WPS7502 Policy Research Working Paper 7502 What’s Left for the WTO? Chad P. Bown Development Research Group Trade and International Integration Team December 2015 Policy Research Working Paper 7502 Abstract Suppose that when addressing the question of “what’s left WTO members set in the presence of large amounts of tariff for the WTO?,” tariff negotiators relied not on the agenda binding overhang. These are almost exclusively the domain established in 2001 but instead on the terms-of-trade of developing countries’ own trade policies and they are theory of trade agreements to identify negotiating priori- collectively important; 3.5 billion people currently live in ties. This paper uses the lens of the terms-of-trade theory to countries in which the WTO has had minimal effect for investigate three areas in which it is frequently alleged that one of these three reasons. This paper builds upon recent currently applied tariffs remain “too high”; the implication developments in the empirical literature to present evi- being that the WTO’s job performance to date is incom- dence—some direct, some indirect—that sheds light on plete. This includes applied tariffs for countries that are each area. It then identifies specific needs for additional not members of the WTO, applied MFN tariffs for WTO research to clarify policy implications for the future role of members that are unbound, and applied MFN tariffs for the WTO in the ever-changing international trading system. 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 author 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 What’s left for the WTO?∗ Chad P. Bown The World Bank, CEPR and CESifo JEL: F13 Keywords: WTO, tariff liberalization, MFN, developing countries, tariff bindings, binding over- hang, non-members ∗ Correspondence: Development Research Group (DECTI); The World Bank, 1818 H Street, NW, MSN MC3 303, Washington, DC 20433 USA; Tel: +1.202.473.9588, fax: +1.202.522.1159, email: chad.p.bown@gmail.com, web: https://sites.google.com/site/chadpbown/. Special thanks to Kyle Bagwell and Robert Staiger for useful discussions. Maurizio Zanardi, Anna Maria Mayda, Ben Zissimos, Mostafa Beshkar, Rick Bond, Kamal Saggi, and participants at the CESifo Venice Summer Institute provided insightful comments on an earlier draft. Thanks also to Alessandro Nicita, Marcelo Olarreaga, and Peri Silva for graciously sharing their estimated trade elasticities. Research for this paper has been supported in part by the World Bank’s Multidonor Trust Fund for Trade and Development and through the Strategic Research Partnership on Economic Development. Semira Ahdiyyih provided outstanding research assistance. All remaining errors are my own. 1 Introduction While the WTO may appear ubiquitous, in reality there are substantial segments of the interna- tional trading system that remain seemingly untouched by its reaches. This paper utilizes the lens of the terms-of-trade theory of trade agreements and insights from recent empirical developments to investigate three of these areas in particular. First, at least 36 countries are not WTO members. The people living in these countries do not enjoy the basic rights and obligations of the multilateral system for 100 percent of the products that they might trade. Second, another 25 countries have now been full WTO members for more than 20 years and yet their governments have not taken on even the minimal legal commitment of binding the upper limit of their import tariffs for more than two thirds of manufactured products. Third, even for the WTO members that have legally bound their tariffs, another 45 countries have committed to binding rates that convey limited economic meaning. On average, the binding commitments are more than 15 percentage points above these countries’ applied MFN tariff rates; put differently, these countries could immediately and perma- nently raise their applied MFN tariffs by an average of 400 percent with only minimal notification to other WTO members and with no required compensation. Combined, more than 3.5 billion people live under one of these three sets of conditions in what are predominantly developing countries. One reason why I highlight and choose to investigate these three areas is because of the cross- roads at which the WTO currently finds itself. On one hand, trade policy negotiators seem to be moving on past the WTO. The WTO’s current and ongoing Doha Round of multilateral trade negotiations is seemingly dead; albeit, its failures may be at least partially laid at the feet of those who established the Doha negotiating agenda in 2001. Many have argued the agenda and approach to be fundamentally ill-fitted to deliver any sort of successful outcome along the lines of what the previous institutional and reciprocal negotiating frameworks of the GATT had repeatedly delivered over eight previous Rounds and more than 50 years of negotiations (Bagwell and Staiger, 2014).1 Perhaps more threatening to the stasis that currently plagues the WTO, however, is that many important WTO members are turning their negotiating efforts away from the multilateral sys- tem and toward something else. This includes moving away from the GATT/WTO’s historical, “shallow”-integration approach of negotiating over tariffs and market access in favor of the “deeper” integration and direct negotiation over behind-the-border policy instruments through the “mega- regional” negotiations of the Trans-Pacific Partnership (TPP) and Transatlantic Trade and Invest- ment Partnership (T-TIP), as well as a potential Regional Comprehensive Economic Partnership (RCEP) or Free Trade Area of the Asia-Pacific (FTAAP)(Bagwell, Bown and Staiger, forthcom- ing).2 This movement away from the WTO and toward these mega-regional efforts is being led by both historical champions of the multilateral system, such as the United States, European Union, and Japan, as well as other recent and chief beneficiaries, such as China. 1 Other explorations behind the stalled Doha Round and its ineffectiveness include Martin and Mattoo (2011) and Jones (2010). For a behind-the-scenes perspective of many of the personalities involved, see Blustein (2009). 2 Krishna (2014) also provides a skeptical view of the proliferation of preferential trade agreements and its impli- cations for the multilateral trading system. See also Maggi (2014). 1 On the other hand, despite the trading system’s frustration with the lack of negotiating ad- vancements taking place under the WTO, economics research has made improved strides toward clarifying the core microeconomic and institutional underpinnings behind the GATT/WTO’s rela- tively successful achievement of reaching and sustaining levels of import tariffs that are historically low, even despite massive macroeconomic shocks to the system (Bown, 2011a). In particular, the terms-of-trade literature of trade agreements, most closely associated with the theoretical develop- ments introduced by Bagwell and Staiger (1999, 2002), as well as the inaugural empirical work of ao and Weinstein (2008), has ushered in a number of recent theoretical and empirical Broda, Lim˜ advancements. In Section 2, I survey the key aspects of this literature that have significantly helped clarify our understanding of the “status quo” of trade policy under the multilateral system. Many of these insights rely on understanding the fundamental role played by the WTO as a voluntary trade agreement between countries seeking to address the prisoner’s dilemma outcome of terms-of-trade externalities. One of the primary insights from the theory is that, in order for the GATT/WTO to work at getting significant areas of the global economy to internalize such externalities, it has focused on shallow integration and the reduction of border barriers (tariffs), relied on fundamental principles such as reciprocity, most-favored nation (MFN) treatment and national treatment, and secured market access commitments implied by tariff reductions through a legal system of tariff bindings that is backed up by third party dispute settlement. The research that I review in Section 2 not only sheds light on some of the successes of this approach at getting countries to internalize terms- of-trade externalities, but it has also usefully begun to reveal specific places where the impact of the historical approach has proven incomplete, and potential explanations behind why any failures have arisen. My approach is to extend this analysis of the WTO and with a particular focus on three areas of tariffs that are particularly critical to the interests of developing countries. In Section 3, I begin this paper’s empirical contribution by introducing the applied tariffs for the 36 countries and 500 million people that are not yet a part of the WTO system. I choose this as my launching point not only because this is where the WTO has had the least impact to date, but also because this is one of the least studied areas of international trade policy. As such, much of my effort here is expositional - i.e., a contributing reason as to why so little has been studied for these countries is due to a combination of data limitations (some of which I am able to overcome) but also because these particular countries have many other economic and social problems to address in the global community that may outweigh the importance of international trade agreements. Nevertheless, this section also provides me the opportunity to compare the applied tariffs and political-economic characteristics of the current WTO non-member countries with a group of nearly 30 other countries that recently acceded to the WTO. Furthermore, I am able to utilize both newly available data and newly constructed measures of importer market power - taking advantage of newly available foreign export supply estimates provided by Nicita, Olarreaga and Silva (2013) - to reassess (and ultimately largely confirm) previously provided evidence in the terms-of-trade literature on the tariff-setting behavior for a subsample of these recent accession countries, and the role of market 2 power in affecting the changes to their trade policies upon accession to the WTO. I then turn to a more formal empirical investigation of two areas in which the applied tariffs of current WTO members are sometimes alleged to be too high. Section 4 focuses on the applied tariffs for the products that are currently “unbound” in the WTO system. I examine a set of 25 countries (and more than 700 million people), mostly concentrated in Sub-Saharan Africa, that are longstanding WTO members that have nevertheless not yet taken on the legal commitment to bind the upper limit of their tariffs at any level for more than two thirds of their manufactured import products. Nevertheless, while there may be non-terms-of-trade motivated arguments for the WTO to encourage these countries to bind the tariffs of these currently unbound products, I fail to find evidence that the currently applied tariffs for these unbound products are positively related to the importing country’s ability to exert market power. There is, however, evidence linking import market power influences and applied MFN tariffs for countries that have legally bound their tariffs under the WTO and yet which retain considerable discretion as to the level at which they would be applied due to the existence of “tariff overhang.” In Section 5, I illustrate the 45 countries (and more than 2.4 billion people) where substantial tariff overhang still remains in the WTO system, and I provide some evidence identifying this area as potentially one in which the terms-of-trade theory could motivate the WTO as useful forum to facilitate additional tariff liberalization. Finally, in Section 6, I conclude by integrating this evidence alongside related work that high- lights the difficulties confronting negotiators seeking to utilize the WTO system to facilitate addi- tional tariff liberalization. As such, I also highlight priority areas and some remaining unanswered questions for policy-related research. Before delving into the formal theoretical and empirical analysis, it is worth acknowledging two additional points. First, my focus on tariffs and the terms-of-trade theory is limited by design so as to keep the empirical analysis manageable, but it is admittedly incomplete.3 Second, the role of the WTO in the multilateral trading system goes well beyond it serving as a forum for reciprocal tariff cutting. Put differently, even if the evidence were to indicate that the WTO’s tariff-liberalization function were now somehow complete - which even the evidence that I review and provide below suggests is not yet the case - the WTO institution makes other substantive contributions to the system that are not currently provided by any other entity. These include it providing fora for the peaceful resolution of bilateral trade disputes between countries over its commitments and obligations (Maggi and Staiger 2011, 2015; Bown 2009, Bown and Reynolds 3 In addition to the terms-of-trade theory described in more detail below, there are other prominent theories of trade agreements that I will not integrate into my formal analysis but which also deserve mention. The first alternative approach to trade agreements is the commitment theory (Maggi and Rodriguez-Clare 1998, 2007; Staiger and Tabellini, 1987; Lim˜ ao and Tovar, 2011) in which governments may seek an external agreement to tie their own hands vis-a-vis their private sectors. Other recent alternative theories include consideration of potential other international externalities aside from the terms-of-trade externality, e.g., that may arise through firm delocation (Ossa 2011, 2012). A third theory is motivated by the rise of offshoring (Blanchard 2007, 2010; Antr` as and Staiger 2012a,b). Bagwell, Bown and Staiger (forthcoming) provide a more extensive survey of theoretical and empirical advances in these areas as well as the terms-of-trade literature. Bown and Crowley (forthcoming) survey the empirical landscape of tariffs and other trade policy instruments in historical perspective and in more detail. 3 2015a,b) and for transparency and the dissemination of information - e.g., the Trade Policy Review Mechanism and other reporting requirements - regarding how governments make changes to their trade policies in ways that affect trading partners’ market access (Maggi, 1999). 2 The Terms-of-Trade Approach to Trade Agreements My analysis of “where to look” for evidence that the WTO’s tariff liberalization performance-to- date may be incomplete is guided by the terms-of-trade theory of trade agreements and a number of recent pieces of empirical evidence. This section provides a brief description of the core insights of the terms-of-trade theory of trade agreements and recent empirical research that searches for evidence of this theory inside and outside of the GATT/WTO system. Its main purpose is to survey the state of the art of the existing research literature in this area in order to establish expectations for my formal empirical analysis that follows. I begin with the theory of the terms-of- trade motivations for trade agreements, before I turn to evidence on how this affects trade policy determination for countries outside of the GATT/WTO, for countries that change their tariffs in order to enter the WTO through accession, and for countries that have been more longstanding participants of the GATT/WTO regarding their applied and binding tariffs. 2.1 The terms-of-trade theory of trade agreements Here I review the basics of the terms-of-trade theory of trade agreements introduced in Bagwell and Staiger (1999). In a noncooperative setting characterizing the absence of a trade agreement, two large countries each have a unilateral incentive to impose import tariffs at Nash levels that are too high, relative to the jointly efficient outcome. Each Nash tariff is too high because it shifts some of the cost of the tariff - by reducing the price received by the trading partner’s exporters of the product - onto the trading partner via a terms-of-trade externality. The result of each country setting its tariff at an excessively high level is the classic, terms-of-trade driven prisoner’s dilemma outcome. Bagwell and Staiger then compare this outcome with an outcome whereby they suppose that each government was not motivated by terms-of-trade considerations in its objective function when setting its tariff, but that each government was only (potentially) concerned with the domestic price effects of its tariff choice. In this way, their model allows for the consideration of political-economy influences; e.g., a government may be interested in using its tariff to redistribute income from one group in the domestic economy to another.4 The Bagwell and Staiger (1999) approach generates a number of insights that have subsequently had implications for empirical analysis.5 First, a trade agreement like the GATT/WTO can be used 4 In this way it allows for political-economy influences of many different classes of models, including Grossman and Helpman (1994). 5 Some of the terms-of-trade externality analysis in the context of trade policy was provided by Johnson (1953-54). Bagwell and Staiger (2002) provides a book-length treatment that considers a number of alternative applications of the model to trade agreements under different settings, including consideration of some forms of nonpecuniary exernalities and domestic policy instruments. 4 to coordinate tariff reductions for the governments of two large countries, neither of which would have a unilateral incentive to reduce tariffs because it would suffer terms-of-trade losses. They interpret the GATT principle of reciprocity as providing a framework for the mutual reduction of import tariffs that serves to expand trade volumes from inefficient levels of market access when under Nash tariffs to jointly efficient levels. Reciprocity allows for the mutual reduction in tariffs that serves to neutralize the impact on each country’s terms-of-trade so that neither country experiences a negative price effect of its own tariff liberalization. A second important insight, and one that often goes overlooked, is that the only role for the GATT/WTO in this framework is to reduce tariffs to a level that eliminates the international (terms-of-trade) externality impact of each government’s tariff choice. I.e., in the trade agreement equilibrium, the “politically optimal” trade agreement tariffs that the government imposes may still be positive. In this case, once the terms-of-trade externality has been neutralized, the jointly efficient equilibrium tariffs arising under the trade agreement may still be positive and the GATT/WTO under the terms-of-trade theory will have nothing left “to do” in terms of facilitating additional tariff liberalization. The key implication of the theory is that when empiricists begin to examine the tariff data, the existence of positive tariffs is not, by itself, evidence that job performance of the WTO is incomplete. Under a strict interpretation of the terms-of-trade theory, the WTO only has work to be done if any non-zero tariff is positive because the country is exercising its import market power - i.e., if, for some reason, the country is a member of the agreement but the terms-of-trade component to its tariff has not been fully exorcised. Put differently, if the non-zero tariff is positive for political or redistributive purposes (in light of the government’s preferences), and all of the import market power exertion motives have been extinguished (e.g., either through reciprocal bargaining under GATT rounds or through WTO accession negotiations), then the terms-of-trade motive for the WTO would indicate that its tariff-reducing job is done. From the perspective of this basic theory, I use the next two subsections to review recent devel- opments in the empirical literature on trade agreements. A number of recent contributions provide evidence supporting key elements of this basic theory. However, the evidence is also beginning to shed light on particular areas where, within the international trading system, the GATT/WTO has failed to deliver evidence consistent with the baseline theory, thus identifying potential limits as to what the GATT/WTO and the terms-of-trade approach might be able to achieve. 2.2 The first wave of evidence on applied and bound tariffs for countries outside and inside the WTO In light of the main theoretical predictions of the terms-of-trade theory described above, what is the empirical evidence? As this recent and evolving literature covers a number of different trade policy environments, samples of countries, and historical moments in time, I also use Table 1 to briefly summarize. When contemplating whether the terms-of-trade externality is a serious problem that countries 5 seek to solve by establishing a trade agreement like the GATT/WTO, a first question to consider is what are the determinants of tariffs that countries set when they are not constrained by such agreements? Is there evidence that tariffs are influenced by import market power, or is the variation in import tariffs driven simply by domestic political economy influences? ao and Weinstein (2008) were the first to provide an empirical approach to directly Broda, Lim˜ examine whether the tariffs set by a number of countries outside of the WTO - and thus countries unencumbered by (multilateral) trade agreement constraints - were influenced by market power motives. Their benchmark analysis focused on the applied tariffs set by 15 countries listed in Table 1 during the 1993-2000 period when they were not GATT Contracting Parties or (at the time) members of the WTO. They first construct estimates of foreign export supply elasticities facing those importing countries, and they then provide strong evidence that governments impose higher import tariffs in products where they are found to have market power, as captured by the inverse of the foreign export supply elasticity that their consumers face, just as is predicted by the canonical optimal tariff formula. Their first round of evidence was thus consistent with the potential terms- of-trade motive for the GATT/WTO - i.e., in the absence of such agreements, governments set import tariffs that reflect their market power and a result is that some of the externality costs of those higher tariffs are imposed on trading partners through reductions in those partners’ exporter- received prices. ao and Weinstein (2008) also examine the rela- To further support their analysis, Broda, Lim˜ tionship between these measures of a country’s import market power and a number of different trade policy instruments utilized by the United States. The US is different from the 15 countries in their baseline sample in that it is a country within the GATT/WTO and one that the theory would predict trading partners would have been motivated to seek the terms-of-trade component ao and Weinstein (2008) find no statistical evidence of its tariffs extinguished. Indeed, Broda, Lim˜ of market power affecting US applied MFN import tariffs; this is consistent with an interpreta- tion of decades of GATT/WTO tariff reduction negotiations having eliminated the terms-of-trade cost-shifting component from the applied US tariff. Furthermore, they do find evidence that mar- ket power considerations affect US trade policies in two other places: first, the US application of non-tariff measures - or the policies less constrained by GATT/WTO negotiations and rules;6 and second, the US’s statutory (or “column 2”) tariffs, which are the tariffs that the United States applied to a number of countries that were not members of the WTO and with which the US did not have normal trading relations. Given that countries outside of the GATT/WTO agreement may impose import tariffs in a way that reflects their market power, is there other evidence that such market power is neutralized (or at least reduced) when they eventually join the WTO? Bagwell and Staiger (2011) examine this question by empirically examining the determinants of the tariff cuts made by a group of 16 countries 6 As I describe in more detail below, Bown and Crowley (2013a) provide a separate empirical analysis of a particular class of non-tariff barriers for the United States. That study covered a different time period and assessed the terms-of- trade implications of a slightly different theoretical model (Bagwell and Staiger, 1990), but it also provides evidence consistent with the terms-of-trade theory. 6 that acceded to the WTO between 1995 and 2005, five of which (including China) overlapped with ao and Weinstein sample of non-GATT countries. Unlike countries that had long the Broda, Lim˜ been members of the GATT/WTO but whose tariff levels may have gradually been brought to more globally efficient levels over time, the Bagwell-Staiger framework investigates whether these new members brought their tariffs down from unbound (Nash-like) levels to bound (politically optimal and efficient) levels in one shot upon accession and in accordance with the terms-of-trade theory’s core predictions.7 The Bagwell-Staiger evidence is broadly consistent with the theory; i.e., there is a strong positive relationship between the magnitude of tariff cuts negotiated under the WTO and the pre-negotiation volume of imports. Furthermore, for the five countries with which they ao-Weinstein sample, their evidence also holds when specifically have overlap with the Broda-Lim˜ ao-Weinstein estimated trade controlling for the import market power as measured by Broda-Lim˜ elasticities. While these first two papers present evidence that is consistent with the terms-of-trade theory, the bulk of that evidence admittedly derives from countries either outside of the GATT/WTO ao and Weinstein) or that only recently acceded to the WTO (Bagwell and Staiger). (Broda, Lim˜ What about the trade policymaking behavior of the major economies that are both “inside” the GATT/WTO system and are the ones that have driven the GATT/WTO through sixty years of reciprocal tariff cutting under multilateral negotiating rounds? Furthermore, with the exception of ao-Weinstein evidence for the United States, and the Bagwell-Staiger evidence for the Broda-Lim˜ China, most of the countries in these samples were not major trading economies in the international system. This has the potential to raise concerns about the external validity for the terms-of-trade theory of trade agreements if, for some reason, these countries did not exhibit behavior consistent with that of the major players. Ludema and Mayda (2013) provide one approach to address these concerns by examining the applied MFN tariffs under the WTO at the conclusion of the GATT’s Uruguay Round of negoti- ations for a larger sample of 26 countries, including most of the major economies.8 In particular, they explore whether variation in these countries’ applied MFN tariffs is related to variation in these countries’ import market power and their trading partners’ (exporters’) industrial concentra- tion. They find that the concentration of trading partner exporter interests at the product level, as measured by the Herfindahl-Hirschman Index (HHI), helps explain applied MFN tariff variation - i.e., products with a combined situation of (i) foreign export suppliers that are less concentrated and (ii) an importer with more market power tend to have higher tariffs even after GATT/WTO negotiations. The Ludema-Mayda evidence is that there is variation in the extent to which the terms-of-trade component of a country’s tariff may be negotiated away under the WTO and that can be linked to the free rider problem arising from the GATT/WTO’s MFN rule. First, this empirical result 7 To clarify, Bagwell and Staiger (2011) compare a country’s unbound (applied MFN) tariff rates before the country’s WTO accession with its legally binding tariff commitment post-WTO accession, and not its post-accession applied MFN rate. 8 See also Ludema and Mayda (2009) for an alternative approach focused exclusively on the United States. 7 is intuitive in that it may help to explain the relatively high applied tariffs remaining under the WTO in sectors such as agriculture, textiles, and footwear that continue to persist because the exporting interests behind these products are diffuse. A limitation of the historical framework for conducting negotiations may have arisen because negotiations were voluntary and the tariff liberalization outcome would be extended to all members under the MFN rule of nondiscrimination. However, because the existence of MFN implied that countries could free ride in the negotiations, sometimes a critical mass of exporting interests may not have bothered to show up at the negotiating table in the first place. Second, an important insight arising from this research is the recognition that not all terms-of-trade effects may be fully neutralized even upon a country’s entry into the WTO, a point to which I return below. I.e., Ludema-Mayda’s results identify one potential area in which there may be more tariff-liberalizing work (for terms-of-trade neutralizing reasons) to be “done”; nevertheless in discovering it, they also identify how the historical GATT/WTO approach of relying on voluntary negotiations and MFN may have contributed to the process by which tariff liberalization (for terms-of-trade neutralizing reasons) remains currently incomplete.9 Finally, given the evidence that the terms-of-trade effects matter for trade policy determination, and that the GATT/WTO system may be working to at least partially neutralize such externalities through negotiations, how economically important is the job that the WTO has done for the major economies of the system? One way to address this issue is to ask how large Nash tariffs - i.e., the combination of best-response tariffs that countries would use - would be in a trade war, and what the economic costs of eliminating current levels of trade policy cooperation would be. Using a quantitative approach, Ossa (2014) constructs counterfactual estimates for the size of Nash tariffs in a model featuring seven regions (including the US, EU, Japan, China, India, Brazil, and rest of the world) and finds the median to be 58.1 percent across countries and industries.10 The quantitative model suggests substantial gains from the imposition of the tariffs that are currently in place, relative to the levels of welfare that would arise were countries to resort to imposing their Nash tariffs under a trade war. 2.3 Additional evidence on applied tariffs, bindings, and tariff overhang for countries inside the WTO The next framework that I explore is the recent theoretical and empirical contribution of Nicita, Olarreaga and Silva (2013), which examines the relationship between a WTO member’s applied tariffs and the role of import market power, contingent on whether those tariffs are constrained 9 Bown and Crowley (2013a) provide additional evidence that terms-of-trade motives continue to affect trade policy decisions for WTO members, albeit in a different trade policy setting. They provide evidence consistent with the Bagwell and Staiger (1990) repeated game model of trade agreements by focusing on the US use of antidumping and safeguards over 1997-2006. They find for a country like the US (with applied tariffs virtually at their binding level), the flexibility of antidumping and safeguards can be seen as allowing the government to raise import protection levels in response to trade volume shocks arising for terms-of-trade motivations. 10 This is notably higher than the estimates of the tariffs applied at the height of the Great Depression in the 1930s, after the US imposition of the Smoot-Hawley tariff in 1930 and international retaliatory response. See Bown and Irwin (2015) for a discussion of the range of tariffs more likely to have been in effect just prior to the GATT’s starting point in 1947, which they put at around 22 percent. 8 by WTO tariff binding legal commitments. First, they develop a theoretical model that allows for the political influence of not only import-competing sectors but also exporting sectors. In an environment in which export policies are constrained - as under the WTO, where export subsidies are illegal - they provide a theory that predicts an exporting country government will negotiate larger tariff reductions exactly where that importing country has the most market power. Their model predicts that in the instances in which applied tariffs are at their WTO binding rates, and countries are cooperating under the WTO, there will actually be a negative relationship between the importer’s market power and its negotiated tariff. The intuition is that in these instances, not only does the trade agreement get the country to cooperatively reduce its tariff (so as to neutralize the terms-of-trade externality) but in equilibrium the negotiation “overshoots” and the tariff ends up even lower so as to compensate the politically organized exporters in the trading partner. Furthermore, the theoretical prediction of the positive relationship between applied tariffs and market power also arises in their model, but it only arise for applied tariffs that are well below tariff binding rates - i.e., applied tariffs in the presence of sufficiently large amounts of tariff binding “overhang.” The second major contribution of Nicita, Olarreaga and Silva (2013) is empirical. First, they construct estimates of “foreign” export supply elasticities for 101 WTO member economies at the 6-digit Harmonized System (HS06) level, resulting in a database of hundreds of thousands of importing country-product-specific elasticities. (I will draw heavily on these elasticities in the formal empirical analysis that I introduce below.) Second, they utilize these estimated elasticities to empirically investigate their model’s theoretical predictions for applied tariffs imposed between 2000 and 2009. They find evidence that the inverse foreign export supply elasticity has a negative relationship with applied MFN tariffs when there is zero tariff overhang - i.e., when countries are “cooperating” in that applied rates are set at binding levels - and they find a positive relationship between the importer’s market power and the applied tariff when tariff overhang levels are positive. I further investigate empirically below this second result; i.e., for “tariff overhang” products, are their currently un-checked terms-of-trade externalities that countries are imposing through their applied tariffs that the WTO could potentially be used as a negotiating forum to eliminate? In related work, Behskar, Bond and Rho (2015) provide a terms-of-trade based theory exploring the question of where a country might set its tariff binding in relationship to its applied tariff under a trade agreement. Their theoretical model predicts that governments will seek to retain flexibility and thus bind their tariffs significantly above the applied rates where the importer has little market power. They conduct an empirical examination of product-level tariff data for a sample of 108 WTO member economies over the period 1995-2007; they also partially rely on the inverse foreign export supply elasticities generated by Nicita, Olarreaga, and Silva (2013) as the measure of import market power in their sensitivity analysis. First, they find that newly acceding WTO members bind a larger share of their product lines than the historical GATT members under the WTO. Second, their various measures of import market power are negatively related to the level of the bindings 9 that countries take on, as well as the size of the tariff binding overhang.11 A final stream of recent research that I briefly highlight explores additional economic impli- cations of countries failing to constrain their applied tariffs by leaving sufficient tariff overhang between the applied rates and their tariff bindings.12 Handley and Lim˜ ao (2015) develop a dy- namic, heterogeneous firms model with sunk costs of exporting and show that investment and entry into export markets is reduced when trade policy is uncertain. Furthermore, they show how a credible commitment implied by a trade agreement (e.g., reducing tariff bindings) can increase trade even if applied trade barriers are already low.13 Handley (2014) provides an application of some of the key elements of this theory to the context of WTO tariff bindings and the case of Aus- tralia, finding that growth of exporter-product varieties would have been 7 percent lower between 1993 and 2001 without the binding commitments that Australia took on upon its WTO entry. While the Handley results suggest gains (to the exports) of a trading partner, one would expect that the reciprocal reduction of uncertainty - i.e., two countries jointly eliminating uncertainty by simultaneously binding their applied tariffs at low levels - could lead to analogous joint gains that accrue under the distinct exercise of two countries simultaneously lowering those applied rates under a terms-of-trade neutralizing trade agreement in the first place.14 11 To clarify, Beshkar, Bond and Rho (2015) focus on the determinants of the level tariff bindings (taking applied rates as given) whereas Nicita, Olarreaga and Silva (2013) focus on the determinants of the level of applied tariffs (taking binding rates as given). Nicita, Olarreaga and Silva do not investigate the impact of import market power on either the level of tariff bindings or the amount of overhang between the binding and the applied tariff; an IV for the amount of overhang is interacted with the measure of importer market power. 12 Separately, there is some empirical evidence related to the commitment theory of trade agreements, however, it is much less developed in the literature. Examples include Tang and Wei (2009) which finds some evidence of a positive impact of WTO accession on economic growth. Bown and Crowley (2014) find evidence for some developing countries that WTO entry has committed them to change how they implement increases to their levels of import protection (in response to macroeconomic shocks) by switching to different (and WTO-sanctioned) trade policy instruments, and this is both different from how they operated before the WTO and it is similar to the commitments and trade policy use of higher income WTO members. See also Staiger and Tabellini (1999) for evidence on the role of the GATT in allowing the United States to make trade policy commitments during the Tokyo Round of negotiations. 13 Handley and Lim˜ ao (2015) provide a structural approach to estimating the model and apply it to Portuguese firm-level data. Their policy environment does not entail the binding of tariffs under the WTO, instead they examine the 1986 Portuguese trade agreement accession to the European Economic Community which reduced trade policy uncertainty by locking in zero import tariffs from European trading partners. Francois and Martin (2004) provide an alternative theoretical approach examining the role of tariff bindings in reducing the uncertainty associated with market access. Lim˜ ao and Maggi (2015) provide a more general theory examining when trade agreements can provide gains through the reduction of trade-policy uncertainty. Conditional on the level of income risk aversion in a country, gains from reducing trade policy uncertainty are more likely to arise for economies that are more open and specialized and that have lower export supply elasticities. 14 See also Handley and Lim˜ ao (2014) for a an examination of the resolution of trade policy uncertainty facing Chinese firms resulting from accession to the WTO in 2001 and the reduction of uncertainty surrounding US applied tariffs that had persisted during the 1990s through the annual Senate debate on whether to renew China’s MFN treatment. They find that the effect of the WTO on reducing the threat of a trade war explains 22 percent of China’s export growth to the US, and that the reduction in policy uncertainty lowered U.S. prices and increased consumers’ income by the welfare equivalent of an 8 percentage point tariff decrease. 10 3 WTO Non-Members (and recently acceded members) This section focuses attention on the current WTO non-member countries in the international trading system. One ultimate question of interest - to which I will admittedly only be able to provide very indirect evidence - is whether such countries apply import tariffs that reflect market power motives and whether those would be neutralized should those countries accede to the WTO. First, I introduce the current WTO non-members and their political economic characteristics. Then I examine a comparison group of countries that recently acceded to the WTO. I then investigate empirically the implications of the term-of-trade theory of trade agreements for that second group of countries by applying the Nicita, Olarreaga and Silva (2013) foreign export supply elasticities to the basic estimation approach introduced by Bagwell and Staiger (2011). 3.1 Introduction and political-economic characteristics As Figure 1 illustrates, the current non-members of the WTO are found throughout the world; nevertheless, they are disproportionately concentrated in the Middle East and North Africa, East Africa, and Central Asia. Table 2 provides summary data for key economic characteristics of these countries, as well as comparable data for a separate list of important comparison countries that recently acceded to the WTO - i.e., between 1998 and 2014.15 For ease of exposition, I rank the countries in each group by Gross National Income (GNI) per capita, and I split them roughly into three categories based on GNI per capita - I refer to the three groups as low income, lower-middle income, and middle and higher income.16 For countries that are not yet members of the WTO, I also provide information on whether they have formally been granted “observer” status by the WTO.17 Table 2 reveals a number of stylized facts about the current WTO non-members. First, they are disproportionately poor countries - at least 28 of the 38 countries have GNI per capita in 2013 that was less than the world average of $10,683. Second, there is a wide range in the size of the populations of these countries. Some are tiny (and relatively wealthy) city-states or islands, with less than a million people. Others are poorer and larger countries in Africa - the largest is Ethiopia at 94 million people. Combined, 490 million people live in these WTO non-member countries, or 6.9 percent of the total world population. Most of the WTO non-member countries had imports that were greater than exports in 2013. The exceptions are mostly made up of major energy (oil and/or natural gas) producers/exporters - e.g., Algeria, Azerbaijan, Equatorial Guinea, Iran, Iraq, Kazakhstan, Libya, and Turkmenistan. For 15 I utilize data on accessions starting only in 1998 (instead of, for example, 1996) because some of the initial wave of WTO accession countries in 1996 and 1997 were countries that may have simply waited to begin the domestic legal process to formally ratify WTO membership until after the major WTO members had done so, i.e., recalling the US experience at failing to ratify the ITO in the 1940s, which led to the GATT. 16 Note that these three country groupings do not correspond to the World Bank’s official categories. 17 Governments with WTO observer status are not members but they are granted limited WTO rights, such as access to certain WTO meetings, but they are also expected to uphold other obligations, such as making some (minimal) contributions to the WTO’s operating budget. 11 the rest of the countries with imports substantially larger than exports, this is potentially notable for two reasons. First, the expectation might be that their imports would be limited because their import policies are legally unaffected and undisciplined by the WTO system. Second, many of the non-members are relatively poor and are therefore likely (at least in principle) to be beneficiaries of unilateral preference programs offered by WTO member countries. Ceteris paribus, their firms may face lower-than-MFN tariffs for their sales to those markets which would tend to encourage their exports. Nevertheless, at least at a first glance, the data does not suggest this to be the case. Finally, I mention briefly some other geo-political factors that are likely contributors to the question of why these countries are not (yet) members of the WTO. First, fourteen of these countries can be characterized as states in Fragile and Conflict Affected Situations (FCS) (World Bank, 2014) - these are areas affected by civil war or other forms of violence and strife. Second, while Russia finally acceded to the WTO in 2012 and a handful of former Soviet Republics became members earlier, five of the former Soviet Republics (Azerbaijan, Belarus, Kazakhstan, Turkmenistan and Uzbekistan) have not yet gained entry. Next compare the current WTO non-members with the list on the right-hand-side of Table 2, which includes the countries that acceded to the WTO between 1998 and 2014. The recently acceded countries are also disproportionately poor and include a range of small and large countries by population. The recent accession list also includes countries with geopolitical constraints, such as Russia and other former Republics of the Soviet Union (Armenia, Georgia, Kyrgyz Republic, Moldova, Tajikistan, and Ukraine, as well as Estonia, Latvia and Lithuania that have since also acceded to the European Union), and also FCS countries such as Nepal and Yemen. Overall, I conclude that these sets of WTO non-member and recent WTO accession countries have a number of similarities. 3.2 Establishing a benchmark: The experience of recently acceded WTO mem- bers What might accession to the WTO for current non-member countries mean? To provide context, in this section I benchmark these non-member countries’ current applied tariffs against the tariffs of a set of recently acceded WTO member countries. Table 3 introduces the most recently available information on the applied tariffs for these WTO non-member countries. The table documents the mean of their applied rates, as well as their minimum and maximum rates, and the standard deviation of applied tariffs across import products. The average tariff of these countries ranges from a high of 35.1 percent (Bahamas) to a flat import tariff of 2.5 percent applied to every imported product (Timor-Leste). Some of these countries do have tariffs that peak at rates higher than 100 percent. Table 3 also provides important summary statistics for the tariffs of the recently acceded WTO members, as a point of comparison. For these recently acceded countries, I present four pieces of information: (i) the tariffs they applied five years prior to the their WTO membership, (ii) the share of imported products over which the country agreed to bind its tariffs upon accession to the 12 WTO, (iii) the average tariff binding rate that the country committed not to exceed when joining the WTO, and (iv) the MFN tariff rate that the country applied to all other WTO members in 2013. First, Table 3 indicates that even the poorest recently acceded countries have bound almost 100 percent of their tariffs at some level. As I will observe in Section 4.1, this is very different from many developing countries at similar levels of income per capita that joined the WTO upon its inception in 1995 or which had previously been a Contracting Party to the GATT 1947, and which did not similarly bind all of their products’ tariffs. (I investigate and address this issue for such countries separately below.) Second, for a number of recent WTO accession countries, they were not forced to make substan- tial cuts (on average) to their applied tariffs upon entry into the agreement. Indeed, for more than half of the 27 recently acceded WTO members listed in Table 3, their average binding commitment under the WTO is actually higher than the average tariff the country applied five years prior to WTO entry, meaning that the country could (on average) increase its applied tariffs upon entry into the WTO and still be in compliance with its obligations. Major exceptions include a number of large economies such China, Saudi Arabia, Taiwan (China), and Ukraine. However, a notable characteristic of all of the recently acceded WTO members is the relatively limited amount of aver- age tariff overhang between binding rates and applied MFN tariffs in 2013. With the exception of Vanuatu (30.6 percentage points), no newly acceded member has an average level of tariff overhang exceeding 13.8 percent in 2013 (Nepal) - as Section 5.1 reveals, this is also substantially different from countries that acceded to the WTO upon its entry into force in 1995; i.e., there are 45 WTO members with more than 15 percentage points of average tariff overhang in 2013. Figure 2 illustrates the industry-level variation for these tariff data summarized by Table 3. The three panels represent the average tariffs by sector for three groupings of countries - low- income countries, lower-middle-income countries, and middle- and higher-income countries. For each sector, there are two sets of bars - the first set reflects the average tariffs for the recently acceded WTO members, and the second set reflects the average tariffs for the current WTO non- members. Finally, for WTO members, for each sector there are three pieces of information - the grey bar reflects the average MFN applied rate in 2013, the white bar reflects the tariff binding overhang (or water) above the current applied rate, and the black star reflects the average applied tariff that was in place five years prior to the country’s WTO accession. For the WTO non-member countries, the black bar represents the average tariff in the sector that the countries in that income group applied in 2013. First compare the black stars with the black bars - i.e., compare the average applied tariffs for the recent accession countries five years prior to their WTO membership with the average applied tariffs of the current non-members. Overall, Figure 2 suggests the patterns are quite similar (conditional on income group) across industries; on average at least, the “future” WTO accession countries currently apply import tariffs that are similar to the applied tariff starting point of the recent accession countries before they gained WTO entry. And while there is variation across sectors and 13 income groups, if anything, the evidence would suggest that WTO non-members currently apply rates that are slightly higher than the applied rates of the recently acceded countries five years prior to their joining the WTO. Second, focus attention on the applied tariff changes for the countries that recently acceded to the WTO - i.e., the difference between the star (applied tariff level 5 years prior to WTO accession) and the grey bar (applied tariff level in effect in 2013). The pattern across industries and country groupings is that applied rates do tend to fall on average upon joining the WTO. In levels, the average changes are largest for the lower-middle-income group of countries in the middle panel - this reflects the fact that both more tariff cutting is likely expected of them (relative to low income countries) and they are starting from higher tariff levels (relative to higher income countries). Third, consider the differences in tariff binding overhang that results upon entry into the WTO. On average in 2013, there is more tariff overhang remaining upon WTO accession for low-income countries in comparison to higher income countries. While Table 3 and Figure 2 illustrate a suggestive path forward for current WTO non-member countries - if what is expected of them roughly corresponds to what has been the impact of WTO accession on the tariffs of recently acceded members - I have not yet provided any evidence that this is linked to the terms-of-trade theory of trade agreements. In the next section I consider the potential implications of WTO accession for current non-members through the lens of this theory and drawing from evidence arising from the experience of recent accession countries. 3.3 Empirical evidence from tariff bindings for recent WTO accession countries The first empirical question is whether it is likely that accession to the WTO by these current non-members would neutralize any terms-of-trade externalities that their currently applied tariffs impose on trading partners. Because I do not have the ability to test this counterfactual, instead I examine whether there is evidence from the group of recently acceded WTO member countries to suggest that terms-of-trade externalities of their import tariffs were reduced when they joined the WTO. The alternative - i.e., that there is no relationship between their post-WTO accession tariffs and market power influences - would suggest that these countries joined the WTO with something else in mind, and thus some other approach aside from the terms-of-trade theory would be required to motivate why they find the WTO valuable. In order to specifically investigate this question, I broadly follow the Bagwell and Staiger (2011) estimation approach described earlier. In particular, I examine whether there is a rela- tionship between the binding rate that country c adopts for HS06 product g after WTO accession, W T O−binding (τgc ), and two theoretically-motivated determinants: (i) the pre-accession applied tariff pre−W T O ∗ ). I thus estimate rate (τgc ), and (ii) the inverse of the foreign export supply elasticity (1/ωgc models of the form W T O−binding ∗ pre−W T O ln(1 + τgc ) = αg + αc + β0 ln(1/ωgc ) + β1 ln(1 + τgc )+ gc (1) 14 where αc is importing country fixed effect, αg is the HS06 product fixed effect, and gc is the iid error term. The Bagwell-Staiger theory clearly predicts β1 > 0 and β0 < 0, or that the post-WTO pre−W T O binding rate will be positively related to τgc and negatively related to the measure of the ∗ ). importer’s market power (1/ωgc My estimation exercise serves to complement the original Bagwell-Staiger approach in a number of ways. First, I utilize a slightly different sample of countries (see Table 1 for the list), though notably my additional countries include a number of relatively large (by population) importers - such as Russia, Saudi Arabia, and Ukraine - that acceded to the WTO only after the Bagwell- Staiger sample period. Second, here I rely heavily on the export supply elasticities provided by Nicita, Olarreaga and Silva (2013) that were not available at the time of the original study. Third, I utilize fixed effects to address other potential determinants of tariffs.18 Before turning to the estimates, I also explain here the general approach that I take throughout the paper to address potential data limitations.19 For example, one potential concern is that the elasticities are themselves estimates, and some of the estimated values are extreme.20 First, I winsorize the data set of the elasticities by setting the extreme values to be the values at the 10th and 90th percentiles of the distribution. Second, in the baseline specifications to each of the regressions, I will take the log of the inverse of foreign export supply elasticity, and I will utilize as a robustness check either an indicator for “high elasticity” products (defined as those above the median of the distribution) or the level of inverse of the foreign export supply elasticity. Third, I will also use as my measure of import tariffs ln(1 + τ ), though I frequently report as a robustness check a measure of the tariff that is simply the level of the tariff, τ . Table 4 provides evidence of the expected strong negative relationship between the inverse foreign export supply elasticity and the WTO tariff binding commitment taken upon accession for this sample of 12 countries that recently acceded to the WTO. I.e., ceteris paribus, newly acceding members are requested (through WTO negotiations) to take on lower tariff binding commitments in products for which they have higher market power and thus where their tariffs (if left unchecked) would result in larger terms-of-trade externality losses for trading partners. Note that I also find a strong positive relationship between the pre-WTO applied tariff and the WTO tariff binding commitment, in line with the theoretical prediction. In column (2), I show the robustness of the results by replacing the log of the inverse foreign export supply elasticity with an indicator that takes on the value of one if the elasticity is “high” (above the median value) and zero otherwise, and again the estimated size of the coefficient is negative. Specification (3) substitutes the levels of the tariffs and the elasticities for the log levels that are used in the baseline specification and elsewhere in the table. In column (4), I add importing country fixed effects. Columns (5) and (6) split the sample in two depending on whether the importing country was large (by population) 18 Finally, my estimation exercise here and below relies only on OLS. Unlike the prior literature, I do not imple- ment instrumental variables estimation; thus the estimates reported here should not be interpreted as identifying magnitudes associated with causal effects. 19 The Appendix provides a full description of the data and its sources. 20 For a discussion of a variety of potential approaches to adopt to assess the robustness of results, see Broda, Lim˜ao, and Weinstein (2008) and Nicita, Olarreaga and Silva (2013). 15 - i.e., China, Russia, Saudi Arabia, and Ukraine - or small. While both sets of estimates on the elasticity are negative, as predicted by the theory, the estimate on the elasticity is no longer significant for the small (by population) country subsample. Nevertheless, even this nonresult is somewhat reassuring, given that I would expect the results to be more likely to break down in the small country subsample. Overall this section suggests evidence consistent with the terms-of-trade theory of trade agree- ments and that the pre-existing WTO membership has negotiated tariff binding commitments for newly acceding WTO non-members that serves to reduce the negative (terms-of-trade) externality impact of their tariffs on trading partners. Again, to the extent that there are similarities between the current WTO non-members ’ applied tariffs and the tariff-setting behavior of these recently acceded WTO members before their WTO accession, any future WTO accession by the current non-members could also be expected to have them take on lower tariff binding commitments where they would otherwise have more import market power. 4 WTO members with unbound tariffs This section begins my examination of the tariffs that current WTO members apply, and in particu- lar whether there is scope for the WTO to “provide” a forum for additional terms-of-trade-motivated applied tariff reductions for these countries. Put differently, my approach for the next two sections is to examine different areas in the WTO system where speculation has been that applied tariffs remain “too high,” and I ask whether the level of applied tariffs in each area continues to remain influenced by measures of import market power. Evidence of such a relationship would be con- sistent with identification of additional tariff-reduction work for countries to utilize the WTO to potentially pursue under the terms-of-trade theory of trade agreements. However, an alternative may be that, while applied tariffs in one or more areas may appear “too high” (or otherwise uncon- strained by the WTO); nevertheless, the applied tariffs are not related to product-level measures of the importing country’s market power. If this is the case, there may be little scope to engage the WTO in a terms-of-trade neutralizing attempt to get the country to reduce its tariffs further. This section begins by focusing on the issue area of unbound tariffs. These are the products for which countries have not taken on the legal commitment to set any upper limit for their MFN applied import tariffs. I first introduce where it is that unbound tariffs are most prevalent in the WTO system, and then in Section 4.2 I investigate whether there is evidence linking import market power motives and applied tariff levels in the areas where tariffs are currently unbound. 4.1 The countries and the unbound products - descriptive Table 5 introduces the WTO member countries with the largest share of products for which their applied import tariffs are unbound. Given that a condition of WTO entry for all countries was the expectation that they would agree to bind all tariffs for their agricultural products, I rank the countries in the table by the share of their non-agricultural tariff lines that are bound. The left 16 half of the table lists the 25 WTO member countries (“Group A”) that will serve as the main sample for the regression analysis that I describe in the next section; these are countries that have bound fewer than one third of their non-agricultural import products. Cameroon has committed to a legally binding upper limit on the smallest share of imported products at 1.7 percent, followed by Tanzania and Gambia. An examination of the 25 WTO members with less than 33 percent of bound non-agricultural products suggests a number of common characteristics. First, they are disproportionately poor, as only one (Macao SAR, China) has GNI per capita in 2013 greater than the world average of $10,683. Second, with only a handful of exceptions (Bangladesh, Macao SAR (China), Cuba, Sri Lanka, Suriname), Figure 1 reveals that the vast majority of unbound tariffs are geographically located in Sub-Saharan Africa. Third, while there is also a range of large and small (by population) countries with substantial unbound tariffs, in total the numbers add up: more than 700 million people - or 10% of the world’s population - live in WTO member countries that have bound fewer than one third of their non-agricultural tariffs at any level. Finally, the last column on the left half of Table 5 does suggest relatively little variation in average applied tariffs across these countries - with the exception of Macao SAR, China and Mauritius, the average applied MFN tariff (over all products) for the other 23 WTO member countries ranges between 10 and 20 percent. A major element of this is due to the fact that many of these countries are part of the ECOWAS (Economic Community of West African States), which has been developing a customs union arrangement and thus a common external tariff against non-participants, including the MFN tariff that each would apply against imports arising from all other (non-participant) WTO members. The right side of Table 5 provides similar summary statistics for WTO member countries that have bound between 33 percent and 95 percent of their non-agricultural product tariff lines. These 14 countries (“Group B”) will be used in robustness checks in the formal regression analysis in the next section, but a cursory examination of their economic characteristics suggests that they are much more diverse. At the extremes, some countries on the list are very poor (Central African Republic) and others very rich (Singapore), and with populations that are very small (Brunei and Iceland) or very large (India). The 2013 average applied MFN tariff also ranges substantially from free trade (Hong Kong SAR, China) to 18 percent (Central African Republic). Finally, a country like Turkey in particular is also notable in that - while it may have bound relatively few (only 35 percent) of its non-agricultural products legally at the WTO, it has constrained its applied MFN tariffs through other trade agreement means, i.e., by forming a customs union arrangement with the European Union covering most of its non-agricultural products, with the exception of steel and textiles. Before moving on, the last note that I make about Table 5 concerns those countries that are not found in the table. I.e., the rest of the WTO membership (more than 100 WTO members) that are not listed in the table have bound 95 percent or more of their non-agricultural products. I have already illustrated the tariff data for some of these countries - i.e., the recently acceded WTO members - in Table 3. 17 Finally, consider Figure 3 which illustrates the average MFN applied tariffs by sector for the 25 WTO members with less than 33 percent of their non-agricultural products that are bound. Much of the cross-industry pattern is similar to what is commonly observed in other settings for low– income countries (see again Figure 2, for the comparable tariffs for low-income WTO non-members and recently acceded members) - e.g., relatively higher applied tariffs in sectors such as footwear, textiles, hides and skins, and lower applied tariffs for fuel, chemicals and machinery. 4.2 Empirical evidence for unbound tariffs To my knowledge, there is no theoretical or empirical work exploring the finer question of why a WTO member would choose to bind some products and yet leave other products unbound. Nevertheless, in this section I use the following model to examine empirically the question of whether measures of importer market power are related to applied tariffs for these unbound products W T O−applied ∗ ln(1 + τgc ) = αg + αc + γ0 ln(1/ωgc )+ gc . (2) If importing countries continue to exert market power over their applied MFN import tariffs W T O−applied (τgc ) for these unbound products, the theoretical expectation is that γ0 would be positive. Table 6 presents the results. The general finding is that there is no evidence that market power considerations are driving applied tariff rates for unbound products when the model is estimated on the 25 countries (“Group A”) that have committed to bind their tariffs for less than 33 percent of their non-agricultural products. The first column is the baseline specification which indicates no statistically significant relationship between the log of the inverse of the foreign export supply ∗ ) and the applied MFN tariff rate, given by ln(1 + τ W T O−applied ). In fact, when I elasticity ln(1/ωgc gc introduce importing country fixed effects in column (2), there is actually a negative and statistically significant relationship between the measures of import market power and applied MFN tariffs. While, to my knowledge, no one has previously investigated this particular area of unbound tariffs for WTO member countries, these results have some similarities to the pattern of results found by Beshkar-Bond-Rho (described earlier) that examine binding tariff levels for 108 WTO members. They find tariff binding levels are negatively related to market power, especially in the presence of substantial amounts of tariff overhang (what they refer to as “weak bindings”). Their theoretical model interprets this negative relationship between import market power and tariff binding levels (in the presence of tariff overhang) as allowing countries flexibility to raise their applied rates in response to shocks. While speculative, a similar motivation could also be at work explaining the applied tariffs for products that are unbound in the WTO system. Indeed, the last two columns of Table 6 provide additional evidence of this negative relationship between importer market power and applied MFN tariffs for unbound products by altering the sample of unbound products on which the model is estimated. In column (4), I also include in the sample the unbound products for the 14 WTO member countries (in “Group B”) of Table 5 that had (overall) between 33 percent and 95 percent of their non-agricultural products bound. 18 In column (5) I estimate the model on only the subsample of data from those 14 WTO member countries. In both cases, the estimate of γ0 is negative and statistically significant. To conclude this section, I am unable to find evidence to suggest that the applied MFN tariff levels for unbound products under the WTO are positively associated with importer market power considerations. Under the basic terms-of-trade theory of trade agreements, if countries with un- bound tariffs are not applying them to exert market power and impose externalities on trading partners, this suggests little role for the WTO to facilitate applied tariff reductions in this area. While there may be other theories that would motivate welfare improvements arising from countries voluntarily binding these tariffs through the external commitment of a trade agreement - e.g., the ao (2014, 2015), Handley trade policy and uncertainty literature associated with Handley and Lim˜ ao and Maggi (2015) - in this instance, the motivation may not arise from the basic (2014), or Lim˜ terms-of-trade theory itself. 5 WTO members with bound tariffs but substantial tariff over- hang A second contentious area within the WTO system involves countries that, while having taken on the legal commitments to bind their tariffs at some upper limit, have set the upper limit so high relative to the MFN tariff that the country ultimately applies that the binding level is economically meaningless. The difference between the legally binding commitment and the applied tariff is, again, defined as the amount of tariff overhang. In this section I examine whether applied import tariffs are positively associated with importer market power considerations for products which are characterized by substantial tariff overhang. My approach in this section follows the theoretical insights and empirical framework introduced by Nicita, Olarreaga and Silva (2013) described above. To summarize, they study the applied tariffs for roughly 100 WTO member countries and provide two key empirical results. First, when applied tariffs are constrained by WTO binding commitments - e.g., in the extreme, suppose that the applied rate is equal to the binding commitment, so there is zero tariff overhang - then there is a negative relationship between importer market power and the applied tariff. Second, when applied tariffs are unconstrained by WTO binding commitments - e.g., in the extreme, suppose that there is substantial tariff overhang because tariff bindings have not been negotiated down close to applied levels - then there is a positive relationship between importer market power and the applied tariff. It is this second result in particular that I investigate in more detail. 5.1 The countries and the products with overhang - descriptive First I need to identify the set of WTO member countries with bound tariffs but with significant amounts of tariff overhang remaining between their tariff binding commitments and their applied rates. Table 7 provides the list of WTO member countries that each have at least 15 percentage points of average tariff overhang. First, it is interesting to note that almost all of the countries in 19 Table 7 acceded to the WTO at the time of its inception in 1995. As is apparent from the data in Table 3 for countries that acceded to the WTO sometime later - i.e., in 1998 or after - they were only allowed to enter the WTO with much less tariff overhang in place. Second, it is important to clarify that none of the countries listed in Table 7 overlap with the “Group A” countries (of Table 5) that had bound less than 33 percent of their non-agricultural products - i.e., these two lists are mutually exclusive. However, a handful of countries do appear in both Table 7 and on the “Group B” list of countries in Table 5, i.e., those with less than 95 percent of their non-agricultural products being bound.21 While these countries’ unbound products were included as part of the robustness checks provided in columns (4) and (5) of Table 6, here I only consider the countries’ bound products. Therefore, because the unbound products are dropped from the analysis here, the country-product pairs included in the robustness check regressions of Table 6 and those presented next are mutually exclusive. The countries in Table 7 share some similarities, but also a number of notable differences, with the WTO non-members and recently acceded members (see again Tables 2 and 3) and the list of WTO members with substantial unbound tariffs (see again Table 5) discussed thus far. Like the earlier lists, the countries with substantial tariff overhang are also developing countries - e.g., nearly three quarters of the 45 countries have a 2013 GNI per capita at or below the world average. Nevertheless, these developing countries with substantial tariff overhang on average do have higher GNI per capita than the developing countries that are WTO non-members, WTO members that recently acceded, or WTO members with substantial unbound products. Next, to the extent that the countries with substantial unbound products were geographically concentrated in Sub-Saharan Africa, the countries with substantial tariff overhang tend to be geo- graphically concentrated in Latin America (see again Figure 1). Nevertheless, there are important exceptions, including countries with substantial overhang arising in South and East Asia and North Africa. Furthermore, while relatively large population countries such as Egypt, Philippines, Brazil, Mexico, Indonesia, and India are notably on the list of countries with substantial tariff overhang, this list also contains a number of countries with tiny populations - e.g., eleven of the 45 have less than one million people - including a number of small island economies of the Caribbean. Never- theless, the combined population of these 45 countries is over 2.4 billion people, or more than one third of the global population. Figure 4 illustrates the average MFN applied tariffs and tariff bindings by sector for these 45 WTO members that average more than 15 percentage points of tariff overhang. The average applied tariffs exhibit cross-industry patterns similar to the other settings for developing countries - e.g., relatively higher applied tariffs in sectors such as footwear, textiles, hides and skins, and lower applied tariffs for fuel, chemicals and machinery. There are significant differentials for the binding levels across sectors, however. Tariff binding levels average over 60 percent in animals, vegetables, and foodstuffs, whereas they are closer to 40 percent for all other (non-agricultural) sectors. 21 These countries are Israel, Turkey, Central African Republic, Philippines, Bahrain, India and Tunisia. 20 5.2 Empirical evidence for bound tariffs with substantial overhang In this section I follow a modified version of Nicita, Olarreaga, and Silva (2013) to examine em- pirically the question of whether measures of importer market power are related to applied tariffs for the countries identified in Table 7 as having substantial tariff overhang, or an average of more than 15 percentage points between their tariff bindings and their applied MFN tariffs. In the es- timation, I also condition on the country-product pairs that have 15 percentage points or more of tariff overhang as well.22 The basic model that I estimate is again simply W T O−applied ∗ ln(1 + τgc ) = αg + αc + γ0 ln(1/ωgc )+ gc , (3) where if importing countries continue to exert market power over their applied import tariffs W T O−applied (τgc ) for this subset of bound products over which there is substantial tariff overhang, I expect γ0 to be positive. The main difference from the approach described in the last section is not the model, it is simply the subsample of countries and products (those with bound tariffs and tariff overhang) over which the model is estimated. Table 8 presents the results. The general finding confirms the Nicita, Olarreaga and Silva evidence for this particular subsample of countries that market power considerations are positively related to applied MFN tariff rates in 2013 for these products. The first column of Table 8 is the baseline specification which indicates a positive and statisti- cally significant relationship between the log of the inverse of the foreign export supply elasticity, ∗ ), and the measure of the applied MFN tariff rate, given by ln(1 + τ W T O−applied ). given by ln(1/ωgc gc In column (2) I introduce importing country fixed effects, and in column (3) I utilize the high inverse elasticity indicator variable in lieu of the continuous measure. The results are robust to these different specifications. The next three columns of Table 8 examine subsamples of these data. Column (4) focuses on where tariff overhang is the greatest by changing the threshold from 15 percentage points to 25 percentage points, thereby reducing the sample almost in half.23 The size of the estimated impact of market power is even larger in the subsample of countries and products where tariff overhang is largest. Columns (5) and (6) split the original baseline sample in two depending on whether or not the products fall into agriculture. Interestingly, the potential influence of market power is not found in the agricultural product subsample of the data in column (6), though admittedly this is a much smaller sample of observations. Finally, and as a last “consistency check” with expectations, the very last column of Table 8 presents estimates from the same model on a completely different subsample of data - i.e., the twelve 22 That is, I drop from the sample all products within these 45 countries that have bound tariffs but applied MFN tariffs that are within 15 percentage points (or less) of the binding rate. Because I am therefore conditioning on a sample of countries and products that only have tariff overhang, I do not need to include interaction terms a la Nicita, Olarreaga and Silva (2013) so as to thereby separate out the potential negative relationship between measures of import market power in the absence of such overhang (i.e, when the applied MFN tariff is equal to the binding rate). 23 For the countries involved in this subsample, see again Table 7, and the bottom two thirds of the listed countries, beginning with Peru (26.1 percent). 21 countries that recently acceded to the WTO that were part of the formal econometric analysis of tariff bindings presented in Table 4. Not surprisingly, the relationship in column (7) between the inverse foreign export supply elasticity and applied import tariffs for these twelve countries is not only not positive, but it is negative and statistically significant. Recall from Table 3 that upon entry to the WTO, countries like China, Russia and Ukraine not only took on nearly universal tariff binding coverage, but they bound their tariffs at relatively low levels compared to their applied rates. I.e., average tariff overhang for the countries and products in the column (7) sample is only 3.6 percentage points, and less than 5 percent of observations in that sample have 15 percentage points or more of tariff overhang.24 The applied tariffs for the recently acceded WTO members thus have a very different empirical relationship with measures of import market power than the applied tariffs for the WTO members that have been around since the agreements’ inception and which continue to have large amounts of tariff overhang. The evidence from this section suggests that products for countries that have taken on WTO bindings but for which substantial tariff overhang remains have applied MFN import tariffs that continue to reflect import market power considerations. As such, this may constitute an area where additional WTO-facilitated negotiations for applied MFN tariff reductions would be consistent with the insights of the terms-of-trade theory of trade agreements. 6 Conclusions and Policy Implications This paper uses the lens provided by the terms-of-trade theory of trade agreements, as well as recent empirical and data advances arising in the literature, to assess whether there may be a market power neutralization motive for the WTO to facilitate additional tariff reductions in three distinct areas: (i) applied tariffs for current WTO non-members, (ii) applied tariffs for current members where they are unbound, and (iii) applied tariffs for current members where there is substantial tariff overhang. An open policy question is how could the WTO be redeployed to address these areas where additional terms-of-trade motivated liberalization might take place? While I have provided a mix of direct and indirect evidence for where there remains a positive relationship between import market power and applied import tariffs, nevertheless, I have refrained from assessing why it is that currently “high” applied import tariffs (that reflect terms-of-trade motives) have yet to be extinguished even by WTO negotiations, as well as whether institutional impediments might be overcome that would allow for their negotiated reduction. A first promising line of research involves the Bagwell, Staiger, and Yurukoglu (2015) examina- tion of the historical process of reciprocal trade negotiations that took place product-by-product under the early GATT Rounds. There may be lessons to be learned from the details of such experiences for any additional liberalization remaining to be undertaken today. 24 While not presented in the Table, I can also confirm another relationship identified by Nicita, Olarreaga, and Silva (2013) for this particular sample of countries - that when applied rates are equal to binding rates (so “cooperation” is the strongest), the relationship between market power and the applied MFN tariff is still negative. 22 Nevertheless, one additional possible starting point arises out of the results that I have developed here in Section 5.2. WTO members that currently retain substantial amounts of tariff overhang and have applied MFN tariffs that continue to reflect market power influences could potentially be grouped with one another to identify reciprocal liberalization matches in the spirit suggested by the Bagwell and Staiger (1999) theory. While obviously these regression results are only suggestive of where negotiators could potentially look in greater detail, the countries in this sample include Argentina, Brazil, India, Indonesia, and Mexico - all members of the Group of 20 (G20) and potential future leaders with a vested interest in sustaining the multilateral trading system.25 On the other hand, the last set of results of Section 5.2 presents no evidence that, on average, currently applied MFN tariffs and market power remain positively related for the set of recently acceded WTO members that includes China and Russia. Such evidence would tend to suggest that the recent WTO accession countries may not be great candidates to lead a new set of reciprocal tariff liberalization negotiations. Furthermore, I have already noted one particularly important strand of research in the terms- of-trade literature that identifies variation in the concentration of export interests across countries as presenting an additional bottleneck that may mitigate the effectiveness of the GATT/WTO’s reciprocal, shallow-integration approach to tariff cutting (Ludema and Mayda, 2013). The Ludema- Mayda evidence was based on a 26 country sample that included of a number of high-income countries and it does suggest that not all of the terms-of-trade motives may have (as yet) been exercised for the high-income economy applied MFN tariffs. While this would imply that such countries could also plausibly be part of future reciprocal bargains still to be struck, the current difficulty for the WTO and trade negotiators may rest in how to make those matches and strike those bargains. Put differently, the second insight from the Ludema-Mayda evidence is that the real world of trade negotiations is certainly even more complicated than simply getting two large importing countries together to reciprocally reduce their import tariffs. The potential asymmetry of exporters in a many-country world, or the concentration (or lack thereof) of exporting interests for a particular product, may make implementation of the GATT/WTO’s historical “principal supplier rule” approach to pairing negotiating interests difficult. To what extent might third party intermediaries (such as an institution like the WTO) be needed to organize triangular liberalization efforts, say, if bilateral trade liberalization opportunities between partners are unlikely due to trade imbalances or other asymmetries? More research is certainly required to further investigate all of these questions. An additional and potentially related concern requiring additional exploration is that the im- porting countries that continue to impose positive tariffs reflecting their market power incentives may also not face significantly large “foreign” tariffs on their exported products to generate the trade-off necessary for the neutralization of the terms-of-trade cut under the traditional, reciprocal 25 However, this is complicated by the fact that many of the countries on this list - e.g., Mexico, Colombia, Peru, and Chile - are actively involved in the formation of preferential tariff agreements with major high-income economies. These agreements may serve as an alternative to neutralizing the terms-of-trade motives associated with certain applied bilateral tariffs (vis-a-vis major trading partners at least) if not their applied MFN tariffs. 23 approach. This may be because the importing country receives preferential tariff treatment from trading partners for its exports, either through unilateral preferences such as GSP, or through reciprocal preferential trade agreements. Alternatively, “intermediate” (but not “latecomer” non- member) countries to the system may find that they already receive MFN treatment of very low applied tariffs from the major markets of other WTO members for their exports. While the WTO system has seemingly been able to overcome this hurdle when it comes to neutralizing the terms- of-trade motives behind recently acceded WTO member countries (Bagwell and Staiger 2011, see also the results in Section 3.3), it appears that it may have been much less successful in doing so when it allowed in the initial tranche of acceding members in 1995, when it did not require these countries especially to take on particularly stringent tariff binding commitments (see the results of Section 5 and Nicita, Olarreaga and Silva 2013). There are other complications to the historical GATT/WTO approach to reciprocal liberal- ization that the theoretical literature has begun to identify and explore that may also serve as impediments for future liberalization. These include trade in products where prices are determined as and Staiger 2012a,b) and by bilateral bargaining and not market-clearing conditions (e.g., Antr` environments characterized by cross-border ownership and foreign direct investment (Blanchard 2007, 2010). While these particular impediments may be more suited to the relatively complex trade in parts and tasks that is commonly associated with high-income countries, nevertheless, as Johnson and Noguera (2014) document, the importance of such trade is increasing almost every- where over time. Finally, I conclude by pointing out that even once countries are inside of the WTO and the terms-of-trade incentives may have been extinguished from their applied MFN tariffs, significant institutionally-provided flexibilities exist so that trade policy is not truly and permanently locked in at levels that may turn out to be too low in the face of political-economic shocks. Bown and Crowley (2013a), for example, provide evidence consistent with the Bagwell and Staiger (1990) theoretical, repeated-game framework of trade agreements that interprets some use of antidumping and safeguards as governments managing the terms-of-trade pressure - even once they have bound their applied MFN tariffs at low levels - associated with trade volume shocks.26 Thus while the WTO may still have some work to do, so as to more completely exorcise the terms-of-trade incentives from its members’ applied MFN tariffs, even after potential completion of those efforts, some trade policy flexibility (and influence of terms-of-trade motives affecting the use of such flexibility) may likely remain. 26 For evidence that macroeconomic shocks - real exchange rate shocks, real GDP and unemployment shocks - also trigger new import protection under such temporary trade barrier policies permitted under the WTO, see Bown and Crowley (2013b,2014) for cross-country studies on high-income and emerging economies. respectively, in the spirit of the Bagwell and Staiger (2003) theoretical framework. Vandenbussche and Zanardi (2008) describe motivations for the rise of antidumping laws - the most commonly invoked temporary trade barrier policy - across the WTO membership over time, Bown (2011b) provides a recent empirical accounting of use of the policies across countries over time. 24 References as, Pol and Robert W. Staiger. 2012a. “Trade agreements and the nature of price determination.” Antr` American Economic Review: Papers and Proceedings 102(3): 470-6. as, Pol and Robert W. Staiger. 2012b. “Offshoring and the role of trade agreements.” American Antr` Economic Review 102(7): 3140-3183 e?” Journal of Bagwell, Kyle, Chad P. Bown and Robert W. Staiger. forthcoming. “Is the WTO Pass´ Economic Literature. Bagwell, Kyle and Robert W. Staiger. 1990. “A Theory of Managed Trade.” American Economic Review 80(4): 779-95. Bagwell, Kyle and Robert W. Staiger. 1999. “An Economic Theory of GATT.” American Economic Review 89(1): 215-48. 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. 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. 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, Robert W. Staiger, and Ali Yurukoglu. 2015. “Multilateral Trade Bargaining: A First Look at the GATT Bargaining Records.” NBER Working Paper No. 21488, August. Beshkar, Mostafa, Eric W. Bond, and Youngwoo Rho. 2015. “Tariff binding and overhang: theory and evidence.” Journal of International Economics 97(1): 1-13. Blanchard, Emily J. 2007. “Foreign direct investment, endogenous tariffs, 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. Blustein, Paul. 2009. Misadventures of the Most-Favored Nations: Clashing Egos, Inflated Ambitions, and the Great Shambles of the World Trade System. New York: Public Affairs. 25 Bown, Chad P. 2009. Self-Enforcing Trade: Developing Countries and WTO Dispute Settlement. Wash- ington, DC: Brookings Institution Press. Bown, Chad P., Ed. 2011a. The Great Recession and Import Protection: The Role of Temporary Trade Barriers. London, UK: CEPR and World Bank. Bown, Chad P. 2011b. “Taking Stock of Antidumping, Safeguards and Countervailing Duties, 1990-2009.” The World Economy 34(12): 1955-1998. 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 Meredith A. Crowley. forthcoming. “The Empirical Landscape of Trade Policy.” In The Handbook of Commercial Policy, edited by Kyle Bagwell and Robert W. Staiger. Amsterdam, NL: Elsevier Bown, Chad P. and Douglas A. Irwin. 2015. “The GATT’s Starting Point: Tariff Levels circa 1947.” World Bank Policy Research Working Paper. Bown, Chad P. and Kara M. Reynolds. 2015a. “Trade Flows and Trade Disputes.” Review of International Organizations 10(2): 145-177. Bown, Chad P. and Kara M. Reynolds. 2015b. “Trade Agreements and Enforcement: Evidence from WTO Dispute Settlement.” World Bank Policy Research Working Paper No. 7242, April. ao, and David E. Weinstein. 2008. “Optimal tariffs and market power: the Broda, Christian, Nuno Lim˜ evidence.” American Economic Review 98(5): 2032-65. Francois, Joseph F. and William J. Martin. 2004. “Commercial policy variability, bindings, and market access.” European Economic Review 48(3): 665-679. Grossman, Gene M. and Elhanan Helpman. 1994. “Protection for sale.” American Economic Review 84(4): 833-50. Handley, Kyle. 2014. “Exporting under trade policy uncertainty: Theory and evidence.” Journal of International Economics 94(1): 50-66. ao. 2014. “Policy Uncertainty, Trade and Welfare: Theory and Evidence for Handley, Kyle and Nuno Lim˜ China and the U.S.” Mimeogr., University of Maryland. 26 ao. 2015. “Trade and Investment under Policy Uncertainty: Theory and Firm Handley, Kyle and Nuno Lim˜ Evidence.” American Economic Journal: Economic Policy 7(4): 189-222. Johnson, Harry G. 1953-4. “Optimum Tariffs and Retaliation.” Review of Economic Studies 21(2): 142–53. Johnson, Robert C. and Guillermo Noguera. 2014. “A Portrait of Trade in Value Added over Four Decades.” Mimeogr., Dartmouth College. Jones, Kent. 2010. The Doha Blues: Institutional Crisis and Reform in the WTO. Oxford, UK: Oxford University Press. Krishna, Pravin. 2014. “Preferential Trade Agreements and the World Trade System: A Multilateralist View.” In Globalization in an Age of Crisis: Multilateral Economic Cooperation in the Twenty-First Century, edited by Robert C. Feenstra and Alan M. Taylor, 131-160. Chicago, IL: University of Chicago Press for the NBER. ao, Nuno and Giovanni Maggi. 2015. “Uncertainty and Trade Agreements.” American Economic Lim˜ Journal: Microeconomics 7(4): 1-42. ao, Nuno and Patricia Tovar. 2011. “Policy Choice: Theory and Evidence from Commitment via Lim˜ International Trade Agreements.” Journal of International Economics 85(2): 186-205. 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 Effects Matter for Trade Agree- ments? Theory and Evidence from WTO Countries.” Quarterly Journal of Economics 128(4): 1837- 93. 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 Rogoff, 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. 2015. “Optimal Design of Trade Agreements in the Presence of Renegotiation.” American Economic Journal: Microeconomics 7(1): 109-43. 27 Martin, William J. and Aaditya Mattoo, Eds. 2011. Unfinished Business? The WTO’s Doha Agenda. London, UK: CEPR and the World Bank. Nicita, Alessandro, Marcelo Olarreaga, and Peri Silva. 2013. “Cooperation in WTO’s Tariff Waters.” CEPR Discussion Paper No. 9529. Ossa, Ralph. 2011. “A ‘new-trade’ theory of GATT/WTO negotiations.” Journal of Political Economy 119(1): 122-52. Ossa, Ralph. 2012. “Profits 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. 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 Commitments?” Economics and Politics 11(2): 109-144. Tang, Man-Keung and Shang-Jin Wei. 2009. “The value of making commitments externally: Evidence from WTO accessions.” Journal of International Economics 78(2): 216-229. Vandenbussche, Hylke and Maurizio Zanardi. 2008. “What explains the proliferation of antidumping laws?” Economic Policy 23(1): 93–138. World Bank. 2014. “Harmonized List of Fragile Situations FY14.” Washington, DC: The World Bank. 28 Data Appendix The sources of the applied MFN tariff data for WTO members, the tariff binding data for WTO members, and the applied tariff data for WTO non-members are a combination of WTO IDB, CTS and UNCTAD TRAINS. Some of the tariff data is more disaggregated than the HS06 level, in which case I first construct means at the HS06 level before further employing it. The data on the inverse export supply elasticities at the HS06 level for 108 WTO member countries is from Nicita, Olarreaga and Silva (2013). The sources of the data on the economic characteristics of countries is primarily the World Bank’s World Development Indicators - for some countries with missing data, estimates were utilized from CIA’s World Factbook. 29 Figure 1: WTO Non-Members, Members with Substantial Unbound Products, and Members with Substantial Tariff Overhang in 2013 Source: Constructed by the author. For the list of WTO non-members, see Table 2. WTO members with substantial unbound products defined as countries with fewer than one third of non-agricultural products with tariff bindings; for list, see Table 5. WTO members with substantial tariff overhang defined as countries with more than one third of non-agricultural products with tariff bindings but with average tariff overhang of 15 percentage points or more; for list, see Table 7. 30 Figure 2: Average Tariffs for WTO Non-Members versus Recently Acceded WTO Members, by Industry and Country Group Source: Constructed by the author from tariff data at the HS-06 level from the WTO IDB, CTS and UNCTAD TRAINS. Constructed from available data and country groupings provided in Table 3. 31 Figure 3: Average Applied MFN Tariffs for WTO Members with Substantial Unbound Tariffs in 2013, by Industry Source: Constructed by the author from tariff data at the HS-06 level from the WTO IDB, CTS and UNCTAD TRAINS. Constructed from the data for the 25 WTO member countries in Table 5 (“Group A”) with less than 33 percent of non-agricultural tariffs that are bound. Figure 4: Average Applied MFN Tariffs and Tariff Bindings for WTO Members with Substantial Tariff Overhang in 2013, by Industry Source: Constructed by the author from tariff data at the HS-06 level from the WTO IDB, CTS and UNCTAD TRAINS. Constructed from the data for the 45 WTO member countries in Table 7 with 15 percentage points or more of average tariff overhang. 32 Table 1: Selected Empirical Studies of Trade Agreements, Import Tariffs and Market Power Paper Trade policy environment Countries Broda, Limo and Weinstein (2008) Applied tariffs set by 15 non-GATT/WTO countries, Algeria, Belarus, Bolivia, China, Czech Republic, as a cross section (at some point over 1993-2000) Ecuador, Latvia, Lebanon, Lithuania, Oman, Paraguay, Russia, Saudi Arabia, Taiwan (China), Ukraine Applied tariffs, statutory tariffs, and non-tariff United States measures set by one major GATT/WTO member Bagwell and Staiger (2011) WTO tariff binding levels upon accession for Albania, Armenia, Cambodia, China, Ecuador, 16 new members that joined over 1995-2005 Estonia, Georgia, Jordan, Kyrgyz Republic, Latvia, Lithuania, Macedonia, Moldova, Nepal, Oman, Panama Ludema and Mayda (2013) Applied MFN tariffs for 26 WTO members at the Argentina, Australia, Bolivia, Brazil, Canada, Chile, conclusion of the Uruguay Round Colombia, Ecuador, European Union, Hungary, Iceland, India, Indonesia, Japan, South Korea, Madagascar, Malaysia, Mauritius, Mexico, Morocco, New Zealand, Norway, Peru, Romania, Thailand, United States Ossa (2014) Quantification of Nash, unilaterally optimal, and Brazil, China, European Union, India, Japan, cooperative tariffs for 7 countries and rest of the world United States and rest of the world Nicita, Olarreaga and Silva (2013) Applied MFN tariffs for 101 WTO members with 101 countries and without binding overhang, 2000-2009 Beshkar, Bond and Rho (2015) Binding levels and tariff overhang for 108 WTO 108 countries members, 1995-2007 Bown and Crowley (2013) Antidumping and safeguard tariffs for a WTO member United States with applied tariffs at the binding level, 1997-2006 The current paper applies the Nicita, Olarreaga, and Silva (2013) export supply elasticities to... Section 3 WTO tariff binding levels for 12 countries upon Albania, Armenia, Cabo Verde, China, Georgia, WTO accession (countries acceded 1998-2012) Jordan, Kyrgyz Republic, Moldova, Nepal, Oman, Russia, Saudi Arabia, Ukraine Section 4 Applied tariffs for unbound products of 25 WTO members 25 countries listed in Table 5 that have bound fewer than one third of nonagricultural products, 2013 Section 5 Applied tariffs for bound products of 45 WTO members 45 countries listed in Table 7 with an average of 15 percentage points or more of tariff overhang, 2013 33 Table 2: Economic Characteristics of WTO Non-Members and Recently Acceded WTO Members, 2013 WTO WTO GNI Imports Exports Recent WTO WTO GNI Imports Exports Non-member Observer per capita Population (billions, (billions, Accession Accession per capita Population (billions, (billions, Country Status (2013 US$) (millions) 2013 US$) 2013 US$) Country Year (2013 US$) (millions) 2013 US$) 2013 US$) Low-income countries Low-income countries Liberia Observer 410 4.3 1.6 0.6 Nepal 2004 730 27.8 7.2 2.1 Ethiopia Observer 470 94.1 13.8 5.9 Cambodia 2004 950 15.1 11.2 10.0 Eritrea No 490 6.3 0.6 0.4 Tajikistan 2013 990 8.2 5.8 1.6 Afghanistan Observer 690 30.6 10.0 1.3 Kyrgyz Republic 1998 1,210 5.7 6.9 3.4 Comoros Observer 840 0.7 0.4 0.1 Yemen* 2014 1,330 24.4 11.0 8.1 South Sudan No 950 11.3 5.1 2.1 Laos 2013 1,450 6.8 5.2 4.2 Sao Tom´ e and Principe Observer 1,470 0.2 0.1 <0.1 Vietnam 2007 1,740 89.7 136.8 143.8 Sudan Observer 1,550 38.0 10.7 6.4 Uzbekistan Observer 1,880 30.2 18.0 15.7 Syria* Observer NA 22.8 7.6 1.9 North Korea* No NA 24.9 4.8 4.0 Somalia* No NA 10.5 0.8 0.6 Lower-middle-income countries Lower-middle-income countries Bhutan Observer 2,330 0.8 1.1 0.7 Moldova 2001 2,470 3.6 6.5 3.5 Kiribati No 2,620 0.1 0.2 <0.1 Vanuatu 2012 3,130 0.3 0.4 0.4 Micronesia* No 3,280 0.1 0.3 0.1 Georgia 2000 3,560 4.5 9.3 7.2 Timor-Leste* No 3,940 1.2 1.6 0.2 Cabo Verde 2008 3,620 0.5 0.9 0.6 Marshall Islands* No 4,310 0.1 0.1 <0.1 Armenia 2003 3,800 3.0 5.0 2.8 34 Bosnia and Herzegovina Observer 4,780 3.8 9.5 5.7 Ukraine 2008 3,960 45.5 98.3 83.2 Algeria Observer 5,330 39.2 63.6 69.7 Samoa 2012 3,970 0.2 0.4 0.2 Iran* Observer 5,780 77.4 60.1 93.0 Tonga 2007 4,490 0.1 0.3 0.1 Tuvalu* No 5,840 0.0 0.2 <0.1 Albania 2000 4,510 2.9 6.8 4.5 Serbia Observer 6,050 7.2 23.6 18.6 Macedonia 2003 4,870 2.1 7.4 5.5 Nauru No NA <0.1 0.1 0.1 Jordan 2000 4,950 6.5 24.0 14.3 Middle- and higher-income countries Middle- and higher-income countries Iraq Observer 6,720 33.4 75.0 77.9 China 2001 6,560 1,357.4 2,203.6 2,440.5 Belarus Observer 6,730 9.5 45.9 43.9 Montenegro 2012 7,250 0.6 2.7 1.8 Turkmenistan No 6,880 5.2 15.6 25.8 Seychelles 2015 13,210 0.1 1.3 1.1 Azerbaijan Observer 7,350 9.4 19.8 35.8 Croatia** 2000 13,420 4.3 24.6 24.9 Lebanon Observer 9,870 4.5 33.8 27.7 Russia 2012 13,850 143.5 471.6 594.8 Palau No 10,970 <0.1 0.2 0.2 Lithuania** 2001 14,900 3.0 33.9 33.2 Kazakhstan Observer 11,550 17.0 61.9 88.7 Latvia** 1999 15,290 2.0 17.8 16.8 Seychelles No 13,210 0.1 1.3 1.1 Estonia** 1999 17,780 1.3 21.2 21.4 Equatorial Guinea Observer 14,320 0.8 10.7 13.8 Oman 2000 25,150 3.6 27.8 48.5 Bahamas Observer 21,570 0.4 4.7 3.5 Saudi Arabia 2005 26,260 28.8 229.3 387.6 Andorra* Observer NA 0.1 1.5 <0.1 Taiwan, China* 2002 NA 23.4 267.4 304.6 Holy See (Vatican City) Observer NA <0.1 NA NA Libya* Observer NA 6.2 26.8 34.9 Monaco* No NA <0.1 1.2 1.1 San Marino* No NA <0.1 2.1 2.6 Subtotal (Non-members) 490.4 534.3 583.9 Subtotal (Recent accession) 1,814.7 3,644.8 4,170.8 World 10,683 7,125.1 22,719.6 23,442.6 World 10,683 7,125.1 22,719.6 23,442.6 Share of world 6.9% 2.4% 2.5% Share of world 25.5% 16.0% 17.8% Share of world (not including China) 6.4% 6.3% 7.4% Sources: World Bank’s World Development Indicators, *data unavailable so supplemented with estimates from the CIA’s The World Fact Book. GNI=Gross national income, NA=not available. Income classifications not based on official World Bank categories. ** indicates country also acceded to the European Union during this period and adopted the EU’s common external tariff. Table 3: Tariff Characteristics of WTO Non-Members and Recently Acceded WTO Members, 2013 MFN applied tariff MFN MFN WTO rate, 2013 applied applied binding WTO WTO WTO Recent WTO WTO tariff rate tariff rate tariff rate binding Non-member Observer Simple St. Accession Accession (simple avg.), (simple avg.), (simple coverage Country Status average Min. Max. Dev. Country Year pre-WTO‡ 2013 avg.) (%) Low-income countries Low-income countries Liberia Observer 10.0 0.0 50.0 6.9 Nepal* 2004 12.3 12.2 26.0 99.4 Ethiopia Observer 17.3 0.0 35.0 11.8 Cambodia 2004 16.4 10.9 19.1 100.0 Eritrea No 7.9 0.0 25.0 8.5 Tajikistan 2013 7.6 7.6 7.9 100.0 Afghanistan Observer 5.9 0.0 40.0 3.9 Kyrgyz Republic* 1998 0.0 4.5 7.4 99.9 Comoros Observer 15.3 0.0 20.0 7.8 Laos 2013 9.7 9.7 18.8 100.0 Sao Tom´ e and Principe Observer 10.2 0.0 20.0 4.1 Vietnam 2007 16.4 9.4 11.4 100.0 Sudan Observer 21.2 0.0 40.0 15.8 Uzbekistan Observer 15.1 0.0 30.0 10.9 Syria Observer 16.5 0.0 80.0 23.2 Lower-middle-income countries Lower-middle-income countries Bhutan Observer 21.9 0.0 100.0 13.7 Moldova* 2001 6.0 8.8 6.7 100.0 35 Timor-Leste No 2.5 2.5 2.5 0.0 Vanuatu 2012 14.0 9.1 39.7 100.0 Bosnia and Herzegovina Observer 6.5 0.0 824.4 13.4 Georgia* 2000 10.6 1.4 7.2 100.0 Algeria Observer 18.6 0.0 30.0 10.3 Cabo Verde* 2008 10.4 10.3 15.8 100.0 Iran Observer 26.6 3.0 400.0 28.7 Armenia* 2003 3.0 3.6 8.5 100.0 Tuvalu No 7.7 0.0 35.0 9.8 Ukraine* 2008 7.0 4.5 5.8 100.0 Serbia Observer 7.4 0.0 30.0 7.3 Samoa 2012 11.0 11.3 21.1 100.0 Tonga 2007 11.7 11.7 17.6 100.0 Albania* 2000 15.9 3.8 7.0 100.0 Macedonia 2003 14.4 6.5 6.9 100.0 Jordan 2000 22.1 9.5 16.2 100.0 Middle- and higher-income countries Middle- and higher-income countries Belarus Observer 8.8 0.0 100.0 6.6 China* 2001 23.7 9.6 10.0 100.0 Turkmenistan No 5.1 0.0 150.0 15.4 Montenegro 2012 4.6 4.2 5.1 100.0 Azerbaijan Observer 9.7 0.0 1478.8 26.0 Croatia** 2000 10.6 4.6 4.1 100.0 Lebanon Observer 6.3 0.0 334.0 13.7 Russia* 2012 9.0 8.8 7.3 100.0 Palau No 4.2 0.0 1370.1 29.1 Lithuania** 2001 3.6 4.6 4.1 100.0 Kazakhstan Observer 8.7 0.0 100.0 6.6 Latvia** 1999 4.3 4.6 4.1 100.0 Equatorial Guinea Observer 17.9 0.0 30.0 9.5 Estonia** 1999 0.1 4.6 4.1 100.0 Bahamas Observer 35.1 0.0 75.0 16.2 Oman* 2000 4.7 4.5 13.6 100.0 Libya Observer 21.3 0.0 3000.0 113.8 Saudi Arabia* 2005 11.9 4.7 10.7 100.0 Taiwan, China 2002 7.8 5.6 5.7 100.0 Sources: Compiled by the author from WTO IDB and CTS and UNCTAD TRAINS made available via WITS. ‡ pre-accession data taken from 5 years prior to WTO accession. **Acceded to the European Union during this period and thus adopted the EU’s common external tariff. Yemen and Seychelles not included because they acceded in 2014 and 2015, respectively. *Countries utilized in the econometric exercise of Table 4. Table 4: Market Power and Post-WTO Accession Import Tariff Bindings for Recently Acceded Countries W T O−binding ∗ ) + β ln(1 + τ pre−W T O ) + Regression equation: ln(1 + τgc ) = αg + αc + β0 ln(1/ωgc 1 gc gc High inv. Level Add Large Small elasticity inverse importer countries countries Baseline indicator elasticity FE only only (1) (2) (3) (4) (5) (6) ∗ ) Log inverse elasticity: ln(1/ωgc -2.39*** -0.66** -1.40*** -0.62 (0.30) (0.29) (0.41) (0.65) Indicator for high inverse elasticity -0.06*** (0.01) ∗ ) Inverse elasticity: (1/ωgc -1.49*** (0.19) pre−W T O Log pre-accession tariff: ln(1+τgc ) 0.26*** 0.26*** 0.31*** 0.35*** 0.27*** (0.01) (0.01) (0.01) (0.01) (0.01) pre−W T O Pre-accession tariff: τgc 0.24*** (0.03) Product level (HS06) fixed effects Y Y Y Y Y Y Importing country fixed effects N N N Y Y Y Observations 26,417 26,417 26,417 26,417 13,659 12,758 R2 0.48 0.48 0.43 0.62 0.68 0.66 Notes: Robust standard errors in parentheses, ***, **, * indicates statistically significant at the 1, 5, or 10 percent levels, respectively. Estimates for the constant term suppressed. Pre-WTO accession tariffs for HS-06 digit product g taken five years prior to accession date for 12 countries (c): Albania, Armenia, Cabo Verde, China, Georgia, Jordan, Kyrgyz Republic, Moldova, Nepal, Oman, Russia, Saudi Arabia, and Ukraine. Large countries in column (5) defined as China, Russia, Saudi Arabia and Ukraine. 36 Table 5: Economic and Tariff Characteristics of WTO Members with Substantial Unbound Tariffs, 2013 Countries with WTO binding coverage that is Countries with WTO binding coverage that is less than 33 percent of all non-agricultural products between 33 and 95 percent of all non-agricultural products (Group A) (Group B) GNI Binding MFN GNI Binding MFN WTO WTO per coverage, applied WTO WTO per coverage, applied Member Accession capita Population non-ag (simple Member Accession capita Population non-ag (simple Country Year (2013 US$) (millions) (%) avg.), 2013 Country Year 2013 US$) (millions) (%) avg.), 2013 Cameroon 1995 1,290 22.3 1.7 18.0 Turkey 1995 10,980 74.9 35.0 10.8 Tanzania 1995 840 49.3 1.8 12.8 Hong Kong SAR, China 1995 38,520 7.2 35.2 0.0 Gambia 1996 500 1.8 2.2 14.1 Tunisia 1995 4,210 10.9 52.7 15.5 Kenya 1995 1,160 44.4 2.3 12.8 Central African Rep. 1995 320 4.6 58.9 18.0 Togo 1995 530 6.8 2.4 11.9 Singapore 1995 54,580 5.4 63.9 0.1 Ghana 1995 1,770 25.9 2.8 12.9 Philippines 1995 3,270 98.4 63.9 3.7 Uganda 1995 600 37.6 4.3 12.7 Thailand 1995 5,360 67.0 68.4 10.4 Bangladesh 1995 1,010 156.6 4.4 14.0 Bahrain 1995 21,330 1.3 71.1 5.4 Congo 1997 2,590 4.4 5.0 18.0 India 1995 1,560 1,252.0 71.2 13.3 Zambia 1995 1,780 14.5 5.5 13.2 Israel 1995 33,930 8.1 72.4 3.4 37 Zimbabwe 1995 860 14.1 5.7 13.2 Malaysia 1995 10,420 29.7 75.4 5.0 Mauritius 1995 9,570 1.3 6.0 1.5 Korea 1995 25,870 50.2 93.5 12.2 Nigeria 1995 2,690 173.6 8.4 11.7 Brunei 1995 NA 0.4 94.1 1.3 Burundi 1995 260 10.2 12.1 12.8 Iceland 1995 46,650 0.3 94.3 5.9 Macao SAR, China 1995 71,270 0.6 12.7 0.0 Suriname 1995 9,370 0.5 13.3 10.4 Malawi 1995 270 16.4 20.8 12.7 Madagascar 1995 440 22.9 21.2 11.7 Cuba 1995 NA 11.3 21.2 10.3 Cote d’Ivoire 1995 1,450 20.3 23.7 11.9 Sri Lanka 1995 3,180 20.5 28.2 10.5 Guinea 1995 460 11.7 30.9 11.9 Burkina Faso 1995 660 16.9 31.3 11.9 Benin 1996 790 10.3 31.4 11.9 Mali 1995 690 15.3 32.0 11.9 Subtotal (Group A) 709.5 Subtotal (Group B) 1,610.5 World 10,683 7,125.1 World 10,683 7,125.1 Share of world 10.0% Share of world 22.6% Sources: World Bank’s World Development Indicators, tariffs constructed by the author with data from WTO CTS, IDB and UNCTAD TRAINS. Ranked by binding coverage of non-agricultural products. GNI=Gross national income, NA=not available. Table 6: Market Power and WTO Members’ Applied Tariffs for Unbound Products, 2013 W T O−applied ∗ )+ Regression equation: ln(1 + τgc ) = αg + αc + γ0 ln(1/ωgc gc Add High inv. Add 33% Alternative importer elasticity to 95% bound unbound Baseline FE indicator to sample sample (1) (2) (3) (4) (5) ∗ ) Log inverse elasticity: ln(1/ωgc 0.44 -1.28*** -0.96*** -2.43*** (0.63) (0.45) (0.34) (0.70) Indicator for high inverse elasticity -0.02 (0.02) Product level (HS06) fixed effects Y Y Y Y Y Importing country fixed effects N Y Y Y Y Observations 25,326 25,326 25,326 36,525 11,199 R2 0.44 0.71 0.71 0.69 0.70 Notes: Robust standard errors in parentheses, ***, **, * indicates statistically significant at the 1, 5, or 10 percent levels, respectively. Estimates for the constant term suppressed. Columns (1), (2) and (3) include only the 25 WTO member countries (Group A) with less than 33 percent of non-agricultural products bound, as listed in Table 5. Column (4) adds 14 countries (Group B of Table 5) that have bound between 33 and 95 percent of non-agricultural products. Column (5) estimates the model on only the 14 Group B countries that have bound between 33 and 95 percent of non-agricultural products. 38 Table 7: Economic and Tariff Characteristics of WTO Members with Substantial Tariff Overhang, 2013 GNI Tariff Tariff MFN WTO WTO per Overhang Binding applied rate Binding Member Accession capita Population (simple avg.), rate (simple avg.), Coverage Country Year (2013 US) (millions) 2013 (simple avg.) 2013 ( %) Panama 1997 10,700 3.9 16.2 23.0 6.8 99.9 Maldives 1995 6,850 0.3 16.2 36.7 20.5 99.1 Israel 1995 33,930 8.1 17.2 20.7 3.4 72.4 Turkey 1995 10,980 74.9 17.5 28.3 10.8 35.0 Brazil 1995 12,550 200.4 17.9 31.4 13.5 100.0 Senegal 1995 1,050 14.1 18.1 30.0 11.9 100.0 Central African Republic 1995 320 4.6 18.1 36.1 18.0 58.9 Argentina 1995 14,590 41.4 18.5 31.9 13.4 100.0 Chile 1995 15,230 17.6 19.1 25.1 6.0 100.0 Egypt 1995 3,140 82.1 20.6 36.1 15.5 99.1 Uruguay 1995 15,640 3.4 21.1 31.6 10.5 100.0 Philippines 1995 3,270 98.4 21.9 25.6 3.7 63.9 Brunei 1995 NA 0.4 22.9 24.2 1.3 94.1 Venezuela 1995 11,730 30.4 23.2 36.5 13.3 100.0 Paraguay 1995 3,980 6.8 23.5 33.5 10.0 100.0 Peru 1995 6,270 30.4 26.1 29.4 3.4 100.0 Honduras 1995 2,120 8.1 26.4 32.1 5.7 100.0 Morocco 1995 3,030 33.0 27.0 41.2 14.3 100.0 Dominican Republic 1995 5,770 10.4 27.0 34.3 7.3 100.0 Papua New Guinea 1996 2,020 7.3 27.1 31.5 4.4 100.0 Mexico 1995 9,880 122.3 27.5 35.2 7.7 100.0 Bolivia 1995 2,550 10.7 28.3 40.0 11.6 100.0 Bahrain 1995 21,330 1.3 28.8 34.2 5.4 71.1 Indonesia 1995 3,760 249.9 29.9 37.1 7.2 96.0 El Salvador 1995 3,720 6.3 30.6 36.6 6.0 100.0 Niger 1996 410 17.8 33.0 44.9 11.9 96.1 Nicaragua 1995 1,750 6.1 35.3 41.1 5.7 100.0 Colombia 1995 7,610 48.3 35.6 42.3 6.8 100.0 Guatemala 1995 3,340 15.5 35.9 41.6 5.7 100.0 India 1995 1,560 1,252.0 36.2 49.6 13.3 71.2 Costa Rica 1995 9,450 4.9 37.2 42.7 5.6 100.0 Jamaica 1995 5,220 2.7 39.4 49.8 10.4 100.0 Tunisia 1995 4,210 10.9 43.0 58.5 15.5 52.7 Trinidad and Tobago 1995 15,640 1.3 45.3 55.8 10.5 100.0 Guyana 1995 3,750 0.8 45.4 56.6 11.2 100.0 Grenada 1996 7,490 0.1 46.3 56.7 10.4 100.0 Belize 1995 4,510 0.3 47.4 58.1 10.7 100.0 Antigua and Barbuda 1995 13,050 0.1 48.1 58.6 10.5 100.0 Dominica 1995 6,860 0.1 48.4 58.7 10.3 96.4 Saint Lucia 1995 7,060 0.2 51.8 62.1 10.3 100.0 Saint Vincent and the Grenadines 1995 6,540 0.1 52.5 62.7 10.2 100.0 Barbados 1995 NA 0.3 65.6 78.2 12.6 100.0 St. Kitts and Nevis 1996 13,760 0.1 65.7 76.0 10.3 100.0 Lesotho 1995 1,590 2.1 70.8 78.4 7.6 100.0 Rwanda 1996 630 11.8 76.6 89.4 12.8 100.0 Subtotal 2,442.0 World 10,683 7,125.1 Share of world 34.3% Sources: World Bank’s World Development Indicators, tariffs constructed by the author with data from WTO CTS, IDB and UNCTAD TRAINS. Members with average tariff overhang greater than 15 percentage points, ranked from lowest to highest. GNI=Gross national income, NA=not available. 39 Table 8: Market Power and WTO Members’ Applied Tariffs for Bound Products with Substantial Tariff Overhang, 2013 W T O−applied ∗ )+ Regression equation: ln(1 + τgc ) = αg + αc + γ0 ln(1/ωgc gc Add High inv. Change to Recent importer elasticity 25 p.p. Non-Agr. Agr. accessions Baseline FE indicator subsample only only only (1) (2) (3) (4) (5) (6) (7) ∗ ) Log inverse elasticity: ln(1/ωgc 1.74*** 0.49** 1.37*** 0.55** 0.31 -1.25*** (0.28) (0.24) (0.32) (0.25) (0.66) (0.38) Indicator for high inverse elasticity 0.03*** (0.01) Product level (HS06) fixed effects Y Y Y Y Y Y Y Importing country fixed effects N Y Y Y Y Y Y Observations 68,355 68,355 68,355 38,710 60,532 7,823 30,096 R2 0.33 0.58 0.58 0.65 0.59 0.57 0.48 Notes: Robust standard errors in parentheses, ***, **, * indicates statistically significant at the 1, 5, or 10 percent levels, respectively. Estimates for the constant term suppressed. With the exception of columns (4) and (7), model estimated on bound products for 45 countries (listed in Table 7) each with tariff overhang greater than 15 percentage points. Column (4) model estimated on bound products for 30 countries (listed in Table 7) with tariff overhang greater than 25 percentage points. Column (7) estimated on bound products for 12 recently acceded WTO countries listed in Table 3. With the exception of Nepal (13.8) and Oman (9.1), the other ten countries have average tariff overhang of 6 percentage points or less. 40