Department of Foreign Affairs and Trade

• Ministers
• International Issues
• Trade Policy
• Country information
• About Australia
• Embassies
• Home

• Travel advice
• Passports
• Visas

• Today is Lebanon Independence Day

Australia-China FTA Conference in Shenzhen

28-29 June 2006

Day 2: Experience and Lessons from other FTAs

Recent Empirical Studies of Preferential Trade Agreements

Dr Russell Hillberry, University of Melbourne

Abstract:  The growing complexity of international goods trade has elevated a variety of legal issues - including intellectual property rights, dispute settlement and investment measures - to the forefront of trade policy negotiations.  It can be difficult to reach agreement on these issues in multilateral fora, so countries anxious to negotiate bilateral commitments on these issues have increasingly turned to preferential agreements.  The economic impact of commitments in these areas is often difficult to quantify.  Overlapping PTA negotiations also complicate efforts to evaluate the agreements.  Proliferating PTAs give economists a moving target, and they increase the risk of systemic problems in the world's trading system.  Standard welfare analyses are also unable to quantify the strategic and non-economic benefits that negotiators often claim.  This survey reviews the recent ex post empirical literature on preferential trade agreements in light of these issues.

Prepared for the Australia-China FTA Conference in Shenzhen, China; 28-29 June 2006.  I would like to thank Witada Anukoonwattaka for valuable research assistance.

The last decade has seen remarkable growth in the number of preferential trade agreements (PTAs).  Figure 1 shows the cumulative number of such agreements notified to the WTO over time.  107 goods agreements and 34 preferential services agreements have been notified since the implementation of the Uruguay Round in 1995.

There are a number of explanations for the rapid growth in the number of PTAs.  First, growth in the number of WTO members, and in the scope of their economic interests, has made it more difficult to identify mutually agreeable multilateral commitments.  As a result, multilateral negotiations have slowed down, and trade ministries have shifted their attention toward PTAs.  Second, international goods trade is becoming more complex, requiring policymakers to address a more comprehensive set of issues than they once did.  Third, PTAs are increasingly used by governments to pursue non-economic objectives.   The United States, for example, appears to have embraced PTAs as foreign policy tool.  Finally, the shift by large countries toward PTAs has had a destabilizing effect on the multilateral consensus.  The subsequent proliferation of PTAs follows, in part, from the absence of credible large country support for the multilateral process.

Both the growth in the number of PTAs and the expansion of their policy scope have complicated economists' efforts to inform trade policy.  When policymakers negotiated occasional, multilateral agreements that focussed primarily on tariff reductions, economists could fully endorse the agreements, and support them with plausible quantitative estimates of their likely impact.  Recent PTAs are much harder to evaluate, for two reasons.  First, the newer agreements are broad in policy scope - including commitments on intellectual property, investment, dispute settlement, and more.  The economic effect of these commitments can be difficult to quantify.  Second, PTAs are no longer occasional one-off agreements that can be treated analytically as policy ends; they are part of a process in which each agreement affects, and is affected by, all other PTAs.  Evaluating the effects of any given negotiation on subsequent negotiations is, so far, beyond the scope of applied theory.

A side benefit of the flurry of recent PTAs is that economists with a variety of experiments with which to evaluate the economic effects of PTAs.  Large, comprehensive agreements like the North American Free Trade Agreement (NAFTA) have now been in place for a sufficiently long time to allow rigorous ex post analysis.  Quantitative tools have been developed for evaluating non-traditional policy commitments in areas such as intellectual property rights.   Hypotheses about the implications of non-economic policy objectives have been developed, and their consequences for trade negotiations evaluated.

The purpose of this paper is to review the recent empirical literature on PTAs.   The emphasis is on ex post analysis of PTAs and involves (primarily) econometric work.  There is a large theoretical literature on PTAs that will only get brief attention.  The vast ex ante literature using applied general equilibrium models as a primary tool for assessing PTAs is referenced only in passing.

Figure 1.  Notifications of PTAs to the WTO

Source:  World Trade Organization

One explanation for the increased scope of modern trade agreements is the growing complexity of international goods trade.  The first section of the paper attempts to illuminate this trend.  Section two reviews the recent empirical literature documenting the effects of PTA policy measures on trade flows.  The third section investigates empirical studies of broader PTA policy commitments, including dispute settlement mechanisms and intellectual property rights.  Section four reviews the literature on how PTAs have affected variables other than trade flows, such as welfare and productivity.  Section five reviews the literature treating PTAs as tools in a broader policy context.

 

I. The increasing complexity of traded goods

There are a number of explanations for the proliferation of PTAs in recent years.  One of these explanations is that the composition of goods has shifted in a way that demands more of national legal systems than in earlier years.  Put simply, traded goods are becoming more complex.  Measuring complexity is not an easy thing to do, but the recent international trade literature provides some insight about how best to illustrate this trend.  This section of the paper applies the Rauch (1999) goods classification system to historical trade flows, in order to show that national export bundles are increasingly composed of differentiated commodities.  Differentiated commodities are frequently sold through negotiated contracts.  As such, trade in these commodities is plausibly more demanding of national legal systems than trade in more homogeneous goods.  Thus, growing trade in differentiated product can help explain the increasing demand for comprehensive legal reforms in trade agreements.

Rauch (1999) categorizes goods according to the manner in which they are sold.  "Homogeneous goods," like gold, are defined as those that are sold on spot markets.  These goods can be defined so precisely that the goods producers' identity (and typically, their national origin) need not be known by the eventual end user.  "Reference priced goods" are defined as goods with prices published in public fora like trade magazines; highly specialized chemicals are an example.

Rauch's third category of goods, "differentiated commodities," are those goods that are not sold in spot markets or by posted prices.  In many cases, these sales are accomplished with negotiated contracts.  One might expect that these sales might require more of national legal systems in terms of enforcement.  Negotiated contracts may include performance provisions such as service after the sale.  Contracts may also stipulate restrictions on the purchaser's use of the product - intellectual property rights laws are an example.  These complexities likely mean that the "differentiated commodities" category of goods is more demanding of national legal systems than Rauch's other two categories.

Figure 2 shows the share of exports that are in Rauch's "differentiated commodities" grouping, over time, for selected countries, and for the world as a whole.  China has seen the fastest growth in the differentiated commodities' share of exports in the group.   While it still has a relatively small share of differentiated commodities in its exports, Australia has also experienced substantial growth in this measure.  Figure 3 shows that the shift toward differentiated commodities has not been as significant in Australia-China trade.

Figure 2. Differentiated commodities in exports (1962-2000)

Source:  Feenstra et al (2004), Rauch (1999), author's calculations

Figure 3.  Differentiated commodities in Australia-China Trade (1962-2004)a

Source:  Feenstra et al (2004), OECD trade statistics, Rauch (1999) and author's calculations

a Data for 2001-2004 from OECD.

The increased share of differentiated products in world trade coincides with the expansion of policy scope of most trade agreements.  It is likely that this relationship is, in part, causal.  Increased demand by exporters for government commitments in the areas of investment, services, dispute settlement procedures and intellectual property rights (IPRs) are all plausibly related to the growing share of differentiated commodities in world trade.  Since these more complicated provisions are difficult to negotiate in a large multilateral forum, trade ministries may well have turned to PTAs to push forward a trade agenda that can accommodate these issues.

This broader agenda is, unfortunately, much more difficult to quantify in terms of its economic impact.  One might plausibly estimate tariff equivalent effects for some policy changes, but it is far from clear that such estimates are valid treatments of the economic effects of more comprehensive reforms.  The effects of many of the broader provisions are quite plausibly beyond the scope of economists' ability to quantify them.  There remains considerable scope for political judgement in evaluating both the efficacy and the value of broader policy commitments.

II.  Empirical literature on preferential trade policies and trade

The most straightforward econometric exercise in evaluating a PTA is estimating its effect on trade.  Data on the dependent variable are readily available, the policy instruments aimed at increasing trade (i.e. tariff cuts) are also easily quantified, and theoretic predictions about the relationships between policy changes and the dependent variable are clear.   PTAs have been especially interesting to empirical economists because a) PTAs have ambiguous effects on economic welfare, and b) the Vinerian framework links observable changes in the trade pattern to the unobservable of interest, economic welfare.

Econometric evaluation of PTAs dates from Aitken (1973), who estimated the effect of the European Economic Community and European Free Trade Area agreements on European trade.  Aiken estimated an empirical gravity model, and used dummy variables to denote two countries' joint membership in a PTA.  Aitken found that, in the early years of the two agreements, both agreements were trade creating, but that trade diversion occurred in later years.

Econometric specifications that identify the effects of trade policy off of dummy variables, like Aitken's, are useful in that they give the analyst a framework for quantifying the total effect of the agreement, including difficult-to-quantify provisions like commitments on investment or IPRs. In a properly specified model, a dummy variable denoting membership in the agreement would capture all of the increased trade generated by the agreement, whether it be due to tariff cuts or to other policy commitments.

Unfortunately, it is impossible to know the proper econometric specification.  Dummy variable approaches have a key drawback; they risk conflating the effects of the agreement with other impacts that affect trade.  For example, the North American Free Trade Agreement (NAFTA) took effect in the midst of the Mexican peso crisis.  Econometric specifications that treat the agreement as a simple dummy variable will likely conflate the effects of the peso crisis with NAFTA.

A related problem with studies that employ dummy variables to identify PTA effects is that such studies frequently lack a theoretical grounding for their econometric model.  Given the possibility of numerous plausible econometric specifications, it remains quite possible that analysts will report the results of specifications that confirm their prior beliefs.  In an important paper, Ghosh and Yamarik (2004) investigate the effect of PTAs on trade flows, using extreme bounds analysis to test for specification uncertainty.  Their estimates lead them to conclude that "the pervasive trade creation effect found in the literature reflects not the information content of the data but rather the unacknowledged beliefs of the researchers."

Few areas of study in economics have data sources as rich in variation as international trade data.  Studies that employ dummy variables to study aggregate trade flows discard much of this information.  Econometric evaluations of PTAs have only recently begun to exploit this variation.  Product-level detail is informative in that a) theoretic predictions linking tariff cuts to trade growth are most precise at the product-level, and b) additional degrees of freedom allow econometric estimates to be identified with greater precision.

Clausing (2001) exploits HS-10 digit U.S. import data to evaluate the Canada-U.S. Free Trade Agreement (CUSFTA).  She estimates that U.S. imports from Canada were 26 percent higher in 1994 than they would have been without the agreement.  These estimates attribute approximately half of the growth in U.S. imports from Canada over the period 1989-1994 to tariff changes in the agreement.  Clausing finds no evidence of trade diversion.

Romalis (2005) conducts a similar exercise, evaluating the effects of preferential CUSFTA and NAFTA tariffs on North American trade flows.  Romalis uses EU imports to benchmark changes in trade flows, requiring him to aggregate the data somewhat, relative to Clausing.  Romalis' data are at the HS 6-digit commodity detail.  On the demand side, Romalis estimates a common elasticity of substitution between import sources across commodities.    In U.S. imports, estimates of the elasticity of substitution range between 6 and 10; for Canadian imports 5-8; and for Mexican imports 0.6 to 2.2.   One explanation for the lower estimated Mexican substitution elasticity is that rules of origin requirements for final goods shipped to the U.S. constrain Mexico's sourcing of intermediate goods.

Romalis also estimates inverse supply elasticities, using price and quantity detail from the 10-digit U.S. import data.   The estimates suggest that export supply curves into the U.S. are price elastic.  These estimates, together with the demand elasticities, are used to inform Romalis' counterfactual analysis of trade and welfare.  Romalis calculates substantial trade diversion from both CUSFTA and NAFTA.   His welfare calculations suggest no welfare gain from the NAFTA tariff cuts, and welfare losses for Mexico.

Another study that investigates NAFTA with product level detail is USITC (2004), chapter 6.  This study also exploits time variation in the data.  As in earlier studies, this study finds that preferential tariffs granted in NAFTA had a substantial effect on U.S.-Mexico trade.  This study considers allows the impact of NAFTA preferences to differ from preferences in place before the agreement.   The study finds that Mexican shares of U.S. imports were more responsive to NAFTA tariff preferences than to earlier preference arrangements.  This may indicate that other features of NAFTA - including the permanence of the preferences, a dispute settlement procedure, and non-tariff provisions on related issues like investment - made U.S. preferences in NAFTA more valuable to Mexican exporters than earlier preferences.

Trade growth via the extensive margin

The above studies focus primarily on the effects of PTAs on the intensive margin of trade.  In other words, these studies evaluate how much preferential tariff cuts increase trade, in products that were already traded with the partner country before the agreement.  A recent literature has emphasized that the extensive margin of trade is also important for welfare.  Broda and Weinstein (2004), for example, estimate that growth in the number of imported varieties since 1972 has raised real U.S. income by 3 percent.  These estimates are 6 times larger, for example, than USITC (2003) estimates of the U.S. gains from all the trade agreements negotiated over a similar period.  If Broda and Weinstein are correct, and the gains from variety growth are that large, then the effects of trade policy on the extensive margin should be a question of keen interest.

Hummels and Klenow (2005) provide a decomposition framework that is useful for quantifying trade growth via the extensive margin.  Hillberry and McDaniel (2002) find that most of the U.S. trade growth with NAFTA partners occurred among already-traded goods, but that the value of U.S. imports from Mexico included some growth along the extensive margin.  Kehoe and Ruhl (2003) show that the extensive margin was important in the case of Canada-Mexico trade.  More broadly, Kehoe and Ruhl point out that, across several trade liberalization episodes, there has been more trade growth in the goods that were least traded before the policy change.

None of the above papers link growth along the extensive margin directly to trade policy changes.  One paper which does so is Debaere and Mostashari (2006), who estimate the contribution of tariff cuts to growth along the extensive margin of trade.  Using 1989-2001 U.S. import data, they find evidence that trade grows along the extensive margin when preferential tariff cuts are awarded (this is likely trade diversion).  There is mixed evidence that across-the-board tariff cuts generate growth in the extensive margin.  Overall, DeBaere and Mostashari find little evidence that tariff cuts are economically significant in inducing new trade.  They do point out, however, that other policy changes frequently embedded in PTAs, like investment guarantees, may have effects on the extensive margin that are not observed in their estimates.

PTA effects on trade with excluded countries

One of the generally overlooked effects of PTAs is their effects on excluded countries.  If preferential access reduces third-country exporters' supply prices, then those third countries are harmed by PTAs.  In a world of proliferating preferential arrangements, such effects are particularly important, as subsequent agreements by a PTA partner may dilute the effects of any given agreement.  Evidence of third country effects can also be used to justify agreements with countries that have made agreements with other third parties.

In an application that specifically investigates the effects of PTAs on third countries, Chang and Winters (2002) explore variation in Brazilian import prices following MERCOSUR.   Chang and Winters find that prices charged by non-MERCOSUR exporters in Brazil fall more when preferential tariff cuts are larger.  This suggests that that preferential tariff cuts are reducing the mark-up that non-MERCOSUR exporters receive in the Brazilian market.  In the context of Chang and Winters' model, where price reductions are interpreted as reduced mark-ups for non-MERCOSUR exporters, this is evidence of third country harm.  Chang and Winters point out that these price reductions imply additional benefits to Brazil from MERCOSUR.  In Chang and Winters' model, these benefits from trade liberalization arise through increased competitive pressure.

Trade and Rules of Origin

A necessary feature of PTAs are rules of origin (RoOs), which define the set of goods that can be traded under the preferential terms of the agreement.  While necessary for meaningful enforcement of the agreement, such rules are problematic in that they increase firms' administrative costs, and they limit the degree to which an agreement increases trade among the parties.  A recent literature has emerged that quantifies some of these effects of rules of origin.

One approach to studying these issues is to inspect the import data, and ask whether exporters are taking advantage of the preferential tariffs awarded in the PTA.  If the process of establishing origin is costly, firms may choose not to do so, and to pay the MFN tariff instead.  Herin (1986) showed that around 21.5% of EFTA imports from the EC, and  27.6% of EC imports from EFTA paid non-preferential tariff rates.

Anson et al (2005) calculate implied administrative costs for NAFTA rules of origin.  They show that NAFTA utilization rates are well below 100% on U.S. imports from Mexico.  For aggregate trade, on products for which there is a positive preference for NAFTA-qualifying imports, Anson et al calculate a NAFTA utilization rate of 83 percent.  Utilization rates vary across sectors, and are low in some sectors (e.g. Textiles and apparel, Leather goods) where NAFTA tariff preferences are high.  Anson et al. combine information on utilization rates and tariff preferences to infer administrative costs.  They estimate that the costs of meeting NAFTA RoOs are equivalent to a tariff of about 1.8 percent.  This amounts to about 1/3 of the preferential tariff margin the U.S. awards Mexican imports.

The proliferation of bilateral agreements signed by the U.S. and EU have produced a number of hub-and-spoke arrangements, with smaller countries entered into agreements with one or both of the larger partners.  One way in which rules of origin inhibit trade is that they preclude, for example, Turkey using Norwegian parts in products sold into the EU, even though Turkey and Norway both have PTAs with EU.  Faced with growing complaints from industry about the complexity of the system, the EU, EFTA, and several European countries agreed to allow "diagonal cumulation" of rules of origin across partners.  This means that parts originating in a different "spoke" of a hub and spoke arrangement would be able to enter under the preferential tariff.

Augier et al (2005) point out that the change in cumulation rules is a useful natural experiment in which to evaluate the trade-reducing consequences of overlapping rules of origin.  The change in rules of origin was not, in this case, accompanied by a change in tariffs, so that changes in the trade pattern might reasonable be linked to changes in rules of origin alone.  Augier et al use a gravity framework to evaluate the impact of this rules change on intra-European trade.   Relatively faster spoke-to-spoke trade growth after diagonal cumulation is taken as evidence of a trade restricting impact in the more restrictive rules.  Augier et al calculate a lower bound on the trade restricting impact of rules of origin at 10%, and an upper bound at 70%.

III. Quantifiable effects of policy commitments other than tariffs

One of the motivations for the recent spate of PTAs is the desire among many countries to undertake a more ambitious trade policy agenda.  Developed country tariffs are now quite low, especially on the goods exported by other developed countries.   As a result, developed country trade ministries interested in removing hurdles to exporting have shifted their negotiating efforts away from tariff reduction and toward other measures.  The effects of these measures can often be difficult to evaluate empirically, as - unlike tariffs and trade - neither the policy measure itself, nor the dependent variable of interest is quantifiable.  This section discusses some attempts to quantify impacts of policy changes other than tariffs.

Dispute settlement and administered protection

One of Canada's primary concerns in negotiating a free trade agreement with the United States was to limit the arbitrary use of US administered protection laws.  In negotiations for the CUSFTA, Canada proposed that the U.S. and Canada eliminate anti-dumping (AD) and countervailing duty (CVD) actions against one another; the U.S. rebuffed this suggestion.  Instead, the two nations agreed to create a binational appellate panel to determine whether or not a country had made errors "in fact or law" when reaching an AD/CVD decision.  Similar language was included in NAFTA and Mexican law was also brought under the purview of a review panel.

Blonigen (2002) investigates the effects of review panels on two types of activity: the filing of AD/CVD petitions by parties in the U.S., and the number of affirmative decisions by U.S. authorities.  Blonigen shows that, after controlling for relatively large Mexican and Canadian shares of U.S. imports, Mexico and Canada face relatively fewer AD/CVD actions.  The calculations suggest that the two countries both experience approximately six fewer AD/CVD cases than would otherwise be expected.

Blonigen argues that the agreements could have affected U.S. behaviour through two channels: AD/CVD activity could have increased because imports from Mexico and Canada grew after NAFTA; and/or the new appeals process could have reduced such AD/CVD activity.   He finds relatively little evidence that either channel affects either measure of AD/CVD activity.  Import penetration is not statistically significant in his regressions, meaning that one cannot conclude that increased imports from the NAFTA partners led to increased AD/CVD activity.  His measures of panel activity appear not to affect either AD/CVD filings or affirmative determinations, with one exception.

Blonigen's empirical framework asks whether past action by the appellate panel affects either the number of filings by domestic parties in the U.S. or affirmative determinations by U.S. authorities.  When only the previous year's activity is included in the regression, he finds no statistically significant effects between remands and either measure of AD/CVD activity.  When cumulative remands are included as the independent variable, he finds generally the same pattern.  The one exception is that the coefficient on the cumulative number of Canadian remands is negative and statistically significant for the affirmative determinations measure of AD/CVD activity.  This is limited evidence that, in the case of Canada, the appellate panel has reduced the number of affirmative findings by U.S. authorities.  There is no such evidence for filings by U.S. domestic petitioners.

Because the U.S. steel industry is a frequent complainant in U.S. administered protection hearings, Blonigen also considers the effect of the Nafta review panels on AD/CVD activity by the steel industry.  The estimates suggest that Canada and Mexico experience approximately 1.5 fewer AD/CVD cases than otherwise would be expected.  As with aggregate trade, there is little evidence that NAFTA affects AD/CVD activity in the steel industry.

Blonigen concludes that the appellate panels have had relatively little impact.  He argues that it would be better to pursue a broader agenda of incorporating anti-dumping and countervailing duty laws into competition policy.  Because this is politically difficult, he argues that perhaps a better option would be to encourage safeguard actions as an alternative to AD/CVD cases.

Intellectual property rights

One of the more contentious trade policy issues in recent years has been the incorporation of commitments on intellectual property rights (IPRs) into trade agreements.  The most visible and controversial of these commitments is the incorporation of an agreement on Trade-Related aspects of Intellectual Property Rights (TRIPs).  PTAs now frequently include policy commitments in IPRs

The inclusion of IPRs in such agreements has not yet afforded the opportunity of an investigation into the effects of specific PTA provisions on economic behaviour.  There are, however, some studies that are informative about the effects of such agreements on technology transfer, investment in research and development, and income transfers between countries.

McCalman (2005) conducts a structural estimation of a dynamic model of innovation and growth.  He fits the model to data on the stock of domestic and foreign patents in each country, an index measuring IP protection, other determinants of patenting behaviour (i.e. distance between innovator and the country where the patent is registered).  He then uses structural estimates to conduct a number of policy experiments.

McCalman first evaluates the net transfers from TRIPs under the assumption that TRIPs has no impact on the innovative activity.  In this scenario, approximately half of the countries (including Australia) are net losers from the TRIPs agreement.   The United States, Germany and France gain the most from the agreement, while India, Brazil and Canada lose the most.

In the second scenario, McCalman allows innovation to respond to the increased global value of patents.  In this scenario, the central point estimates of the value of TRIPs to each country imply a net gain for all countries considered in the analysis. While all countries are winners, the gains from TRIPs remain lopsided, with developed country innovators benefiting much more than poorer countries that are net payers of patent royalties.

In the final scenario, McCalman considers the effects of unilateral IPR reforms by each country in the sample.  Approximately half of the countries show a net benefit from unilateral reform, while half show a net loss.  The gains to most countries from unilateral reforms are considerably smaller than the gains from universal reform.  This suggests that there are gains from coordinating IPR reforms.

Branstetter, Fisman and Foley (2004) conduct a somewhat different exercise.  They consider the impact of IPR reforms around the world on the foreign patenting, technology transfer, and research and development activities of U.S. multinationals.  One reform episode they consider is China's IPR reforms of 1993.  They find that, among firms that use patents intensively, IPR reforms have a significant and positive impact on patenting behaviour, royalty payments from affiliates (an indicator of technology transfer), and research and development spending by affiliates (such spending is often associated with technology transfer).  There does not appear to be a similar increase in multinational's arm's length licensing behaviour.  Despite these reforms, multinationals prefer to keep their technologies within the firm.

Branstetter, Fisman and Foley also consider the effect of China's 1993 reforms alone.  They find that the results from the broader sample do not generalize to China.  U.S. multinationals do not appear to have responded to Chinese reforms by increasing technology transfers.  The authors argue that this is likely because Chinese enforcement of IPRs was lacking.

IV. Trade agreements' estimated effects on the broader economy

In this section we turn to assessments of the effects of trade agreements on economic outcomes that matter: welfare, productivity and the distribution of income.  Many econometric studies focus on the response of trade flows to trade policy changes, but increased trade volumes are not a meaningful end in themselves.  Presumably the purpose of trade agreements is to improve economic well being, rather than simply increasing trade.

The difficulty in such assessments is that trade policy is only one of many variables that bear upon the variables we care most about.  While the effect of trade policy on trade might well be observable, it is much harder to tease out effects on other variables that are only indirectly affected by trade policy.  The challenge in such assessments is to avoid the post hoc ergo propter hoc fallacy (if an event follows A it is therefore caused by A) in making these assessments. USITC (2003) notes, for example, that some economic changes that are frequently attributed to trade agreements in the U.S. political debate might also be plausibly attributed to increased trade with developing countries like India and China.    This review considers only those studies that isolate the effect of specific policy changes on the variables of interest.

Welfare

The most important variable to consider in assessing trade agreements is economic welfare.  But, as welfare is unobservable, assessments of welfare are also tenuous.  What is typically required is theoretical model that allows estimated changes on import prices (and/or) quantities to be translated into welfare changes.  We have already reviewed Romalis (2005), who uses models of supply and demand at the detailed product level to make inferences about the welfare consequences of tariff changes in CUSFTA and NAFTA.  As noted above, he finds no welfare changes for the U.S. and Canada in NAFTA.  Mexico's welfare change is -0.3% of GDP.  Romalis is quick to point out that his partial equilibrium framework may miss some important aspects of welfare change.

Most welfare assessments of trade agreements are calculated ex ante in simulation analysis of AGE models.  In its assessment of the effects of 5 trade agreements on the United States, USITC (2003) conducted an ex post simulation analysis in which the tariff cuts from each agreement were reimposed on the 2001 model of the U.S. economy.   The estimated welfare changes from those simulations suggested that U.S. welfare in 2001 was approximately 0.6 percent higher than it would have been in the absence of tariff cuts embodied in these agreements.

As with other AGE models, the estimates of welfare changes in these simulations are dependent on the model structure, and on the parameterization of that structure.  The value of these particular simulations is that they consider the effects of 5 separate agreements using the same model and parameterization, allowing a transparent comparison of the relative effects of the agreements.   Figure 4 shows the estimated welfare effects from removing each of the 5 agreements (i.e. putting the tariffs back in place) in 2001.

These simulations are most useful in demonstrating the relative sizes of the economic impacts of the PTAs, relative to each other and relative to multilateral agreements.  NAFTA and CUSFTA were significant agreements for the United States; large preferential tariff cuts were awarded to two of the U.S. largest trading partners.  Nonetheless, these agreements appear to be substantially less significant, in terms of welfare, than multilateral rounds.  The simulated gains from the Uruguay and Tokyo rounds appear to have been at least 3 times as large as either of the two preferential arrangements.

Figure 4.  USITC ex post estimates of welfare changes in 5 trade agreements

Source: USITC (2003)

Another lesson to take from Figure 4 is the very small gains estimated from the U.S.-Israel FTA.  In recent years, the U.S. has negotiated a number of PTAs with countries that trade even less with the U.S. than does Israel (e.g. Bahrain, Jordan, Morocco).  As these figures make clear, the economic gains to the United States from such agreements are quite small.  These estimates help to emphasize a point discussed below, that countries like the U.S. frequently negotiate PTAs in which economic benefits to the U.S. are a secondary consideration.

Productivity

Welfare estimates like those of Romalis (2005) and USITC (2003) may understate the gains from trade liberalization if trade liberalization increases productivity.  One motivation for Canada negotiating a PTA with the United States was the hope that Canadian firms would be able to increase productivity by increasing production scale.  In light of this argument, a question of interest in the empirical literature has been whether, ex post, Canadian productivity can be shown to be higher because of the agreement.

Trefler (2004) studies plant level productivity and employment in Canadian manufacturing over a 16 year period centered around the Canada-U.S. Free Trade Agreement. He estimates separate impact of both Canadian and U.S. tariff changes, and combines these estimates to consider effects of the agreement on employment and labour productivity.  He finds that the FTA reduced manufacturing employment by 5 percent, though this number masks significant differences.  Exporting industries increased employment and import-competing industries reduced employment.

Given plant level output and employment levels, Trefler is able to identify effects on productivity.  He finds large productivity increases in import-competing industries (where employment is falling).  It appears that this productivity shift happens in large part because the least productive Canadian firms decrease their share of domestic production.  On the export side, Trefler finds that the FTA increases productivity at the plant level; industry level productivity is less because less productive new firms entered in the export-intensive industries.

While Bernard et al (2003) are not looking at specific agreements, but rather changes in trade costs, their results are broadly consistent with Trefler's.  Those U.S. industries that experience the largest reductions in international trade costs experience the largest increases in productivity.  It appears that the productivity change occurs in large part through the exit of less productive firms.

V.  PTAs and the broader policy agenda

A notable contributor to world wide growth in the number of PTAs has been the United States.  Recent U.S. bilateral arrangements would seem difficult to justify on economic grounds alone.  The trading partners are distant from the U.S., scattered around the globe, and are generally not large U.S. trading partners.  Instead, the U.S. trade negotiatiors appear to be pursuing a number of non-economic objectives such as cooperation in military matters and in drug interdiction.  The U.S. Trade Representative also argues that the negotiations are motivated by a strategy of "competitive liberalization."  The idea being that U.S. participation in bilateral arrangements will push forward negotiations in multilateral fora.

What recent experience with the U.S. makes clear is that countries do not necessarily choose FTA partners with an eye toward maximizing economic welfare in the short run.  In some cases, trade ministries are pursuing non-economic objectives.  In others, they might be choosing partners in a manner devised to affect the negotiating environment in another forum.  In any case, standard simulation analyses and measures of economic welfare are not accurately identifying the full impact of the agreement.

Whalley (1996) asks the question "Why do countries seek regional trade agreements?"  Whalley notes a number of explanations for regional agreements.  In addition to standard gains from trade liberalization, countries may seek a) to strengthen domestic policy reforms by making binding international commitments, b) to affect trade negotiations in other fora, c) to ensure access to markets that might serve as a safe haven in a multilateral crisis, and d) to encourage cooperation on non-economic objectives.

Two recent papers discuss objectives of this sort.  The first paper looks at the issue of PTAs as possible anchors for domestic policy reforms, especially in developing countries.  The second considers a theoretical model of the implications of PTAs on multilateral negotiations, when non-economic objectives are taken into consideration in PTA negotiations.  The implications of the theoretical model - that PTAs can undermine multilateral progress when they are negotiated with non-economic objectives in mind - are validated in the data.

Policy anchors

Ferrantino (2006) considers the possibility that recent trade negotiations have improved governance in developing countries.  One of Mexico's motivations for signing the NAFTA agreement was to anchor the reform program it had put in polace over the previous decade.  Ferrantino considers developing country partners of the U.S. and developing countries seeking to join the WTO.  He asks whether participating in negotiations has improved these countries' measures of governance.  Ferrantino uses quality of governance measures taken from the Heritage Foundation's Index of Economic Freedom, and the World Bank's Governance Matters indicators.

Ferrantino notes that US PTA partners score better on these measures than recent applicants to join the WTO.  He also notes that U.S. PTA negotiations are considerably shorter than WTO working parties.   The median US PTA negotiation takes only 2.5 years, while the median length of WTO working party negotiations is 6.8 years.  This much shorter period may reflect an ability of one-on-one negotiations to deal with country specific peculiarities more quickly.

Ferrantino finds relatively little evidence that governance systematically improves, either through negotiating a PTA with the U.S., or through accession to the WTO.  There are anecdotes of such success stories, including Mexico with Nafta and China with the WTO.  However it does not appear that such negotiations can be said to improve governance as a matter of course.  In fact, several measures of governance worsen across the period of negotiation.  Ferrantino emphasizes that one should not jump too quickly to attributing causation to this result.  It may be that a deteriorating policy environment was a backdrop against which the negotiations took place.

Non-economic objectives

Whalley argues that countries may choose to enter in to PTAs in order to get cooperation from their PTA partners on other issues.  Limao (2005) notes that tariff preferences given on such a basis will implies a concern about preference erosion in subsequent negotiations.  Reducing the trade preference by reducing the multilateral tariff in the next round may reduce the level of cooperation on the non-economic issues.

Limao (2005) investigates U.S. tariff cuts in the Uruguay Round.  He finds that U.S. tariff cuts are systematically smaller on goods that it trades with PTA partners.  Limao and Karacaovali (2005) find similar results in European Union data. This suggests that PTA preferences are reducing the degree of liberalization that large countries like the EU and the US conduct through multilateral liberalization.

VI. Conclusion

There are several overlapping reasons for the recent proliferation of PTAs.  One reason is the increasing complexity of international trade, measured here as an increased share of differentiated commodities.  The increase in the share of differentiated commodities also helps to explain the tendency of recent agreements to include such commitments as intellectual property rights, investment, dispute settlement mechanisms, and more.  The difficulty in negotiating commitments in these politically sensitive areas is another reason for the shift toward PTAs, for it is perhaps implausible that wider ranging commitments can be effectively negotiated in the current multilateral setting.

Economists' standard methods for quantifying the effects of PTAs are well suited for occasional agreements in which tariffs are the primary policy instrument.  As trade agreements have begun to incorporate broader policy measures, it has become harder to provide good summary measures of the agreements' effect on economic welfare.  These difficulties are compounded by the interactions between multiple PTAs  - across both space and time.  Each agreement that is signed affects the operation of other agreements, as well as the substance of future negotiations.

One area of particular policy concern to trade economists is the degree to which rules of origin in PTAs constrain optimal sourcing behaviour.  The effect of rules of origin are likely to be even more problematic as PTAs begin to overlap.  As this problem became evident in Europe, steps were taken to minimize the losses due to rules of origin.  These efforts appear to have been successful in increasing intra-European trade.

Countries may enter into to PTA negotiations with objectives other than maximizing a comparative static welfare measure.  PTA negotiations may well influence other negotiations in ways that are perceived to be beneficial to the PTA parties.   PTAs may also be part of a broader foreign policy strategy, where tariff preferences are awarded in exchange for cooperation on other issues.   Evidence from

EU and US tariff cuts in the Uruguay Round suggests that those countries PTAs appeared to negatively affect their willingness to reduce multilateral tariffs.

References

Aitken, Norman (1973) "The effect of the EEC and EFTA on European trade: A Temporal Cross-section Analysis," American Economic Review, 63: 881-892.

Anson, Jose´, Olivier Cadot, Antoni Estevedeordal, Jaime de Melo, Akiko Suwa-Eisenmann and Bolormaa Tumurchudur (2005) "Rules of Origin in North-South Preferential Trading Arrangements with an Application to NAFTA,"  Review of International Economics, 13: 501-517.

Augier, Patricia, Michael Gasiorek, and Charles Lai-Tong (2005), "Rules of Origin" Economic Policy, (July) 567-624.

Blonigen (2002) "The Effects of CUSFTA and NAFTA on Antidumping and Countervailing Duty Activity," University of Oregon, mimeo.

Branstetter, Fisman and Foley (2004)

 "Do Stronger Intellectual Property Rights Increase International Technology Transfer? Empirical Evidence from U.S. Firm-Level Panel Data" World Bank Policy Research Working Paper 3305

Broda, Christian and David Weinstein (2004) "Globalization and the Gains from Variety," Federal Reserve Bank of New York, Staff Report 180.

Brown, Drusilla (1987) "Tariffs, the Terms of Trade and National Product Differentiation," Journal of Policy Modeling, Fall, 1987.

Chang, Won and Alan Winters (2002) "How Regional Blocs Affect Excluded Countries:  The Price Effects of MERCOSUR," American Economic Review, 92: 889-904

Clausing, Kimberley, (2001) "Trade Creation and Trade Diversion in the Canada-United States Free Trade Agreement", Canadian Journal of Economics, 34: 677-696.

DeBaere, Peter (2006) "Small Fish, Big Issues:  The Effect of Trade Policy on the Global Shrimp Market," mimeo, University of Texas.

DeBaere, Peter and Shalah Mostashari (2006) "Do Tariffs Matter for the Extensive Margin of International Trade? An Empirical Analysis" mimeo, University of Texas.

Ferrantino, Michael (2006) "Policy Anchors: Do Free Trade Agreements And WTO Accessions Serve As Vehicles For Developing-Country Policy Reform?" U.S. International Trade Commission Working Paper 2006-04-A.

Feenstra, Robert C., Robert E. Lipsey, Haiyan Deng, Alyson C. Ma, and Hengyong Mo, (2004) "World Trade Flows: 1962-2000," NBER Working Paper no. 11040.

Ghosh, Sucharita and Steven Yamarik (2004), "Are Regional Trading Arrangements Trade Creating?  An Application of Extreme Bounds Analysis." Journal of International Economics, 63: 369-395.

Gresser, Edward (2002) "Toughest on the Poor," Foreign Affairs, 81: 9-14.

Herin, Jan (1986) "Rules of Origin and Differences Between Tariff Levels in EFTA and the EC,"  EFTA occasional paper 13.

Hillberry, Russell and Christine McDaniel (2002), "A Decomposition of North American Trade Growth Since NAFTA," USITC Working Paper 02-12-A.

Hummels, David and Peter Klenow (2005), "The Variety and Quality of a Nation's Exports," American Economic Review 95: 704-723.

Kehoe, Timothy and Kim Ruhl (2003), "How Important is the New Goods Margin in International Trade?" Federal  Reserve Bank of Minneapolis Staff Report 324.

Limao (2005) "Preferential Trade Agreements as Stumbling Blocks for Multilateral Trade Liberalization: Evidence for the U.S.,"  CEPR Working Paper 4884.

Limao and Karacaovali (2005) "The Clash of Liberalizations: Preferential vs. Multilateral Trade Liberalization in the European Union" World Bank Policy Research Working Paper No. 3493

Rauch, James, (1999) "Networks vs Markets in International Trade," Journal of International Economics, 48: 7-35.

Rodrik, Dani (2006) "What's So Special about China's Exports?" NBER Working Paper 11947.

Romalis, John (2005), "NAFTA's and CUSFTA's Impact on International Trade" NBER Working Paper 11059.

Schott, Peter (2006), "The Relative Sophistication of Chinese Exports" NBER Working Paper 12173.

Trefler, Daniel (2004) "The Long and the Short of the U.S.-Canada Free Trade Agreement" American Economics Review 94: 870-895.

U.S. International Trade Commission (2003), The Impact of Trade Agreements:  Effect of the Tokyo Round, U.S.-Israel FTA, U.S.-Canada FTA, NAFTA, and the Uruguay Round on the U.S. Economy, Publication 3621.