role of trade agreements as a commitment device when governments can use both tari¤s and behind-the-border policies to redistribute to favored groups. But the focus of Limão and Tovar is on whether a government might wish to constrain its use of a more e¢ cient instrument (in their model the tari¤) knowing that this would result in more reliance on a less e¢ cient instrument (in their model behind-the-border non-tari¤ barriers). As they show, a government might …nd such a commitment desirable despite the associated e¢ ciency costs because the commitment can improve its bargaining power relative to the lobby. Limão and Tovar do not consider the possibility that international commitments might be extended to cover behind-the-border non-tari¤ barriers, so unlike Brou and Ruta their model does not yield insights about the desirability of deep integration. 119 Maggi and Rodriguez-Clare (2007) show that in a large-country dynamic version of the Maggi and Rodriguez-Clare (1998) model which combines the commitment and terms-of-trade theories, it is optimal for governments to implement liberalization in two phases, a …rst (and in their model instantaneous) phase in which liberalization re‡ ecting the 74 Finally, Ethier (1998) employs the commitment theory to address whether the emergence of PTAs following a period of multilateral liberalization might be viewed as a positive development for the world trading system. Actually, like Maggi and Rodriguez-Clare (2007), Either combines two reasons for a trade agreement into one model. One of these reasons can be interpreted as s model this is the motive that best describes why govern- the commitment motive, and in Ethier’ ments might be interested in PTAs (see especially Ethier’s discussion on pp. 1240-41). The second reason is an international externality, but it is not the terms-of-trade externality; rather, it is a (Marshallian) scale economy that operates at the world-wide level and creates a positive interna- tional (non-pecuniary) externality associated with greater investment. This form of international s model. externality is what underpins the purpose of a multilateral trade agreement in Ethier’ Ethier’s (1998) model is meant to capture the forces behind the growth in numbers of PTAs beginning in the 1990s that involved large developed countries forming PTAs with small reforming developing countries. In Ethier’s model, foreign investment from the developed world is by assump- tion necessary for successful reform in a developing country, and PTAs are ways in which developing countries compete among themselves for the required foreign investment. In essence, a PTA with a large developed country can enable the developing country to credibly “lock in”its reforms with commitments to deep integration: these deep integration commitments attract foreign investors to sink capital in the developing country which in turn, by creating natural interests that will push the foreign government to enforce the developing country’s commitments, ensures the success of the reform. As Ethier argues, once multilateral liberalization among developed countries has occurred and makes entry into the global economy attractive for developing countries, the commitment role of the PTAs can lead to a greater level of reform and investment world-wide than would otherwise occur. The international scale economy externality then implies that the greater investment in and scale of the successfully reformed developing countries leads to gains for everyone.120 Viewed together, these papers support the potential appeal of deep integration in trade agree- ments as a way to solve commitment problems. Three key issues remain. First, from the perspective of the commitment theory, are there good reasons to think that the required deep integration is best carried out in PTAs rather than in the WTO? Second, when it comes to tari¤ commitments, what does the commitment theory say about the appeal of preferential tari¤ cuts? And third, is there evidence that commitment motives are important for understanding real-world trade agreements? Regarding the …rst question, the WTO treatment of domestic subsidies illustrates that deep integration is possible in the WTO. But the failure of the WTO’s Doha Round to gain traction on the elimination of terms-of-trade motives occurs, and then a second (and in their model gradual) phase in which further liberalization to handle the domestic commitment motive occurs. When Brou and Ruta’ s (2013) …nding is viewed alongside these results it is tempting to conjecture that, if behind-the-border policies such as domestic subsidies were added to the Maggi-Rodriquez-Clare (2007) model, the resulting model might yield predictions that could support, as an optimal development, the gradual spread of deep integration, but only after terms-of-trade considerations had been removed from tari¤ choices. We leave this conjecture as a potential topic for future research. 120 We have con…ned our discussion here to the economic bene…ts that commitments via a trade agreement may provide, but there may also be important political bene…ts, as the recent paper by Liu and Ornelas (2014) suggests. Liu and Ornelas show that PTAs can serve as commitment devices for future governments that lower the probability of democracy failure, and they show as well that unstable democracies are more likely to join PTAs as a result. 75 deep integration components of the so-called Singapore issues points to severe limits to how far deep integration is likely to proceed in the WTO. So to put the …rst question slightly di¤erently: Does the commitment theory literature provide reasons to believe that the WTO cannot generate enough deep integration and that PTAs should be called upon to shoulder the load? Here the literature does not provide a direct answer, but there are suggestions of a partial answer: the WTO may be ine¤ective at helping small countries make commitments, and the preferential nature of PTAs may allow PTAs to be more e¤ective for small countries in serving this role. The reason is that, as Bown and Hoekman (2008) argue, a small country is likely to face challenges in getting trading partners to utilize the WTO dispute settlement system when it violates a trade-policy commitment; however, a small country may become “large”to foreign exporters who enjoy preferential access to its market, and those foreign exporters then have an incentive to push even a small PTA partner- country to follow through on its commitments.121 This suggests in turn that, at least when it comes to PTAs between large developed countries and small developing countries (small because of the small-country enforcement issues associated with the WTO, developing because commitment issues are arguably most germane for developing countries), the commitment theory could provide a reason to look more favorably on PTAs than does the terms-of-trade theory.122 Regarding the second question raised above, we have just pointed out one possible reason that the commitment theory could lend support to the preferential tari¤ cuts that de…ne PTAs, namely to aid in the enforcement of commitments for small countries. But beyond this, the commitment theory of trade agreements does not display any particular a¢ nity to PTAs over the GATT/WTO. To the contrary, as can be seen with reference to the Maggi and Rodriguez-Clare (1998) workhorse model, a preferential tari¤ cut could be completely ine¤ective in reducing the ex-ante distortions that the domestic government is seeking to address with its commitment to (in the model, multilateral) free trade.123 More generally, the domestic distortions that the government is attempting to reduce 121 This reasoning is not without caveats, however. For example, the rents created for the foreign exporters of a large country by preferential access to a small PTA partner market may only exist in the short run, and be dissipated by supply responses as the foreign exporter prices return to their long run MFN world-price levels. Notice also that if Ethier’ s (1998) position – that it is sunk foreign investment rather than preferential trade access per se that creates the natural interests to make sure the foreign government enforces the deep integration commitments of its PTA partner – is correct, then it is not obvious why the preferential nature of a PTA should matter for enforcement, and therefore not clear why a PTA would be more e¤ective than the WTO in this capacity. 122 A second suggestive possibility comes from a particular interpretation of the …ndings of Aghion, Antràs and Helpman (2007). They identify circumstances under which global free trade may not be achievable via multilateral negotiations but could be achievable if FTAs are permitted; however, this possibility arises only when governments do not maximize welfare (so that global free trade is not e¢ cient relative to governments’own objectives). If we make a distinction between governments’ex ante and ex post preferences and evaluate the institution relative to the former, then this result suggests a possible commitment story that could support the creation of an institution that permits PTAs. The interpretation is that governments seek to maximize overall welfare at the ex ante stage but anticipate that they may be captured by lobbies in the ex post stage. To maximize their ex ante objectives, governments then might set up an institution that permits FTAs as a potential bulwark against ex post political motivations. We regard this interpretation as leading only to a suggestive possibility, however, since Aghion et al. also identify circumstances under which politically motivated governments could achieve global free trade only when PTAs are banned. 123 This would be the case as long as the domestic-country imports from non-PTA countries were not driven to zero in the presence of the preferential tari¤ cut (because then the unchanged MFN tari¤ of the domestic country together with its small size in world markets and the arbitrage condition would ensure that the local-market prices in the domestic economy are una¤ected by its tari¤ preferences and so the sectoral allocation of domestic productive 76 with commitments under a trade agreement are related to local prices in the domestic economy, and any set of local prices that can be achieved via preferential tari¤ cuts can be achieved as well with appropriate MFN tari¤s. Finally, we turn to the third question: Is there evidence that commitment motives are impor- tant for understanding real-world trade agreements? Here the evidence is thin but supportive. Staiger and Tabellini (1999) o¤er early empirical support for the commitment theory. They look for evidence that governments make di¤erent tari¤ choices across GATT environments that are distinguished by the degree of commitment that GATT rules provide for those choices. Focusing on US tari¤ choices made under the GATT Escape Clause (where GATT should not have helped provide commitment) and made in the Tokyo Round of GATT negotiations (where GATT rules could have helped provide commitment), Staiger and Tabellini …nd that US tari¤ decisions were more responsive to the production distorting impacts of the tari¤s in the latter decisions as com- pared to the former decisions, in line with what their commitment model would predict. Tang and Wei (2009) adopt a di¤erent approach. They consider the di¤erences in the growth performance of developing countries that joined the GATT/WTO under two di¤erent kinds of accession rules, rules that applied in GATT prior to the creation of the WTO and that did not require acceding countries to undertake extensive policy commitments, and rules that applied subsequent to the creation of the WTO in 1995 and that typically required substantial policy commitments by the acceding government. Tang and Wei …nd that the post-accession growth performance of the de- veloping countries that acceded to the GATT/WTO under the latter set of rules was signi…cantly better than that of non-acceding countries and countries that acceded to GATT under the former rules, and they attribute this di¤erence to the domestic commitment role played by the WTO. Further evidence lending some support to the commitment theory of trade agreements is provided by Limão and Tovar (2011), Liu and Ornelas (2014) and Bown and Crowley (2014).124 Summarizing, we conclude from our survey of the commitment theory strand of the literature that this theory provides some reason to be more supportive of PTAs than our survey of the terms- of-trade theory alone would suggest, though it provides no reason to be less supportive of the WTO. We thus see the commitment theory as moving the answer to the motivating question of our survey toward a view that PTAs and the WTO may be complementary, but in light of the relatively thin s role in actual empirical support to date on the widespread importance of the commitment theory’ trade agreements, probably not as yet moving the answer very far in this direction. resources would be una¤ected as well). 124 Limão and Tovar (2011) (see note 118) employ data on Turkish tari¤s and non-tari¤ barriers and …nd evidence in line with their theory that Turkey was more likely to bind its tari¤s in the WTO and to bind them more tightly in industries where it had low bargaining power relative to the lobbies it faced. Liu and Ornelas (2014) (see note 120) …nd evidence supporting the two key predictions of their model that PTAs are more likely to be formed by unstable democracies and participation in PTAs helps to stabilize these democracies. Finally, in their cross-country study of emerging economies, Bown and Crowley (2014) provide evidence that these countries changed how they conduct their trade policy (through antidumping and safeguards) by taking on tari¤ binding commitments when joining the WTO. 77 6.2 The Delocation/Pro…t-Shifting Theory Another alternative to the terms-of-trade theory can be found in a growing body of literature emphasizing …rm delocation and pro…t-shifting as sources of international externalities that might give rise to and shape the design of trade agreements. This strand of the literature shares with the terms-of-trade theory a focus on the internalization of international policy externalities as the primary task of trade agreements; but the delocation/pro…t-shifting theories reject the implication of the terms-of-trade theory that terms-of-trade externalities are the only – or even the most important – cause of the ine¢ ciency that a trade agreement can correct. Instead, according to these theories, non-cooperative trade policy choices are ine¢ cient because, when left on their own, governments use trade protection to ine¢ ciently “delocate” …rms or shift …rm-pro…ts from foreign locations to the domestic market. The delocation/pro…t-shifting theories build on the unilateral incentives for trade policy inter- vention that arise when the assumption of perfect competition is relaxed, incentives that were …rst identi…ed by Brander and Spencer (1984, 1985), Spencer and Brander (1983) and Venables (1985, 1987). Brander and Spencer argued that trade policy intervention could be used to shift …rm prof- its toward the intervening country when …rms possess market power and use their market power in the presence of entry barriers to earn pro…ts in equilibrium. Venables showed that in a world where pro…ts are dissipated by free entry it is still possible to gain unilaterally with trade policy intervention as long as there are international transport costs, due to the delocation/entry-exit e¤ect of this intervention on foreign and domestic …rms.125 Ossa (2011) was the …rst to explore the consequences of the delocation e¤ect for the purpose and design of trade agreements, while Mrázová (2011) develops the pro…t shifting rationale for trade agreements.126 Ossa (2011) considers a monopolistically competitive setting in which …rms producing di¤eren- tiated products compete for sales in both the home and foreign markets under conditions of free entry, and where exporting the product abroad involves shipping costs. Venables (1987) establishes that a …rm-delocation motive for trade policy arises in such an environment: if the home country o¤ers protection to its importers or a subsidy to its exporters, foreign …rms can be “delocated” to the home market and home consumers save on trade costs and enjoy a lower overall price index as a result. This home gain, however, comes at the expense of foreign consumers, whose price index rises. Hence, the …rm-delocation e¤ect represents a negative international policy externality. What Ossa demonstrates is that in his model the transmission of the …rm-delocation e¤ect from the home to the foreign country can be interpreted as traveling through local prices, not the terms of trade. Intuitively, in Ossa’s model each country is impacted directly by the local price in the other country’s market, because each country could enjoy the savings in transport costs if it could have s …rms (and the production of their individual varieties) located locally rather more of the world’ 125 As we discuss in note 75, the problem for a trade agreement to solve in models featuring pro…t shifting or delocation e¤ects can be given a terms-of-trade interpretation under certain conditions. But the papers we next discuss impose conditions so that this interpretation is not valid, which is why we discuss these papers here rather than in the context of our earlier discussion of the terms-of-trade theory. We return to this point further below. 126 Brander and Spencer (1984) o¤er an early analysis of the bene…ts of a trade agreement in a pro…t-shifting model. 78 than abroad; and the equilibrium pattern of …rm location across countries depends on local prices in both countries via the free-entry condition. A main thrust of Ossa (2011) is therefore that one does not have to believe that terms-of- trade e¤ects of trade policy are important in order to understand the purpose of trade agreements. In fact, Ossa argues that a number of the prominent design features of the GATT/WTO (e.g., reciprocity and MFN) can be equally interpreted as sensible if governments are instead attempting to internalize delocation externalities with their trade agreements.127 Accordingly, and based on Ossa’s results, if anything the delocation theory of trade agreements could be said to strengthen support for the GATT/WTO as a well-designed institution, as it broadens the interpretation of the problem that the GATT/WTO is well-designed to solve. Still, while the delocation theory does not appear to undercut support for the GATT/WTO in some fundamental way, might it nevertheless swing support from shallow to deep integration, in which case like the commitment theory the delocation theory might be interpreted as supporting the need for PTAs to complement the GATT/WTO? This is still an open question in the literature, but the results of DeRemer (2013a) suggest that the answer to this question may be a quali…ed “No.” DeRemer demonstrates that shallow integration can work in a delocation model where governments have both tari¤s and wage subsidies at their disposal, but DeRemer also shows that the particular form of the “market access preservation rule” that makes shallow integration work in his model bears little relationship to any of the closest analogies in the GATT/WTO. Together with Ossa’s (2011) …ndings, DeRemer’s results therefore suggest that the GATT/WTO is probably less-well designed to solve problems associated with international delocation externalities than it is to solve the terms-of-trade problem, but with some selective …xes there is no reason in principle that it could not be optimized in this direction, and thus no particular reason to believe that PTAs are needed to help shoulder the load. Similarly, Mrázová (2011) develops a model of trade agreements based on the notion that such agreements help to internalize an international pro…t-shifting externality, and uses this model to explain the WTO ban on export subsidies that is puzzling from the perspective of the terms-of-trade s argument is based on self-enforcement considerations: she shows that it can be theory. Mrázová’ easier to enforce commitments in a repeated game model of trade agreements when import tari¤s are the only trade policy instrument allowed under the agreement. O¤ering a di¤erent perspective, DeRemer (2013b) abstracts from self-enforcement issues but employs a pro…t-shifting model of trade agreements to argue that the evolution of GATT/WTO domestic and export subsidy rules can be better understood from the perspective of pro…t-shifting models of trade agreements than from the perspective of the terms-of-trade theory. To generate his domestic subsidy result, DeRemer assumes that governments wield a particular form of entry subsidy, and it is an open question whether the result would hold for more general forms of domestic subsidy (or for domestic policy 127 This is not to say that the predictions of the terms-of-trade theory and the delocation theory are the same. For example, as Ossa (2011) demonstrates, the particular de…nition of reciprocity that is suggested by his model di¤ers from that suggested by the terms-of-trade theory. 79 instruments more generally).128 But at a minimum DeRemer’s result illustrates that pro…t-shifting externalities can be usefully employed to help interpret the evolution of GATT/WTO subsidy rules. And …nally, Ossa (2014) assesses the importance of pro…t-shifting and terms-of-trade externalities in a calibrated quantitative model of trade agreements, and …nds that together these externalities are sizable enough to account broadly for the observed magnitude of multilateral tari¤ liberalization under the GATT/WTO. Like the delocation theory, the results from the pro…t-shifting theory seem to provide further support for the GATT/WTO approach to liberalization. And at least to date, the pro…t-shifting theory has not generated results that would indicate speci…c support for PTAs. These conclusions are further supported from the perspective of the …ndings of Bagwell and Staiger (2012a, 2015). Ac- cording to their …ndings, as long as governments have both import tari¤s and export tax/subsidies at their disposal, the underlying problem for a trade agreement to solve in the delocation and pro…t-shifting models can still be given a terms-of-trade externality interpretation, though novel local-price externalities do arise when export policies are not available to governments. This means that we should expect key results across these theories to be similar, at least when the use of export policies is not ruled out, and Bagwell and Staiger con…rm this for a number of the standard predictions of the terms-of-trade theory.129 We conclude from our survey of the delocation/pro…t-shifting theories of trade agreements that this strand of the literature, if anything, bolsters the case for the GATT/WTO approach to liberalization, as it provides a broader base from which to interpret as sensible many of the core design features of the GATT/WTO; and while more work is needed to tease out the implications of these theories for the desirability of PTAs as a form of liberalization, at present these theories provide no speci…c reasons to think that PTAs o¤er an attractive alternative or complement to the s (2014) GATT/WTO approach. Combined with the fact that, with the notable exception of Ossa’ calibration exercise, there is to date no evidence on the empirical importance of these theories, we do not view the delocation/pro…t-shifting strand of the trade agreement literature as providing further quali…cations to our conclusions concerning the relative merits of PTAs and the GATT/WTO. 6.3 The O¤shoring Theory It is by now well-documented that trade in intermediate inputs dominates modern trade ‡ows, that many of these inputs appear to be highly specialized to their intended use, and that this 128 It is for this reason that we interpret Deremer’ s (2013b) domestic subsidy result somewhat more narrowly as applying to certain subsidies but probably not to domestic policies more generally, while we view the analogous results of Brou and Ruta (2013) concerning domestic subsidies as suggesting (from the perspective of the commitment theory) broader implications for deep integration. 129 To be clear, the delocation and pro…t-shifting models do o¤er important new insights. First, to the extent that they are successful, the GATT/WTO attempts to limit export subsidies make analyses of these models without export policies an empirically relevant case to consider. Second, even when export policies are available and a standard terms-of-trade interpretation can be given to the problem, the deviation from perfect competition that is featured in the delocation and pro…t-shifting models can yield novel predictions, as discussed in Section 4.3. And as Ossa (2014) shows in the context of a pro…t-shifting model, the existence of product di¤erentiation can interfere with the ability of MFN and reciprocity to neutralize third-party spillovers in bilateral tari¤ negotiations. 80 has not always been so (see Johnson and Noguera, 2014, as well as the discussion in Antràs and Staiger, 2012b, and Baldwin, 2014). This rise in the importance of “o¤shoring”raises the question whether the rules and norms of the GATT/WTO, conceived at a time when the nature of trade was quite di¤erent, are still relevant today. There are two issues that have been addressed in the literature. First, the rise in o¤shoring has been accompanied by a signi…cant rise in foreign investment. Adopting the perspective of the terms-of-trade theory of trade agreements, Blanchard (2007, 2010) argues that this investment might naturally reduce the magnitude of the terms-of-trade externality problem, and in this way o¤shoring may act to reduce the need for a GATT/WTO-type institution. We discuss Blanchard’s work in Section 4.5 in the context of our survey of the terms- of-trade literature. In this section, we focus on a second issue associated with the rise in o¤shoring: its potential to alter the mechanism of international price determination. This issue is emphasized by Antràs and Staiger (2012a, 2012b), who argue that the rise of o¤shoring may have changed the way that international prices are determined and thereby the nature of the international externality that a trade agreement must address, with implications for the design of e¤ective trade agreements. Antràs and Staiger (2012a, 2012b) distinguish between international prices that are determined by standard market clearing conditions and prices that are determined by bilateral bargaining between foreign suppliers and domestic buyers. The former mechanism is featured in the terms- of-trade theory, and it underpins a property that is key for the terms-of-trade theory’s a¢ nity with shallow integration: the tari¤ is the …rst-best policy for terms-of-trade manipulation and the international cost-shifting that is implied. This is a key property because, as we discuss earlier in our survey, according to the terms-of-trade theory the only “problem” for a trade agreement to “solve” is terms-of-trade manipulation, and as the tari¤ is the …rst-best policy for terms-of-trade manipulation it then follows that the tari¤ is in fact the only policy that is distorted in the Nash equilibrium: behind-the-border measures are set e¢ ciently under Nash choices. But when international prices are determined by bilateral bargains between buyers and sellers located in di¤erent countries and these bargains are undisciplined by market clearing conditions, as Antràs and Staiger (2012a, 2012b) argue is increasingly the case with the rise in o¤shoring, the tari¤ is no longer the …rst-best policy for cost-shifting, and governments typically …nd it unilaterally optimal to distort many of their policy choices – border but also behind-the-border – in an e¤ort to manipulate international prices and shift costs onto their trading partners. This leads to Antràs and Staiger’s …rst claim: through its implications for international price determination, the rise in o¤shoring is likely to erode the e¤ectiveness of the GATT/WTO shallow integration approach. As Antràs and Staiger show, a second claim follows when governments have political economy motives: in the presence of o¤shoring, and in stark contrast to the predictions of the terms-of-trade theory, these motives introduce additional policy ine¢ ciencies that a trade agreement can address. Taken together the implication of Antràs and Staiger’s …ndings is that the rise of o¤shoring may usher in a new world in which a collection of individualized deep-integration agreements is needed to guide governments to e¢ cient policy choices. Strikingly, that sounds a lot like the recent wave of PTAs. Evidently, the o¤shoring theory of trade agreements has strong implications for the relative mer- 81 its of PTAs and the GATT/WTO, both diminishing the appeal of the GATT/WTO and elevating the appeal of PTAs. In comparison to the other theories reviewed in this section, the o¤shoring theory seems more fundamentally at odds with the terms-of-trade theory and the implied support for the GATT/WTO approach to liberalization. A key question for our survey is thus whether there is empirical support for the o¤shoring theory. In fact we are unaware of any direct empirical evidence relating to this theory, and only a few pieces of indirect evidence, which are supportive and suggestive, but hardly de…nitive.130 Hence, while the o¤shoring theory of trade agreements has the potential to substantially alter our conclusions about the relative merits of PTAs and the GATT/WTO, in light of the lack of empirical evidence to date on the relevance of the theory it would be premature to place much weight on its implications at this time. Clearly, however, empirical research in this area is warranted. 6.4 Summing Up Thus Far Having surveyed the three additional strands of the trade agreement literature, we are therefore left with the following view: while quali…ed along some important dimensions, the terms-of-trade s implication, that strong support for the GATT/WTO is warranted while a cautious view of theory’ PTAs seems appropriate, survives largely intact. The commitment theory serves to elevate support for PTAs in certain circumstances beyond what the terms-of-trade theory would suggest, while s support for the the delocation/pro…t-shifting theory seems to reinforce the terms-of-trade theory’ GATT/WTO. The o¤shoring theory could potentially overturn the implications of the terms-of- trade theory in this regard, but there is to date insu¢ cient evidence to justify a change in position based on the predictions of this theory. The view that we put forth here re‡ects our assessment of research to date, and as we note further research is required in many areas. Research on deep integration, in particular, is at an early stage. The potential coordination bene…ts of regulatory harmonization, as well as the potential third-party costs, represent an important direction for future research, for example. As we also note in Section 5.3, another important direction is to include bargaining frictions and explore the potential bene…ts that such frictions suggest for negotiations among smaller groups of countries as PTA members or in WTO plurilateral or critical-mass agreements. 7 Dispute Settlement Such is the apparent success of WTO dispute settlement that it is often referred to as the “crown jewel” of the multilateral trading system. Over a relatively short period, economists, political scientists, and legal scholars have developed a range of positive and normative approaches to explore 130 We are aware of two pieces of evidence that o¤er some indirect support for this theory. A …rst is presented in Antràs and Staiger (2012b), and relates to apparent di¢ culties in achieving negotiated reductions of tari¤s on imported di¤erentiated inputs for a set of countries acceding to the WTO. And a second is contained in Ore…ce and Rocha (2014), who …nd evidence that an important predictor of whether two countries sign a deep-integration PTA is the share of their bilateral trade that is comprised of parts and components. 82 important research questions in this area.131 For example, legal scholars now parse the language of each newly-arriving WTO legal decision – so as to draw potential precedent implications for international and domestic law, as well as for public policy –with the same voracity as the American bar devours fresh Supreme Court rulings or Europeans tackle judgments from the European Court of Justice. Furthermore, legal and economic scholars have already teamed for more than a decade to annually publish joint inter-disciplinary assessments of each year’s new WTO jurisprudence.132 We begin this section by describing how the WTO dispute settlement system has been used to date. We then survey the theory on the role of the WTO dispute settlement system before examining the relationship between PTA implementation and WTO dispute settlement activity. We also describe recent PTA disputes associated with the enforcement of non-trade policies. Finally, we conclude with thoughts on the relative merits of a multilateral dispute settlement system. 7.1 WTO Dispute Settlement: Patterns in Use WTO disputes involve state-to-state level interaction. Here we characterize some of the data on the nearly 500 formal disputes initiated between 1995 and 2014 in order to …rst clarify what the disputes tend to be about, who they tend to be between, how the process works, and the typical outcomes that arise. Due to a number of potential issues related to sample selection, a central argument of the literature described below is that dispute settlement cannot be appreciated or evaluated based solely on an appeal to information regarding the disputes that are observed –i.e., the procedures also have important “o¤ equilibrium” implications that do not necessarily arise through formal disputes. Nevertheless, evidence that the WTO members are putting trust in the system by frequently triggering its use over important policies or signi…cant amounts of trade is potentially supportive of the idea that dispute settlement is playing a signi…cant role.133 7.1.1 WTO disputes are most frequently allegations of excessive import protection We begin by appealing to an assessment drawn from a database of WTO dispute settlement infor- mation compiled and made publicly available by Horn and Mavroidis (2011). Of the formal disputes that have arisen to date, the typical topic concerns a plainti¤ (“complainant”) member alleging that the defendant (“respondent”) country has imposed an excessive level of import protection. Horn, Johannesson, and Mavroidis (2011) examine disputes taking place between 1995 and 2010 and report that nearly 95 percent of all WTO disputes concern trade in goods, with the remainder 131 Examples of seminal legal scholarship over GATT law begin with Dam (1970), Jackson (1969) and Hudec (1990). The WTO’ s legal process is explained in more detail by Palmeter and Mavroidis (2004). Davis (2012) provides a recent overview of political science research on WTO dispute settlement. 132 Beginning in 2001, Horn and Mavroidis (2003) initiated a series of annual assessments with the backing of the American Law Institute, subsequently extended by Bown and Mavroidis (2013), which pairs legal scholars with economists to jointly analyze each of the formal WTO Appellate Body (and non-appealed Panel) decisions that arise. Cumulatively these legal-economic assessments have covered nearly 100 di¤erent individual dispute decisions to date. 133 Put di¤erently, at this moment in time, the WTO members do not seem to be avoiding use of the system or clamoring to develop an alternative system. The GATT experience of the 1980s in particular suggests that that is not necessarily always the case (Bhagwati and Hudec, 1990). 83 concerning services trade or intellectual property rights protection. The most frequently invoked legal agreements in these disputes include the baseline 1947 GATT, followed by the separate 1995 WTO Agreements on Antidumping, Subsidies and Countervailing Measures, Agriculture, TBT, Safeguards, and SPS Measures. s accession terms China, which acceded to the WTO in 2001, is an illustrative case study. China’ required it to take on signi…cant commitments to domestic (economic) and trade policy reform. And because China is such a large importer and exporter, and has so many trading partners, there were expectations that it would become signi…cantly involved in WTO litigation (Bown, 2010). Within a relatively short period of time, China has become an important WTO litigant; more than 40 percent of disputes initiated between 2007 and 2011, for example, featured China as either a respondent or complainant, with an almost 2 to 1 ratio of respondent to complainant. China’s WTO disputes are also quite diverse. Some disputes …led against China …t the “typical” mold –i.e., they feature an allegation of excessive import protection in a politically sensitive sector such as autos, steel, or agriculture, and address a commonly-challenged policy such as antidumping. However, China has also faced a number of disputes in relatively new issue areas. These include trading partners using the WTO to legally challenge China’s export restrictions over various raw materials and “rare earth” metals that are especially important in electronic goods. There have been protests over China’s tax and subsidy policies and its allegedly lax protection of intellectual property rights. Finally, trading partners have objected to China’s import restrictions on foreign services providers; examples include …nancial information services (e.g., Bloomberg, Dow Jones, Thomson-Reuters), electronic payment services (e.g., Visa, MasterCard, American Express), and audio-visual services (e.g., movie studios, media and publishers, software providers). s More generally, WTO disputes are rarely as simple as one country challenging another country’ increase of an applied MFN tari¤ above its legal binding. Instead, the complainant country’s typical allegation is that the respondent has implemented excessive import protection through some non- tari¤ policy. At least in part, the dispute can be understood to arise because of a disagreement in interpretation of whether the respondent’s policy was implemented in order to achieve some other (non-trade) objective that may nevertheless be justi…able under the rules or exceptions of the WTO agreement. The appearance –if not reality –of two countries having di¤erent interpretations of the WTO raises the possibility that dispute settlement may be doing more than simply enforcing the agreement, but it may also address the incompleteness of the GATT/WTO contract. Section 7.2.2 describes advances in this particular area of research. An early dispute settlement literature was motivated by recognition that most disputes initiated immediately in the aftermath of implementation of the Uruguay Round commitments (e.g., 1995- 1998) seemed to involve the trading interests of only high income countries. The policy concern was that the newly arising legal and institutional costs of engaging dispute settlement –given the increased “legalization”of the dispute settlement process under the WTO relative to its immediate GATT predecessor –might be too burdensome for developing country members with limited legal 84 capacity to enforce their market access interests.134 Beginning with Horn, Mavroidis and Nordstrom (2005), the evidence indicates strong correla- s exports, and the diversity of its trading tions between dispute settlement use, the level of country’ partners. The more a country traded and the more bilateral trading relationships that it had, the greater the scope for potential frictions to arise that would result in a formal dispute. Subsequent research on disputes from the early WTO period that also explored the potential role of other factors - such as legal and retaliatory capacities - that might a¤ect dispute initiation tended to con…rm the central importance of the result that high levels of trade were positively correlated with the triggering of disputes.135 Nevertheless, Bown and Reynolds (2015) characterize the bilateral trade in disputed products for a sample of disputes making up more than 70 percent of the WTO caseload between 1995 and 2011 and …nd evidence of vast heterogeneity in the levels of market access at stake. Roughly 14 percent of WTO disputes involve bilateral trade in disputed products of tiny amounts – e.g., less than $1 million per year - and yet 15 percent of disputes involve bilateral trade in products over more than $1 billion per year. The heterogeneity in trade stakes across the dispute data raises research questions regarding determinants of dispute settlement use. Some of the new theoretical advances described below may improve our understanding of these forces. The Bown and Reynolds (2015) evidence also suggests that WTO member countries in the aggregate have entrusted the dispute settlement system to assess policies that cover signi…cant amounts of trade. Over 1995-2011, WTO dispute settlement investigations collectively scrutinized nearly $1 trillion in goods imports, an average of $55 billion per year, or roughly 0.5 percent of world imports in 2011.136 These are arguably signi…cant amounts of trade for dispute settlement to address, without even yet considering the impact of the potential precedent arising through WTO jurisprudence, as well as the “o¤-equilibrium” impacts of the system. 7.1.2 The WTO legal process and outcomes If the mandatory consultations with the respondent resulting from the initiation of a formal dispute fails to resolve the issue, the complainant country can trigger a formal legal process whereby the legal representatives of the two sides make arguments and present evidence to a WTO dispute settlement Panel. Nearly 200 of the roughly 500 WTO disputes initiated to date have resulted 134 Bown (2009) provides a more comprehensive and in depth treatment of these and related issues confronting developing country access to WTO dispute settlement. 135 Bown (2005a), for example, provides evidence linking higher import penetration ratios to US antidumping duties in the …rst stage, and higher levels of bilateral trade a¤ected by those US duties subsequently positively associated with the trading partners decision of whether to formally challenge them through GATT/WTO dispute settlement in a second stage. In a separate study examining a cross-country sample of WTO disputes that concern policies imposed on a national treatment basis –in which the policy negatively a¤ected all trading partners –Bown (2005b) …nds that higher pre-policy levels of bilateral exports of the disputed products are positively associated with potential litigants’ decision of whether to formally engage in the dispute settlement process. 136 These are conservative statistics because they do not include roughly 30 percent of the WTO caseload of disputes that are either tied to export policies, services policies, TRIPS, or policies that a¤ect all imports and are not linked to speci…c product codes. The overall level of trade directly a¤ected by WTO disputes is likely to be much higher. 85 in formal legal rulings through Panel Reports. Of these, more than half have subsequently been appealed and received rulings from the WTO’s standing Appellate Body. Each WTO dispute can contain numerous “claims” made by the complainant against the re- spondent; the data also reveal substantial variation across disputes as to the total number of claims …led. There is a “pro-trade bias” in WTO legal rulings, in the sense that the Panel or Appellate Body …nds that the respondent has done something wrong in almost every dispute that receives a formal ruling. Nevertheless, there are also selection issues associated with this interpretation that the theory that we introduce below has begun to tackle.137 Horn, Johannesson, and Mavroidis (2011) also provide information on the average time that disputes take to work their way through the sequential steps of the WTO dispute settlement process. On average, the process takes almost three years between the initiation of the dispute, the issuances of the Panel report and Appellate Body report, and …nally the outcome whereby the respondent country is required to bring itself into compliance with legal rulings or face authorized retaliation.138 Finally, the default form of compensation in WTO disputes in the event of noncompliance is tari¤ retaliation by the complainant subject to limits determined by WTO arbitrators; and authorized retaliation can only begin after the legal process described above is exhausted. Overall, and despite the increased frequency of disputes arising during the WTO period, retaliation has rarely been an equilibrium outcome. Fewer than 15 disputes have resulted in the WTO judges even having to articulate the permissible level for formal retaliation by the complainant in the event of noncompliance. Many fewer cases have resulted in the complainant country actually implementing the WTO’s authorized retaliation.139 Some high pro…le exceptions notwithstanding, there are very few examples of egregious non-compliance with rulings of the sort that result in the explicit WTO authorization of retaliation.140 Nevertheless, some of these exceptions identify limits to the system and have become subject of areas of formal research. 137 Furthermore, at the level of legal rulings over particular claims made, Horn, Johannesson, and Mavroidis (2011) report that the complainant “wins” only 57 percent of the claims over which the Panel ultimately rules. There are also often many claims in these disputes over which Panel declines to rule for reasons of judicial economy. 138 On average, disputing countries remain in consultations for nearly 6 months, and then another 15 months elapse before the WTO’ s …rst Panel ruling. While the cases that are appealed receive an Appellate Body report relatively quickly (3 months) thereafter, another 9-11 months typically elapse before expiration of the reasonable period of time necessary for the respondent to bring its disputed policy into compliance with rulings and before the potential for discussion of compensation due to noncompliance can occur. 139 Bown and Ruta (2010) describe the calculation of economic retaliation levels for the roughly 10 disputes that had made it all the way through the WTO dispute settlement process between 1995 and 2008 whereby arbitrators determined the upper limits to the level of permissible retaliation. The Bown and Pauwelyn (2010) volume provides a broader set of research contributions on the retaliation-setting experiences under the WTO from other perspectives in law, political science, and economics, as well as from policymakers. 140 The EC –Beef Hormones dispute involved a sustained period of retaliation by the US, as did EC-Banana Regime dispute between the EU, US and Latin American banana exporting countries before it was eventually resolved. US – Upland Cotton has not resulted in compliance but in a payo¤ (…nancial transfer) from US to Brazil. US – Internet Gambling, discussed by Irwin and Weiler (2008), has not resulted in compliance or implemented retaliation by the tiny island nations of Antigua and Barbuda against the US. Finally, Bown and Prusa (2011) describe the repeated challenges - 15 di¤erent disputes over 10 years - brought by a number of di¤erent WTO members to the US use of “zeroing” in antidumping investigations. 86 7.1.3 Dispute use and retaliatory capacity The terms-of-trade theory suggests that trade agreements are designed to facilitate an escape from a terms-of-trade driven Prisoners’ Dilemma. Cooperation in trade policies is achieved in a repeated game, where observed deviations may lead to a severe o¤-equilibrium-path punishment such as Nash reversion that corresponds to an unraveling of the agreement. A di¤erent kind of retaliation is featured in the WTO dispute settlement system, as authorized retaliation in this context is arguably both on the equilibrium path (i.e., something that is part of the agreement and occurs in practice) and commensurate in nature.141 The terms-of-trade theory thus suggests that governments’ decisions to impose certain trade policies along the equilibrium path may be in‡uenced by their perceptions of the retaliatory capacities of their a¤ected trading partners. We next discuss empirical approaches providing evidence broadly supportive of this perspective. Bown (2002, 2004b) provides theoretical and empirical approaches, respectively, to explore the potential enforcement implications of the terms-of-trade theory in the setting of dispute settlement. The theory exploits di¤erences across the GATT/WTO-mandated limits to the compensatory re- taliation response under di¤erent legal provisions and their implications in an environment char- acterized by trading partners with di¤erent capacity constraints on retaliation. The empirical application examines a cross-country sample of policies imposed between 1973 and 1994 and a framework in which governments are given the choice between GATT-consistent and inconsistent policies of import protection, the latter which would subsequently result in a formal dispute. Con- ditional on choosing to impose additional import protection, the evidence suggests that heightened retaliation capacity by a¤ected trading partners makes the policy-imposing country more likely to implement policy changes in a way that conforms to the rules of trade agreements.142 Blonigen and Bown (2003) present a related empirical analysis that focuses on US antidumping policy and provides further evidence that an enforcement system based on retaliation capacity is likely to endogenously a¤ect the structure of policies that countries impose in the …rst place. That study examines US policies implemented during the GATT and early WTO period (1980-1998) and …nds that bilateral retaliation capacity in‡uenced how new US import restrictions were implemented along two di¤erent dimensions. First, US industries were less likely to request antidumping import restrictions be imposed against trading partners for which they have industry-level export exposure to retaliation; this likely a¤ects the potential policy actions that a government is ultimately asked to consider implementing. Second, conditional on receipt of a request for additional import protection, the US government was less likely to impose duties on trading partners for which overall US exports would subsequently be exposed to bilateral retaliation under a potential WTO dispute. 141 For discussion of interpretations of retaliation in GATT/WTO, see Bagwell and Staiger (2002, Chapter 6). In addition, as discussed in Section 4.2, commensurate retaliation may occur along the equilibrium path in less formal ways, as for example when a privately informed government imposes an anti-dumping duty while recognizing that doing so increases the likelihood that its exporters will face a similar duty in the future. 142 Bown (2004c) presents related evidence on retaliation capacity impacting the outcomes of trade disputes as well, which is consistent with the notion that such incentives are likely to a¤ect policy choices ex ante. See also Bown (2004a) for evidence that the outcomes of such bilateral disputes are extended to third country exporters in a manner consistent with successful application of the MFN rule. 87 7.2 WTO Dispute Settlement: Theoretical Perspectives The terms-of-trade theory suggests that the WTO represents a codi…cation of a set of coopera- tive strategies for governments engaged in the repeated play of a terms-of-trade driven Prisoners’ Dilemma game. Cooperation in such a setting is possible only if governments understand that o¤-equilibrium-path deviations may lead to a breakdown in the agreement and a corresponding reduction in cooperation. In this general context, what, then, is the role of the WTO dispute set- tlement system? This is an important and under-studied question. Here we highlight two theoretical perspectives on this question: the system may enhance cooperation by increasing transparency (i.e., by generating and disseminating information) or by helping to “complete”the WTO contract. We also discuss research concerning possible reforms of the dispute settlement system. 7.2.1 Transparency and Information Cooperation is easier to achieve in Prisoners’ Dilemma settings when behavior is transparent or public.143 The decision of GATT Contracting Parties to concentrate protection mainly into tar- i¤s (rather than quotas) can be understood in this light. Nevertheless, governments may face monitoring impediments and may not be perfectly informed about the full range of trade-policy conduct of other governments. Consequently, WTO rules that generate and disseminate public information about trade-policy conduct may facilitate greater cooperation. The WTO Trade Pol- icy Review Mechanism (TPRM), under which the WTO Secretariat conducts periodic reviews of the trade policies of member governments, may be evaluated in this context. We now brie‡y de- scribe research in which the WTO dispute settlement system likewise may facilitate cooperation by generating and disseminating information. Consider again Maggi (1999), which provides a model where cooperation can be achieved under a multilateral enforcement mechanism in which third-party transparency is present and third coun- tries stand ready to retaliate. If a trade agreement is designed to ensure that any deviation would be observed by all member governments, then the resulting o¤-equilibrium-path punishment could take an immediate multilateral form and thus be more severe. A dispute settlement body that publicly identi…es an o¤-equilibrium-path deviation could play a role in facilitating cooperation by ensuring that all member governments would be aware of the transgression. Even when “bilateral” deviations would be made observable at a multilateral level, governments may be able to achieve more cooperative tari¤s along the equilibrium path as part of a self-enforcing agreement. Park (2011) develops a two-country model to explore a related information role dispute set- tlement. Each government has a publicly observed trade policy (e.g., a tari¤) and a “hidden” trade policy (e.g., a non-tari¤ barrier) that is imperfectly observed by the other government, and each government privately observes a signal of the other’s hidden trade policy. Governments at- tempt to cooperate in this “private monitoring”setting even in the absence of a dispute settlement system, and a key feature of optimal cooperation is that, after observing a suspicious signal, a 143 See again the discussion of these and related insights for the models surveyed in Section 4.2. 88 government increases its publicly observed trade policy, which publicly initiates a trade-war phase. The prospect of a trade-war phase discourages opportunistic behavior with respect to the hidden policy; however, suspicious signals are sometimes observed even when the hidden trade policy is set at cooperative levels, and so the trade-war episodes are a feature of optimal cooperation in the absence of a dispute settlement system. Relative to this benchmark, a dispute settlement system can facilitate greater cooperation by changing the information structure of the game through its provision of a public signal and enriching the forms of feasible punishment. Empirically, a potentially useful environment to examine implications of the WTO’s information- dissemination role may turn out to be the Great Recession of 2008-2009. For despite the highly synchronized and sudden global collapse in economic activity and trade ‡ows, it is now well under- stood that a global surge in new trade protection of the scale of even earlier recessions, let alone the Great Depression of the 1930s, did not occur (Bown and Crowley, 2013b). Given the relatively moderate trade policy response, it is unsurprising that the WTO has also not been ‡ooded with a subsequent onslaught of newly initiated disputes. However, one contributing explanation may be the sharp increase starting in 2009 of the multilateral monitoring e¤orts to improve information dissemination on trade policy changes, including by the WTO Secretariat (through its Trade Policy Review Body), the World Bank, and establishment of the independent Global Trade Alert.144 7.2.2 Contract completion Disputes sometimes arise as a result of disagreement about whether particular legal conditions are met. Legal scholars especially stress the view that countries invoke dispute settlement to address instances in which the original terms of the agreement were vague or incomplete, perhaps because it was too costly to write all possible contingencies into the original agreement. Dispute settlement may present a forum to address legitimate di¤erences of opinion about what behavior the contract was intended to induce. We brie‡y describe here research that adopts this perspective.145 Maggi and Staiger (2011) provide a …rst analysis of roles that dispute settlement might play in a trade agreement modeled as an incomplete contract. Trade takes place between two countries in a single industry. The importing country has the policy option of free trade or protection, and makes its choice after both countries and the “court”(i.e., the dispute settlement process) observe realization of state variables that a¤ect welfare levels. However, it is too costly to describe the states in an ex-ante contract, and the court does not observe the joint payo¤ to the two countries under the realization of the state variables. The two countries thus cannot write a complete state- contingent agreement. Maggi and Staiger focus on three forms of contract incompleteness: the 144 Bown (2011b) provides a more complete discussion of these three enhanced initiatives for additional trade policy monitoring and surveillance that arose during the Great Recession, with emphasis on motivations underlying the World Bank’ s TTBs information dissemination activities in particular. 145 There is also a growing literature that explores the role of dispute settlement procedures in facilitating ex-post renegotiation of trade agreements where commitments may be viewed as either property rules or liability rules. See Maggi and Staiger (forthcoming) where there are no disputes in equilibrium and Beshkar (2010, 2013) and Maggi and Staiger (2013) where disputes arise in equilibrium. See also Bagwell (2008), Lawrence (2003) and the legal discussion of Schwartz and Sykes (2002). 89 contract may leave gaps, it may be overly rigid, or it may use vague language which leaves the obligations under the contract ambiguous in some states. The theory explores ways in which the agreement might articulate various roles for the court, including its degree of “activism,” such as limiting the court to enforce clearly stated obligations, interpreting ambiguous obligations, …lling gaps, and modifying clearly stated obligations. The model assumes the court is costly to use and makes errors in decisions, and the contract and the court mandate are chosen together as part of the optimal design of the institution. A two-period extended model is also considered to explore implications of allowing the court to establish precedent. Maggi and Staiger (2011) derive a number of results assessing di¤erent degrees of court activism. First, it is never optimal to allow the court to modify obligations that are clearly stated in the con- tract. Second, as court decisions become more accurate, its mandate changes from non-involvement (beyond a pure enforcement role) to a more activist role, such as interpreting vague clauses and even …lling in the contract where it is silent. Third, while precedent can improve e¢ ciency by reducing expected future litigation costs on issues that would have been litigated in any event, o¤setting negative e¤ects include the costly increase of additional disputes arising over issues that would not otherwise have been litigated. Overall, the bene…ts of precedent outweigh the costs when governments are impatient and when the court is more likely to make mistakes. Finally, this model can also be used to describe other important features of the disputes that arise, especially in relation to the characteristics of the court. First, there is a tendency of the court to exhibit a pro-trade bias in its legal rulings if the litigation costs to the complainant (exporter) are high relative to the respondent. However, this pattern arises due to selection e¤ects associated with the importing country being more likely to actually have been at fault in the …rst place. Second, and following the same intuition as arises in the enforcement literature, the o¤-equilibrium in‡uences of the court are what generate its bene…cial impacts. Third, and also in parallel with the enforcement literature, the frequency with which countries trigger disputes provides little information on the performance of the court or the value of the dispute settlement process to the overall agreement, a result that occurs in the model because of the interaction between the optimal choice of the contract and the optimal mandate of the court. 7.2.3 Dispute Settlement Reform Proposals Despite the apparent success of WTO dispute settlement, there have been a number of proposals for reform. Here we consider the literature that has evaluated some of these proposals. For the purpose of evaluation, it is also important to consider why such reforms may be desirable. One concern is that tari¤ retaliation, when implemented, may generate e¢ ciency losses. A second concern may be the “equity”properties of dispute settlement that arise for countries with bilateral trade imbalances that may not have a su¢ ciently credible enforcement threat to induce policy compliance. A …rst proposal is to replace the current system of trade retaliation with a system of …nancial compensation; this could also be motivated by the realization that …nancial transfers have emerged 90 in a couple of instances as part of voluntary settlements in actual WTO disputes.146 To explore this proposal, Limão and Saggi (2008)147 analyze the potential role of “…nes” (…nancial transfers) in a self-enforcing, repeated-game framework. In order for …nes to support low cooperative tari¤s, they must be backed by an additional policy instrument that is not controlled by the deviating country. The natural instrument to consider is a tari¤, and thus tari¤ retaliation remains the central backstop of the system. Accordingly, Limão and Saggi …nd that a system in which retaliatory tari¤s enforce the payment of …nes yields no more cooperation than a system that relies directly on retaliatory tari¤s. While the two systems are equivalent in the absence of disputes, Limão and Saggi argue that a system with …nes o¤ers an advantage when shocks occur that lead to disputes. The intuition is simply that …nes are a more e¢ cient transfer instrument.148 Mexico introduced a separate reform proposal based on the idea that countries could be allowed to trade their right to retaliate to other trading partners in lieu of implementing retaliation them- selves (WTO, 2002). This option may be of special interest to smaller countries which perceive less bene…t from retaliating on their own. Bagwell, Mavroidis and Staiger (2007) consider di¤erent auction formats to analyze tradeable retaliation rights in a trade agreement setting.149 They begin with a basic auction, in which two competing importers of the product on which retaliation can take place bid for the right to retaliate. The basic auction has positive externalities: the losing country prefers that the other bidding country win and impose a retaliatory tari¤ in comparison to the scenario in which no retaliation occurs.150 Intuitively, the retaliatory tari¤ imposed by the winning country lowers the world price of the common import good and thereby generates a terms-of-trade gain for the losing country. Due to this positive externality, free-riding is a potential concern, and auction failures (where neither country bids) and misallocations of retaliation rights (due to pooling at the reserve price) can occur. They then consider an extended auction, in which the respondent country is allowed to bid and potentially retire the right of retaliation against it. Both positive and negative externalities can arise in the extended auction, the respondent country always wins, and the retaliation right is always retired thus without the realization of the ine¢ ciencies. The extended auction thus suggests a potential means through which monetary compensation might be extended from a large (respondent) country to a small (complainant) country. Drawing normative inference from these di¤erent auctions requires further thinking about the underlying motivation of the reform to the enforcement mechanism –e.g., is it to ensure compen- sation, to encourage respondent compliance with rulings, or to enhance e¢ ciency? The results suggest that the basic auction generates lower expected revenue for the complainant than the ex- 146 As a settlement in the US –Upland Cotton dispute, the US agreed to transfer $147.3 million annually as a form of technical assistance and capacity building to the government of Brazil (USTR, 2010). In the US – Section 110(5) of the US Copyright Act dispute, the US payed European copyright holders 1.3 million euros annually in compensation. See Bronckers and van den Broek (2005), which also provides a more complete legal articulation of the proposal. 147 See also Bagwell and Staiger (2005b). Extensions include Limão and Saggi (2013). 148 Limão and Saggi (2008) develop this point in by considering the possibility of unanticipated shocks. Bagwell and Staiger (2005b) also explore the bene…ts of monetary transfers in a repeated game model, although in their model (preference) shocks are anticipated and privately observed along the equilibrium path. 149 See also Chen and Potipiti (2010), who derive the optimal auction design for this setting. 150 Bagwell, Mavroidis and Staiger’s (2007) formal analysis is closely related to that of Jehiel and Moldovanu (2000). 91 tended auction in which the respondent country is also allowed to participate. On the other hand, the compliance and e¢ ciency criteria favor the basic auction under some circumstances. 7.3 WTO Dispute Settlement and PTAs The stumbling-block/building-block literature reviewed in Section 5 indicates that PTA formation can impact the extent to which multilateral tari¤ cooperation can be achieved in a self-enforcing agreement. A related but distinct question concerns the relationship between PTA implementa- tion and WTO dispute settlement activity. While there is little formal econometric work assessing implications of repeated-game models for disputes, there are many examples of actual PTA im- plementation resulting in policy changes that lead PTA non-members to formally challenge those policy changes through GATT or WTO disputes. These examples highlight this as a likely area of continued con‡ict and thus an important area for additional research, especially in light of the proliferation of “WTO-extra” provisions arising under the new wave of PTAs. The GATT period was replete with disputes arising after countries took on new PTA commit- ments involving tari¤s and trade-related policies that led to adjustments of the PTA members’ external trade policy commitments toward non-members. The European Economic Community in particular faced formal GATT disputes in 1973 after the accession of UK, Ireland and Denmark, in 1982 after the accession of Greece, and in 1987 after the accession of Spain and Portugal, where the three disputes were respectively initiated by Canada, the US and Argentina.151 Other and more recent examples of disputes involve new PTA members allegedly adjusting their non-tari¤ policies toward non-members, both immediately as well as long after implementation of the PTA, in the latter case due to PTA rules constraining the conduct of policy toward members in particular. The MERCOSUR tari¤ cuts and customs union between Argentina, Brazil, Paraguay and Uruguay in the early 1990s provide two case studies. Argentina’s preferential tari¤ cuts in footwear under MERCOSUR led to an import surge from Brazil and was the precipitating event behind one important WTO dispute. Argentina’s response to the preferential import surge was to subsequently impose a safeguard on footwear beginning in 1997. However, because of a MERCOSUR legal requirement that safeguards cannot be applied against other PTA members, Argentina exempted imports from Brazil from the policy. Not surprisingly, the policy failed to stem the import surge from Brazil - though it was e¤ective against imports from non-members - and the result was that the EU and Indonesia …led a formal WTO dispute against Argentina. The WTO rulings in the dispute were some of the …rst WTO jurisprudence establishing additional conditions under which countries could viably apply safeguard measures.152 151 Hudec (1993) provides an expanded discussion of Canada’ s dispute over cereals exports (pp. 460-61), US’s dispute over fruit exports (pp. 496-498), and Argentina’ s dispute over corn and sorghum exports (p. 550). 152 Bown, Karacaovali and Tovar (2015, pp. 449-452) discuss the Argentina – Footwear dispute, which is also note- worthy as its legal decisions resulted in the important WTO jurisprudence establishing the principle of “parallelism” whereby safeguard-imposing countries became required to only impose safeguard measures against trading partners whose imports they had actually included in the safeguard’ s injury investigation. We note that there are also impor- tant examples of countries facing WTO disputes brought by PTA partners after they have imposed a safeguard in a way that increased the level of import protection against PTA partners relative to non-partners. An example is the 92 s ban on retreaded tire imports in 2000; Brazil later A second example escalated from Brazil’ claimed a public policy motive based on environmental and human health protection. Uruguay challenged Brazil’s ban under MERCOSUR’s dispute settlement provisions, and the legal ruling re- quired Brazil to remove the ban’s application on retreaded tire imports from MERCOSUR partners. Brazil’s imports from some of the newly exempted MERCOSUR partners subsequently increased, and the EU - another exporter of retreaded tires still subject to the Brazilian import ban - …led a WTO dispute. This dispute also resulted in important WTO jurisprudence – this time for the overlap of trade and environmental policy (Bown and Trachtman, 2009). 7.4 PTA Dispute Settlement and Deep Integration While the GATT approach emphasizes shallow integration, many PTAs pursue deep integration linkages between domestic and trade policies. As we discuss in Section 4.4, these linkages may have s (2005) work implications for the most-cooperative tari¤s that can be enforced; in particular, Limão’ raises the possibility that optimal cooperation in a linked setting with non-pecuniary international spillovers could entail a reduction in trade-policy cooperation, if the linked agreement “borrows” enforcement power from trade policy to achieve non-trade objectives.153 In this section, we focus on a related but distinct theme and consider PTA disputes that are associated with the enforcement of non-trade policies. We regard this area as an especially promising area for future research. Our discussion is motivated by recent developments. For the …rst time, countries are now using trade policy to explicitly enforce some of the new, non-trade policy commitments that countries have undertaken bilaterally, outside of the WTO, in other agreements. Some of these examples stem directly from the “trade and...” provisions - such as environmental or labor standards - that were introduced into the public lexicon and came to prominence in the 1990s when NAFTA was …rst under consideration (Bhagwati and Hudec, 1996). We describe three recent examples here. In 2010, the US initiated its …rst formal dispute for a trading partner’s failure to implement suf- …ciently high labor standards that it had committed to uphold as part of the PTA. As background, the CAFTA-DR that the US signed in 2004 included a number of “WTO-extra” provisions of the kind described by Horn, Mavroidis, and Sapir (2010) - including labor standards, environmental standards, and foreign direct investment provisions - that would subsequently be enforceable under the PTA’s own dispute settlement procedures. When Guatemala allegedly failed to enforce its own labor laws, the US initiated a formal CAFTA-DR dispute (USTR, 2011). Such a dispute has the potential to escalate to tari¤ retaliation. The second example stems from Mexico …ling, winning, and retaliating after a NAFTA dispute over the US failure to liberalize its market for commercial trucking services due to alleged public safety concerns. As background, when NAFTA was implemented in 1994, the US agreed to remove restrictions on commercial truck and cargo shipping services provided by Mexican …rms. However, Dominican Republic - Safeguard Measures dispute; for an analysis see Bown and Wu (2014). 153 There are many high-pro…le examples of countries “borrowing” enforcement through trade retaliation to achieve other objectives, including countries implementing trade sanctions for allegedly political or national security purposes (Hufbauer et al., 2009) – e.g., those imposed recently on North Korea, Iran, and Russia. 93 the US announced in 1995 that it would not remove the restrictions out of alleged safety concerns that the Mexican trucks posed for the US public. Mexico initiated a formal dispute under NAFTA and in 2001 a NAFTA Panel found against the US restrictions. After eight years of the US failing to implement the Panel recommendations, Mexico announced it would seek compensation by implementing NAFTA-authorized tari¤ retaliation over more than $2 billion of imports. Ultimately this retaliation on US exporters galvanized su¢ cient political pressure within the US to allow for policy reform in 2011 whereby Mexican trucks were allowed to service the US market and Mexico ended its period of retaliation (Department of Commerce, 2011). The third example involves the US implementing trade sanctions in 2012 against Argentina by removing the lower tari¤ preferences the US had previously o¤ered under the Generalized System of Preference (GSP) program. The US does not have a PTA with Argentina, and the trade retaliation did not follow from a formal dispute, and it was not authorized by any particular dispute settlement process. The source of friction between the two countries is not even a trade matter; instead, it stems from a dispute over an alleged expropriation of US investments in Argentina, and as such is covered by a bilateral investment treaty between the two countries. The investment dispute was litigated under the International Centre for Settlement of Investment Disputes (ICSID), which determined that Argentina should compensate US investors for damages with a …nancial transfer. The implementation of US trade retaliation beginning in 2012 was an attempt to enforce Argentina’s foreign direct investment commitments and was due to Argentina’s failure to pay roughly $300 million that it owed US investors since 2005-6 (USTR, 2012). The Guatemala (labor standards), US-Mexico (trucking services and safety standards), and Argentina (foreign direct investment) examples highlight the need for further research to evaluate the dispute settlement implications of existing “deeper integration”trade agreements. This is par- ticularly evident if these three recent examples also signal a tendency for such agreements to move over time beyond “soft” law and toward “hard” law, where deep integration policy commitments could then expose highly sensitive behind-the-border measures not only to litigation under formal dispute settlement, but also to enforcement through tari¤ retaliation. 7.5 Summary Our review highlights central features of the WTO dispute settlement system, the theoretical frame- works that attempt to identify a speci…c role for dispute settlement procedures, the complicated relationship between WTO dispute settlement and PTA implementation, and some aspects of PTA dispute settlement in the context of deeper-integration initiatives. In this concluding section, we brie‡y re‡ect on some of the relative merits of the WTO’s multilateral dispute settlement system. While WTO dispute settlement is widely praised, it is challenging to determine a solid bench- mark against which to measure its success. Certainly the willingness of member governments to utilize WTO dispute settlement procedures to examine disputed policies is suggestive that these procedures play an important role. Furthermore, it is also the case that for many bilateral trading relationships, the WTO’s multilateral system may be the only game in town. Indeed, one half to 94 two thirds of world trade occurs between countries that are not in a common PTA and thus between countries without an obvious alternative forum under which to resolve their potential disputes.154 One possible relevant benchmark is to compare the current WTO dispute settlement system to its most immediate predecessor, i.e., the multilateral dispute settlement system under the GATT in the late 1980s. The GATT system, which was relatively toothless and based more on diplomacy than law, spurred the rogue path of unilateralism that the US undertook through its Section 301 actions during the period (Bhagwati and Hudec, 1990). US unilateralism was a signal of its strong displeasure with the old system and at least partially served to illustrate what a world without binding multilateral disciplines could look like. Ultimately these events contributed to ushering in the WTO’s new dispute settlement system in 1995, and the evolving system has arguably performed well in comparison to the GATT system. A second possible benchmark is to compare the WTO dispute settlement system to systems emerging in PTAs. In principle, PTA systems might be customized to better address the deep- integration initiatives that these agreements increasingly pursue. However, it is far from clear that dispute settlement in PTAs can handle the job on its own. First, there are a number of cases in which purely bilateral issues arising between partners in PTAs with relatively well-functioning dispute settlement systems (e.g., NAFTA) could not be resolved internally and ultimately spilled over into WTO dispute settlement anyway. Second, there are other examples (e.g., MERCOSUR), in which PTA dispute settlement decisions pushed PTA members to make policy choices that imposed externalities on non-members, thus leading those non-members to initiate WTO disputes. While the empirical record of PTA dispute settlement use is scant, such examples at least suggest that PTA dispute settlement may cause as many problems as they help resolve. More research is required to better understand the tradeo¤s, incentives, and forces also at work at the particular intersection of preferential and multilateral commitments. Such research may be wherever there of special value now, since it is unlikely that a single overarching rule - such as, ‘ might be a con‡ict in commitments or obligations, WTO law dominates PTA law’- can be relied upon. Indeed, some of the major proponents of new obligations arising under PTAs - such as the US and EU - are also some of the most signi…cant litigants and contributors to WTO jurisprudence. As a thought experiment, consider once again the US-Guatemala dispute over labor standards or the US-Argentina retaliation over investment provisions and compensation. When would it make sense for Guatemala or Argentina to use the WTO to challenge potential US trade sanctions as a violation of its WTO commitments, since there are no explicit WTO provisions authorizing the US to raise its tari¤s for such reasons in the …rst place? Together these considerations lend support to the WTO dispute settlement system, but it is clear that much more research is needed.155 The relative merits of multilateral and preferential dispute 154 The WTO (2011, p. 64) reports that 65 percent (49 percent) of world trade in 2008 was extra-PTA trade excluding (including) the EU. The trade included in the construction of these ratios does include some trade between non-WTO members, so not all of it would therefore be subject to WTO dispute settlement. 155 An additional area in which research is needed is investor-state dispute settlement (ISDS), whereby private …rms in one country have direct access to a separate court system to defend their foreign direct investment from potential government expropriation. One example of a dispute in which ISDS procedures were triggered is the case that 95 settlement systems is thus an important subject for further theoretical and empirical analysis. 8 Conclusion The world trading system seems to be at a crossroads. The emphasis of international cooperation has apparently experienced a momentum shift away from the multilateral and nondiscriminatory framework of the GATT/WTO in favor of discriminatory arenas under new PTAs, and away from negotiations emphasizing shallow integration and toward negotiations stressing increasingly deep integration. Understanding the implications of these shifts is important. What is on the line is a choice over which international institutions will set the future rules of globalization and shape the trade-o¤s we face in a globalized world economy. The best path forward may depend on how we have arrived at this particular crossroads. In broad terms, there are two possibilities suggested by the trade agreements literature. One possibility is that fundamental changes in the global economy, such as the rise in global supply chains and o¤shoring and the emergence of BRICS, have somehow a¤ected the kinds of rules needed for international economic cooperation. According to this possibility, countries must now negotiate deeper constraints in what traditionally was considered the realm of domestic policy making; and as a consequence, greater restrictions on national sovereignty are now an inevitable feature of globalization, with PTAs the most e¢ cacious institution for carrying out the task. The second possibility is that changes in the global economy have created new challenges for the WTO, but the rules needed for international economic cooperation are still fundamentally the same. Under this possibility, the GATT/WTO approach to shallow integration may be in trouble, but its problems are …xable provided that the WTO is supported and its approach strengthened. On balance, our review of the literature to date favors the second possibility. The WTO is not passé. Subject to some caveats, it receives strong support from economic principles. This position is only enhanced when also taking into consideration the WTO’s relatively successful track record of resolving bilateral frictions through its system of dispute settlement. At the same time, our review also suggests directions in which the WTO approach can be improved. It is likely that the task of shallow integration is not yet completed and important work remains to be done, especially for some of the major emerging economies and less developed countries. However, creative adaptations to the GATT/WTO’s historically successful principles of reciprocity and nondiscrimination could seemingly be re-deployed - albeit in a more guided and targeted way - to address these challenges. While our review of the literature to date broadly supports the GATT/WTO shallow integration approach, we also highlight research that raises important caveats. As we discuss, the presence of non-pecuniary international externalities may motivate deeper forms of integration, and it is possi- ble that governments can enforce greater overall cooperation in an agreement that links trade and US investors took against Argentina under the ICSID described above. Sykes (2005b) provides an introduction to bilateral investment treaties (BITs), the distinctive form of dispute settlement arising under such agreements, and how this di¤ers from the system typically arising under trade agreements. 96 domestic policies.156 We also describe research indicating that the presence of private information or commitment problems may provide rationales for certain forms of deep integration. As well, some recent research indicates that the e¤ectiveness of the GATT/WTO shallow integration ap- proach may be undermined by o¤shoring and its implications for international price determination. These research areas are at early stages, however, and their full implications are uncertain. As such, they represent especially important areas for future research. Our review also highlights some important considerations that have received relatively little attention in the formal trade agreements literature. First, actual negotiations may entail bargaining frictions, which are plausibly higher for negotiations that involve many countries and complex issues. Second, gains from international economic cooperation in certain policies may take the form of coordination gains.157 Further work on these topics may o¤er new insights about deep integration initiatives and the possible bene…ts of negotiations among smaller groups of countries. Finally, the e¤ect of deep integration initiatives on third-party countries, and especially developing countries, is an important and under-explored direction for additional research.158 We close our review with a brief discussion of one potential approach to strengthening the WTO in response to these challenges that has received recent attention in the literature, namely, the possibility of pursuing critical mass agreements (CMAs) or plurilateral agreements (PAs). Such agreements can be thought of as a hybrid between the preferential trade agreements and what has s “Single Undertaking”approach.159 A CMA is de…ned as an agreement otherwise been the WTO’ in a WTO-covered area between a subset of WTO member countries whereby, because WTO disciplines apply, the bene…ts the members o¤ered to one another under the CMA must be extended to all other WTO members on an MFN basis. In order to prevent free riding, CMAs therefore may be only likely to occur between major subsets of large countries. The main example of a successfully concluded CMA occurring under the WTO is the 1997 Information Technology Agreement which cut tari¤s to zero in products covered under the agreement, and for which the original negotiations were 156 Indeed, the WTO TRIPS agreement is a deep integration agreement and a possible example of this sort of linkage within the WTO. As we discuss in footnote 1, we do not include in our survey the literature that investigates the purpose and design of the TRIPS agreement, as it is not a market access agreement. Recent work adopting TRIPS as the focus includes Maskus (2000), Grossman and Lai (2004), Scotchmer (2004) and Geng and Saggi (2014). Similarly, we do not address the possible role of trade agreements with regard to other non-pecuniary international externalities, such as global warming, human rights, and geopolitics and global security. 157 See, e.g., Hoekman (2014) on possible coordination gains in the context of the Trade Facilitation Agreement. 158 We mention here two further important issues for future research. The …rst issue concerns the extent to which power asymmetries between developed and developing countries impact deep integration negotiations among par- ticipating countries more readily in PTAs than in the WTO. The second issue is whether deep integration PTAs facilitate or hinder deeper integration at the multilateral level. Deep integration PTAs might o¤er laboratories in which to discover e¤ective disciplines and thereby facilitate multilateral e¤orts, but they might also generate incom- patible standards across PTAs that “lock in” members and thereby inhibit multilateral harmonization. For further discussion of these and related issues, see Bhagwati, Krishna and Panagariya (2014), Trebilcock (2014) and WTO (2011, p. 182). 159 Plurilateral agreements were a common outcome of GATT rounds prior to the Uruguay Round. For example, the conclusion of the Tokyo Round in 1979 led to a number of plurilateral agreements adopted mainly by industrialized countries, in issue areas such as subsidies and countervailing measures, technical barriers to trade (standards), import licensing procedures, government procurement, customs valuation, antidumping, bovine meat, dairy, and trade in civil aircraft. The Kennedy Round of negotiations in the 1960s also brought forward a plurilateral code on antidumping. 97 concluded between only 29 WTO members, though this has subsequently grown to more than 70. A PA, on the other hand, is an agreement between a subset of WTO member countries in an area where either WTO disciplines are not applied (“WTO-extra”) or where they are extended (“WTO-plus”), and to which therefore the bene…ts o¤ered to one another would not need to be extended to other WTO members. Examples under the WTO include the Agreement on Government Procurement and the Agreement on Civil Aviation.160 As Hoekman and Mavroidis (2013, 2014) point out, PAs and PTAs have important similari- ties and di¤erences.161 Similarities include that both can be applied on a non-MFN basis without violating WTO rules and both seem to be going beyond the shallow integration approach of tradi- tional GATT/WTO disciplines and into deep integration. On the other hand, important di¤erences may make PAs more appealing than PTAs from the perspective of the multilateral system. First, PAs are “open” in that other WTO members should (in principle) be allowed an explicit path to accede to the PA in the future, whereas PTAs do not typically have an open accession process for potentially interested trading partners. Second, attempts to bring PAs into the WTO system would potentially make the content of these agreements much more transparent, which may be especially important for non-signatory countries. Third, problems arising between PA signatories would be addressed through litigation taking place under the WTO, thus more likely completing the contract in a coherent way, as opposed to the potential fragmentation of international jurispru- dence that might otherwise arise under PTA dispute settlement provisions. Combined, Hoekman and Mavroidis suggest these features of PAs could make them less likely to impose externalities on third countries than PTAs. Nevertheless, the full theoretical implications of such alternative approaches have yet to be fully explored by the literature, and thus further analysis along these lines is a ripe area for additional research. 9 References Aghion, Philippe, Pol Antràs, and Elhanan Helpman. 2007. “Negotiating free trade.” Journal of Interna- tional Economics 73(1): 1-30. Agur, Itai. 2008. “The U.S. trade de…cit, the decline of the WTO and the rise of regionalism.” Global Economy Journal 8(3): 1-34. Allee, Todd and Manfred Elsig. 2015. “Dispute Settlement Provisions in PTAs: New Data and New Concepts”In Trade Cooperation: The Purpose, Design and E¤ ects of Preferential Trade Agreements, edited by Andreas Dür and Manfred Elsig, 319-352. Cambridge, UK: Cambridge University Press. 160 Negotiations using CMA/PA approaches in new issue areas include attempts to liberalize trade in environmental goods and additional liberalization in services under a proposed TiSA (Trade in Services Agreement). Negotiations outside of the WTO among a small subset of mostly high-income countries have taken place for additional intellectual property rights protection and resulted in the anti-counterfeiting agreement (ACTA). 161 For additional discussions of plurilateral agreements, or what is sometimes referred to as potential “variable geometry” under the WTO, see also Lawrence (2006), Levy (2006) and Trebilcock (2014). 98 Amador, Manuel and Kyle Bagwell. 2012. “Tari¤ revenue and tari¤ caps.” American Economic Review: Papers and Proceedings 102(3): 459-65. Amador, Manuel and Kyle Bagwell. 2013. “The theory of optimal delegation with an application to tari¤ caps.” Econometrica 81(4): 1541-99. Anderson, James E. and Eric van Wincoop. 2002. “Borders, trade and welfare.”In Brookings Trade Forum 2001, edited by Dani Rodrik and Susan M. Collins, 207-44. Washington, DC: Brookings Institution. Anderson, Kym, Gordon Rausser, and Johan Swinnen. 2013. “Political Economy of Public Policies: Insights from Distortions to Agricultural and Food Markets.” Journal of Economic Literature 51(2): 423-477. Antràs, Pol and Robert W. Staiger. 2012a. “Trade agreements and the nature of price determination.” American Economic Review: Papers and Proceedings 102(3): 470-6. Antràs, Pol and Robert W. Staiger. 2012b. “O¤shoring and the role of trade agreements.” American Economic Review 102(7): 3140-3183 Antràs, Pol and Stephen R. Yeaple. 2014. “Multinational …rms and the structure of international trade.” In Handbook of International Economics, edited by Gita Gopinath, Elhanan Helpman, and Kenneth Rogo¤, Volume 4. Amsterdam, NL: Elsevier. Athey, Susan and Kyle Bagwell. 2001. “Optimal collusion with private information.” Rand Journal of Economics 32(3): 428-465. Bacchetta, Marc and Roberta Piermartini. 2011. “The value of bindings.” Sta¤ Working Paper ERSD- 2011-13, World Trade Organization. Bagwell, Kyle. 2008. “Remedies in the WTO: an economic perspective.” In The WTO: Governance, Dispute Settlement & Developing Countries, edited by Merit E. Janow, Victoria J. Donaldson and Alan Yanovich, 733-770. Huntington, New York: Juris Publishing. Bagwell, Kyle. 2009. “Self-enforcing trade agreements and private information.” NBER Working Paper No. 14812. Bagwell, Kyle and Seung Hoon Lee. 2015. “Trade policy under monopolistic competition with …rm selec- tion.” Mimeogr., Stanford University. Bagwell Kyle, Petros C. Mavroidis, and Robert W. Staiger. 2002. “It’s a question of market access.” American Journal of International Law 96(1): 56-76. Bagwell, Kyle, Petros C. Mavroidis, and Robert W. Staiger. 2007. “Auctioning countermeasures in the WTO.” Journal of International Economics 73(2): 309-32. Bagwell, Kyle and Robert W. Staiger. 1990. “A Theory of Managed Trade.” American Economic Review 80(4): 779-95. 99 Bagwell, Kyle and Robert W. Staiger. 1994. “The sensitivity of strategic and corrective R&D policy in oligopolistic industries.” Journal of International Economics 36(1-2): 133-50. Bagwell, Kyle and Robert W. Staiger. 1997a. “Multilateral tari¤ cooperation during the formation of free trade areas.” International Economic Review 38(2): 291-319. Bagwell, Kyle and Robert W. Staiger. 1997b. “Multilateral tari¤ cooperation during the formation of customs unions.” Journal of International Economics 42(1-2): 91-123. Bagwell, Kyle and Robert W. Staiger. 1997c. “Collusion over the business cycle.” Rand Journal of Eco- nomics 28(1): 82-106. Bagwell, Kyle and Robert W. Staiger. 1998. “Will preferential agreements undermine the multilateral trading system?” Economic Journal 108(449): 1162-82. Bagwell, Kyle and Robert W. Staiger. 1999a. “An Economic Theory of GATT.” American Economic Review 89(1): 215-48. Bagwell, Kyle and Robert W. Staiger. 1999b. “Regionalism and Multilateral Tari¤ Cooperation.” In International Trade Policy and the Paci…c Rim, edited by John Piggott and Alan Woodland, 157-85. London: MacMillan. Bagwell, Kyle and Robert W. Staiger. 2001a. “Reciprocity, non-discrimination and preferential agreements in the multilateral trading system.” European Journal of Political Economy 17(2): 281-325. Bagwell, Kyle and Robert W. Staiger. 2001b. “Domestic policies, national sovereignty and international economic institutions.” Quarterly Journal of Economics 116(2): 519-62. Bagwell, Kyle and Robert W. Staiger. 2001c. “Strategic Trade, Competitive Industries and Agricultural Trade Disputes.” Economics and Politics 13(2): 113-28. Bagwell, Kyle and Robert W. Staiger. 2002. The Economics of the World Trading System. Cambridge, MA: The MIT Press. Bagwell, Kyle and Robert W. Staiger. 2003. “Protection and the Business Cycle.” Advances in Economic Analysis & Policy 3(1): Article 3 (available at http://www.bepress.com/bejeap/advances/vol3/iss1/art3/) Bagwell, Kyle and Robert W. Staiger. 2005a. “Multilateral Trade Negotiations, Bilateral Opportunism and the Rules of GATT/WTO.” Journal of International Economics 67(2): 268-94. Bagwell, Kyle and Robert W. Staiger. 2005b. “Enforcement, Private Political Pressure and the GATT/WTO Escape Clause.” Journal of Legal Studies 34(2): 471-513. Bagwell, Kyle and Robert W. Staiger. 2006. “Will International Rules on Subsidies Disrupt the World Trading System?” American Economic Review 96(3): 877-95. 100 Bagwell, Kyle and Robert W. Staiger. 2010a. “The WTO: Theory and Practice.” Annual Review of Economics 2: 223-56. Bagwell, Kyle and Robert W. Staiger. 2010b. “Backward Stealing and Forward Manipulation in the WTO.” Journal of International Economics 82(1): 49-62. Bagwell, Kyle and Robert W. Staiger. 2011. “What do trade negotiators negotiate about? Empirical evidence from the World Trade Organization.” American Economic Review 101(4): 1238-73. Bagwell, Kyle and Robert W. Staiger. 2012a. “Pro…t Shifting and Trade Agreements in Imperfectly Competitive Markets.” International Economic Review 53(4): 1067-1104. Bagwell, Kyle and Robert W. Staiger. 2012b. “The economics of trade agreements in the linear cournot delocation model.” Journal of International Economics 88(1): 32-46. Bagwell, Kyle and Robert W. Staiger. 2014. “Can the Doha Round be a Development Round? Setting a Place at the Table.” In Globalization in an Age of Crisis: Multilateral Economic Cooperation in the Twenty-First Century, edited by Robert C. Feenstra and Alan M. Taylor, 91-124. Chicago, IL: University of Chicago Press for the NBER. Bagwell, Kyle and Robert W. Staiger. 2015. “Delocation and trade agreements in imperfectly competitive markets.” Research in Economics 69(2): 132-56. Bagwell, Kyle, Robert W. Staiger and Alan O. Sykes. 2013. “Border Instruments.”In Legal and Economic Principles of World Trade Law, edited by Henrik Horn and Petros C. Mavroidis, 68-204. New York: Cambridge University Press for the American Law Institute. Bagwell, Kyle and Alan O. Sykes. 2004. “Chile - Price Band System and Safeguard Measures Relating to Certain Agriculture Products.” World Trade Review 3(3): 507-28. Bagwell, Kyle and Alan O. Sykes. 2005. “India - measures a¤ecting the automotive sector.” In The WTO Case Law of 2002: The American Law Institute Reporters’Studies, edited by Henrik Horn and Petros C. Mavroidis, 158-178. New York: Cambridge University Press for the American Law Institute. Bajona, Claustre and Josh Ederington. 2011. “Domestic Policies, Hidden Protection and the GATT/WTO.” Mimeogr., Ryerson University. Baldwin, Richard. 1987. “Politically realistic objective functions and trade policy PROFs and tari¤s.” Economics Letters 24(3): 287-290. Baldwin, Richard. 2014. “WTO 2.0: governance of 21st century trade.” The Review of International Organizations 9(2): 261-283. Battigalli, Pierpaolo and Giovanni Maggi. 2003. “International agreements on product standards: an incomplete contracting theory.” NBER Working Paper No. 9533. 101 Beshkar, Mostafa. 2010. “Trade skirmishes and safeguards: a theory of the WTO dispute settlement process.” Journal of International Economics 82(1): 35-48. Beshkar, Mostafa. 2013. “Arbitration and renegotiation in trade agreements.” Mimeogr., Indiana. Beshkar, Mostafa and Eric W. Bond. 2014. “Cap and escape in trade agreements.”Mimeogr., Vanderbilt. Beshkar, Mostafa, Eric W. Bond, and Youngwoo Rho. Forthcoming. “Tari¤ binding and overhang: theory and evidence.” Journal of International Economics. Bhagwati, Jagdish. 2008. Termites in the Trading System: How Preferential Agreements Undermine Free Trade. Oxford University Press. Bhagwati, Jagdish and Robert E. Hudec, Eds. 1990. Aggressive Unilateralism: America’s 301 Trade Policy and the World Trading System. Ann Arbor, MI: University of Michigan Press. Bhagwati, Jagdish and Robert E. Hudec, Eds. 1996. Fair Trade and Harmonization: Prerequisites for Free Trade? Volumes I and II. Cambridge, MA: The MIT Press. Bhagwati, Jagdish, Pravin Krishna and Arvind Panagariya. 2014. “The World Trade System: Trends and Challenges.” Mimeogr., Columbia. Bhagwati, Jagdish and V. K. Ramaswami. 1963. “Domestic distortions, tari¤s and the theory of optimum subsidy.” Journal of Political Economy 71(1): 44-50. Blanchard, Emily J. 2007. “Foreign direct investment, endogenous tari¤s, and preferential trade agree- ments.” BE Journal of Economic Analysis & Policy 7. Blanchard, Emily J. 2010. “Reevaluating the role of trade agreements: Does investment globalization make the WTO obsolete?” Journal of International Economics 82(1): 63-72. Blanchard, Emily J. and Xenia Matschke. forthcoming. “US multinationals and preferential market access.” Review of Economics and Statistics. Blonigen, Bruce A. and Chad P. Bown. 2003. “Antidumping and retaliation threats.” Journal of Interna- tional Economics 60(2): 249-73. Blonigen, Bruce A. and Thomas J. Prusa. 2003. “Antidumping.” In Handbook of International Trade, edited by E. Kwan Choi and James Harrigan, 251-284. Oxford, UK: Blackwell Publishers. Bond. Eric W., Raymond G. Riezman and Constantinos Syropoulos. 2004. “A strategic and welfare theoretic analysis of free trade areas.” Journal of International Economics 64: 1-27. Bond, Eric W. and Constantinos Syropoulos. 1996a. “The size of trading blocs: Market power and world welfare e¤ects.” Journal of International Economics 40(3-4): 411-37. 102 Bond, Eric W. and Constantinos Syropoulos. 1996b. “Trading Blocs and the Sustainability of Inter- Regional Cooperation.” In The New Transatlantic Economy, edited by Matthew Canzoneri, Wilfred Ethier, and Vittorio Grilli, 118-141. Cambridge, UK: Cambridge University Press. Bond, Eric. W., Constantinos Syropoulos and L. Alan Winters. 2001. “Deepening of regional integration and multilateral trade agreements.” Journal of International Economics 53(2): 335-61. s Article XXIII and the WTO’ Bown, Chad P. 2002. “The Economics of Trade Disputes, the GATT’ s Dispute Settlement Understanding.” Economics and Politics 14(3): 283-323. Bown, Chad P. 2004a. “Trade policy under the GATT/WTO: empirical evidence of the equal treatment rule.” Canadian Journal of Economics 37(3): 678-720. Bown, Chad P. 2004b. “Trade Disputes and the Implementation of Protection under the GATT: An Empirical Assessment.” Journal of International Economics 62(2): 263-294. Bown, Chad P. 2004c. “On the Economic Success of GATT/WTO Dispute Settlement.” Review of Eco- nomics and Statistics 86(3): 811-823. Bown, Chad P. 2005a. “Trade remedies and World Trade Organization dispute settlement: why are so few challenged?” Journal of Legal Studies 34(2): 515-555. Bown, Chad P. 2005b. “Participation in WTO dispute settlement: complainants, interested parties and free riders.” World Bank Economic Review 19(2): 287-310. Bown, Chad P. 2009. Self-Enforcing Trade: Developing Countries and WTO Dispute Settlement. Wash- ington, DC: Brookings Institution Press. Bown, Chad P. 2010. “China’s WTO Entry: Antidumping, Safeguards, and Dispute Settlement.” In China’s Growing Role in World Trade, edited by Robert C. Feenstra and Shang-Jin Wei, 281-337. Chicago, IL: University of Chicago Press for the NBER. Bown, Chad P. 2011a. “Taking Stock of Antidumping, Safeguards and Countervailing Duties, 1990-2009.” The World Economy 34(12): 1955-1998. Bown, Chad P. 2011b. “Introduction.” In The Great Recession and Import Protection: The Role of Temporary Trade Barriers, edited by Chad P. Bown, 1-51. London, UK: CEPR and World Bank. Bown, Chad P. 2014. Temporary Trade Barriers Database. The World Bank. Available at http:/econ. worldbank.org/ttbd/, last accessed July 25, 2014. Bown, Chad P. and Meredith A. Crowley. 2006. “Policy externalities: how U.S. antidumping a¤ects Japanese exports to the EU.” European Journal of Political Economy 22(3): 696-714. Bown, Chad P. and Meredith A. Crowley. 2007. “Trade de‡ection and trade depression.” Journal of International Economics 72(1): 176-201. 103 Bown, Chad P. and Meredith A. Crowley. 2010. “China’s Export Growth and the China Safeguard: Threats to the World Trading System?” Canadian Journal of Economics 43(4): 1353-1388. Bown, Chad P. and Meredith A. Crowley. 2013a. “Self-Enforcing Trade Agreements: Evidence from Time-Varying Trade Policy.” American Economic Review 103(2): 1071-1090. Bown, Chad P. and Meredith A. Crowley. 2013b. “Import Protection, Business Cycles, and Exchange Rates: Evidence From the Great Recession.” Journal of International Economics 90(1): 50-64. Bown, Chad P. and Meredith A. Crowley. 2014. “Emerging Economies, Trade Policy, and Macroeconomic Shocks.” Journal of Development Economics 111: 261-273. Bown, Chad P. and Bernard M. Hoekman. 2008. “Developing Countries and Enforcement of Trade Agree- ments: Why Dispute Settlement Is Not Enough.” Journal of World Trade 42(1): 177-203. Bown, Chad P., Baybars Karacaovali, and Patricia Tovar. 2015. “What Do We Know About Preferential Trade Agreements and Temporary Trade Barriers?” In Trade Cooperation: The Purpose, Design and E¤ ects of Preferential Trade Agreements, edited by Andreas Dür and Manfred Elsig, 433-462. Cambridge, UK: Cambridge University Press. Bown, Chad P. and Petros C. Mavroidis, Eds. 2013. The WTO Case Law of 2011. Cambridge, UK: Cambridge University Press. Bown, Chad P. and Joost Pauwelyn, Eds. 2010. The Law, Economics and Politics of Retaliation in WTO Dispute Settlement. Cambridge, UK: Cambridge University Press. Bown, Chad P. and Thomas J. Prusa. 2011. “U.S. Antidumping: Much Ado About Zeroing.”In Un…nished Business? The WTO’s Doha Agenda, edited by William J. Martin and Aaditya Mattoo, 355-392. London, UK: CEPR and World Bank. Bown, Chad P. and Kara M. Reynolds. 2015. “Trade Flows and Trade Disputes.” Review of International Organizations 10(2): 145-177. Bown, Chad P. and Michele Ruta. 2010. “The Economics of Permissible WTO Retaliation.”In The Law, Economics and Politics of Retaliation in WTO Dispute Settlement, edited by Chad P. Bown and Joost Pauwelyn, 149-193. Cambridge, UK: Cambridge University Press. Bown, Chad P. and Alan O. Sykes. 2008. “The Zeroing Issue: A Critical Analysis of Softwood V.” World Trade Review 7(1): 121-142. Bown, Chad P. and Patricia Tovar. 2011. “Trade Liberalization, Antidumping, and Safeguards: Evidence s Tari¤ Reform.” Journal of Development Economics 96(1): 115-125. from India’ Bown, Chad P. and Joel P. Trachtman. 2009. “Brazil – Measures A¤ ecting Imports of Retreaded Tyres : A Balancing Act.” World Trade Review 8(1): 85-135. 104 Bown, Chad P. and Mark Wu. 2014. “Safeguards and the Perils of Preferential Trade Agreements: Do- minican Republic - Safeguard Measures.” World Trade Review 13(2): 179-227. Brainard, Lael. 1994. “Last one out wins: trade policy in an international exit game.” International Economic Review 35(1): 151-72. Brander, James A. 1995. “Strategic trade policy.”In Handbook of International Economics, edited by Gene M. Grossman and Kenneth Rogo¤, 3: 1395-1455. Amsterdam, NL: North-Holland. Brander, James A. and Barbara Spencer. 1981. “Tari¤s and the extraction of foreign monopoly rents under potential entry.” Canadian Journal of Economics 14(3): 371-89. Brander, James A. and Barbara Spencer. 1984. “Tari¤ Protection and Imperfect Competition.”In Monopo- listic Competition and Product Di¤ erentiation and International Trade, edited by Henryk Kierzkowski, 194-206. New York: Oxford Economic Press. Brander, James A. and Barbara Spencer. 1985. “Export subsidies and market share rivalry.” Journal of International Economics 18(1-2): 83-100. Brock, William A. and Stephen P. Magee. 1978. “The Economics of Special Interest Politics.” American Economic Review 68(2): 246-250. Broda, Christian, Nuno Limão, and David E. Weinstein. 2008. “Optimal tari¤s and market power: the evidence.” American Economic Review 98(5): 2032-65. Bronckers, Marco and Naboth van den Broek. 2005. “Financial Compensation in the WTO: Improving the Remedies of WTO Dispute Settlement.” Journal of International Economic Law 8(1): 101-126. Brou, Daniel and Michele Ruta. 2013. “A commitment theory of subsidy agreements.” BE Journal of Economic Analysis & Policy 13(1): 239-270. Cadot, Olivier, Jaime de Melo and Marcelo Olarreaga. 1999. “Regional integration and lobbying for tari¤s against non-members.” International Economic Review 40(3): 635-57 Caliendo, Lorenzo and Fernando Parro. 2015. “Estimates of the trade and welfare e¤ects of NAFTA.” Review of Economic Studies 82(1): 1-44. Calvo-Pardo, Hector, Caroline Freund, and Emanuel Ornelas. 2011. “The ASEAN Free Trade Agreement: Impact on Trade Flows and External Trade Barriers.” In Costs and Bene…ts of Economic Integration in Asia, edited by Robert J. Barro and Jong-Wha Lee, 157-186. Oxford, UK: Oxford University Press. Caplin, Andrew and Kala Krishna. 1988. “Tari¤s and the most-favored-nation clause: A game-theoretic approach.” Seoul Journal of Economics 1(3): 267-89. Carmichael, Calum M. 1987. “The control of export credit subsidies and its welfare consequences.” Journal of International Economics 23(1-2): 1-19. 105 s tari¤ structure.”Canadian Journal Caves, Richard E. 1976. “Economic models of political choice: Canada’ of Economics 9(2): 278-300. Chang, Pao-Li and Myoung-Jae Lee. 2011. “The WTO Trade E¤ect.” Journal of International Economics. 85(1): 53-71. Chang, Won and L. Alan Winters. 2002. “How regional blocs a¤ect excluded countries: the price e¤ects of MERCOSUR.” American Economic Review 92(4): 889-904. Chase, Claude, Alan Yanovich, Jo-Ann Crawford, and Pamela Ugaz. 2013. “Mapping of Dispute Settlement Mechanisms in Regional Trade Agreements: Innovative or Variations on a Theme?” WTO Sta¤ Working Paper, No. ERSD-2013-07. Chen, Bo and Tanapong Potipiti. 2010. “Optimal selling mechanisms with countervailing positive exter- nalities and an application to tradable retaliation in the WTO.” Journal of Mathematical Economics 46(5): 825-43. Chen, Maggie X. and Aaditya Mattoo. 2008. “Regionalism in standards: good or bad for trade?” Canadian Journal of Economics 41(3): 838-63. Choi, Jay Pil. 1995. “Optimal Tari¤s and the Choice of Technology: Discriminatory Tari¤s vs. the Most Favored Nation Clause.” Journal of International Economics 38(1-2): 143-60. Clausing, Kimberly A.. “Trade creation and trade diversion in the Canada-United States Free Trade Agreement.” Canadian Journal of Economics 34(3): 678-96. Copeland, Brian R. 1990. “Strategic interaction among nations: negotiable and non-negotiable trade barriers.” Canadian Journal of Economics 23(1): 84-108. Corden, W. Max. 1971. The Theory of Protection. Oxford: Oxford University Press. Costinot, Arnaud. 2008. “A comparative institutional analysis of agreements on product standards.” Journal of International Economics 75(1): 197-213. Costinot, Arnaud, Dave Donaldson, Jonathan Vogel and Iván Werning. 2015. “Comparative advantage and optimal trade policy.” Quarterly Journal of Economics 130(2): 659-702. Crowley, Meredith A. 2011. “Cyclical dumping and US antidumping protection: 1980-2001.” Federal Reserve Bank of Chicago Working Paper : WP-2007-21. Curzon, Gerard. 1965. Multilateral Commercial Diplomacy: The General Agreement on Tari¤ s and Trade and Its Impact on National Commercial Policies and Techniques. London, U.K.: Michael Joseph. Dai, Mian, Yotov, Yoto V., and Thomas Zylkin. 2013. “On the trade diversion e¤ects of Free Trade Agreements.” Economics Letters 122(2): 321-325. 106 Dam, Kenneth W. 1970. The GATT: Law and International Economic Organization. Chicago, IL: University of Chicago Press. Damuri, Yose Rizal. 2012. “How Preferential are Preferential Trade Agreements?: Analysis of Product Exclusions in PTAs.” CTEI Working Paper No. CTEI-2012-3, Geneva. Davey, William J. and André Sapir. 2009. “The Soft Drinks Case: The WTO and Regional Agreements.” World Trade Review 8(1): 5-23. Davis, Christina. 2012. Why Adjudicate? Enforcing Trade Rules in the WTO. Princeton, NJ: Princeton University Press. Demidova, Svetlana. 2015. “Trade Policies, Firm Heterogeneity, and Variable Markups.” Mimeogr. Mc- Master. Department of Commerce. 2011. “Foreign Retaliations.” International Trade Administration. Available at http://www.trade.gov/mas/ian/tradedisputes-enforcement/retaliations/tg_ian_002094.asp , last accessed July 25, 2014. DeRemer, David R. 2013a. “Domestic Policy Coordination in Imperfectly Markets.” ECARES Working Paper 2013-46. DeRemer, David R. 2013b. “The Evolution of International Subsidy Rules.” ECARES Working Paper 2013-45. Dhingra, Swati. forthcoming. “Reconciling Observed Tari¤s and the Median Voter Model.” Economics and Politics. Disdier, Anne-Celia, Lionel Fontagne, and Olivier Cadot. 2015. “North-South standards harmonization and international trade.” World Bank Economic Review 29(2): 1-26. Dixit, Avinash. 1987. “Strategic Aspects of Trade Policy.” In Advances in Economic Theory: Fifth World Congress, edited by Truman F. Bewley, 329-62. New York, NY: Cambridge University Press. Dür, Andreas, Leonardo Baccini, and Manfred Elsig. 2014. “The Design of International Trade Agreements: Introducing a New Database.” Review of International Organizations 9(3): 353-375. Dutt, Pushan, Ilian Mihov and Timothy Van Zandt. 2013. “The e¤ect of WTO on the extensive and the intensive margins of trade.” Journal of International Economics 91(2): 204-219. Dutt, Pushan and Devashish Mitra. 2002. “Endogenous trade policy through majority voting: an empirical investigation.” Journal of International Economics 58(1): 107-133. Eaton, Jonathan and Gene M. Grossman. 1986. “Optimal Trade and Industrial Policy under Oligopoly.” Quarterly Journal of Economics 101(2): 383-406. 107 Eaton, Jonathan and Samuel Kortum. 2002. “Technology, geography, and trade.” Econometrica 70(5): 1741-79. Ederington, Josh. 2001. “International coordination of trade and domestic policies.” American Economic Review 91(5): 1580-93. Eicher, Theo S. and Christian Henn. 2011. “In search of WTO trade e¤ects: Preferential trade agreements promote trade strongly, but unevenly.” Journal of International Economics 83(2): 137-53. Eschenbach, Felix, and Bernard Hoekman. 2006. “Services policies in transition economies: on the EU and WTO as commitment mechanisms.” World Trade Review 5(3): 415-43. Estevadeordal, Antoni, Caroline Freund, Emanuel Ornelas. 2008. “Does regionalism a¤ect trade liberal- ization toward nonmembers?” Quarterly Journal of Economics 123(4): 1531-75. Ethier, Wilfred J. 1998. “Regionalism in a multilateral world.” Journal of Political Economy 106(6): 1214- 1245. Feenstra, Robert C. 1989. “Symmetric pass-through of tari¤s and exchange rates under imperfect compe- tition: An empirical test.” Journal of International Economics 27(1-2): 25-45. Feenstra, Robert C. 1995. “Estimating the e¤ects of trade policy.”In Handbook of International Economics, edited by Gene M. Grossman and Kenneth Rogo¤, 3: 1553-1595. Amsterdam: North Holland. Feenstra, Robert C. and Jagdish N. Bhagwati. 1982. “Tari¤ Seeking and the E¢ cient Tari¤.” In: Import Competition and Response, edited by Jagdish N. Bhagwati, 245-262. Chicago, IL: University of Chicago Press for the NBER. Feinberg, Robert M. and Kara M. Reynolds. 2006. “The spread of antidumping regimes and the role of retaliation in …lings.” Southern Economic Journal 72(4): 877-890. Findlay, Ronald and Stanislaw Wellisz. 1982. “Endogenous Tari¤s, the Political Economy of Trade Restric- tions, and Welfare.” In: Import Competition and Response, edited by Jagdish N. Bhagwati, 223-244. Chicago, IL: University of Chicago Press for the NBER. Francois, Joseph and Bernard Hoekman. 2010. “Services trade and policy.” Journal of Economic Literature 48(3): 642-92. Frazer, Garth and Johannes Van Biesebroeck. 2010. “Trade Growth under the African Growth and Opportunity Act.” Review of Economics and Statistics 92(1): 128-144. Freund, Caroline. 2000a. “Multilateralism and the endogenous formation of free trade agreements.”Journal of International Economics 52(2): 359-76. Freund, Caroline. 2000b. “Di¤erent Paths To Free Trade: The Gains From Regionalism.” Quarterly Journal of Economics 115(4): 1317-1341. 108 Freund, Caroline and Emanuel Ornelas. 2010. “Regional trade agreements.” Annual Review of Economics 2(1): 139-166. Fudenberg, Drew, David Levine and Eric Maskin. 1994. “The folk theorem with imperfect public informa- tion.” Econometrica 62(5): 997-1039. Furusawa, Taiji. 1999. “The Negotiation of Sustainable Tari¤s.” Journal of International Economics 48(2): 321-45. Furusawa, Taiji and Hideo Konishi. 2007. “Free trade networks.” Journal of International Economics 72(2): 310-35. Geng, Difei and Kamal Saggi. 2014. “The case for non-discrimination in the international protection of intellectual property.” Mimeogr., Vanderbilt. Goyal, Sanjeev and Sumit Joshi. 2006. “Bilateralism and free trade.” International Economic Review 47(3): 749-78. Grossman, Gene M. 1981. “The theory of domestic content protection and content preference.” Quarterly Journal of Economics 96(4): 583-603. Grossman, Gene M. and Elhanan Helpman. 1994. “Protection for sale.” American Economic Review 84(4): 833-50. Grossman, Gene M. and Elhanan Helpman. 1995a. “Trade wars and trade talks.” Journal of Political Economy 103(4): 675-708. Grossman, Gene M. and Elhanan Helpman. 1995b. “The politics of free-trade agreements.” American Economic Review 85(4): 667-90. Grossman, Gene M., Henrik Horn and Petros C. Mavroidis. 2013. “National treatment.” In Legal and Economic Principles of World Trade Law, edited by Henrik Horn and Petros C. Mavroidis, 205-345. New York: Cambridge University Press for the American Law Institute. Grossman, Gene M. and Edwin Lai. 2004. “International protection of intellectual property.” American Economic Review 94(5): 1635-53. Goldberg, Pinelopi Koujianou and Michael M. Knetter. 1997. “Goods Prices and Exchange Rates: What Have We Learned?” Journal of Economic Literature 35(3): 1243-1272. Gulotty, Robert. 2014 “Governing Trade Beyond Tari¤s: The Politics of Multinational Production and Its Implications for International Cooperation.” Mimeogr., Stanford University. Handley, Kyle. 2014. “Exporting under Trade Policy Uncertainty: Theory and Evidence,” Journal of International Economics 94(1): 50–66. 109 Handley, Kyle and Nuno Limão. forthcoming. “Trade and Investment under Policy Uncertainty: Theory and Firm Evidence,” American Economic Journal: Economic Policy. Hillman, Arye L. 1982. “Declining Industries and Political-Support Protectionist Motives.” American Economic Review 72(5): 1180-87. Hochman, Gal and Ella Segev. 2010. “Managed trade with imperfect information.” International Economic Review 51(1): 187-211. Hoekman, Bernard M. 2008. “The General Agreement on Trade in Services: doomed to fail? Does it matter?” Journal of Industry, Competition and Trade 8(3-4): 295-318. Hoekman, Bernard M. 2014. “The Bali Trade Facilitation Agreement and Rulemaking in the WTO: Milestone, Mistake or Mirage?” EUI Working Paper No. RSCAS 2014/102. à la carte’ or WTO ‘ Hoekman, Bernard M. and Petros C. Mavroidis. 2013. “WTO ‘ menu du jour’? Assessing the case for Plurilateral Agreements.” EUI Working Paper No. RSCAS 2013/58. Hoekman, Bernard M. and Petros C. Mavroidis. 2014. “Embracing Diversity: Plurilateral Agreements and the Trading System.” CEPR Discussion Paper No. 10204. Hoekman, Bernard M. and Caglar Ozden. 2007. “Introduction.” In Trade Preferences and Di¤ erential Treatment of Developing Countries, edited by Bernard M. Hoekman and Caglar Ozden, xi-xlii. Chel- tenham, UK: Edward Elgar. Holmstrom, Bengt. 1977. “On incentives and control in organizations.” PhD dissertation, Stanford Uni- versity. Horn, Henrik. 2006. “National treatment in the GATT.” American Economic Review 96(1): 394-404. Horn, Henrik, Louise Johannesson, and Petros C. Mavroidis. 2011. “The WTO Dispute Settlement System: 1995–2010: Some Descriptive Statistics.” Journal of World Trade 45(6): 1107-1138 Horn, Henrik, Giovanni Maggi, and Robert W. Staiger. 2010. “Trade agreements as endogenously incom- plete contracts.” American Economic Review 100(1): 394-419. Horn, Henrik and Petros C. Mavroidis. 2001. “Economic and Legal Aspects of the Most-Favored Nation Clause.” European Journal of Political Economy 17(2): 233-79. Horn, Henrik and Petros C. Mavroidis, Eds. 2003. The WTO Case Law of 2001. Cambridge, UK: Cambridge University Press. Horn, Henrik and Petros C. Mavroidis. 2011. “WTO Dispute Settlement Database.” The World Bank. Available at http://www.worldbank.org/trade/wtodisputes, last accessed July 25, 2014. 110 Horn, Henrik, Petros C. Mavroidis, and Hakan Nordstrom. 2005. “Is the Use of the WTO Dispute Settlement System Biased?” In The WTO and International Trade Law/Dispute Settlement, edited by Petros C. Mavroidis and Alan O. Sykes. Cheltenham, UK: Edward Elgar. Horn, Henrik, Petros C. Mavroidis, and André Sapir. 2010. “Beyond the WTO? An Anatomy of EU and US Preferential Trade Agreements.” The World Economy 33(11): 1565-1588. Hudec, Robert E. 1990. The GATT Legal System and World Trade Diplomacy, 2nd edition. New Hamp- shire: Butterworth Legal Publishers. Hudec, Robert E. 1993. Enforcing International Trade Law: The Evolution of the Modern GATT Legal System. Salem, NH: Butterworth Legal Publishers. Hufbauer, Gary C. Je¤rey J. Schott , Kimberly Ann Elliott, and Barbara Oegg. 2009. Economic Sanctions Reconsidered, 3rd Edition. Washington, DC: Peterson Institute for International Economics. Irwin, Douglas A. 1995. “The GATT’s Contribution to Economic Recovery in Post-War Europe.” In Europe’s Postwar Recovery, edited by Barry Eichengreen, 127-150. New York: Cambridge University Press. Irwin, Douglas A. 2011. Peddling Protectionsim: Smoot-Hawley and the Great Depression. Princeton and Oxford: Princeton University Press. Irwin, Douglas A. 2012. Trade Policy Disaster: Lessons from the 1930s. Cambridge, MA: The MIT Press. Irwin, Douglas A. 2015. “Tari¤ Incidence: Evidence from U.S. Sugar Duties, 1890-1930.” NBER Working Paper No. 20635. Irwin, Douglas A., Petros C. Mavroidis, and Alan O. Sykes 2008. The Genesis of the GATT. Cambridge, UK: Cambridge University Press. Irwin, Douglas A. and Joseph Weiler. 2008. “Measures A¤ecting the Cross-Border Supply of Gambling and Betting Services (DS 285).” World Trade Review 7(1): 71-113. Itoh, Motoshige and Kazuharu Kiyono. 1987. “Welfare-enhancing export subsidies.” Journal of Political Economy 95(1): 115-137. Jackson, John H. 1969. World Trade and the Law of GATT. New York: Bobbs-Merrill. Jehiel, Philippe and Benny Moldovanu. 2000. “Auctions with downstream interaction among buyers.” Rand Journal of Economics 31(4): 768-92. Jensen, J. Bradford. 2011. Global trade in services: fear, facts and o¤ shoring. Washington D.C.: Petersen Institute for International Economics. Johnson, Harry G. 1953-4. “Optimum Tari¤s and Retaliation.”Review of Economic Studies 21(2): 142–53. 111 Johnson, Robert C. and Guillermo Noguera. 2014. “A Portrait of Trade in Value Added over Four Decades.” Mimeogr., Dartmouth College. Karacaovali, Baybars and Nuno Limão. 2008. “The Clash of Liberalizations: Preferential vs. Multilateral Trade Liberalization in the European Union.” Journal of International Economics 74(2): 299-327. Keck, Alexander and Andreas Lendle. 2014. “New Evidence on Preference Utilization.” Mimeogr., WTO. Kennan, John and Raymond Riezman. 1988. “Do Big Countries Win Tari¤ Wars?”International Economic Review 29(1): 81-85. Kennan, John and Raymond Riezman. 1990. “Optimal tari¤ equilibria with customs unions.” Canadian Journal of Economics 23(1): 70-83. Kim, In Song. 2013. “Political cleavages within industry: …rm level lobbying for trade liberalization.” Mimeogr., MIT. Kowalczyk, Carsten and Raymond Riezman. 2011. “Trade agreements” in Palgrave Handbook of Inter- national Trade, edited by Daniel Bernhofen, Rod Falvey, David Greenaway and Udo Kreickemeier. Macmillan, U.K. Kreinin, Mordechai. 1961. “E¤ect of tari¤ changes on the prices and volume of imports.” American Economic Review 51(3): 310-24. Krishna, Pravin. 1998. “Regionalism and multilateralism: a political economy approach.” Quarterly Jour- nal of Economics 113(1): 227-51. Krishna, Pravin and Devashish Mitra. 2005. “Reciprocated unilateralism in trade policy.” Journal of International Economics 65(2): 461-87. Krugman, Paul R. 1991. “Is bilateralism bad?”In International Trade and Trade Policy, edited by Elhanan Helpman and Assaf Razin, 9-23. Cambridge, MA: The MIT Press. Lapan, Harvey E. 1988. “The optimal tari¤, production lags, and time consistency.” American Economic Review 78(3): 395-401. Lawrence, Robert Z. 2003. Crimes and Punishments? Retaliation under the WTO. Washington, DC: Institute for International Economics. Lawrence, Robert Z. 2006. “Rulemaking Amidst Growing Diversity: A ‘club of clubs’approach to WTO reform and new issue selection.” Journal of International Economic Law 9(4): 823-35. Lee, Gea M. 2007. “Trade agreements with domestic policies as disguised protection.” Journal of Interna- tional Economics 71(1): 241-59. Lee, Gea M. 2014. “Optimal International Agreement and Restriction on Domestic E¢ ciency.” Mimeogr., Singapore Management University. 112 Levy, Philip I. 1997. “A political-economic analysis of free-trade agreements.” American Economic Review 87(4): 506-19. Levy, Philip I. 2006. “Do We Need an Undertaker for the Single Undertaking? Angles of Variable Geom- etry.” In Economic Development and Multilateral Trade Cooperation, edited by Simon Evenett and Bernard Hoekman, 417-437. London: Palgrave Macmillan. Limão, Nuno. 2005. “Trade policy, cross-border externalities and lobbies: do linked agreements enforce more cooperative outcomes?” Journal of International Economics 67(1): 175-99. Limão, Nuno. 2006. “Preferential Trade Agreements as Stumbling Blocks for Multilateral Trade Liberal- ization: Evidence for the U.S.” American Economic Review 96(3): 896-914. Limão, Nuno 2007. “Are Preferential Trade Agreements with Non-Trade Objectives a Stumbling Block for Multilateral Liberalization?” Review of Economic Studies 74(3): 821-55. Limão, Nuno and Giovanni Maggi. forthcoming. “Uncertainty and Trade Agreements.”American Economic Journal: Microeconomics. Limão, Nuno and Kamal Saggi. 2008. “Tari¤ retaliation versus …nancial compensation in the enforcement of international trade agreements.” Journal of International Economics 76(1): 48-60. Limão, Nuno and Kamal Saggi. 2013. “Size inequality, coordination externalities and international trade agreements.” European Economic Review 63(C): 10-27. Limão, Nuno and Patricia Tovar. 2011. “Policy Choice: Theory and Evidence from Commitment via International Trade Agreements.” Journal of International Economics 85(2): 186-205. Liu, Xuepeng and Emanuel Ornelas. 2014. “Free Trade Agreements and the Consolidation of Democracy.” American Economic Journal: Macroeconomics 6(2): 29-70. Ludema, Rodney D. 1991. “International Trade Bargaining and the Most-Favored Nation Clause.” Eco- nomics and Politics 3(1): 1-20. Ludema, Rodney D. and Anna Maria Mayda. 2009. “Do Countries Free Ride on MFN?” Journal of International Economics 77(2): 137-50. Ludema, Rodney D. and Anna Maria Mayda. 2013. “Do Terms-of Trade E¤ects Matter for Trade Agree- ments? Theory and Evidence from WTO Countries.” Quarterly Journal of Economics 128(4): 1837- 93. Magee, Christopher and Stephen Magee. 2008. “The United States is a Small Country in World Trade.” Review of International Economics 16(5): 990-1004. Maggi, Giovanni. 1996. “Strategic trade policies with endogenous mode of competition.” American Eco- nomic Review 86(1): 237-58. 113 Maggi, Giovanni. 1999. “The Role of Multilateral Institutions in International Trade Cooperation.” Amer- ican Economic Review 89(1): 190-214. Maggi, Giovanni. 2014. “International trade agreements.”In Handbook of International Economics, edited by Gita Gopinath, Elhanan Helpman, and Kenneth Rogo¤, Volume 4: 317-90. Amsterdam, NL: Elsevier. Maggi, Giovanni and Andres Rodriguez-Clare. 1998. “The value of trade agreements in the presence of political pressures.” Journal of Political Economy 106(3): 574-601. Maggi, Giovanni and Andres Rodriguez-Clare. 2007. “A political-economy theory of trade agreements.” American Economic Review 97(4): 1374-1406. Maggi, Giovanni and Robert W. Staiger. 2011. “The Role of Dispute Settlement Procedures in International Trade Agreements.” Quarterly Journal of Economics 126(1): 475-515. Maggi, Giovanni and Robert W. Staiger. 2013. “Trade disputes and settlement.” Mimeogr., University of Wisconsin. Maggi, Giovanni and Robert W. Staiger. forthcoming. “Optimal Design of Trade Agreements in the Presence of Renegotiation.” American Economic Journal: Microeconomics. Mai, Joseph and Andrey Stoyanov. forthcoming. “The e¤ect of the Canada-US free trade agreement on Canadian multilateral trade liberalization.” Canadian Journal of Economics. Martin, Alberto and Wouter Vergote. 2008. “On the role of retaliation in trade agreements.” Journal of International Economics 76(1): 61-77. Maskus, Keith E. 2000. Intellectual Property Rights in the Global Economy. Washington DC: Institute for International Economics. Matsuyama, Kiminori. 1990. “Perfect equilibria in a trade liberalization game.” American Economic Review 80(3): 480-92. Mavroidis, Petros C. 2000. “Remedies in the WTO Legal System: Between a Rock and a Hard Place.” European Journal of International Law 11(4): 763-814. Mayer, Wolfgang. 1980. “Theoretical Considerations on Negotiated Tari¤ Adjustments.” Oxford Economic Papers 33(1): 135-153. Mayer, Wolfgang. 1984. “Endogenous Tari¤ Formation.” American Economic Review 74(5): 970-85. McCalman, Phillip. 2002. “Multi-lateral Trade Negotiations and the Most-Favored Nation Clause.”Journal of International Economics 57(1): 151-76. McLaren, John. 1997. “Size, sunk costs, and Judge Bowker’s objection to free trade.” American Economic Review 87(3): 400-20. 114 McLaren, John. 2002. “A theory of insidious regionalism.” Quarterly Journal of Economics 117(2): 571- 608. McMillan, John. 1986. Game Theory in International Economics. New York: Harwood. McMillan, John. 1989. “A Game-Theoretic View of International Trade Negotiations: Implications for De- veloping Countries.”In Developing Countries and the Global Trading System, edited by John Whalley, 1. London, UK: MacMillan. Melitz, Marc and Gianmarco Ottaviano. 2008. “Market size, trade, and productivity,” The Review of Economic Studies 75(1): 295-316. Mitra, Devashish. 1999. “Endogenous lobby formation and endogenous protection: a long-run model of trade policy determination.” American Economic Review 89(5): 1116-34. Mitra, Devashish. 2002. “Endogenous political organization and the value of trade agreements.” Journal of International Economics 57(2): 473-85. Mrázová, Monika, David Vines and Ben Zissimos. 2013. “Is the GATT/WTO’s Article XXIV bad?” Journal of International Economics 89: 216-32. Mrázová, Monika. 2011. “Trade agreements when pro…ts matter.”Mimeogr., London School of Economics. Nicita, Alessandro, Marcelo Olarreaga, and Peri Silva. 2014. “Cooperation in WTO’s Tari¤ Waters.” Mimeogr., University of Geneva. Olarreaga, Marcelo, Isidro Soloaga, and L. Alan Winters. 1999. “What’s Behind Mercosur’s CET?” CEPR Discussion Paper No. 2310. Olson, Mancur. 1965. The Logic of Collective Action: Public Goods and the Theory of Groups. Cambridge, MA: Harvard University Press. Ore…ce, Gianluca and Nadia Rocha. 2014. “Deep integration and production networks: an empirical analysis.” The World Economy 37(1): 106-136. Ornelas, Emanuel. 2005a. “Trade creating free trade areas and the undermining of multilateralism.” European Economic Review 49(7): 1717-35. Ornelas, Emanuel. 2005b. “Rent destruction and the political viability of free trade agreements.” Quarterly Journal of Economics 120(4): 1475-1506. Ornelas, Emanuel. 2005c. “Endogenous free trade agreements and the multilateral trading system.”Journal of International Economics 67(2): 471-97. Ornelas, Emanuel. 2008. “Feasible multilateralism and the e¤ects of regionalism.” Journal of International Economics 74(1): 202-24. 115 Ossa, Ralph. 2011. “A ‘new-trade’ theory of GATT/WTO negotiations.” Journal of Political Economy 119(1): 122-52. Ossa, Ralph. 2012. “Pro…ts in the “new trade” approach to trade negotiations.” American Economic Review: Papers and Proceedings 102(2): 466-69. Ossa, Ralph. 2014. “Trade Wars and Trade Talks with Data.” American Economic Review 104(12): 4104-46. Palmeter, David and Petros C. Mavroidis. 2004. Dispute Settlement in the World Trade Organization: Practice and Procedure. Cambridge, UK: Cambridge University Press. Panagariya, Arvind. 2000. “Preferential Trade Liberalization: The Traditional Theory and New Develop- ments.” Journal of Economic Literature 38(2): 287-331. Park, Jee-Hyeong. 2011. “Enforcing International Trade Agreements with Imperfect Private Monitoring.” Review of Economic Studies 78(3): 1102-34. Petersmann, Ernst-Ulrich. 1997. The GATT/WTO Dispute Settlement System: International Law, Inter- national Organization and Dispute Settlement. London: Kluwer Law International. Pierce, Justin and Peter K. Schott. 2015. “The Surprisingly Swift Decline of U.S. Manufacturing Employ- ment.” Mimeogr., Yale University. Potipiti, Tanapong. 2012. “Import tari¤s and export subsidies in the WTO: a small country approach.” Mimeogr., Chulalongkorn Univ. Prusa, Thomas J. and Susan Skeath. 2002. The economic and strategic motives for antidumping …lings.” Weltwirtschaftliches Archive 138(3): 389-413. Prusa, Thomas J. and Robert Teh. 2010. “Protection reduction and diversion: PTAs and the incidence of antidumping disputes.” NBER Working Paper No. 16276. Richardson, Martin. 1993. “Endogenous protection and trade diversion.” Journal of International Eco- nomics 34: 309-24. Richardson, Martin. 1995. “Tari¤ revenue competition in a free-trade area.” European Economic Review 39(7): 1429-37. Riezman, Raymond. 1985. “Customs unions and the core.” Journal of International Economics 19 (3-4): 335-65. Riezman, Raymond. 1999. “Can bilateral trade agreements help to induce free trade?” Canadian Journal of Economics 32(3): 751-66 s and CUSFTA’ Romalis, John. 2007. “NAFTA’ s impact on international trade.” Review of Economics and Statistics 89(3): 416-43. 116 Rose, Andrew. 2004. “Do we really know that the WTO increases trade?” American Economic Review 94(1): 98-114. Rotemberg Julio J. and Garth Saloner. 1986. “A supergame-theoretic model of price wars during booms.” American Economic Review 76(3): 390-407. Saggi, Kamal. 2004. “Tari¤s and the Most Favored Nation Clause.” Journal of International Economics 63(2): 341-68. Saggi, Kamal. 2006. “Preferential trade agreements and multilateral tari¤ cooperation.” International Economic Review 47: 29-57. Saggi, Kamal, Alan Woodland, and Halis Murat Yildiz. 2013. “On the relationship between preferen- tial and multilateral trade liberalization: the case of customs unions.” American Economic Journal: Microeconomics 5(1): 63-99. Saggi, Kamal and Halis Murat Yildiz. 2010. “Bilateralism, multilateralism, and the quest for global free trade.” Journal of International Economics 81(1): 26-37. Sauré, Philip. 2014. “Domestic policies in self-enforcing trade agreements.” European Economic Review 68: 19-30. Schi¤, Maurice and Won Chang. 2003. “Market presence, contestability, and the terms-of-trade e¤ects of regional integration.” Journal of International Economics 60(1): 161-75. Schwartz, Warren F. and Alan O. Sykes. 1997. “The Economics of the Most Favored Nation Clause.” In Economic Dimensions in International Law: Comparative and Empirical Perspectives, edited by Jagdeep S. Bhandari and Alan O. Sykes, 43-79. Cambridge, UK: Cambridge University Press. Schwartz, Warren F. and Alan O. Sykes. 2002. “The economic structure of renegotiation and dispute resolution in the World Trade Organization.” Journal of Legal Studies 31(1): S179-204. Scotchmer, Suzanne. 2004. “The political economy of intellectual property treaties.” Journal of Law, Economics and Organization 20(2): 415-37. Seidman, Daniel J. 2009. “Preferential trading agreements as strategic positioning.” Journal of Interna- tional Economics 79(1): 143-59. Shirono, Kazuko. 2004. “Are WTO Tari¤ Negotiations Reciprocal? An Analysis of Tari¤ Liberalization.” Mimeogr., Columbia University. Spagnola, Giancarlo. 1999a. “On interdependent supergames: multimarket contact, concavity and collu- sion.” Journal of Economic Theory 89(1): 127-39. Spagnola, Giancarlo. 1999b. “Issue linkage, delegation and international policy coordination.” Mimeogr. Stockholm School of International Economics. 117 Spamman, Holger. 2006. “The myth of ‘re-balancing’ retaliation in WTO dispute settlement practice.” Journal of International Economic Law 9(1): 31-79. Spearot, Alan. 2015. “Unpacking the Long Run E¤ects of Tari¤ Shocks: New Structural Implications from Firm Heterogeneity Models.” Mimeogr., UC Santa Cruz. Spencer, Barbara and James A. Brander. 1983. “International r&d rivalry and industrial strategy.” Review of Economic Studies 50(163): 707-22. Staiger, Robert W. 1995. “International rules and institutions for trade policy.” In Handbook of Inter- national Economics, edited by Gene M. Grossman and Kenneth Rogo¤, 3: 1495-1551. Amsterdam: North Holland. Staiger, Robert W. and Alan O. Sykes. 2011. “International trade, national treatment, and domestic regulation.” Journal of Legal Studies 40(1): 149-203. Staiger, Robert W. and Alan O. Sykes. 2013. “Non-Violations.” Journal of International Economic Law 16(4): 741-775. Staiger, Robert W. and Alan O. Sykes. 2015. “How important can the non-violation clause be for the GATT/WTO?.” Mimeogr., Dartmouth College. Staiger, Robert W. and Guido Tabellini. 1987. “Discretionary trade policy and excessive protection.” American Economic Review 77(5): 823-37. Staiger, Robert W. and Guido Tabellini. 1999. “Do GATT rules help governments make domestic com- mitments?” Economics and Politics 11(2): 109-44. Subramanian, Arvind and Shang-Jin Wei. 2007. “The WTO Promotes Trade, Strongly But Unevenly.” Journal of International Economics 72(1): 151–175. Protectionism’as a safeguard: a positive analysis of the GATT ‘ Sykes, Alan O. 1991. “‘ escape clause’with normative speculations.” University of Chicago Law Review 58(1): 255-305. Sykes, Alan O. 2003. “The Safeguards Mess: A Critique of WTO Jurisprudence.” World Trade Review 2(3): 261-295. Sykes, Alan O. 2005a. “The economics of WTO rules on subsidies and countervailing measures.” In The World Trade Organization: Legal, Economic and Political Analysis, edited by Arthur E. Appleton, Patrick F. J. Macrory, and Michael G. Plummer, 2. Springer-Verlag. Sykes, Alan O. 2005b. “Public versus Private Enforcement of International Economic Law: Standing and Remedy.” Journal of Legal Studies 34(2): 631-666. Syropoulos, Constantinos. 1999. “Customs unions and comparative advantage.” Oxford Economic Papers 51: 239-56. 118 Tabakis, Chrysostomos. 2010. “Customs unions and special protection.” BE Journal of Economic Analysis & Policy 10(1). Tallberg, Jonas and James McCall Smith. 2014. “Dispute Settlement in World Politics: States, Suprana- tional Prosecutors, and Compliance.” European Journal of International Relations 20(1): 118-144. Tang, Man-Keung and Shang-Jin Wei. 2009. “The value of making commitments externally: evidence from WTO accessions.” Journal of International Economics 78(2): 216-29. Tornell, Aaron. 1991. “On the ine¤ectiveness of made-to-measure protectionist programs.”In International Trade and Trade Policy, edited by Elhanan Helpman and Assaf Razin, 66-79. Cambridge: The MIT Press. Tintlenot, Felix. 2014. “Global production with export platforms.” Mimeogr., University of Chicago. Tovar, Patricia. 2012. “Preferential Trade Agreements and Unilateral Liberalization: Evidence from CAFTA.” World Trade Review 11(4): 591-619. Trebilcock, Michael. 2014. “Between theories of trade and development: the future of the world trading system.” Mimeogr. University of Toronto Law. Trebilcock, Michael, Robert Howse and Antonia Eliason. 2013. The Regulation of International Trade, 4th edition. Routledge. Tre‡er, Daniel. 2004. “The long and short of the Canada-U.S. free trade agreement.” American Economic Review 94(4): 870-95. USTR. 2010. “U.S., Brazil Agree on Memorandum of Understanding As Part of Path Forward To- ward Resolution of Cotton Dispute.” Available at http://www.ustr.gov/about-us/press-o¢ ce/press- releases/2010/april/us-brazil-agree-memorandum-understanding-part-path-f, accessed July 25, 2014. USTR. 2011. “USTR Kirk Seeks Enforcement of Labor Laws in Guatemala.”Available at http://www.ustr.gov/ about-us/press-o¢ ce/press-releases/2011/may/ustr-kirk-seeks-enforcement-labor-laws-guatemala, ac- cessed July 25, 2014. USTR. 2012. “U.S. Trade Representative Ron Kirk Comments on Presidential Actions Related to the Generalized System of Preferences.” Available at http://www.ustr.gov/about-us/press-o¢ ce/press- releases/2012/march/us-trade-representative-ron-kirk-comments-presidenti, accessed July 25, 2014. Venables, Anthony J. 1985. “Trade and trade policy with imperfect competition: The case of identical products and free entry.” Journal of International Economics 19(1-2): 1-19. Venables, Anthony J. 1987. “Trade and Trade Policy with Di¤erentiated Products: A Chamberlinian- Ricardian Model.” Economic Journal 97(387): 700-717. Viner, Jacob. 1950. The Customs Union Issue. New York: Carnegie Endowment for International Peace. 119 Vousden, Neil. 1990. The Economics of Trade Protection. Cambridge: Cambridge University Press. Wauters, Jasper M. 2010. “The safeguards agreement - an overview.”In Law and Economics of Contingent Protection in International Trade, edited by Kyle W. Bagwell, George A. Bermann, and Petros C. Mavroidis, 334-366. Cambridge, UK: Cambridge University Press. Winkelmann, Liliana and Rainer Winkelmann. 1998. “Tari¤s, quotas and terms-of-trade: The case of New Zealand.” Journal of International Economics 46(2): 313-332. Winters, L. Alan and Won Chang. 2000. “Regional integration and the prices of imports: an empirical investigation.” Journal of International Economics 51(2): 363-77. WTO. 2002. “Dispute Settlement Body — Special Session — Negotiations on Improvements and Clari- …cations of the Dispute Settlement Understanding — Proposal by Mexico.” November 4, catalogue record TN/DS/W/23. WTO. 2007. World Trade Report 2007: Six Decades of Multilateral Trade Cooperation: What Have We Learnt? Geneva: WTO. WTO. 2011. World Trade Report 2011: The WTO and Preferential Trade Agreements: From Co-Existence to Coherence. Geneva: WTO. WTO. 2012. World Trade Report 2012: Trade and Public Policies: A Closer Look at Non-Tari¤ Measures in the 21st Century. Geneva, WTO. WTO. 2014a. “Regional Trade Agreements.”Available at http://www.wto.org/english/tratop_e/region_e/ region_e.htm, last accessed July 25, 2014. WTO. 2014b. “Current Situation of Schedules of WTO Members.”Available at http://www.wto.org/english/ tratop_e/schedules_e/goods_schedules_table_e.htm, last accessed July 25, 2014. WTO, ITC and UNCTAD. 2013. World Tari¤ Pro…les 2013. Geneva: WTO, ITC and UNCTAD. Yi, Sang-Seung. 1996. “Endogenous formation of customs unions under imperfect competition: open regionalism is good.” Journal of International Economics 41(1-2): 153-77. 120 Table 1: GATT/WTO - 60 Years of Tari¤ Reductions Implementation Round Weighted tari¤ Weights based on period covered reduction MFN imports (year) 1948 Geneva (1947) -26 1939 1950 Annecy (1949) -3 1947 1952 Torquay (1950-51) -4 1949 1956-58 Geneva (1955-56) -3 1954 1962-64 Dillon Round (1961-62) -4 1960 1968-72 Kennedy Round (1964-67) -38 1964 1980-87 Tokyo Round (1973-79) -33 1977 (or 1976) 1995-99 Uruguay Round (1986-94) -38 1988 (or 1989) Source: WTO (2007), Table 5. Notes: MFN tari¤ reduction of industrial countries for industrial products, excluding petroleum. Tari¤ reductions for the …rst …ve rounds refer to the United States only. The calculation of average rates of reductions are weighted by MFN import values. Table 2: Applied tari¤ rates of selected GATT/WTO Members, 1952 and 2005 Economy 1952 2005 Austria 17 4.2 Benelux 9 4.2 Denmark 5 4.2 France 19 4.2 Germany 16 4.2 Italy 24 4.2 Sweden 6 4.2 United Kingdom 17 4.2 European Union (EU-25) - 4.2 Canada 11 3.8 United States 16 3.7 Source: WTO (2007), Appendix Table 7. Notes: computed as simple average of 52 products in 1952 and of all tari¤ lines in 2005. 121 Table 3: Trade Policy under the WTO for Selected Economies, 2012 WTO member economy MFN Bind- Binding Cov- Max- MFN TTB applied ing coverage erage imum applied coverage rate, rate, of ap- applied rate, simple simple plied rate agricul- average average duties ture only > 15 per- cent (1) (2) (3) (4) (5) (6) (7) G20 High-income Australia 2.7 10.0 97.1 0.1 28.0 1.2 1.0 Canada 4.3 6.9 99.7 6.9 551.0 16.2 1.2 European Union 5.5 5.2 100.0 5.1 605.0 13.2 3.0 Japan 4.6 5.2 99.7 3.8 692.0 16.6 <0.1 Saudi Arabia 5.1 11.3 100.0 0.4 427.0 6.2 NA South Korea 13.3 16.6 94.6 10.4 887.0 52.7 0.5 United States 3.4 3.5 100.0 2.7 350.0 4.7 6.8 G20 Emerging Argentina 12.5 31.9 100.0 36.0 35.0 10.5 3.2 Brazil 13.5 31.4 100.0 36.2 55.0 10.1 1.9 China (2011) 9.6 10.0 100.0 14.6 65.0 15.6 1.3 India 13.7 48.6 73.8 19.6 150.0 33.5 6.5 Indonesia 7.0 37.1 96.6 1.6 150.0 7.9 1.7 Mexico 7.8 36.1 100.0 13.8 254.0 21.2 1.1 Russia 10.0 7.8 100.0 11.7 292.0 13.3 NA South Africa 7.6 19.0 96.4 20.6 >1000 8.4 0.6 Turkey 9.6 28.6 50.3 10.8 225.0 41.2 4.9 Developing, other* Bangladesh (2011) 14.4 169.2 15.5 40.1 25.0 17.2 ** Burma 5.6 83.4 17.6 5.0 40.0 8.6 ** DR of the Congo NA 96.2 100.0 NA NA NA ** Egypt 16.8 36.7 99.4 19.2 >1000 66.7 NA Ethiopiay (observer only) 17.3 ** ** 50.8 35.0 22.4 ** Irany (observer only, 2011) 26.6 ** ** 45.7 400.0 30.4 ** Nigeria (2011) 11.7 119.1 19.1 39.0 35.0 15.5 ** Pakistan 13.5 59.9 98.7 36.1 100.0 15.5 0.3 Philippines 6.2 25.7 67.0 3.1 65.0 9.8 0.1 Thailand 9.8 27.8 75.0 22.6 142.0 21.8 0.7 Vietnam 9.5 11.4 100.0 24.7 135.0 16.1 NA Source: tari¤ data taken from WTO, ITC, and UNCTAD (2013) and temporary trade barrier (TTB) data taken from Bown (2014). Notes: parentheses indicate data availability for year other than 2012. *selected other developing countries chosen as those with 2012 populations greater than 50 million. **indicates non-user (or un-reported user) of the policy instrument. NA = not available. G20=Group of 20. y indicates WTO non-member. Columns (1), (2), (5) and (6) are ad valorem rates, and columns (3), (4), and (7) are shares of import products. 122