Clarifications

• Trade is a generator of economic well-being, enriches nations and allows companies and workers to specialise in doing what they do best.  Competition forces them to become more productive. Consumers reap the bounty of cheaper and better goods and services.

• Imports are the real fruits of trade because the end goal of economic activity is consumption. Exports represent resources we don't consume at home, allowing us to pay for what we buy abroad. 

• Mercantilism: economic practice from the 16th to 18th century, where nations sought to increase wealth through export surplus and controlled trade.

Comparative Advantage was introduced by David Ricardo in The Principles of Political Economy and Taxation (1817), who stressed that potential gains from international trade were not confined to Adam Smith's absolute advantage.

Goal of this chapter is show that contrary to the Mercantilist thinking, that trade is a positive-sum game in which all trading partners benefit.

Positive Sum Game: where the net result is greater than zero, even though there may be some winners and losers. In economics, trade and exchange are thought to be examples of a positive sum game.

Assumptions of the Basic Ricardian Model

1. Each country has a fixed endowment of resources, and all units of each particular resource are identical.

2. The factors of production are completely mobile between alternative uses within a country. This assumption implies that the prices of factors of production also are the same among these alternative uses.

3. The factors of production are completely immobile externally; that is, they do not move between countries. Therefore, factor prices may be different between countries prior to trade.

4. A labour theory of value is employed in the model.

Labour theory of value: based solely on its relative labour content.
From a production standpoint, this implies that either:
4.1. No other inputs are used in the production
4.2. Any other inputs are measured in terms of the labour embodied in their production.
4.3. The other inputs/labour ratio is the same in all industries.
In simple terms, this assumption means that a good embodying two hours of labour is twice as expensive as a good using only one hour.

5. The level of technology is fixed for both countries, although the technology can differ between them.

6. Units costs of production are constant. Thus, the hours of labour per unit of production of a good do not change, regardless of the quantity changed. This means that the supply good of any good is horizontal.

7. There is full employment.

8. The economy is characterised by perfection competition. No single consumer or producer is large enough to influence the market; hence all are price takers:

8.1. All participants have full access to market information

8.2. There is free entry to and exit from an industry
8.3. All prices equal the marginal cost of production.

9. There are no government-imposed obstacles to economic activity.

10. Internal and external transportation costs are zero.

11. The analysis is confined to a two-country, two-commodity "world' to simplify the presentation. This simplification will be dropped later to make the model more realistic.

David Ricardo's Mini-Bibliography

David Ricardo (1772-1823): was born in London, the son of wealthy Jewish immigrants. He began a career independently in securities making him a fortune. After reading Adam Smith's The Wealth of Nations in 1799 whilst on vacation he gradually made economics his avocation. Ricardo was opposed to government's gold policies and the corn laws. He began readings into questions of profits and income distribution. 

In 1817, Ricardo published The Principles of Political Economy and Taxation. He became a member of Parliament in 1819, and became a great educator to the House of Commons on economic questions. Credited with originating the concept of comparative advantage, and created an entire model of the economic system in which growth rests on capital accumulation and profits and the law of diminishing returns eventually leads to a stationary state with zero profits and affluent landlords.

Ricardian Comparative Advantage

Ricardo began by noticing that Smith's idea of absolute advantage determined the pattern of trade and production internal to a country when factors were perfectly mobile.

Industry locates where the greatest absolute advantage exists and that labour and capital move to the area where productivity and returns are the greatest. 

The movement would continue until factor returns were equalised. While international trade can take place on the basis of absolute advantage (trade between tropical and temperate zones), given the international immobility of the factors of production, gains from trade on the basis of comparative advantage can occur as well.

England's Cloth & Portugal's Wine for Comparative Advantage

Ricardo presented a case describing the production of two commodities, wine and cloth, in England and Portugal. The labour requirements per unit of production presented below reflect the technologies in each country and imply the relative value of each commodity.

In this example, Portugal has an absolute advantage in the production of both commodities. From Smith's perspective, there is no basis for trade between these countries because Portugal is more efficient in the production of both goods; England has an absolute disadvantage in both goods.

However, Ricardo pointed out that Portugal is relatively more efficient in the production of wine than of cloth and that England's relative disadvantage is smaller in cloth. The figures show that the relative number of hours needed to produce wine (80 in Portugal, 120 in England) is less than that of cloth (90 in Portugal, 100 in England). Due to these relative cost differences, both countries have an incentive to trade.

Autarky (pre-trade) price ratios: the price ratios when the country has no international trade. Refers to the state of self-sufficiency and is typically used to describe nations or economies that have the goal of reducing their dependency on international trade.

Consider the autarky price ratios

• Within England, 1 barrels of wine would exchange for 1.2 yards of cloth (because the same labour time is embodied in each quantity)

• While in Portugal, 1 barrel would exchange for a measly 0.89 yard of cloth

• Thus, Portugal can gain if it can specialise in wine, acquiring cloth from England at a ratio of 1 barrel of wine : 1.2 yards of cloth 

• Similarly, England specialising in cloth production would receive 0.98 barrels of wine per yard of cloth instead of 1.2 barrels of wine per yard of cloth at home.

Even though trade is unrealistically restricted to two goods, similar potential gains also occur in more comprehensive analysis. The main point is that the basis for and the gains from trade rest on comparative, not absolute, advantage.

• With England in autarky, 1 barrel of wine exchanges by the labour theory of value for 1.2 yards of cloth, so any price ratio in which less than 1.2C has to be given up for 1W is desirable for England. 

• Similarly, the autarkic price ratio in Portugal is 1W:0.89C.

• Thus, Portugal will gain if its wine can command in trade more than 0.89 units of cloth. With an international price ratio between these autarkic price ratios, both countries will gain.

How Price Ratios Converge

Ricardo did not examine the precise determination of the international price ratio or the terms of trade, but after trade there will be a common price of wine in terms of cloth in the two countries. 

• As wine comes into England from Portugal, and the Portuguese demand English cloth, the relative price of English cloth in terms of wine will rise (less cloth for a unit of wine than previous 1.2C). 

• In Portugal, something similar occurs where the relative price of wine will rise. Thus the price will rice above 1W:0.89C toward more cloth being given up to obtain a unit of wine converging towards the other pre trade ratio of 1W:1.2C through trade. 

This is the economic phenomenon of two separate markets (autarky) unifying into a single market (trade). A single price will then prevail. With trade, prices are no longer determined solely by the labour theory of value but also by the relative demands in the two trading countries.

How Trade Saves Time

Ricardo arbitrarily assumed that the terms-of-trade ratio was 1W:1C. 

• With trade, England could devote 100 hours of labour to producing cloth (its comparative advantage good), and get 1C. This 1C could be exchanged with Portugal for 1W. Thus, 100 hours of labour in England have indirectly produced 1 unit of wine. 

• If England had chosen to produce 1W at home directly, the cost involved would have been 120 hours of labour. Therefore, trade saves 20 hours of labour for each unit of its imported good. 

Ricardo expressed the gains in terms of labour time saved because he viewed trade as a mechanism for reducing the outlay of labour necessary for obtaining goods, for such labour implied work effort and "real costs". Therefore, with trade more goods can be obtained for the same amount of labour time than is possible in autarky.

• Portugal can take 80 hours of labour and produce 1 unit of wine, then obtain 1 unit of cloth through trade. 

• Alternatively then can engage in direct production of 1 unit of cloth, requiring 90 hours of labour. 

• Trade has enabled the country to gain or save 10 hours of labour per unit of its imported good. Thus, unlike the zero-sum game of the Mercantilists, international trade is a positive-sum game.

Terms of trade are important for the distribution of gains between the two countries.

• Suppose that we specify the terms of trade as 1W:1.1C instead of 1W:1C. 

• We can expect Portugal to gain more in this case because its export good is now commanding a greater volume of the English good. 

• England experiences smaller gains in the second case.

The closer the terms of trade are to a country's internal autarky price ratio, the smaller the gain for that country from international trade. At the limits, the country whose prices in autarky equaled the terms of trade would get no gain and would be indifferent to trade. The other would obtain all the gains from trade.

Equilibrium terms of trade: those that bring about balanced trade for each country, exports = imports in total value. 

If the Ricardian 1W:1C ratio left Portugal with a balance-of-trade surplus the terms of trade would shift toward relatively more expensive wine, say 1W:1.1C. This shift occurs because the price-specie-flow mechanism raises prices and wages in the surplus country, Portugal, and depresses them in the deficit country, England.

Price-specie-flow mechanism: illustrates how trade imbalances can self-correct and adjust under the gold standard.

Trade moves production in all countries toward a more specialised production pattern than would be the case in autarky.

Comparative Advantage & The Total Gains from Trade

The essence of Ricardo's argument is that international trade does not require different absolute advantages and that it is possible and desirable to trade when comparative advantages exists.

Comparative Advantage: exists whenever the relative labour requirements differ between the two commodities. Simply, when the relative labour requirements are different, the internal opportunity cost of the two commodities is different in the two countries, that is the internal price ratios are different between the two countries prior to trade. 

When trade is initiated between the two countries, it will take place at international terms of trade that lie within the limits set by the price ratios for each country in autarky. If trade takes place at one of the limiting autarky price ratios, one country reaps all the benefits.

• Country A has comparative advantage in the production of cloth, and country B for wine. 

• The relative labour cost for country A (1/2) in cloth is less than that in wine (3/4). 

• The basis for trade is also evident in the fact that the autarky price ratios in each country are different.

• Trade commences at international terms of trade of 1W:3C.

• Then country B gains 1 yard of cloth per each 1 barrel of wine exchanged, while country A gains nothing because it pays the same relative price that it faces in autarky. 

• Thus, for both countries to gain, the international terms of trade must lie between the autarky price ratios.

Reciprocal Demand: the actual location of the equilibrium terms of trade between the two countries is determined by the comparative strength and elasticity of demand of each country for the other's product a concept developed by John Stuart Mill in 1848.

Ricardo's argument is that international trade doesn't require different absolute advantages and that it is possible and desirable to trade when comparative advantage exists.

Resource Constraints

To demonstrate total gains from trade it is necessary to establish the amount of the constraining resource, labour, available to each country:

• Suppose that Country A has 9,000 labour hours

• Country B has 16,000 hours available. 

• This alongside the production information above in table 3 permits us to establish the production possibilities open to these two countries in autarky.

• Country A can produce 9,000 yards of cloth and no wine, or 3,000 barrels of wine and no cloth, or any combination which absorbs the 9,000 hours of labour.

• Country B can produce 8,000 yards of cloth and no wine, or 4,000 barrels of wine and no cloth.

• Assuming that country A produces 6,000 yards of cloth and 1,000 barrels of wine prior to trade and that country B produces 3,000 yards of cloth and 2,500 barrels of wine. 

• Suppose that the two countries exchange goods at the terms of trade of 1W:2.5C, and that country A exchanges 2,500 yards of cloth for 1,000 barrels of wine from country B, but the two countries do not alter their production. How will the post trade and pre trade scenarios compare?

• Keeping with Ricardo's emphasis on labour time, we examine the equivalent quantity of domestic labour services consumed before and after trade for each country. We use labour hours because wine and cloth cannot be added meaningfully without weighting for relative importance. 

• After trade, country A consumes 3,500C (from 6,000C), and 2,000W (from 1,000W), a combination that would have required 9,500 labour hours if produced at home. Country A has thus gained the equivalent of 500 labour hours trough trade. 

• Similar for country B, resulting in a combination which would've costed 17,000 labour hours.

Complete Specialisation

In the previous example, both countries gained from trade even though neither altered its production of cloth or wine, but this is an incomplete picture. With the new prices determined by trade, producers will necessarily increase the production of the good that has a comparative advantage because this good gets a relatively higher price on the world market than it did in autarky.

Complete Specialisation: all resources are devoted to the production of one good, with no production of the other good.

Both countries now alter their production patterns and engage in complete specialisation in the commodities in which they have a comparative advantage. Each experiences even greater gains from trade:

• If country A only produced cloth, and B only wine, they exchange 2,000 barrels of wine for 5,000 yards of cloth. 

• A would consume consume 4,000C (9,000 yards produced), and 2,000W (all imported). This combination has a labour value in country A of 10,000 hours. 

• Country B is also better off with 18,000 hours.

• This contrasts with a labour value of 16,000 in autarky, and 17,000 in trade with incomplete specialisation of production. 

The Classical writers concluded that if there is a basis for trade, it automatically leads a country toward a specialisation in the commodity in which it has the comparative advantage. Consumption remains diversified across goods as dictated by consumer preferences.

Representing the Ricardian Model with Production Possibilities Frontiers

The basis for trade and the gains from trade can also be demonstrated with the production possibilities frontier (PPF) concept. 

Production possibilities frontier: reflects all the combinations of two products that a country can produce at a given point in time given its resource base, level of technology, full utilisation of resources, and economically efficient production.

As all of these conditions are met in the list of assumptions earlier presented, it is clear that the Classical model assumes the participating countries to be producing and consuming on their PPF's in autarky. Furthermore, the constant-cost assumption implies that the opportunity cost of production is the same at the various levels of production. Therefore, the PPF is a straight line whose slope represents the opportunity cost of economy-wide production. 

The shift into this framework presents a graphical picture of the Ricardian Model, and provides a means for escaping from the limitations of the labour theory of value while retaining the comparative-advantage conclusions about the basis for trade.

Slope of the PPF: indicates the opportunity cost amount of production of one commodity must be given up to obtain one additional unit of the other commodity, the values that lie at the basis of this calculation can reflect the cost of all inputs, not only labour, that go into the production of the commodities.

This realisation makes the concept of comparative advantage more realistic and interesting, implying that the basic idea is sufficiently general to cover a wide range of production scenarios, whilst a labour theory of value is only one.

Production Possibilities Example

The figures on labour hours and production for countries A & B make it possible to display the PPF's for each country. A production possibilities schedule can be calculated and the respective PPFs can be inferred from those schedules:

Introduction

Contrasting Vignettes on Trade Policy

In January 2004, U.S. Trade Representative Robert Zoellick proposed that all agricultural export subsidies by countries be eliminated and that all tariffs on food and other agricultural imports be cut dramatically. These and other proposals were made in a letter to the 148 member countries of the World Trade Organization. The intent of the letter and the proposed measures was to put new life into multilateral trade negotiations to reduce trade barriers worldwide:

In November 2003 the chief trade official of the European Union (EU), Pascal Lamy, threatened that the EU would impose steep tariffs on $6 billion of U.S. exports to the EU unless the United States altered two of its trade policies:

•  First, the World Trade Organisation (WTO) had ruled that U.S. tariffs imposed in March 2002 on imported steel were illegal. 

• Second, the WTO had also ruled that favourable U.S. tax treatment of special divisions of U.S. companies set up to promote exports was illegal, since the treatment constituted an export subsidy not permitted by international trade rules. 

• Lamy indicated that the EU was seeking “compliance” with agreed rules, and that the EU intended to carry out its threat if no U.S. changes were made.

These contrasting vignettes indicate that a country, in these cases the United States, can be pushing for:

• Freer trade (first vignette)

• Employing policies that depart from free trade according to comparative advantage, second vignette

Trade policy can involve conflicting and complex economic and political forces, and outcomes are not so clear-cut as traditional trade theory would suggest. 

This chapter first addresses the question of how the setting of trade policy is influenced by institutions and the political process and then summarises U.S. and multilateral trade policy developments during the last several decades:

• We begin with a brief discussion of some common analyses of the interaction between citizens and their elected representatives/policymakers in the process of policy formulation and of the concerns reflected in a country’s trade stance. 

• As will be seen, a minority of the population is often successful in procuring policies that benefit that minority at the expense of general social welfare. 

• We then present an overview of U.S. trade policy since the 1930s. A central point is that the past approximately 80 years have seen a dramatic liberalisation of trade in the United States and other industrial countries.

The Political Economy of Trade Policy

As has often been pointed out, the one issue on which economists are in almost unanimous agreement is the social gains to be made by specialising and trading on the basis of comparative advantage and, correspondingly, opposition to protectionism. 

In spite of this view, the world continues to experience pressures to restrict the movement of goods, services, and factors between countries. Indeed, countries seem to continue to find new and novel ways to restrict these economic activities. Too often as one trade-restricting instrument falls into disfavour and is reduced or eliminated, new trade-restricting provisions seem to pop up. 

It is little wonder, then, that one is often asked, “If free trade is so beneficial to a country, why are so many groups or individuals trying their best to reduce trade?” Why is it that, as Robert Baldwin (1989, p. 119) so aptly pointed out, “international trade seems to be a subject where the advice of economists is routinely disregarded”? 

The answer to this question lies in what is referred to as “the political economy of trade policy”. 

In reality, trade policy takes place within a political-social milieu and is influenced by individuals and groups who feel that they will be better off with restricted trade even though the country as a whole may be worse off. As we have noted in previous chapters, reducing the barriers to trade may make the country better off, but as the corresponding structural adjustments are made, some individuals will be made better off and some worse off. 

Politicians thus find themselves confronted with a vast array of groups attempting to influence trade policy, and, consequently, they often ignore the advice of economists when establishing the country’s trading regime. In the past several decades, an area of research has emerged that focuses on analysing the actual determinants of trade policy in the political environment within which it is developed. We now turn to a brief discussion of several of the more important strands of research on this topical issue.

The Self-Interest Approach to Trade Policy

The study of the political factors influencing trade policy has proceeded along two major fronts:

1. The first, and perhaps the most pervasive, focuses on the economic self-interest of the political participants. Much of this literature is embedded in public-choice economics, which essentially uses economic models to analyse governmental decision-making behaviour. In this approach, government decision makers are essentially utility maximisers whose level of satisfaction is dependent upon being reelected and who act in a manner that maximizes the probability that this will in fact take place. An immediate implication of this approach is that the majority of the public will be served by public decision makers who enact legislation to maximise their chances of remaining in office.

This is the focus of the median-voter model, which holds that the decision maker who votes in such a way as to satisfy the median voter will maximise his or her reelection possibilities:

• In this approach each individual voter is assigned a position along an array based on the expected costs or benefits of a particular policy. 

• The median voter is in the center of this array, that is, the middle voter. 

• Should the majority expect to benefit from a particular policy, the median voter will be in favour of the policy and support the politician who favours it. 

• Should the majority feel that they will be harmed by a particular action, the median voter will not favour the policy and will not support a political candidate who attempts to make a case for it. 

• This is a natural model to use in studying international trade policy because, as we have noted in previous theory chapters, trade policy inevitably results in different welfare effects for different groups in the economy.

• In the median-voter framework, if a greater number of voters expect to benefit from a particular trade policy than incur a loss, the median voter would be in favour of the proposed legislation and the policymaker would presumably vote in favour of the policy measure. 

• Should the policymaker not support the legislation in this case, the median voter is likely to vote against the legislator in the next election. 

• This approach would thus seem to ensure that the will of the majority would be followed. Unfortunately, however, there are a number of practical problems, which, if they arise, can circumvent the preferences of the median voter and result in policies that are inconsistent with majority benefits. 

• The median-voter model rests on the assumptions:

  • That the voters have full information regarding any gains or losses resulting from a particular policy

  • That they will actually vote consistent with their preferences.

Inasmuch as neither of these two assumptions always holds in the real world, it is clearly possible that the preferences of the median voter do not win out. For example, a tariff benefiting only a small group of individuals may end up being supported by a voting majority.

World Attitudes Towards Foreign Trade

A Pew Research Center survey in 2011, part of the Pew Global Attitudes project, gathered opinions regarding international trade and business ties from citizens of 21 countries. A sample of the results is given in Table 1. It is interesting that, just when the world seems to be accepting and supporting globalisation, households in the United States, a country that has been a longtime advocate of open markets and freer trade, seem to be less favourably inclined toward trade and business ties than the people of other developed countries and of some emerging-market countries:

• In fact, the U.S. summed score of 67 was the lowest of the 21 countries. 

• Another interesting finding with respect to attitudes toward trade has been offered by David Coe (2008). According to Coe:

  • Public opinion surveys show that in industrial countries, higher levels of income inequality are linked to less positive attitudes toward the benefits of trade. 

  • Similar surveys from the Euro area suggest that globalisation is viewed more positively if there has been substantial redistribution toward the lowest 30 percent of the population through government policies. 

• In general, public opinion surveys suggest that pro-trade attitudes are strongly positively correlated with an individual’s economic status level. 

• Coe concludes that redistribution policies to compensate those hurt by an expansion of trade are important for gaining wider popular support for trade liberalisation.

There are a number of situations that can produce these seemingly contradictory results:

1. In the case where there are information-gathering costs as well as opportunity costs associated with actually voting, some potential voters may simply choose not to participate, particularly if expected gains are small and hence the expected net benefits negative. 

2. If, in addition, the voter feels that one vote will not actually influence or “swing” the result, he or she may simply accept the outcome without actually engaging in the political process. In this case, voters are acting as “free riders,” that is, accepting the outcome without expending any effort or costs. Although this would not be a problem if every voter had an equal probability of not participating, in reality, the differences in the amount of expected benefits and costs (asymmetric gains and losses) strongly suggest that individuals will have different degrees of incentive and that interest groups will form. 

3. A particular policy action such as sugar quotas may have only a minor individual or per capita impact on a large group in society (e.g., consumers) but a large per capita impact on a minority (e.g., sugar producers). The matter becomes even more complicated if allegations are true that recent changes in some U.S. states’ voting laws regarding voter identification and early voting deter participation by some potential voters. 

4. Interest groups can influence political outcomes in a variety of ways because the costs (benefits) of a policy are relatively great to these groups, they have an incentive to influence political action and are more likely to overcome the free-rider problem of their relatively small number of members.

1.  A small group of individuals who stand to gain much from a policy intervention in the market thus obtain a certain group solidarity, participate in the political process, and vote for the candidate who supports their position on protection. 

2. The larger group of diverse consumers who stand to lose with protectionism expect to gain little individually by making the effort to acquire information and/or vote; consequently they do not participate. 

• As a result, there is a small voter turnout and the candidate espousing the views of the solid minority block wins. 

• This phenomenon can lead to a status quo bias against liberalising trade policy through lower levels of protection even though doing so carries the promise of improving aggregate welfare. 

• For a large number of people the net personal gains are so small (perhaps even negative) or uncertain that they choose not to participate actively in the political process, and the minority interest group gets its way. 

• There is no doubt that groups such as this are very influential in the conduct of trade policy formation. 

• As Vousden (1990, p. 198) points out, interest or pressure groups tend to be more successful: 

  • If they are large enough to be visible but small enough to control the free riding of their members

  • There is a well-defined commonality of interest

  • The per capita organisational and information-gathering costs are relatively low.

U.S. Attitudes Towards Foreign Trade

The lack of enthusiastic support for trade by U.S. citizens is not only evident in the worldwide Pew survey on page 362. This absence of solid approval of international trade was also emphasised: 

• In a Federal Reserve Bank of Dallas annual report, according to which a November 2010 NBC poll showed that Americans, by 23-47%, judged that free trade “hurt” rather than “helped” the United States:

• Another November 2010 poll, conducted by the Opinion Research Corporation for CNN, revealed that:

  • Half of the U.S. respondents judged that the threats posed by imports outweighed the benefits of those imports

  • 41% thought that trade represented an opportunity. 

• A particularly interesting point made in the Federal Reserve Bank of Dallas report was that attitudes of U.S. consumers toward trade seemed to vary according to trading partner. An indication of this factor was that the average American was almost twice as likely to think that trade with Canada was good, in comparison to trade with China. Further, U.S. citizens looked more favourably toward increasing the amount of trade with Mexico than toward increasing the amount of trade with South Korea. Hence, attitudes toward trade may involve more than just the price and quality of the goods themselves, as feelings toward the country source of imported goods also seem to play a role in consumer satisfaction. 

• In more recent studies, Gallup public opinion polls in 2013 and 2014 indicated that the views of Americans began to change as of 2013. Gallup has asked respondents, in a long series of polls, whether they view international trade as an opportunity for economic growth because of increased exports or as a threat to the U.S. economy because of imports:

  • Over 20 years ago, in 1992, 48% of the respondents had said a “threat” and 44% had said an “opportunity”

  • The “opportunity” responses then exceeded the “threat” responses until about the beginning of 2005, with the high point for trade being 56%-36% in one poll. 

  • The “threat” responses then prevailed until the February 2013 poll, when the result became a majority for trade as 57% saw it as an “opportunity” and only 35% as a “threat”. 

  • This basic result continued in the February 2014 poll with a 54%-38% margin for the “opportunity” category.

  • Finally, another result in 2014 was that individuals with a bachelor’s degree or higher were considerably more favourably inclined toward trade than were individuals without a college degree. 

Sources: “Public Perception of Globalisation’s Impact Shapes Trade Realities,” Globalization and Monetary Policy Institute, 2011 Annual Report, Federal Reserve Bank of Dallas, February 2012, pp. 21–25; Jeffrey M. Jones, “Americans Shift to More Positive View of Foreign Trade,” February 28, 2013, and Lydia Saad, “Americans Remain Positive about Trade,” February 21, 2014, both obtained from www.gallup.com.

A second way in which special interest groups can influence political outcomes is through the funding of political campaigns-both candidates’ immediate campaign funding and the funding of well-financed political action committees (super PACs). 

1. Funding of campaigns:

• Contributes to candidate and issue visibility

• Can be a relatively low cost way of providing interest-group-centered information to potential voters to motivate them to participate in the political process. 

• Allows withholding campaign funding or making implied threats to fund an opposition candidate also are effective in getting a politician’s attention and support for a particular policy position. Groups that attempt to influence policy in their favor through the use of campaign contributions are said to be carrying on rent-seeking activity because the group is committing resources to the pursuit of benefits from protection:

  • Note that the group would not rationally expend resources in excess of the expected benefits that it would receive from the policies in question. 

  • Rent-seeking activity can, of course, extend beyond simple campaign contributions to the use of corruptive practices such as bribes to political decision makers to influence their votes. Because the resources used in this type of activity are not producing any good or service but are merely influencing the distribution of income, these actions are often referred to as directly unproductive activity. 

  • Rent-seeking activity can be further complicated when an interest group tacitly agrees to support continued protection in other sectors, even if it means a loss in welfare for its own members, in exchange for support for one’s own protection:

    • The idea here is that the combined support of several groups for protection provides a sufficient political critical mass to get the protectionist candidates elected and the trade restrictions maintained. 

    • For example, textile and apparel workers (along with sector-specific capital owners in textiles and apparel) might well support the status quo on sugar and steel protection in exchange for the sugar and steel sectors’ support of protection in the textile and apparel industry. 

    • In this instance, the loss to their membership through higher prices for sugar and steel-using products might well be small compared with the gains obtained through the continued protection of textiles and apparel. 

    • This is another example, often called “logrolling,” of status quo bias in which a group (or several groups) benefits at the expense of society as a whole. 

• The median voter has again been supplanted in terms of the policy action undertaken because of a large uninvolved free-riding majority and the efforts of an interest-centered pressure group. Before leaving this discussion, it is important to note that while the self-interest approach has proved quite useful in better understanding the trade policy issue, it ignores the fact that people do things that do not appear to be in their pure economic self-interest. 

• No one would dispute the fact that, although it may be relatively small, individuals have often demonstrated the willingness to sacrifice some of their real income to improve overall welfare, whether it be in their community, their country, or even the world. It is for this reason that we turn to other approaches to political economy.

The Social Objectives Approach

From this perspective, trade policy is conducted taking into account the well-being of different groups in society along with various national and international objectives. In this environment, trade policy is promoted to the public at large in terms of broader social goals such as:

• Income distribution

• Increased productivity

• Economic growth

• National defence

• Global power and leadership

• International equity

With regard to domestic income distribution, Corden (1974, p. 107) suggests that trade policy appears to have a conservative bias in that governments often seem to give more weight to avoiding real income losses in a particular segment of the economy, and assign less weight to increasing the real income of a particular group.

Other social objectives which have been discussed in the literature involve:

• Minimising consumer loss

• Improving the real income of the lowest-income groups

• Minimising or delaying adjustment costs for particular industries

• Protecting the relative income level of specific socioeconomic groups

Politics puts the Squeeze on Tomato Exports

Actions by the Clinton administration in the 1990s to negotiate a floor price on tomato imports from Mexico constitute an excellent example of how the political process can result in a protectionist measure that circumvents the median-voter principle and rewards a small vocal minority (a small number of large Floridian growers) at the expense of the large mass of unorganised free-riding tomato consumers:

• Led by producer Paul J. DiMare, the Florida tomato producers, who are controlled by a handful of wealthy growers, had argued for years for protection from cheap foreign imports that “are driving Florida farmers out of business”.

• Interestingly, however, the Florida tomato industry has not collapsed even though hurricanes, cold snaps, and other weather-related woes have hindered production. 

• Of concern to the growers was the increase in imports from Mexico, which they feared would become even larger as the North American Free Trade Agreement (NAFTA, discussed in Chapter 17) accords came into place.

• Even though the tariff on Mexican tomatoes was low (1.4 cents/pound) prior to NAFTA, DiMare and others argued that the resulting increase in Mexican tomato imports endangered thousands of U.S. jobs and that without help the Florida tomato industry was “going down the tubes.” The Florida growers got little sympathy from their counterparts in California, whose summer crop competes directly with Mexico’s crop. 

• According to Ken Moonie, vice president of the tomato operations of California-based Calgene Inc., “All this is about protecting four big guys … it is not like corn or any other agricultural commodity where thousands of growers are involved.” 

• Imports of tomatoes from Mexico have cut into Florida’s share of the winter market. Some of this market penetration reflects the fact that the taste of the hand-picked, vine-ripened Mexican tomatoes is thought to surpass that of the Florida tomatoes, which are picked green and then gas-ripened prior to shipping. 

• DiMare did not see that consumer preference for the juicier, more tasty vine-ripened fruit had anything to do with the increase in imports, allowing that “it really doesn’t matter” how tomatoes taste since they are condiments, seldom eaten alone. 

• Under the accord reached in October 1996, Mexican growers agreed not to sell tomatoes in the United States below a certain price. Antidumping duties were put in place on November 1, 1996, to ensure that prices would not go below the level. 

• The U.S. Department of Commerce stated that this price would be the lowest average price during a recent period when there was a clear absence of any “price suppression” on the part of Mexican producers. 

• This accord was negotiated by Commerce Secretary Mickey Kantor and the Clinton administration because Florida was considered essential in the November 1996 elections. One Clinton strategist worried not so much about losing the small number of votes of tomato growers, but rather that negative publicity could result if no action were taken against the tomato imports. In later years, the antidumping duties and investigations were suspended. 

Sources: Helene Cooper and Bruce Ingersoll, “Playing Catch-Up: With Little Evidence, Florida Growers Blame Tomato Woes on NAFTA,” The Wall Street Journal, April 3, 1996, p. A1; Robert S. Greenberger, “Mexico Agrees to Temporary Floor on Price of Tomatoes Sold in U.S.,” The Wall Street Journal, October 14, 1996, p. B3; George Anthan, “Politics Put Squeeze on Tomato Imports; U.S. Growers Prevail,” The Des Moines Register, October 29, 1996, p. 3A; United States International Trade Commission, “Antidumping and Countervailing Duty Orders in Place as of September 10, 2008, by Country,” obtained from www.usitc.gov.

Such a macro approach does, however, create problems. If a country talks a “free-trade talk” and then proceeds not to “walk the free-trade walk” by protecting certain import competing sectors for reasons such as those given earlier, it quickly loses credibility with the voters. 

Once pressure groups learn that the government is concerned about income distribution, the verbalised commitment to free trade and structural adjustment is compromised and structural adjustment by both labor and capital slows. The expectation of trade policy relief thus:

• Reduces the outward movement of factors from any given declining industry, slowing down the necessary structural change

• Economic conditions in the industry continue to worsen, loss in overall economic efficiency

• Greater and greater pressures are put on the government to intervene to maintain relative income levels and the status quo, little or no change in equity

The inability to commit to a free-trade policy hence forces the country whose trade policy is influenced by income distribution concerns to maintain protection. It is often argued that this type of analysis is useful in explaining the high level of protection offered to textiles and apparel over time. 

Deardorff (1987) also suggests that such concerns help explain why governments preferred voluntary export restraints (VERs) to tariffs. 

Finally, Baldwin (1989, p. 129) argues that income distribution concerns can also help explain why governments use protection instead of domestic subsidies to help import-competing firms. 

With a tariff, quota, or VER, consumers who use the product essentially bear the burden of the policy through higher prices: 

• Inasmuch as they were the group that initially benefited from the import-induced lower prices

• An increase in domestic price that returns them more-or-less to the initial higher price leaves them no worse off than they were previously—even if the tariff revenue is not returned to them. 

• Were a production subsidy used to support the affected industry, the burden of the tax (to pay for the subsidy) would presumably fall on all taxpayers, thus reducing the relative income of those who do not consume the product and were not affected by the lower product price. 

Foreign trade policy has also been used as part of the total foreign policy package. As such, foreign policy concerns have been used to support both increased protection and increased trade liberalisation. 

Since World War II, the United States has been the major hegemonic power in the world. During much of that time, U.S. foreign policy was directed toward limiting the spread of communism and strengthening the noncommunist world economically. U.S. trade and aid policy was clearly influenced by these concerns, and they perhaps explain why the United States chose not to use its hegemonic power to improve its international terms of trade. In more contemporary times, U.S. sanctions on trade with Iran and the controversy surrounding their removal have been foreign-policy-related (seeking to prevent Iran from becoming nuclear armed) rather than concerned with economic impacts on the United States. 

In addition to hegemonic concerns, there is also evidence that a number of countries have concerns about the international distribution of income that go beyond their own self-interest. Examples of this include:

• The amount of untied foreign aid (i.e., aid that can be spent on goods from any country, not just from the donor country) that has been given to the developing countries.

• The reduced trade restrictions that have been granted both unilaterally and multilaterally through such programs as the Generalised System of Preferences (GSP). Analysis of this type has been relatively common among political scientists in recent years.

A Review of U.S. Trade Policy and Multilateral Negotiations

The liberalisation of trade can be divided into several stages:

1.  Beginning after the Tariff Act of 1930, usually called the Smoot-Hawley Tariff, which established extremely high protection in the United States (an average tariff level of about 50%). 

2. Other countries retaliated with their own stiff tariffs, and world trade shrank dramatically. An estimate by Jakob Madsen (2001, p. 865) is that:

•  41% of the decline in world trade was due to the imposition of the new trade barriers

• 59% was due to falling incomes that resulted in less purchasing power to buy imports.

• Recognising that there is imprecision in such estimates and that tariffs and income are interrelated (higher tariffs can reduce national income, and falling national income can increase the calls for more protection from imports), Madsen concluded (p. 867) that “the contribution of the imposition of the discretionary trade barriers was about as important as the output decline in explaining the contraction in world trade.

3. In any event, economists and policymakers generally agree that this tariff worsened the Great Depression of the 1930s. 

4. The tariff legislation since that time has occurred largely in response to this impact on the Great Depression and the increasing economic interdependence of countries. 

We now give a brief overview of trade agreements from 1934 to 1960, followed by examination of the Kennedy Round (1960s) and the Tokyo Round (1970s) of trade negotiations. We next discuss the most recent round that has been completed (1994), the Uruguay Round. We then look at prospects for the current round of trade negotiations, the Doha Development Agenda, and review several recent trade policy actions.

Reciprocal Trade Agreements and Early GATT Rounds

The long process of tariff reduction began with the passage by Congress of the Reciprocal Trade Agreements Act of 1934. This act authorised the executive branch to engage in bilateral negotiations with individual trading partners on tariff reductions:

• The act was renewed every three years until the end of World War II. 

• A particular feature of the negotiation process was its employment of an item-by-item approach, meaning that rate reductions on goods were bargained individually rather than uniformly agreed upon for a broad range of categories. While significant reductions were made on many goods and on the overall U.S. tariff level, the item-by-item approach does not permit smooth and quick negotiation when many goods are encompassed by the proceedings. 

• At the end of World War II, tariff bargaining took the form of multilateral negotiations, meaning that many countries took part simultaneously. The General Agreement on Tariffs and Trade (GATT) took effect in 1947. Under this agreement, countries committed themselves to multilateral bargaining for the purpose of easing trade restrictions in all of the participating countries. GATT became an ongoing organization that sponsored regular negotiations of this type. 

• From 1947 until 1962, five GATT rounds of trade negotiations were held, in which the participating countries hammered out mutually acceptable reductions of various barriers. The first round, held in Geneva in 1947, was reasonably successful. However, economists did not judge the next four rounds in 1949, 1951, 1956, and 1962, as having attained much success. 

• Nevertheless, some multilateral tariff reductions were achieved, and all these early rounds embodied the most-favoured-nation (MFN) principle (discussed in Chapter 13), as did those that followed.

The Kennedy Round of Trade Negotiations

To put new life into the trade negotiation process and to avoid being shut out by the newly forming European Economic Community, the United States led the way into a new round of negotiations from 1962 to 1967. 

The key stimulus for the round was the U.S. Trade Expansion Act of 1962:

• This legislation authorised the president to negotiate tariff reductions of up to 50%, and these reductions could be negotiated through an across-the-board approach rather than an item-by-item approach. Broad categories of goods could be discussed all at once and a given rate reduction could apply to the whole group—a more streamlined approach. 

The Trade Expansion Act also introduced a feature of trade policy known as trade adjustment assistance (TAA):

• Trade adjustment assistance: if a tariff reduction injures workers or industries by causing an inflow of imports, displaced workers can, for example, petition for additional unemployment compensation or for help in retraining for other types of jobs. To most economists, this was a marked step forward for public policy, because previously the only alternative considered was to reimpose the tariff (the “escape clause”):

• Thus, TAA in a broad sense tries to promote internal adjustment to changing international conditions and reduction in protection. It is an attempt to facilitate the movement of the economy along the production-possibilities frontier (PPF). 

• Without this assistance, a country’s production point might initially move inside the PPF and get back onto the frontier only after a considerable period of time has passed. 

The Trade Adjustment Assistance Reform Act of 2002 created an additional program for a health coverage tax credit and added a new benefit for older workers:

• Workers who lost their jobs because their companies shifted production abroad to countries that were part of a free trade agreement with the United States, or to some African and Western Hemisphere countries, were also made eligible for TAA.

With the passage of the Trade Expansion Act, the United States moved into multilateral negotiations in Geneva in what became known as the Kennedy Round of trade negotiations:

• Seventy countries participated

• Tariffs on manufactured products were reduced by an average of 35%, with at least some reduction occurring in 64% of manufactured goods with tariffs. (See Ellsworth and Leith, 1984, p. 230.) 

• Note that this success was in terms of cuts achieved in manufactured goods

• Kennedy Round achieved little progress in reducing barriers on agricultural products.

• Kennedy Round did little to ease non-tariff barriers (NTBs).

The Tokyo Round of Trade Negotiations

With the completion of the Kennedy Round of tariff cutting in 1967, no further steps were taken until 1973, when preliminary moves were initiated at a multilateral meeting in Tokyo toward beginning another round of negotiations. A main impetus for the new round was that, while tariff rates had been moving downward, non-tariff barriers (NTBs) had been rising and had offset some of the benefits of the tariff reductions. 

The Trade Act of 1974 enabled the United States to participate in this new round, the Tokyo Round of trade negotiations. This act of Congress authorised the president to enter into trade negotiations for the purpose of reducing tariff and nontariff barriers:

• Reductions of up to 60% were authorised on existing duties greater than 5%, and tariffs could be eliminated altogether on goods whose existing duties were less than 5%.

• In addition, authorisation was granted to enter into individual-sector negotiations to work for the liberalization of NTBs. 

• Other features of interest in this bill were the provision for introduction of the U.S. Generalised System of Preferences (GSP) for the products of developing countries (see Chapter 13) and the provision for more generous treatment of claims for TAA. 

• Finally, the Trade Act of 1974 attempted to systematise procedures on claims for import relief by import-competing firms. Prominent provisions of the act in this respect (some of which predate 1974) were as follows: 

1. Section 201: This part of U.S. legislation permits import-competing firms to petition for relief from rapidly increasing imports (or “surges” in imports). The U.S. page 369 International Trade Commission (USITC) then investigates whether the rapid increases in imports are causing “substantial injury” to the U.S. industry and, if so, makes a recommendation for protection to the president. The president may or may not accept the recommendation.

2. Section 232: The president is authorised to restrict any good that “is being imported into the United States in such quantities or under such circumstances as to threaten to impair the national security.” This provision has seldom been used (principally on petroleum in earlier years). In 1986, it was used to limit machine tool imports, which resulted in VERs with Japan and Taiwan.

3. Section 301: This unfair-trade portion of U.S. law permits the president to take retaliatory action in response to unjustifiable, unreasonable, or discriminatory restrictions on U.S. exports by foreign countries:

• Unjustifiable refers to any action that violates the international legal rights of the United States

• Unreasonable refers to unfair and inequitable practices, although these are obviously hard to define

• Discriminatory means actions that deny national or most-favoured nation (MFN) treatment to the United States.

The range of U.S. possible actions is broad, and includes (among others):

• Suspending trade agreement concessions

• Imposing duties or other import restrictions

• Entering into agreements with the other country to eliminate the behaviour or to provide compensation to the United States. 

In addition to this Section 301 provision, there is a Special 301 provision of U.S. trade law whereby the U.S. Trade Representative (USTR), a cabinet-level officer in the executive branch, identifies countries that are denying adequate or effective intellectual property rights to U.S. industries or persons. Such identification can lead to a designation of the offending countries as “Priority Foreign Countries” and be followed by USTR initiation of Section 301 action.

4. Section 701: This covers subsidies to exports by foreign countries. Petition is made by import-competing firms to the Department of Commerce to ascertain the existence of the subsidy; if affirmed, the USITC determines whether or not injury is occurring. 

5. Section 731: This portion of trade law consists of the antidumping provisions discussed in Chapter 15, pages 335–38. 

With this enabling legislation in hand, U.S. negotiators took part in the multilateral negotiations in Geneva between 1974 and 1979, which resulted in agreement (see Allen, 1979) that:

1. Tariff rates on manufactured goods were to be reduced by an average of about one-third in a phase-in process over eight years.

2. New codes of behaviour concerning several non-tariff barriers (NTBs) were to be adhered to with respect to:

• Government procurement procedures

• Subsidies and countervailing duties

• Valuation of goods for customs duties purposes.

• Tariff preferences should be given by developed countries to various manufactured exports from the developing countries

• That the nonreciprocity principle should apply to developing countries. 

Nonreciprocity principle: even though developed countries may reduce barriers on exports from developing countries, no corresponding behaviour is required by developing countries on exports to them from developed countries. 

By embracing tariff preferences and nonreciprocity, the world community basically affirmed the value judgment that the plight of the developing countries in the international economy and the relatively low levels of their per capita income required special, discriminatory measures in favour of these countries.

Trade Adjustment Assistance and Its Implementation

Although controversial, trade adjustment assistance (TAA) has been supported politically for more than 50 years since its inclusion in the Trade Expansion Act of 1962 and the Trade Act of 1974. 

More recently, it was reauthorised and expanded within The Trade and Global Adjustment Act of 2009 as part of President Obama’s Economic Stimulus package. 

• This authorisation lapsed in February 2011, but in October was restored through 2014 with the Trade Adjustment Act of 2011. 

After considerable political controversy, centred on program costs and effectiveness, the Congress passed the Trade Adjustment Act of 2015:

• • It amends the Trade Act of 1974 to extend through December 31, 2020

  • Renews the 2009–2010 eligibility and benefit levels

  • Authorises funding throughout the period

It thus represents long-term support for this program in contrast to the short term focus of previous legislation.

The longer-term focus is designed to:

• Improve the administration of the program

• Reduce worker uncertainty about being able to financially participate in the program

• Provide a more effective underpinning for all participants—firms, workers, farmers, and communities. 

Eligibility requirements to participate in this assistance program have always been somewhat of an issue. 

It may have been unknown which economic situations seemed to be most likely to result in the applicants being accepted into the trade adjustment program because of expanding international trade? 

Christopher Magee (2001) investigated empirically the determinants of any given worker’s chances of being certified by the U.S. Department of Labor to receive trade adjustment assistance (TAA):

• The actual statutory procedure for seeking TAA is that a group of three or more workers from a firm first files a petition with the Department of Labor. 

• Then the Department of Labor ascertains whether:

  •  A significant percentage of the workers in the petitioners’ firm has become partly or wholly separated from the firm or is under a potential threat of being separated

  • Whether sales or production (or both) of the firm have declined in absolute terms

  • Whether increases in imports of the petitioners’ firm’s product contributed in an important way to the separation or threat of separation. 

• If the petition is approved, workers received unemployment compensation for 12 months (rather than for the 6 months in normal unemployment situations), or 18 months if the workers are enrolled in a retraining program.

• Between 1975 and 1994, the percentage of applicants who were certified annually ranged from more than 70% to 10%, depending on the year. 

• Magee’s empirical tests attempted to ascertain the actual variables over the 1975–1992 period that seemed to yield a good chance for workers to be certified to receive TAA:

  • His first result was that a higher level of tariff protection for the industry in which the workers have been engaged yields a greater chance of approval of petitions. His reasoning for this finding is that:

    • Industries with higher tariffs are causing more of a production distortion than are industries with low tariffs. With greater protection and distortion, therefore, there is an incentive for officials to look more favourably upon providing TAA because the TAA might facilitate the removal of the distortion. This is an efficiency-type argument for TAA. 

    • Another way that Magee states this reasoning is that more TAA must be given to get tariff reductions in the industries with high tariffs; that is, “higher tariff industries require a greater payoff in order to reduce protection” (p. 119). 

    • Empirically, and as expected, there was a negative relationship between TAA certification and the tariff change in an industry. In other words, a greater tariff reduction (a more negative tariff change) was associated with a greater positive chance of receiving TAA. This is common sense because the greater tariff change will, other things equal, lead to more workers being displaced. 

    • All in all, this paper indicated that separated workers likely to receive TAA come from:

      • High-tariff industries

      • Industries in which tariff reductions have been great

      • Industries with high unemployment rates 

• Also, workers are more likely to receive certification:

• If they belong to a union

• If they are laid off from a low-wage industry

• If the industry’s market has a large import component.

Source: Congressional Research Service, “Library of Congress Summary,” in “Summaries for the Trade Adjustment Assistance Act of 2015,” obtained from www.govtrack.us/ Christopher Magee, “Administered Protection for Workers: An Analysis of the Trade Adjustment Assistance Program,” Journal of International Economics 53, no. 1 (February 2001), pp. 105–25.

After the implementation of the Tokyo Round tariff cuts, it was estimated that the average tariff level on manufactured products had been reduced to: 

• United States 4.3%

• Canada 5.2%

• France 6.0%

• Japan 2.9%

• United Kingdom 5.2%

• West Germany 6.3%

Except for Japan, all of these rates were slightly higher than the rate in the United States, but the differences were minor. The low Japanese rate points out one difficulty with these calculations, because they do not adequately include nontariff barriers. 

Nevertheless, the broad level of tariff rates was substantially lower than the levels existing at the time of Smoot-Hawley.

The Uruguay Round of Trade Negotiations

The First Four Years, 1986-1990

Despite the presence of relatively low tariff rates in the industrialized world, a new round of trade negotiations began in September 1986 and was to be completed by December 1990. These talks were initiated in Punta del Este, Uruguay, and became known as the Uruguay Round of trade negotiations. Major objectives of this new round included a continuation of the attempt to reduce NTBs, an enlargement of the negotiations to embrace trade in services in addition to the traditional emphasis on trade in goods, and a determination to deal with restrictions on agricultural trade. The Uruguay Round set forth an ambitious agenda. Members established 15 groups to work on reducing restrictions in the following areas: (1) tariffs, (2) NTBs, (3) tropical products, (4) natural resource–based products, (5) textiles and clothing, (6) agriculture, (7) safeguards (against sudden “surges” in imports), (8) subsidies and countervailing duties, (9) trade-related intellectual property restrictions, (10) trade-related investment restrictions, and (11) services, as well as four other areas dealing with GATT itself (e.g., dispute-settlement procedures and implementation of the NTB codes of the Tokyo Round, especially on dumping).12 The biggest controversies in the negotiations involved services, intellectual property, and agriculture. The most heated controversy by far concerned agriculture. Most countries use a wide variety of policies to assist the agriculture sector: price supports, direct production subsidies, export subsidies, quotas on imports, acreage restrictions to raise commodity prices, and others. These interventions lead away from the free-market and free-trade allocation of resources, since they introduce price distortions. The United States initially proposed a 10-year phaseout of all subsidies that affect agricultural trade and of all agricultural import barriers. The proposal was similar to one made by the Cairns Group, a collection of 14 developing and developed countries with agricultural interests (e.g., Argentina, Australia, Brazil, Canada, New Zealand). The European Community (EC) wanted to go more slowly and to moderate the extent of reduction in agriculture support.13 By 1990, the wide disparity in subsequent proposals overshadowed all other aspects of the negotiations, and the four-year effort had seemingly ended with no signed agreement on the liberalization of trade.

Continued Negotiations Lead to Success, 1993

Despite the failure of the Uruguay Round talks to meet the originally scheduled completion deadline of December 1990, negotiations continued. President George H.W. Bush requested and received from Congress in 1991 a two-year extension to continue the talks under the fast-track procedure, also known as the Trade Promotion Authority. Under this procedure, which has characterized past trade negotiations, Congress must simply vote yea or nay on a negotiated agreement. No amendments can be made. The debate over fast-track was heated because the authorization also applied to the negotiations for the North American Free Trade Agreement (NAFTA), which is discussed in Chapter 17. Hope of success for the Uruguay Round began to reappear at the end of 1992 owing to a trade policy threat. The United States had been concerned about a EC agricultural support program that harmed U.S. exports of oilseeds (e.g., soybeans) and had twice received backing from GATT that the EC program should be modified. In retaliation for the EC subsidy, the United States threatened to impose 200 percent tariffs on EC exports to the United States, valued at $300 million; if the EC in turn retaliated against this tariff, the United States was ready to impose a second round of tariffs on $700 million of manufactured exports from the EC to the United States.14 With this stimulus to renewed negotiations on agriculture, an accord was eventually reached by which the oilseeds export subsidies were to be reduced 36 percent by value and 21 percent by quantity over a six-year period. This positive development then set off activity to work again on many other aspects of the Uruguay Round. Finally, after intense discussions, the 117 participating countries in the Uruguay Round reached agreement on December 15, 1993 (the deadline date), and the signing took place on April 15, 1994, in Marrakech, Morocco. After ratification by participating countries, the agreement took effect on January 1, 1995.

Provisions of the Uruguay Round Agreement

Only a brief look is provided here of the features of this broad agreement.15 First, tariffs on average were cut by 34 percent (39 percent by developed countries) and were dropped altogether in the developed countries on a variety of products such as pharmaceuticals, construction and agricultural equipment, furniture, paper, and scientific instruments. Second, the value of agricultural export subsidies was to be cut by 36 percent and most domestic support for agriculture by 20 percent, and average developed-country agricultural tariffs would be reduced by 36 percent over a six-year period. Third, textiles and apparel trade were to be moved from the existing quota framework of the Multi-Fiber Arrangement into the GATT framework, with tariffs to be phased out over 10 years. In fact, the quotas were ended on January 1, 2005, but tariffs remain. In addition, Japan and South Korea promised to open their markets to some extent for rice imports. Revised rules were also adopted regarding dumping and export subsidies, and VERs were to be eliminated. Further, action on trade-related intellectual property rights (TRIPs) provided for minimum standards for trademarks, patents, and copyrights. (For example, patents are now in force for 20 years. In the United States, the previous length had been 17 years.) Some trade-related investment measures (TRIMs), such as local content requirements for foreign investors, were to be eliminated within two years by developed countries, five years by developing countries, and seven years by “least developed” countries. page 373 In retrospect, there is also a widely shared view pertaining to the Uruguay Round that the results were relatively unbalanced and were detrimental to the developing countries. One reason for this judgment was that the TRIPs features of the agreement hurt the developing countries through the payments that were expected to be made by them in the future to the developed countries for the use of intellectual property. A second reason was that the phase-out of the Multi-Fiber Arrangement and its replacement by tariffs caused the developing countries to lose the quota rents that they previously had been receiving. In addition, the developed countries’ changes with respect to agriculture seemed very minor and were not of much benefit to the developing countries (see Finger, 2008). Considerable friction prevented attainment of some of the goals for services. An important controversy involved France’s refusal to permit the import of U.S. motion pictures on the scale the United States wanted. Nevertheless, a general agreement on trade in services (GATS) calls for “national treatment” in services, meaning that any country is to treat foreign service providers in the same fashion as domestic service providers, as well as for MFN treatment for services. New procedures were also adopted for the settlement of disputes. Finally, GATT itself was replaced by a new organization, the World Trade Organization (WTO), which supervised the implementation of the Uruguay Round agreement and is handling trade disagreements.

Trade Policy Issue after the Uruguay Round

With the successful conclusion of the Uruguay Round, governments in the next several years centered on the implementation of the measures that had been adopted. In addition, further sectoral agreements were pursued (such as completed agreements in telecommunications and financial services) that would put specifics into the broad general framework that had been laid out in the Uruguay Round for particular sectors. However, as the 1990s drew to a close, there emerged a desire on the part of many countries to begin a new round of multilateral trade deliberations. Many countries wished to attain further relaxation of trade-restricting measures in agriculture and services, to reduce remaining tariffs further, and to consider a variety of other matters pertaining to areas such as antidumping procedures, procedures within WTO, and intellectual property rights. Further, there was a desire by developed countries—but decidedly not by developing countries—to discuss the broad general area known as “labor standards.” In addition, and again mainly by developed countries, there was pressure to include consideration of the environmental impact of trade and international shifts in production location by firms. The labor standards debate revolves around several issues—most importantly child labor, health and safety features of the workplace, and hours of work per day and days of work per week. For example, campus groups in the industrial countries have protested the use of child labor (often children in the lower-teen years or younger) in assembly plants in developing countries. Out of humanitarian concern, these groups demand that their universities and colleges not deal with companies that employ such labor. Similar protests are made, including objections by labor union groups and their employers, in industrial countries regarding the dangerous work environments in developing countries, where little attention is paid to safety procedures, properly operated equipment, and sanitary conditions in the plants. Developed-country labor and employers in labor-intensive industries, of course, claim that it is “unfair” that they have to compete with goods made under such “sweatshop” conditions. The view is that, were developing-country firms forced (by threat of sanctions against their goods by developed countries) to provide a safe, healthful environment and to adhere to the length of workday and workweek adhered to by developed-country firms, the result would be that all firms are competing on an equal basis.

Tariff Reductions Resulting from the Uruguay Round

Given the broad guidelines of the Uruguay Round with respect to tariff reductions, it is useful to examine the actual differences between pre-Uruguay and post-Uruguay Round tariffs at both the country/region level and the commodity group level. WTO estimates of the pre- and post-Uruguay Round weighted-average tariff levels are found in Table 2. Note that developing countries have substantially higher protection (both pre- and post-Uruguay Round) than developed countries and that, among products, textiles and clothing have the highest weighted-average tariff rates.

In 2015, the world as a whole produced goods and services worth about $74 trillion at current prices, 30% being sold across national borders. World trade in goods and services exceeded $21 trillion.

Gravity Model: helps to make sense of the value of trade between any pair of countries and sheds light on the impediments that continue to limit international trade.

Who Trades with Whom?

Figure 2.1 shows the total value of trade in goods (exports + imports) between the US and its top 15 trading partners in 2015, taken together, these 15 countries accounted for 75% of the value of U.S. trade in that year.

Size Matters: The Gravity Model

The top 3 U.S. trading partners are Europe's three largest economies: Germany, The United Kingdom, France. Therefore, they have the highest values of GDP (measures the total value of all goods and services produced in an economy). There is a strong empirical relationship between the size of a country's economy and the volume of both its imports and its exports.

Figure 2.2 illustrates this relationship by showing the correspondence between the size of different European economies (America's 15 most important Western European trading partners in 2012):

• On the horizontal axis is each country's GDP, expressed as a percentage of the total GDP of the European Union

• On the vertical axis is each country's share of the total trade of the U.S. with the EU.

The scatter of points is clustered around the 45º line, each country's share of U.S. trade with Europe was roughly equal to that country's share of Western European GDP; it also accounts for 24% of U.S. trade with the region.  Sweden has a much smaller economy (3.2% of European GDP and 2.3% of U.S. - Europe trade).

Looking at world trade as a whole, economists have found an equation which predicts the volume of trade between any two countries fairly accurately, this is known as the Gravity Model of World Trade:

                        Tij = A x Yi x Yj/Dij (2.1)

where
A is a constant
Tij is the value of trade between country i and country j
Dij is the distance between the two countries

That is the value of trade between any two countries is proportional, other things equal, to the product of the two countries' GDP and diminishes with the distance between the two countries.

The reason for the name is the analogy to Newton's law of gravity. Just as gravitational attraction between two objects is proportional to the product of their masses and diminishes with distance, the trade between two countries is, other things equal, proportional to the product of their GDPs and diminishes with distance.

Broadly speaking, large economies tend to spend large amounts on imports because they have large incomes. They also tend to attract large shares of other countries' spending because they produce a wide range of products. So, other things equal, the trade between any two economies is larger, the larger is either economy.

In practice countries spend much more or most of their income at home. The U.S. and the European Union each account for about 25% of the world's GDP, but each attracts only about 2% of the other's spending. To make sense of actual trade flows, we need to consider the factors limiting international trade. 

Using the Gravity Model: Looking for Anomalies

It's clear from Figure 2.2 that a gravity model fits the data on U.S. trade with European countries pretty well, but not perfectly.

One of the principal uses of gravity models is that they help us to identify anomalies in trade. Indeed, when trade between two countries is either much more or much less than a gravity model predicts, economists search for the explanation.

We see that the Netherlands, Belgium, and Ireland trade considerably more with the U.S. than a gravity model predicts:

• For Ireland, the answer lies with cultural affinity. They share a language and tens of millions are descended from Irish immigrants. Ireland plays a special role as host to many U.S.-based corporations

• For Netherlands and Belgium, geography and transport costs probably explain it. Both countries are located near the mouth of the Rhine, Western Europe's largest river, which runs pasts the Ruhr, Germany's industrial heartland. Therefore making both the Netherlands and Belgium traditionally the point of entry to much of northwestern Europe. Rotterdam is the most important port in Europe (measured by tonnage handled), and Antwerp ranks second. So, an important role in the determining the volume of international trade is both transport costs and geography. 

Impediments to Trade: Distance, Barriers, and Borders

Figure 2-3 shows the same data as Figure 2-2 with U.S. trade as a percentage of total trade with Western Europe in 2012 versus GDP as a percentage of the region’s total GDP— but adds two more countries:

• Canada and Mexico. 

• As you can see, the two neighbours of the United States do a lot more trade with the United States than European economies of equal size. 

• In fact, Canada, whose economy is roughly the same size as Spain’s, trades as much with the United States as all of Europe does. Why does the United States do so much more trade with its North American neighbours than with its European partners? One main reason is the simple fact that Canada and Mexico are much closer.

• All estimated gravity models show a strong negative effect of distance on international trade; typical estimates say that a 1 percent increase in the distance between two countries is associated with a fall of 0.7 to 1 percent in the trade between those countries. This drop partly reflects increased costs of transporting goods and services. 

Economists also believe that less tangible factors play a crucial role. Trade tends to be intense when countries have close personal contact, and this contact tends to diminish when distances are large. 

For example, it’s easy for a U.S. sales representative to pay a quick visit to Toronto, but it’s a much bigger project for that representative to go to Paris. Unless the company is based on the West Coast, it’s an even bigger project to visit Tokyo.

In addition to being U.S. neighbors, Canada and Mexico are part of a trade agreement with the United States, the North American Free Trade Agreement, or NAFTA, which ensures that most goods shipped among the three countries are not subject to tariffs or other barriers to international trade. We’ll analyze the effects of barriers to international trade in Chapters 8 and 9, and the role of trade agreements such as NAFTA in Chapter 10. For now, let’s notice that economists use gravity models as a way of assessing the impact of trade agreements on actual international trade: If a trade agreement is effective, it should lead to significantly more trade among its partners than one would otherwise predict given their GDPs and distances from one another. It’s important to note, however, that although trade agreements often end all formal barriers to trade between countries, they rarely make national borders irrelevant. Even when most goods and services shipped across a national border pay no tariffs and face few legal restrictions, there is much more trade between regions of the same country than between equivalently situated regions in different countries. The Canadian–U.S. border is a case in point. The two countries are part of a free trade agreement (indeed, there was a Canadian–U.S. free trade agreement even before NAFTA); most Canadians speak English; and the citizens of either country are free to cross the border with a minimum of formalities. Yet data on the trade of individual Canadian provinces both with each other and with U.S. states show that, other things equal, there is much more trade between provinces than between provinces and U.S. states. Table 2-1 illustrates the extent of the difference. It shows the total trade (exports plus imports) of the Canadian province of British Columbia, just north of the state of Washington, with other Canadian provinces and with U.S. states, measured as a percentage of each province or state’s GDP. Figure 2-4 shows the location of these provinces and states. Each Canadian province is paired with a U.S. state that is roughly the same distance from British Columbia: Washington State and Alberta both border British Columbia; Ontario and Ohio are both in the Midwest; and so on. With the exception of trade with the far eastern Canadian province of New Brunswick, intraCanadian trade drops off steadily with distance. But in each case, the trade between British Columbia and a Canadian province is much larger than trade with an equally distant U.S. state. Economists have used data like those shown in Table 2-1, together with estimates of the effect of distance in gravity models, to calculate that the Canadian–U.S. border, although it is one of the most open borders in the world, has as much effect in deterring trade as if the countries were between 1,500 and 2,500 miles apart. Why do borders have such a large negative effect on trade? That is a topic of ongoing research. Chapter 21 describes one recent focus of that research: an effort to determine how much effect the existence of separate national currencies has on international trade in goods and services.

The Changing Pattern of World Trade

World trade is a moving target. The direction and composition of world trade is quite different today from what it was a generation ago and even more different from what it was a century ago. Let’s look at some of the main trends.

Has the World Gotten Smaller? 

In popular discussions of the world economy, one often encounters statements that modern transportation and communications have abolished distance, so that the world has become a small place. There’s clearly some truth to these statements: The Internet makes instant and almost free communication possible between people thousands of miles apart, while jet transport allows quick physical access to all parts of the globe. On the other hand, gravity models continue to show a strong negative relationship between distance and international trade. But have such effects grown weaker over time? Has the progress of transportation and communication made the world smaller? The answer is yes—but history also shows that political forces can outweigh the effects of technology. The world got smaller between 1840 and 1914, but it got bigger again for much of the 20th century. Economic historians tell us that a global economy, with strong economic linkages between even distant nations, is not new. In fact, there have been two great waves of globalization with the first wave relying not on jets and the Internet but on railroads, steamships, and the telegraph. In 1919, the great economist John Maynard Keynes described the results of that surge of globalization: What an extraordinary episode in the economic progress of man that age was which came to an end in August 1914! . . . The inhabitant of London could order by telephone, sipping his morning tea in bed, the various products of the whole earth, in such quantity as he might see fit, and reasonably expect their early delivery upon his doorstep. Notice, however, Keynes’s statement that the age “came to an end” in 1914. In fact, two subsequent world wars, the Great Depression of the 1930s and widespread protectionism, did a great deal to depress world trade. Figure 2-5 shows one measure of international trade: the ratio of an index of world exports of manufactured goods to an index of world industrial production. World trade grew rapidly in the decades leading up to World War I but then fell significantly. As you can see, by this measure globalization didn’t return to pre-World-War-I levels until the early 1970s. Since then, however, world trade as a share of world production has risen to unprecedented heights. Much of this rise in the value of world trade reflects the so-called “vertical disintegration” of production: Before a product reaches the hands of consumers, it often goes through many production stages in different countries. For example, consumer electronic products—cell phones, iPods, and so on—are often assembled in low-wage nations such as China from components produced in higher-wage nations like Japan. Because of the extensive cross-shipping of components, a $100 product can give rise to $200 or $300 worth of international trade flows.

What Do We Trade? 

When countries trade, what do they trade? For the world as a whole, the main answer is that they ship manufactured goods such as automobiles, computers, and clothing to each other. However, trade in mineral products—a category that includes everything from copper ore to coal, but whose main component in the modern world is oil—remains an important part of world trade. Agricultural products such as wheat, soybeans, and cotton are another key piece of the picture, and services of various kinds play an important role and are widely expected to become more important in the future. Figure 2-6 shows the percentage breakdown of world exports in 2015. Manufactured goods of all kinds make up the lion’s share of world trade. Most of the value of mining goods consists of oil and other fuels. Trade in agricultural products, although crucial in feeding many countries, accounts for only a small fraction of the value of modern world trade. Meanwhile, service exports include traditional transportation fees charged by airlines and shipping companies, insurance fees received from foreigners, and spending by foreign tourists. In recent years, new types of service trade, made possible by modern telecommunications, have drawn a great deal of media attention.

The most famous example is the rise of overseas call and help centers: If you call an 800 number for information or technical help, the person on the other end of the line may well be in a remote country (the Indian city of Bangalore is a particularly popular location). So far, these exotic new forms of trade are still a relatively small part of the overall trade picture, but as explained below, that may change in the years ahead. The current picture, in which manufactured goods dominate world trade, is relatively new. In the past, primary products—agricultural and mining goods—played a much more important role in world trade. Table 2-2 shows the share of manufactured goods in the exports and imports of the United Kingdom and the United States in 1910 and 2015. In the early 20th century, Britain, while it overwhelmingly exported manufactured goods (manufactures), mainly imported primary products. Today, manufactured goods dominate both sides of its trade. Meanwhile, the United States has gone from a trade pattern in which primary products were more important than manufactured goods on both sides to one in which manufactured goods dominate. A more recent transformation has been the rise of third-world exports of manufactured goods. The terms third world and developing countries are applied to the world’s poorer nations, many of which were European colonies before World War II. As recently as the 1970s, these countries mainly exported primary products. Since then, however, they have moved rapidly into exports of manufactured goods. Figure 2-7 shows the shares of agricultural products and manufactured goods in developingcountry exports from 1960 to 2001. There has been an almost complete reversal of relative importance. For example, more than 90 percent of the exports of China, the largest developing economy and a rapidly growing force in world trade, consists of manufactured goods.

Service Offshoring 

One of the hottest disputes in international economics right now is whether modern information technology, which makes it possible to perform some economic functions at long range, will lead to a dramatic increase in new forms of international trade. We’ve already mentioned the example of call centers, where the person answering your request for information may be 8,000 miles away. Many other services can also be done in a remote location. When a service previously done within a country is shifted to a foreign location, the change is known as service offshoring (sometimes known as service outsourcing). In addition, producers must decide whether they should set up a foreign subsidiary to provide those services (and operate as a multinational firm) or outsource those services to another firm. In Chapter 8, we describe in more detail how firms make these important decisions.

In a famous Foreign Affairs article published in 2006, Alan Blinder, an economist at Princeton University, argued that 

“in the future, and to a great extent already in the present, the key distinction for international trade will no longer be between things that can be put in a box and things that cannot. It will, instead, be between services that can be delivered electronically over long distances with little or no degradation of quality, and those that cannot.” 

For example, the worker who restocks the shelves at your local grocery has to be on site, but the accountant who keeps the grocery’s books could be in another country, keeping in touch over the Internet. The nurse who takes your pulse has to be nearby, but the radiologist who reads your X-ray could receive the images electronically anywhere that has a high-speed connection. At this point, service outsourcing gets a great deal of attention precisely because it’s still fairly rare. The question is how big it might become, and how many workers who currently face no international competition might see that change in the future. One way economists have tried to answer this question is by looking at which services are traded at long distances within the United States. For example, many financial services are provided to the nation from New York, the country’s financial capital; much of the country’s software publishing takes place in Seattle, home of Microsoft; much of America’s (and the world’s) Internet search services are provided from the Googleplex in Mountain View, California, and so on. Figure 2-8 shows the results of one study that systematically used data on the location of industries within the United States to determine which services are and are not tradable at long distances. As the figure shows, the study concluded that about 60 percent of total U.S. employment consists of jobs that must be done close to the customer, making them nontradable. But the 40 percent of employment that is in tradable activities includes more service than manufacturing jobs. This suggests that the current dominance of world trade by manufactures, shown in Figure 2-6, may be only temporary. In the long run, trade in services, delivered electronically, may become the most important component of world trade. We discuss the implication of these trends for U.S. employment in Chapter 8.

Abstract

The empirical literature on the effect of trade agreements has reached a consensus that WTO effects are insignificant or modest at best especially compared to the robust and strong positive effect of preferential trade agreements (PTAs) on bilateral trade (Rose, 2004; Eicher and Henn, 2011; Baier and Bergstrand, 2007, 2009). 

In this paper, we show that previous work that reports “average treatment effects” on trade flows fails to account for the heterogeneity in the effects of trade arrangements over time, leading to an underestimation of the WTO effect on bilateral trade. 

We present semi-parametric estimates of the impact of WTO membership, allowing for heterogeneity of WTO effects over time. 

We find evidence for strong WTO effects over time, with the WTO effects increasing almost monotonically with years of membership.

In fact, in the long-term, the magnitude of the WTO effect exceeds those of the PTA effects.

We also find that the strongest WTO effect over time is for destinations that are developing countries that underwent rigorous accession procedures to join the WTO. 

Disaggregating PTAs into bilateral, multilateral, and deep integration (customs union, common markets, and economic union), we find that the long-term effects of WTO membership dominate bilateral and multilateral PTAs but fall short of deep integration arrangements.

Finally, we show that WTO membership increases both the extensive and intensive margins of trade over time, with a stronger impact on the former. 

Our results are consistent with gradual trade liberalization and declining trade policy uncertainty over time, following WTO membership.

Introduction

Shortly after his inauguration, US President Donald Trump withdrew from the Trans-Pacific Partnership (TPP), a colossal 12-nation free trade deal designed to boost trade ties between the US and Pacific Rim countries. 

The fate of the Transatlantic Trade and Investment Partnership (TTIP) designed to reduce trade barriers between the US and the European Union remains uncertain. Even NAFTA has been renegotiated. 

The US withdrawal from the TPP and the UK’s Brexit vote were greeted with dismay by heads of state, policymakers, and op-ed writers alike, who regretted the apparent reversal of trade liberalization and economic integration negotiated via bilateral and ever-larger regional trade agreements. 

In an atmosphere of rising hostility towards trade, notably in the US and the UK, many reassure themselves that multilateral trade liberalization via WTO remains an option.

However, with the Doha round of trade talks frozen and gridlocked, multilateral liberalization via the WTO is seen as a less preferred or perhaps less feasible option (Baldwin, 2016). 

The WTO has taken a shallow integration approach focusing mainly on market access for traded goods. At the same time, preferential trading agreements (PTAs) are increasingly complex (covering provisions on liberalising trade in services, copyright laws, patent protection, environmental regulations, labor standards, and dispute mechanisms), many of which have little to do with trade directly.

While PTAs are going beyond market access and formal trade barriers, the consensus among trade economists is that bilateral and regional trading arrangements dominate the WTO multilateral mechanism in terms of their effect on bilateral trade. 

More precisely, previous work has shown that preferential trading arrangements have had a stronger impact on trade than WTO membership since they afford deeper integration. 

• For instance, in a meta-analysis by Cipollina and Salvatici (2010) on the trade effects of PTAs report a mean effect of 0.59 and a median effect of 0.38 on logged bilateral trade over 1867 estimates. 

• Baier and Bergstrand (2007) and Anderson and Yotov (2016) use different specifications, OLS on panel data with fixed effects and Poisson pseudo-maximum likelihood (PPML) estimators respectively, to find that a PTA doubles bilateral trade after 10–15 years.

• In contrast, the empirical literature finds that:

  • WTO membership has had a weak or small impact on bilateral trade flows (Rose, 2004; Rose 2005), 

  • that the impact is mainly confined to developed countries (Subramanian and Wei, 2007), 

  • on the extensive partner margin of trade (Felbermayr and Kohler, 2006, 2011; Liu, 2009), 

  • and that some if not all of the WTO effect on trade can be attributed to PTAs (Eicher and Henn, 2011).

• Herz and Wagner (2011) combine many of the methodological contributions of previous studies and find a strong WTO impact but do not account for the multilateral trade resistance terms.

• Lawrence (1996) argues that this is not surprising since tariffs are only one of many trade barriers, and even if these are removed following WTO accession, the complexity of different regulatory policies among nations remains.

PTAs that go beyond the border barriers that the WTO purports to address, afford deeper integration via changes in intellectual protection, labor standards, and investment measures. These are also more suited to the rise of offshoring and global value chains (Baldwin, 2016). The belief in the greater efficacy of PTAs is reflected in the proliferation of PTAs. 

In 1991 there were six bilateral trading arrangements in Asia. Today there are 166 such arrangements with 60 more being negotiated, a veritable noodle bowl of trade arrangements. Worldwide there are nearly 300 PTAs prompting Bagwell et al. (2016) to ask whether the WTO is now passé and whether PTAs are a more reliable vehicle for deeper integration.

Headlines following the Brexit vote in 2016 reflected a similar understanding – that Brexit was likely to reduce both trade and welfare. A “hard Brexit,” whereby the United Kingdom leaves the Single Market and trades under WTO rules, is predicted to have a strong negative impact on value-added UK exports, amounting to almost 18% (Brakman et al., 2018). Similarly, a report by the H. M. (Treasury, 2016; Treasury, 2016) estimates that leaving the EU and reverting to WTO trade barriers will lead to an annual GDP loss equivalent to £5200 per household. Underlying these calculations are estimates of WTO and PTA effects within a gravity model. Not surprisingly, such predictions and counterfactuals hinge on the reliability of WTO and PTA estimates. This paper examines whether the pessimism about WTO effects on trade flows is indeed warranted. We revisit the magnitude of the impact of WTO on bilateral trade and subsequently, compare this impact to that of PTAs. Our approach contrasts with the traditional way researchers have estimated the effect of trade arrangements within a gravity framework – using dummy variables that take the value 1 if both countries in a country-pair are part of a particular trade arrangement (WTO, PTA, etc.) and 0 otherwise. This formulation conceptualizes trading arrangements as a simple binary treatment and calculates the average treatment effect of such trading arrangements. While previous work has recognized the heterogeneity of PTA and WTO effects, the focus has been on differential effects for different pairs of countries (advanced vs. emerging, early vs. late joiners of WTO; see Subramanian and Wei, 2007; Eicher and Henn, 2011), different types of WTO membership (formal vs. nonmember participants as in Tomz et al., 2007), rigor in accession procedures (Allee and Scalera, 2012), different types of trade arrangements (Kohl, 2014; Kohl et al., 2016; Baier et al., 2019), and different margins of trade (Liu, 2009; Dutt et al., 2013; Baier et al., 2014). Almost all this work essentially imposes homogeneity of impact of trading arrangements over time.3 However, such an approach fails to account for the heterogeneity in WTO treatment effects over time, that there may be diminishing vs. increasing effects over time for WTO membership as well as for different trading arrangements. Unlike previous work, we adopt a flexible semi-parametric approach to the effect of trading arrangements and allow these effects to vary over time. First, we measure the number of years a country-pair has been part of a particular trading arrangement (WTO, PTA) as a time-varying pair-specific variable. This captures a country-pair’s intensity of treatment with respect to a trading arrangement. Next, we create a set of dummy variables spanning all possible years of membership. For instance, with WTO membership spanning 65 years since the inception of GATT in 1948 (precursor to the WTO), we create 65 separate dummy variables to capture the effect of the WTO. We do the same for PTAs. Instead of a binary treatment, we estimate distinct coefficients on these dummies, allowing the effect of WTO and PTA to vary over time. This formulation eschews functional form assumptions and allows for fully flexible treatment effects over time. We find that while the effect of the WTO on bilateral exports is weak and even negative in initial years, it strengthens monotonically over time. In contrast, while the impact of PTAs is stronger initially, the effects stabilize and then diminish over time. Overall, WTO effects strengthen over time, eventually dominating PTA effects in terms of magnitude. These findings are consistent with two plausible mechanisms: First, the simple empirical regularity of gradual trade liberalization over multiple rounds of multilateral trade negotiations, especially for developing countries. Second, the findings are compatible with the decline in trade policy uncertainty and improvement in trade policy predictability that accompanies WTO membership (Handley, 2014; Handley and Limão, 2017). To the extent that rules governing PTAs are relatively easier to renegotiate (e.g., NAFTA) compared to the WTO, we would expect a more substantial decline in trade policy uncertainty over time for WTO membership. Such a decline, in turn, should manifest as a stronger role for WTO over time. With this new methodology, we examine the sensitivity of these findings along various dimensions of heterogeneity, by types of WTO membership, by levels of development, the rigor of accession procedures, and for extensive and intensive goods margin of trade. This facilitates comparison to the prior literature on the heterogeneity of WTO effects. Comparing our results with Tomz et al. (2007), we observe strong positive effects when both countries in a dyad are formal WTO members, negative effects when both are nonmember participants, and insignificant effects when one country in a dyad is a formal member, and the other is a nonmember participant. In contrast to Subramanian and Wei (2007), we find that the WTO effects over time are stronger when the destination country is a developing country. Third, aligned with Allee and Scalera (2012), we find the strongest WTO effects when the destination had to undertake rigorous procedures when acceding to the WTO. In contrast to Allee and Scalera (2012), we find a stronger effect on bilateral exports when the origin acceded to the WTO automatically. Unlike Dutt et al. (2013), we find a positive effect on the intensive goods margin. Like them, we find that WTO effect over time is stronger on the extensive goods margin. These results align with the two mechanisms noted above. First, developing countries upon joining GATT/WTO initially had far fewer obligations to liberalize tariff barriers but did so only gradually over time. This was especially the case for countries that eventually opted for formal membership and had to undergo rigorous accession procedures. Second, to the extent that there is a gradual decline in variable trade costs trade and/or policy uncertainty, we should observe the entry of new firms and new products, increasing the extensive margin of bilateral exports. Next, we distinguish between various types of PTAs estimating separate effects for bilateral arrangements (e.g., KORUS, the FTA between US and Korea), multilateral arrangements (e.g., ASEAN), and deep trade arrangements (e.g., common market such as the EU). We find that the WTO effects dominate both bilateral and multilateral arrangements in the long-run, but fall short of Deep PTAs. These findings are again consistent with the reduction in trade policy uncertainty that accompanies both Deep PTAs and WTO membership. Finally, we show that the WTO has a positive effect on bilateral trade whether or not a country-pair shares PTA membership. PTA effects, on the other hand, are strong only when the pair also shares WTO membership. The remainder of the paper is organized as follows. Section 2 replicates the traditional gravity parametric specification with binary dummies for various trading arrangements. We use both an OLS specification with a comprehensive set of dummies and the methodology of Santos Silva and Tenreyro (2006), which treats bilateral trade as a count variable and uses Poisson Pseudo-Maximum Likelihood (PPML) to estimate the coefficients; Section 3, the heart of the paper, presents arguments for time-varying heterogeneity for WTO membership, the semi-parametric specification to capture these effects and our baseline findings; Section 4 examines additional sources of heterogeneity; Section 5 concludes.

Trade Arrangements & The Gravity Equation

Baseline Specification

The gravity equation is the current workhorse for estimating the importance of trade costs and trading arrangements for bilateral trade. Several theoretical frameworks support the gravity specification (see Costinot and Rodriguez-Claire, 2014), with exports from country o (exporter/origin) to country d (importer/destination) at time t, denoted by Xod, t, given as

μo, t and μd, t are exporter and importer-year dummies that capture attributes of the exporting- and the importing-country, respectively, including size and multilateral trade resistance terms (Anderson and van Wincoop, 2003). Tod, t measures bilateral trade costs, with −θ as the elasticity of exports with respect to trade costs. Next, ln Tod, t is specified in terms of bilateral gravity variables, as shown below.

where zm od,t are M gravity variables and γ m denotes the parameters to be estimated. Head and Mayer (2014) perform a metaanalysis and identify as main variables WTO, PTA, and currency union (CU) dummies capturing pair-specific, time-varying integration effects, while distance, contiguity, shared language, and colonial links, measure time-invariant geographic, cultural, and historical barriers.4 We also include a dummy for Generalized System of Preferences (GSP) where advanced countries provide preferential market access to developing countries (see Hakobyan, 2019 who finds that the 2011 GSP expiration with the US as the provider of such one-way preferences had a detrimental impact on exports of developing countries to the US). Substituting (2) into (1) yields the following estimable specification:

Baier and Bergstrand (2007) emphasize the importance of accounting for unobserved heterogeneity among country-pairs when estimating the importance of PTAs. Country-pairs self-select into PTA membership, so PTA membership is not randomly assigned. Baier and Bergstrand (2007) argue that not accounting for such selection may lead to an underestimation of the PTA coefficient when two countries who face unmeasured trade barriers, that inhibit bilateral trade, are more likely to select into and form an FTA. They find that a panel-data approach with country-pair fixed effects accounts for the endogeneity of PTA formation and yields a strong role for PTAs in increasing bilateral trade (see also Kohl, 2014 who first-differences the data to account for dyadic effects). Others, such as Magee (2008) and Krugman (1991) proposed that country-pairs who are natural trading partners are more likely to form bilateral or regional agreements if they already have significant bilateral trade. This would tend to overestimate the effect of PTAs. In either case, omitted variables raise endogeneity concerns leading to biased estimates for the coefficient on the PTA dummy. Magee (2003), Baier and Bergstrand (2007) and Egger et al. (2011) all use instruments to account for selection into trading arrangements in cross-sectional data. However, these rely on assumptions about exclusion restrictions, that these instruments are uncorrelated with the gravity equation error, which may be difficult to justify. Baier and Bergstrand (2004, 2007) question the plausibility of these exclusion restrictions, highlight the instability of the estimated treatment effects and instead recommend using time-invariant country-pair dummies in a panel data setting. The use of asymmetric dyadic fixed effects eliminates the need to choose controls for Eqs. (3) and (4), thereby alleviating the criticism of Ghosh and Yamarik (2004) that the researcher’s prior beliefs influence the exact results presented and accounts for the fragility of estimated PTA treatment effects. While Ghosh and Yamarik (2004) focused on PTA treatment effects, their criticism extends to the impact of the WTO as well. Therefore, all our specifications include time-invariant country-pair dummies od in addition to exporter-year and importer-year dummies.5 These also absorb all time-invariant gravity variables such as distance, contiguity, shared language, and colonial links. Our baseline specification is as follows:

where βWTO = θγWTO, βPTA = θγPTA, βCU = θγCUand βGSP = θγGSP are the coefficients of interest. WTOod, t access is captured by a dummy variable that takes the value 1 if both trading partners are members of the GATT/WTO at time t and 0 otherwise.6 PTAod, t takes the value 1 if both trading partners are members in a PTA at time t and 0 otherwise. CUod, t is analogously defined for pairs that are part of the same currency union. GSPod, t takes the value 1 if the origin extends one-way trade preferences to the destination and 0 otherwise. od = do is a directional dyadic fixed effect. Helpman et al. (2008) and Haveman and Hummels (2004) highlight the prevalence of zero bilateral trade flows. For the bilateral data used here, 22% of all possible bilateral trade flows show a zero value. Unobserved trade costs can endogenously create zeros and taking logs removes them from the sample, creating sample selection bias. Firms also self-select into exports, but given heterogeneity in productivity, unobserved trade costs can lead firms to remain purely domestic, which Helpman et al. (2008) label as heterogeneity bias. Santos Silva and Tenreyro (2006) also show that the log-linear specification of the gravity model in the presence of heteroskedasticity leads to inconsistent estimates. If the error term in the standard log specification is heteroskedastic, which is well-established in trade data, its log is not orthogonal to the log of the regressors, leading to inconsistent estimates of the gravity elasticities. Following them as well as Liu (2009), we treat bilateral trade as a count variable and use the Poisson Pseudo-Maximum Likelihood (PPML) to estimate the following equation.

Since the dependent variable is trade, rather than the log of trade, it not only eliminates the heteroskedasticity bias but also allows us to include zeros in the trade matrix. We estimate Eq. (4) and (5) using data on bilateral trade flows and the bilateral gravity variables. For bilateral trade flows, our primary data source is the CEPII database from Head and Mayer (2014) that provides data on aggregate bilateral exports (fob) and imports (cif) from 205 exporters to 186 importers over the period 1948–2012. Our measure of bilateral trade is the log of cost-including-freight of imports reported by the destination from the origin. Data on WTO membership are from the CEPII and updated via the WTO website (www.wto.org). Data on PTA membership and GSP are from Jeffry Bergstrand’s website, available for the period 1950–2012. Data on currency unions are from Glick and Rose (2016). Our baseline PPML specification that accounts for zeros in the trade matrix includes 10,017 exporter-year dummies, 8550 importer-year dummies, and 28,532 country-pair dummies. Following Egger and Tarlea (2015), standard errors are adjusted for clustering on exporter, importer, and country-pair.8 Column 1 in Table 1 shows the empirical results from estimating (4). We obtain a coefficient of 0.13 for WTO membership, 0.43 for PTA membership, and 0.49 for a currency union, all of which are significant at 1%. The last two estimates are in line with the meta-analysis presented in Head and Mayer (2014). We find that WTO matters for bilateral trade with membership increasing bilateral trade by 14.0%. At the same time in line with Baier and Bergstrand (2007) and Glick and Rose (2016), we find a stronger role for PTAs and currency unions - PTA membership increases bilateral trade by 54%, while currency unions boost trade by 63%. In fact, the estimated coefficient for βPTA and βCU are close to those in the preferred specification in the above two papers. Like in Herz and Wagner (2011), GSP has a negative impact on bilateral trade. Column 2 in Table 1 shows the PPML estimates of Eq. (5). Here we find that the coefficients on PTA and currency union decline substantially, and the coefficient on the WTO dummy becomes negative, albeit insignificant. These results confirm the original Rose (2004) finding that PTA membership increases bilateral exports while WTO membership has a negligible impact on bilateral exports.

A More Demanding Specification

Our baseline estimates in Columns 1 and 2 in Table 1 use country-year fixed effects that absorb any effects that are particular to changes in variables at the exporter-year level and the importer-year level (e.g., investment in ports and trade infrastructure, regulations that facilitate or impede trade, number of PTAs that the exporter participates in, number of years the exporting (and importing) country has been part of the WTO, etc.) We also account for unobserved time-invariant dyadic effects by including country-pair fixed effects. However, since these dyadic effects are time-invariant, they immunize us to endogeneity concerns only to the extent that any omitted variables are time-invariant (Limão, 2016). If time-varying dyadic effects affect both the likelihood of trade arrangements and the volume of trade, endogeneity concerns persist (Bergstrand et al., 2015). For instance, country-pairs with rising ties, via flows of capital (Di Giovanni, 2005) and/or immigration (Head and Ries, 1998), may observe not just an increase in trade but also an increased likelihood of a bilateral PTA. To the extent that PTAs are anticipated, trade may decline or be delayed in the run-up to implementation so that the increase in trade may not be entirely attributable to a PTA. Therefore, we adopt an even more demanding specification, whereby we also include pair-specific linear time-trends, one for every country-pair. This is the most demanding specification accounting any unobserved country-pair specific variables that evolve in a linear fashion. For example, the pair-specific trends will account for a secular decline in all unobserved trade costs at the country-pair level.9 We estimate the following gravity specifications:

Column 3 in Table 1 estimates Eq. (4a) and shows that adding the pair-specific trends leads to a substantive decline in the coefficient for both the WTO dummy and the PTA dummy as compared to the specification in Column 1. The coefficient on PTA declines by 62% but remains significant while such a comprehensive specification is sufficient to render the coefficient on the WTO dummy insignificant. Currency Union membership continues to have a positive and significant impact on bilateral exports. In Column 4, with the PPML methodology, we find that the WTO dummy remains negative and insignificant, and the PTA dummy remains positive and significant. Overall, these results show that PTA membership is more effective in raising bilateral trade than WTO membership, in line with the deeper integration argument emphasized earlier. Our findings align with those in the original Rose (2004) paper, which found a negligible impact of WTO membership on the volume of bilateral trade flows and questioned its raison d’être as the promoter of world trade

Heterogenous Effects of Trading Over Time

If we think of membership in a trading arrangement as a treatment as in Baier and Bergstrand (2009), then implicit in Eqs. (4a) and (5a) is the assumption of a homogeneous treatment effect. Essentially, we switch on a dummy variable for a trading arrangement, which assumes that the treatment effect is constant across country-pairs and, more importantly, over time. This is a relatively strong assumption. Anderson and Van Wincoop (2004) highlight the implausibly strong assumption of common coefficients imposed on the trade cost function when the effect of trade arrangements is assumed to be uniform for all members. Subsequent work has highlighted the heterogeneous effects of WTO membership (Tomz et al., 2007), of trading arrangements across industries, (Anderson and Yotov, 2016), across countries, (Subramanian and Wei, 2007), across trading arrangements (Kohl, 2014; Eicher and Henn 2011) and even within trading arrangements (Baier et al., 2019). Similarly, Allee and Scalera (2012) show that the more rigorous a country’s accession to WTO, and thus more significant the trade liberalization required to join, the greater the effect on trade from WTO membership. Our contention here is that there is heterogeneity in the treatment effect over time. Moreover, these heterogeneous treatment effects themselves may vary by type of WTO membership, depth of trading arrangements, by levels of development, by the rigor of accession procedures, and differentially impact the extensive vs. intensive margins of trade.

Heterogeneity in WTO Effects over Time

The effect of WTO membership may take time to manifest itself for a variety of reasons. First, eight rounds of trade talks have been successfully concluded under the auspices GATT/WTO that led to reciprocal reductions in tariffs and nontariff barriers over time. In addition, the MFN principle means that concessions made to a trading partner had to be extended to all WTO members, even if these were not negotiated at the multilateral level.10 Earlier rounds of trade talks in the 1960s and the 1970s saw a reduction in tariffs in advanced economies. To a large extent, this was driven by concerns of the United States, Japan, and other large exporters to redress the trade diversion arising from European regional trade liberalization and expansion of the European customs union. At the same time, new forms of protection in the form of non-tariff measures (NTMs include quotas, subsidies, administrative frictions, health, and sanitary measures, etc.) arose as substitutes to offset the competitive effects of earlier tariff cuts. Such non-tariff barriers are generally more disadvantageous to developing countries who lack the capability for meeting the requirements dictated by NTMs and find it challenging to marshal resources needed to analyze and understand the nature and implications of the NTMs for their exports. Starting with the Tokyo Round and culminating in the Uruguay Round, trade talks emphasized the elimination of non-tariff barriers (subsidies and countervailing measures, licensing procedures, intellectual property exceptions) and closing of various escape hatches by tightening dispute settlement measures. New trade-related policy areas were also put on the negotiating table in the Uruguay Round, including TRIPs (Trade-Related Intellectual Property), TRIMs (Trade-Related Investment Measures) and services trade that sought to deepen integration. Many of these new obligations allow gradual phase-in for developing countries, with transition periods varying from two years (sanitary and phytosanitary measures and import licensing) to five years (TRIMs and TRIPs). All these imply that a simple binary variable for WTO membership or even multiple dummies over 5- or 10-year windows may be insufficient to capture such time-varying heterogeneity. Theoretically, a few papers have highlighted why trade liberalization under WTO deepens over time. Maggi and Rodriguez-Claire (2007) model both the terms-of-trade (TOT) and domestic-commitment motivation for joining a trade agreement and show that trade liberalization typically occurs over time. Initially, there is an immediate decline in tariffs, which reflects the TOT motive highlighted first by Bagwell and Staiger (1999, 2002), followed by a gradual tariff reduction, which reflects the domestic-commitment motive used by the government to close the door to importer lobbies. As a result, tariffs decline over time. However, to the extent that a more gradual tariff reduction follows an immediate trade liberalization, we should expect a strong initial impact of WTO membership and a smaller effect over time. Similarly, Burstein and Melitz (2013) model firm dynamics and endogenous innovation in response to trade liberalization and show that the longrun trade elasticities with respect to trade costs are substantially higher than corresponding short-run elasticities. However, this should not differentially affect WTO vs. PTA membership. Second, there are variations in types of GATT/WTO membership across countries and over time. Tomz et al. (2007) make a distinction between formal members and nonmember participants where certain colonies, newly independent states, and provisional members had GATT rights and obligations on a de facto basis, even though they were not formal members. Accounting for both de jure and de facto WTO membership, they find positive WTO trade effects on bilateral trade. Kohl (2017), and Kohl and Trojanowska (2015), on the other hand, show that formal members gained more than non-member participants and that WTO participants’ gains go hand in hand with the extent of their multilateral liberalization commitments. More importantly, these countries graduated from nonmember participants to formal members at various points in time, again creating heterogeneity in WTO effects over time. Therefore, even the procedure in Tomz et al. (2007) that incorporates time-invariant dummies for formal and nonmember participants may fail to capture the heterogeneity of WTO effects over time.11 Third, differences in the level of development among WTO members led to Special & Differential (S&D) Treatment provisions that allowed developing countries to bring down their tariffs gradually over time, as compared to industrialized countries. Over time, even developing countries began to perceive that they were isolating themselves from a global market and that the positive discrimination under S&D had morphed into increasingly negative discrimination towards their trade.12 They shifted towards MFN tariff reductions and adopted the disciplines of GATT. Therefore, we would expect the WTO effect to strengthen over time, and this should be more prominent when the destination is a developing country. Fourth, even within developing countries and de facto members, we observe varying degrees of rigor in accession procedures (Allee and Scalera 2012). For some countries, accession was more or less automatic without significant commitments on reducing trade barriers. These countries were mainly newly independent colonies who joined under Article XXVI: 5(c) primarily in the 1960s and 1970s. They were permitted to retain higher trade barriers. Other countries went through rigorous procedures, reducing not just tariffs but also made a significant commitment to reducing nontariff measures such as quotas, subsidies, health and safety standards, customs procedures, intellectual property, etc. Finally, Handley (2014) emphasizes that binding commitments under WTO reduce trade policy uncertainty and induce entry by exporters. Handley and Limão (2017) show that China’s export boom soon after WTO accession in 2001 eliminated the annual threat of reversion by the US to high tariffs, and this explains 22–30% of the increase in Chinese exports. The mechanism here is that exporters respond not only to changes in the applied levels of trade barriers, but also to changes in the probability that trade barriers might be raised in the future. Under WTO rules, every item of the negotiation is part of an indivisible package (Single Undertaking) and cannot be renegotiated easily or separately. Moreover, WTO operates via consensus of all members, with each member exercising effective veto power. This makes reneging on trade policy commitments following WTO membership harder as compared to PTAs where there are fewer constraints on renegotiation. Such restrictions can induce a decline in trade policy uncertainty, manifesting in a strong and rising impact of WTO membership over time. We should also expect a stronger effect on the extensive goods margin of trade flows as the decline in uncertainty induces entry of new exporting firms and varieties. To summarize, there are multiple mechanisms and channels through which we may expect that the impact of WTO on bilateral trade to change over time. While allowing for heterogeneity in WTO effects as the baseline, we will also allow for heterogeneity over rounds of trade talks, in the type of WTO membership (formal vs. nonmember participant), by levels of development, by the rigor of accession procedures, and over the extensive and intensive margins of trade.

Heterogeneity in PTA Effects over Time

Preferential trading arrangements are also phased in slowly over time. Tariffs are gradually phased out, quotas modified, and regulations adjusted over time. For example, the ASEAN FTA announced in 1992, incorporated an “inclusion list” of goods on which each country agreed to reduce tariffs to 0–5 percent within ten years for some and within five to eight years for others. The agreement also specified a temporary “exclusion list” where countries were permitted to delay tariff reduction on certain goods. Three years after membership, countries had to gradually transfer the goods to the inclusion list (Feridhanusetyawan, 2005). To the extent that we work with aggregate bilateral trade flows, we should expect the PTA impact to vary over time. In multilateral preferential trade agreements, an additional source of heterogeneity is staggered ratification by member countries over time.

PTAs also contain review clauses for goods, and evolutionary provisions regarding non-tariff measures spanning services, investment, competition, and government procurement. Kohl et al. (2016) demonstrate that trade agreements that deepen WTO commitments are particularly trade-promoting while the lack of legal enforceability for policy domains not covered by WTO provisions have a weaker role.13 For instance, the Singapore-EFTA trade agreement came into force in 2003 but included evolutionary clauses that Singapore adhered to later by liberalizing investment and trade in services, passing legislation, and binding these changes into the trade arrangement. Again, we can expect heterogeneity in PTA effects over time. Another reason for heterogeneity may be gradual adjustments of trade flows to terms-of-trade changes following a trade arrangement (Bergstrand et al., 2015). Prior research does account for such differential impact over time. For example, Baier and Bergstrand (2007) use 5-year and 10-year lags and show that accounting for these lags doubles the effect of a PTA. Similarly, Baier et al. (2019) use a 5- year lag. However, such choices are somewhat ad-hoc and insufficiently flexible in capturing rich patterns of heterogeneity over time.

Specification and Baseline Results of Heterogeneity in WTO and PTA Effects over Time

We adopt a flexible semi-parametric approach. We conceptualize the number of years that a dyad has been part of a particular trading arrangement as the intensity of treatment of the country-pair for that trading arrangement, with the treatment intensity increasing in the number of years that the country-pair continue to be members of the same trading arrangement. We code separate dummy variables for this intensity of treatment, one each for the number of years a countrypair has been part of a trading arrangement. We estimate distinct coefficients for each of these dummy variables, which allows us to eschew any functional form assumptions on the effect of trading arrangements and allows us to check for heterogeneity in treatment effects for a particular trading arrangement over time. In other words, we can check if the effect of WTO/PTA increases or decreases with years of membership.14 With such a comprehensive set of dummies, we let the data inform us of the differential impacts over time without resorting to somewhat ad-hoc choices in previous work where membership in trading arrangements impacts trade with a 5- or 10-year lag. Consider the following gravity specification where we use the PPML methodology and estimate the following gravity equation in levels within a Poisson specification:

WTOod, τ t is a dummy that takes the value 1 if both countries in a country-pair od have been in the WTO for τ years in year t. With data on membership spanning 1948–2012, we measure the time dimension of WTO experience with 65 dummies, allowing the effect of the WTO to vary over time. The omitted category is τ = 0 – country-pairs that have been in the WTO for 0 years.15 Therefore, each of the βWTO, τ coefficients measure the impact of membership in WTO for τ > 0 years compared to country-pairs that are not both WTO members. Similarly, we create 62 dummies capturing experience in PTAs and estimate 62 unrestricted coefficients βPTA,τ  allowing its effect to change over time. The oldest continuous PTA in our data is one between Poland and Hungary who, along with other eastern European countries, became members of COMECON in 1949. COMECON was disbanded in Jan 1991, but both Poland and Hungary became members of a new PTA (CEFTA) in 1992.16 Implicit in standard specifications (for example, equations 4 and 5 ´ ) is also an assumption of a linear relation between trading arrangements and (log) bilateral trade. Such a parametric specification is common in prior research. Kohl and Trojanowska (2015), Baier and Bergstrand (2009), and Chang and Lee (2011) are notable exceptions to this parametric approach. These papers rely on non-parametric matching methods to identify the effect of PTAs as in Baier and Bergstrand (2009) and of WTO membership as in Kohl and Trojanowska (2015) and Chang and Lee (2011). Matching in this context compares bilateral trade of a treatment pair (where both countries are members of WTPO/PTA) with a control pair (where at least one is not in WTO/PTA), so identification relies on cross-sectional comparisons. It implicitly assumes that the effect of a treatment is constant over time. In Baier and Bergstrand (2009), matches are for the same year for different dyads, and they perform their analyses separately for each year to estimate the long-run effects of PTAs. As a result, this method is not well suited to uncovering the short-run vs. long-run effects of trading arrangements. In contrast, the specification in Eq. (6) does not impose any functional form restrictions, allows for heterogeneity in treatment effects over time, and is suitable for capturing short vs. medium vs. long-run impact.17 Fig. 1A plots the coefficients for the 65 WTO dummies and 62 PTA dummies respectively from estimating Eq. (6) along with the 90% confidence interval (the shaded region around the coefficient estimate for WTO and bars for PTA dummies). We also plot the WTO coefficient from Column 4 in Table 1 (dashed horizontal line at −0.131), showing the effect of WTO membership when we impose homogeneity of WTO effect over time. In Fig. 1A, we find that the initial impact of the WTO (from Year 1-Year 14 of membership) is insignificant. However, it becomes positive and significant from Year 15 onwards. From Year 15 onwards, the WTO effect rises more or less monotonically to reach a peak in Year 61. By contrast, we find that the effect of PTA is stronger in initial years, rising to a peak for 25 years of PTA membership and then declines subsequently. For the first 20 years, PTA membership has a larger impact on bilateral trade than does WTO membership. From 20–30 years, the impact is nearly identical, and from Year 30 onwards, WTO membership has a larger impact on trade. The sharp downturn in the PTA coefficient starting Year 56 corresponds to calendar years 2011–2012 – the year of the Eurozone crisis. Given the comprehensive dummies used and the three-way clustering strategy, the WTO coefficients are not identified precisely and have large standard errors though they are statistically significant. Hence the confidence intervals for the WTO and PTA coefficients overlap over the entire span of years of membership. This is an important caveat. WTO membership has a stronger impact on trade compared to PTA membership in the long run, but only in terms of magnitude. WTO membership raises trade on average by 1.1% per year, while PTA membership raises trade by 0.2%.18 Fifty-five years of WTO membership increases bilateral trade by 104%. The comparable effect for PTA membership is 58%.19,20 Interestingly, Fig. 1A also shows that the initial impact of WTO membership is negative, though insignificant, from Years 2–6. In 25% of the observations, the number of years of WTO co-membership is less than or equal to 6 years, which is consistent with the negative but insignificant coefficient on the PPML estimate in Column 4 in Table 1. One potential reason for the initial negative impact could be that trade rises in anticipation of WTO membership. We examine this by creating three dummy variables, each taking the value one for two, three, and four years before WTO co-membership and added them to Eq. (6). All coefficients measure the impact of trade relative to the omitted category τ = 0. Fig. 1B shows the WTO coefficients for such a specification. In Fig. 1B, we do find that trade between a country-pair rises significantly two and three years before formal accession by one or both into the WTO, so the negative (albeit insignificant) WTO effects in initial years is consistent with regression to the mean. The coefficient estimates in the post-WTO years are very similar to those in Fig. 1A. The WTO has had various rounds of trade liberalization that affected all existing WTO members, and our estimates may merely be capturing these year-specific WTO effects rather than the number of years of membership. Therefore, as a robustness check, we allowed for common calendar year-specific WTO effects. We interacted the WTOod, t dummy with calendar year dummies, one for each year from 1951 to 2012. We do not find evidence for strong and persistent WTO effects by calendar year. The only significant coefficients are for the calendar years 1994–1998, which coincides with years right after the Uruguay Round of trade talks, and for the initial years (1951–1952) following the founding of GATT. More importantly, compared to Fig. 1A, the coefficients on WTO effects by years of membership increase marginally. Therefore, the effects are not driven by the different rounds of trade liberalization that affects all WTO members but the number of years of shared membership for a country-pair.21 Our results clearly show the heterogeneity in WTO effects over time, accounting for both the lead into WTO membership and post-WTO membership. In the most demanding specification possible and allowing for heterogeneity in WTO and PTA effects over time, WTO impact increases more or less monotonically over time and eventually dominates the PTA effect in terms of magnitude. Given the wide confidence intervals, especially for later years, we cannot reject the null that WTO effects and the PTA effects are significantly different. A conservative interpretation of our results is that we observe strong effects of the WTO, but only in the long run. These findings are consistent with trade liberalization over multiple years of trade negotiations under the WTO, with a gradual decline in both trade and non-trade barriers over time, perhaps for subsets of countries, and with a gradual decline in trade policy uncertainty over time.

Heterogeneity by Depth of Trade Agreements, Level of Development and Margins of Trade

Heterogeneity by Types of WTO Membership

Tomz et al. (2007) find a strong role for the WTO when formal membership in the WTO is supplemented with nonmember participants. Colonies, newly independent states, and provisional members were de facto members subject to obligations similar to formal members in the GATT era. Many transitioned into formal membership at different points in time, potentially creating heterogeneity in WTO effects over time. We follow Tomz et al. (2007) and use their classification of countries into formal members and nonmember participants. We estimate equation (6 ) using PPML but allow for distinct WTO membership effects for three sets of dyads in a single estimation: 1) where both countries in a dyad are formal GATT/WTO members; 2) where both are nonmember participants, and 3) where one country is a formal member, and the other is a nonmember participant. Fig. 2A is the analogue to Fig. 1A, showing the effect of formal WTO and PTA membership over time. The signs, coefficients, and relative magnitudes are nearly indistinguishable from Fig. 1A. All coefficients from Year 15–61 are statistically significant. This is not surprising since the variables measuring years of WTO membership are identical, and the results indicate that formal membership continues to matter even when we account for further degrees of heterogeneity in types of WTO membership. Fig. 2B shows the impact of years of membership when both countries in a dyad are nonmember participants, and where one in a dyad is a formal member and the other a nonmember participant. In neither case, we observe a positive and significant impact of these types of membership. This contrasts with Tomz et al. (2007), who find the strongest effect when both countries in a dyad are nonmember participants. In fact, for nonmember participants, we see a negative and significant impact for almost all years between Year 3 and Year 14 and an insignificant impact subsequently. There are multiple reasons for such a divergent finding. Unlike Tomz et al. (2007), we allow for heterogeneity over time, account for zeros in the trade matrix, and deploy a more comprehensive set of fixed effects.22 The PPML methodology used here puts more weight on larger trade flows. If trade between nonmembers is small but positive, then the coefficient estimates in Fig. 2B for nonmembers are consistent with trade diversion towards formal members. However, the WTO effects where one is a formal member and the other a nonmember is not consistently positive and significant. Therefore, the initial negative effect is not entirely driven by trade diversion from nonmembers towards formal members. Instead, the differences in the specification matter as well.

Heterogeneity by Level of Development

Developed countries undertook far greater trade liberalization under the auspices of the GATT, reducing their average tariffs from 15% in 1947 to about 4.5% (Subramanian and Wei, 2007). In contrast, pre-Uruguay Round, developing countries were granted Special & Differential (S&D) treatment and had far fewer obligations to liberalize tariff barriers.23 This asymmetry implies that we should expect differential effects for GATT/WTO membership for developed vs. developing countries. To the extent that the destination country is a developing economy that reduced trade barriers slowly over time, we would expect a rising WTO effect over time. If the destination country is an advanced economy that quickly lowered trade barriers or had low barriers, to begin with, we should expect a small and even negligible role for WTO membership over time. To the extent that WTO membership is a commitment device and primarily about reducing trade policy uncertainty, we should again expect a bigger impact when the destination country is a developing country. We start with Subramanian and Wei’s (2007) classification of countries as advanced vs. developing economies. These are 25 countries that were OECD members on or before 1973. To this list, we add countries classified as High-Income by the World Bank in 1970. These include 5 major oil-exporting countries (Brunei, Kuwait, Qatar, Saudi Arabia, and the United Arab Emirates), Israel, Liechtenstein, and 8 small, rich island nations. We estimate Eq. (6) using PPML for four sub-samples.24 Fig. 3A shows the PTA and WTO effects over time when both the origin and the destination are advanced economies; in Fig. 3B, both countries in a country-pair are developing countries; in Fig. 3C, the exporter is an advanced country while the destination is a developing country; in Fig. 3D, the exporter is a developing country while the destination is an advanced economy. Fig. 3A shows a negative and insignificant impact of WTO membership when both countries in a pair are advanced economies, regardless of the number of years of WTO membership. However, most of the OECD countries are either founding members of GATT or joined soon after the formation of GATT,25 so their WTO membership is collinear with the countryyear effects. Identification primarily relies on the variation in years of WTO co-membership, where one of the trading partners is one of the small island nations or a major oil producer. This makes it difficult to identify the WTO membership effect over time when both trading partners are advanced economies. For PTA membership, there is sufficient variation within country-pairs over time to identify a strong PTA effect over time, as shown above.26 Fig. 3B shows the effect of WTO and PTA membership over time when both countries in a pair are developing countries. The WTO effect on bilateral trade rises over time, rising to a peak after 58 years, with bilateral exports higher by 248% compared to a pair that have zero years of WTO membership. WTO membership has a small negative effect in initial years but quickly turns positive and significant after 12 years. The effect of PTA membership is negative, and the WTO coefficients are significantly larger than the PTA coefficients. The jump in the coefficient for PTA membership in Year 54 again reflects that identification relies on a very few developing country-pairs with long-lasting PTA membership.27

Fig. 3C shows that a rising magnitude of WTO membership over time, when the exporter is an advanced economy while the importer is a developing economy. Initially, these WTO coefficients are insignificant, but beyond Year 20, they are statistically significant for most years. While PTA membership dominates initially, in terms of magnitude, it is overtaken by WTO membership by year 20. Again, the confidence intervals overlap for the PTA and WTO coefficients.

Comparing Figs. 3B and 3C, we see that for the first 20 years, WTO coefficients are slightly larger when the exporter is a developing country. The average difference between the coefficient estimates βDev,Dev WTO,τ − βAdv,Dev WTO,τ for τ = 1, . . . 20 is 0.02. However, these differences are not statistically significant. Beyond 20 years the magnitude of the WTO effect is stronger when the exporter is an advanced economy with the average of βAdv,Dev WTO,τ − βDev,Dev WTO,τ for τ = 21, . . . 65 equal to 0.39, where βDev,Dev WTO,τ refers to the WTO coefficients in Fig. 3B and βAdv,Dev WTO,τ refers to the WTO coefficients in Fig. 3C. The differences for τ = 36, . . . 65 are statistically significant. Fig. 3D shows the effects of WTO vs. PTA membership when the exporter is a developing country, and the importer is an advanced economy. In this sub-sample, we find a stronger positive effect of PTA membership and an insignificant effect of WTO membership. This is consistent with the fact that advanced economies had low tariff barriers to begin with, with applied tariff rates below WTO mandated bound tariff rates, and extended unilateral preferences to many developing countries in certain sectors. PTA membership on the other hand, extended far greater market access. To summarize our findings, WTO effects are strongest in the long run, if the destination country in a pair is a developing country. Such a finding contrasts sharply with the Subramanian and Wei (2007) results that WTO effects are observable only for advanced economies. Even with the wide confidence intervals, WTO significantly exceeds PTA impact when both countries in a dyad are developing countries. WTO membership does not have a significant influence when the destination country is an advanced economy. Again, this is consistent with developing countries gradually reducing tariff barriers over time, and with a decline in trade policy uncertainty that is particularly salient in developing countries. PTA effects, on the other hand, are stronger when the destination is an advanced economy.

Heterogeneity by rigor of WTO Accession Procedures

Splitting the sample as in the previous sub-section elides differences in accession procedures within and across the broad categories of developed and developing countries. As Allee and Scalera (2012) highlight, some developing countries such as China and Vietnam had a long-drawn accession process and asked to undertake comprehensive trade reform. Others, like Mongolia, had a shorter accession period. Similarly, within advanced economics, countries like Israel, Japan, and Ireland had far more rigorous accession procedures. Finally, more than 60 countries joined GATT/WTO based on Article XXVI:5(c) where a newly independent territory (e.g., Indonesia) could obtain membership in the GATT based on the earlier membership of, and consent by, its former colonial power (Netherlands).28 While most of these are developing countries, one country that we classified as an advanced economy, Qatar, also joined via this route. Overall, categorizing countries as advanced vs. developing omit important differences. The same is true for the broad categorization of countries into de facto vs. de jure WTO members. Automatic accession, for instance, was the route used by newly independent colonies for WTO members, who we classified as nonmember participants in Section 4.1. However, there is considerable heterogeneity in the accession procedures across nonmember participants. In essence, some automatically transitioned into formal members (e.g., Kenya, Uganda, based on Article XXVI:5(c)) while others undertook far more rigorous trade liberalization (e.g., Argentina, Colombia). Therefore, analyzing the effect of WTO by the degree of rigor required in accession procedures is arguably a more accurate way to capture heterogeneity in trade liberalization undertaken by various countries rather than the coarse classifications of developing vs. advanced and formal vs. nonmember participants used so far. Following Allee and Scalera (2012), we classify 48 countries that joined the GATT/WTO after 1953 under Article XXXIII of the GATT or Article XXII of the WTO as rigorous acceders. 65 countries that joined under Article XXVI:5(c) are classified as automatic acceders.29 We estimate Eq. (6) separately for two sub-samples: one where the destination (importer) countries are rigorous acceders and one where the destination countries are automatic acceders. Fig. 4A shows the WTO estimates for each sub-sample. We observe a strong and rising role impact of WTO membership on exports to destinations that underwent rigorous accession procedures. In contrast, we see an initial significant decline, followed by a marginal but insignificant increase in exports into destinations that automatically acceded. The initial drop is consistent with trade diversion for these newly independent colonies as they are no longer captive markets for the exports of the metropole.30 Fig. 4B compares the WTO effect to the PTA effect for rigorous acceders. Here we see that the WTO effects over time easily exceed the PTA estimates both in terms of magnitude and in terms of statistical significance. On average, a one year increase in WTO membership raises exports to the destination by 5.58% while a one year increase in PTA membership raises exports by 0.76%.31 Given that 44 out of the 48 countries that went through rigorous procedures are developing economies, this further refines the findings in the previous section. The strongest WTO effect over time manifest themselves when the destination is a developing country that undertook rigorous trade liberalization as part of WTO accession. If we compare the WTO effects where the origin (exporter) in a country-pair underwent rigorous vs. automatic accession procedures, then we find the opposite pattern, as shown in Fig. 4C. In exporters, who acceded automatically, we find rising WTO effects over time. Exporters who underwent rigorous accession procedures experience a smaller but insignificant rise in exports. A potential explanation is that exporters in these countries faced grater import competition that induced exit. In contrast, exporters in countries that automatically acceded faced less competition in exchange for market access, allowing them to expand their exports.

Heterogeneity at the Goods Margin

Much of the growth of world trade has been on the extensive margin or the new goods margin, especially over longer time horizons. Kehoe and Ruhl (2013) show that this is especially true for significant trade liberalizations such as NAFTA and China’s access to the WTO. Dutt et al. (2013) find that WTO membership affects mainly the extensive margin, while Baier, Bergstrand, and Feng (2015) show that PTA membership increases both the extensive and intensive product margins. However, both use a time-invariant dummy variable for membership in a trading arrangement. Therefore, we look at the effect of these trading arrangements over time on the extensive and intensive goods margins. We decompose bilateral exports as the product of an extensive margin and an intensive margin, using the methodology of Hummels and Klenow (2005) and Feenstra and Kee (2008). Define Jod, t as the set of products exported from origin o to destination d in year t and let J Wd ≡ U o,t Jod,t be the set of all products exported to d from any country in any year in our sample. The index W stands for “World,” i.e., the collection of origin countries. Instead of a simple count measure, Feenstra and Kee (2008) recommend that the number of products be weighted by their importance in the imports by the destination d and that time-invariant weights be used so that extensive margin changes only because of changes in the set of products o exports to d and not due to the change in weights. Define X¯ Wd(j) as the average value of exports from the world (summed over all exporting countries and averaged across years) of product j to d from the world. This is the weight assigned to product j for bilateral exports from any origin country o to d. The extensive margin of exported goods from origin to destination is then defined as the fraction of goods sold by the exporter in the destination, weighing each product by its importance in world exports to this destination, averaged over time. It is given by

The denominator is destination-specific and constant across exporting countries and time. The time-invariant weights X¯ Wd(j) imply that the extensive margin changes over time only when there is a change in the set of goods the origin sells in the destination (Jod, t). If the products were equally weighted, and not by their importance in the destination’s imports from the world, then the extensive margin would simply be a fraction of goods that the origin exports to the destination. The intensive margin of exported goods from o to d is given by

where Xod, t(j) is the value of exports from country o to country d of good j at time t. The intensive margin equals o’s nominal exports relative to W’s average exports (averaged over time) to d in those categories in which o exports to d at time t (Jod, t). Thus, it measures the overall market share country o has within the set of categories in which it exports to d. Note that the product of the two margins is

which equals total bilateral exports from o to d in year t as a fraction of country d’s average imports. A decline in the fixed costs of trade increases the extensive margin by facilitating entry by new firms and new products. It is likely to reduce the intensive margin given that new entrants in most cases will enter at a small scale (relative to the rest of the world’s exports in these new categories in the destination.) A decline in the variable costs of trade will increase both the extensive and the intensive margins of trade as we will observe new entry (with exports becoming more profitable) and incumbent firms increasing their market share in the destination. IMod, t is not defined for pairs with zero bilateral trade since Jod,t = φ (empty set) with a cardinality of zero. This implies that we cannot use the PPML specification. Using OLS, on the other hand, ignores the zeros in the trade matrix. Consequently, as highlighted in Helpman et al. (2008) or HMR, the estimates would be contaminated by selection bias (from dropping the zero or missing observations) and by a heterogeneity bias (from unobserved trade costs that account for firms’ self-selection into exports). Therefore, we apply the HMR methodology used in Dutt et al. (2013) and Baier et al. (2014). The HMR methodology estimates a probit equation with the probability of strictly positive exports from o to d for each year separately. The estimates are subsequently used to predict two values: a latent variable that reflects the ratio of export profits to fixed export costs for the most productive exporter that determines self-selection into exports and the inverse Mills ratio that accounts for zeros and missing observations. As an exclusion restriction, we follow HMR and use an index for common religion defined as  k proportion o f religionk o,t ∗ proportion o f religionk d,t where k is an index for each religion.32 In the second step, these two predicted variables are pooled and included as additional independent variables in the estimating equations for the two margins. Therefore, we estimate the following equations using OLS with an identical set of fixed effects.

The polynomial in zˆod,t approximates the latent variable that controls for heterogeneity bias, and IMR is the inverse Mills ratio that is the standard Heckman correction for sample selection bias. In a log-specification for the margins, adding the coefficients on the extensive and intensive margins will yield the traditional gravity coefficients with the log of trade as the dependent variable, once we include importer fixed effects.33 To ensure that we have sufficient coverage over time and across countries, we use data from the World Trade Flows Database from Feenstra et al. (2005). This database contains information on bilateral exports for more than 150 countries over the period 1962–1999. The data are based on the 4-digit Standard International Trade Classification, Rev. 2, with 790 4-digit categories and account for 98 percent of all world trade.34 Fig. 5A shows the estimates for the extensive margin from Eq. (7); Fig. 5B shows the estimates for the intensive margin from Eq. (8). Several interesting results and implications emerge from this decomposition. First, Fig. 5A and B both indicate that the effect of WTO increases over time. The effects for the extensive margin are statistically significant after Year 3, while those for the intensive margin are significant after Year 8. This is consistent with WTO reducing variable trade costs over time. However, the rising impact on the intensive margin is not inconsistent with WTO membership inducing a decline in uncertainty over time. This would encourage more entry over time of new products and ceteris paribus, reduce the intensive margin. However, if WTO membership also reduces variable trade costs, this negative intensive margin impact could be offset by expanding exports of existing products responding to the decline in these variable trade costs. Comparing the WTO effects for the extensive and intensive margins, as in Dutt et al. (2013), we find that the WTO has a stronger effect on the extensive margin of trade in terms of magnitude. With the wide confidence intervals for the intensive margin, we cannot reject the equality of the WTO coefficients for the extensive and the intensive margin.35 To the extent that growth in world trade in the long run has been along the extensive margin, the WTO seems to have played a key role. Second, PTA has a positive impact on both margins, with a stronger impact on the intensive margin compared to the extensive margin for the first 25 years of membership. The initial strong impact on the intensive margin is in line with Baier et al. (2014) who examine PTA effects up to a 5-year lag. Beyond 25 years, PTA membership increases only the extensive margin with a zero or slightly negative impact on the intensive margin. This is consistent with a decline in variable trade costs in the short to medium run in a PTA and reduction in non-tariff barriers (fixed trade costs) over the long-run. Comparing WTO to PTA, we find that for the extensive margin, WTO effects are consistently stronger than PTA effects, both substantively and statistically, regardless of the number of years of membership. For the intensive margin, PTA dominates WTO membership for the first decade, while beyond that – in the longer term – the WTO plays a stronger role. However, for the intensive margin, the difference in the WTO and PTA coefficients are not statistically significant.

Heterogeneity by type of PTAs

One potential criticism is that we treat all preferential trade arrangements in an equivalent fashion, eliding over differences in their depth of integration. Prior work convincingly demonstrates that deeper trade agreements are more effective at promoting trade (e.g., Kohl, 2014; Baier et al., 2014; Kohl and Trojanowska, 2015; Baier et al., 2019).36 Therefore, a single PTA dummy may not suffice in capturing the heterogeneity across PTAs. Comparison of PTA membership with WTO membership would rely on some ‘average’ PTA effect with individual PTAs exhibiting heterogeneity in the degree of integration. One option would be to have separate time-specific PTA dummies for each PTA. However, this leads to a proliferation of dummies and loss of precision in the coefficient estimates (Baier et al., 2018). Therefore, we follow Baier et al. (2018) and use a fourfold classification of PTAs: one-way trade agreements where the importing country grants unilateral non-reciprocal access to the exporter37; two-way bilateral PTAs, multilateral PTAs that have more than one country in a PTA; and Deep PTAs for country-pairs in customs unions, common markets and/or currency unions. These data are from Jeffry Bergstrand’s website. The expectation is that deeper PTAs such as the EU are likely to have biggest trade impact and may even exceed the WTO in its importance for trade.

We estimate the following specification by PPML

where j = 1 for one-way trading arrangements (GSP), j = 2 for two-way bilateral PTAs, j = 3 for two-way multilateral PTAs and j = 4 for deep PTAs.38 Fig. 6A shows the coefficient estimates for WTO and deep PTAs; Fig. 6B shows the estimates for bilateral and multilateral PTAs. Figs. 6A and 6B show that while WTO membership dominates Multilateral and Bilateral PTAs over time, it is dominated by Deep PTAs. Our estimates imply that bilateral exports are 60% higher after 55 years of WTO membership. For Deep PTAs the corresponding increase is 116%. On average, an extra year in a Deep PTA raises bilateral exports by 1.58% while an extra year of WTO membership raises bilateral exports by 0.8%. By contrast, neither multilateral nor bilateral PTA seems to have a long-run impact on exports, with only a short-term effect of bilateral PTAs on exports.39 Unilateral PTAs (not shown) consistently have a negative and significant impact on bilateral trade. Overall, WTO effects over time are stronger and more persistent than unilateral, bilateral, and multilateral PTAs but fall short of Deep PTAs. The stronger impact of Deep PTAs is not surprising since Deep PTAs are increasingly sophisticated and include integration that goes well beyond simple trade liberalization. Deep PTAs include not just custom unions but also various combinations of currency unions, common markets, regulatory coordination, and even labor mobility. At the same time, enthusiasm for Deep PTAs should be tempered by the fact that they are relatively rare - less than 2% of our observations are part of a Deep PTA. By contrast, 45% of the observations share WTO membership. In 2012, for instance, only 3.77% of country-pairs were members of a Deep PTA. The corresponding number for WTO membership was 69%. Overall, while WTO effects may fall short of Deep PTAs, its role in facilitating bilateral trade flows cannot be discounted, especially when compared to the relatively few Deep PTAs in existence.

Disentangling WTO and PTA Effects

WTO and PTA memberships frequently overlap. Subramanian and Wei (2007) argue that bilateral and multilateral preferences involve different degrees of liberalization, hence defining them as we have done so far contaminates the estimates. They recommend a hierarchical classification to disentangle the two. However, Eicher and Henn (2011) show that Subramanian and Wei’s hierarchical classification of dummies, with PTAs at the top and WTO at the bottom of the classification hierarchy, assumes that PTA membership represents the culmination of trade integration, and produces a WTO effect that is actually a PTA effect. They adopt the Rose (2004) mutually inclusive coding (as we have done so far) that identifies net effects of WTO and PTA membership. To disentangle WTO and PTA effects, we use an alternate specification. We decompose each dummy variable for WTO membership for τ years WTOod, τ t into two terms as

where Prefod, t, that takes the value 1 if o and d share the same PTA at time t while (1 − Pre fod,t) takes the value 1 if o and d are not members of the same PTA at time t. Prefod, t is identical to the standard constant PTA effect over time used in the literature. Eq. (6) constraint the coefficient on these two terms to be equal, but we now allow distinct coefficients on each. Similarly, we decompose the dummies for PTA membership PTAod,τ t as

where Bothod,t = 1, when both countries in a dyad share WTO membership and (1 − Bothod,t) = 1 when at least one country in a dyad is not a WTO member. Bothod, t is identical to the standard, constant WTO effect over time used in the literature. This allows distinct coefficients for τ  years of PTA membership for dyads that share WTO membership vs. those who do not (either one or both are not in the WTO). We estimate the following specification using PPML:

The coefficients βWTO−PTA,τ estimate the WTO effect for pairs that also share PTA membership. β WTO−PTA,τ captures the pure WTO effect, uncontaminated by PTA membership, for country-pairs that are not part of any PTA. The omitted category includes country-pairs that share neither PTA membership nor are WTO members.40

WTO membership coefficients are plotted in Fig. 7A. Fig. 7A shows that a rising effect of WTO over time regardless of whether country-pairs share a PTA or not.41 The monotonic increase in the pure WTO effect, is very similar to the overall effect shown in Figs. 1A and 1B. However, it is smaller in magnitude compared to the WTO effect for pairs that are also part of the same PTA. In fact, for the first 30 years of membership, the WTO effects are significantly stronger when a country-pair also shares a PTA for the first 30 years.

The World Trade Organization (WTO) lacks the power to directly enforce agreements. It is therefore important to understand what role the WTO can play to facilitate international cooperation, and whether a multilateral institution can offer distinct advantages over a web of bilateral agreements. This paper examines two potential benefits of a multilateral trade institution: 

1. First, verifying violations of the agreements and informing third parties, thus facilitating multilateral reputation mechanisms

2. Second, promoting multilateral trade negotiations rather than a web of bilateral negotiations. 

The model suggests that a multilateral approach is particularly important when there are strong imbalances in bilateral trading relationships.

Since 1947, trade barriers between nations have decreased dramatically; for example, the average ad valorem tariff on industrial goods has declined from about 40% to less than 4%. 

Does the success of trade liberalization efforts have something to do with the presence of the WTO (formerly known as the General Agreement on Tariffs and Trade, GATT)? If so, what role has the WTO played in facilitating trade cooperation, given that it has no direct enforcement power? 

The trade literature has only recently started to address this question. Avinash Dixit (1987), Kyle Bagwell and Robert W. Staiger ( 1990 ) , and Raymond Riezman ( 1991 ) , among others, examine issues of trade cooperation by way of two-country models in which governments choose trade policies in a repeated-game setting:

• In these models there is typically a great multiplicity of equilibria, including very inefficient ones. 

• These models suggest that the WTO may help countries coordinate on more efficient equilibria. 

• A second potential role of the WTO, and specifically of its Dispute Settlement Procedure ( DSP ) , is suggested by Thomas Hungerford (1991) and Dan Kovenoch and Marie Thursby (1993): the DSP may act as an information-gathering agency that is able to discern between true violations of the agreement and mistaken perceptions, thus facilitating the use of a bilateral reputation mechanism to support cooperation.

The emphasis here is on bilateral monitoring: the idea is that the DSP can improve monitoring by the country directly affected by the trade policy.

In the present paper I focus on two other roles that the WTO-DSP can perform, both of which are intrinsic to its multilateral nature:

1. First, it can verify violations of the agreement and inform third countries, thus facilitating multilateral enforcement efforts.

2. Second, it can promote a multilateral rule-making procedure in place of a web of bilateral negotiations. 

Since the papers mentioned above all utilise two-country models, they cannot examine these aspects. To start with the informational role, the idea is the following:

• Suppose country A commits a violation against country B, and this violation is observed by country B but not by the rest of the trading community.

• A potential role for the DSP in this case is to identify the violation and bring it to the attention of third countries, exposing the offending country to a loss of reputation in the trading community. Notice the difference with respect to the bilateral-monitoring role discussed above: the idea here is that the institution can improve monitoring by third countries, rather than by the ‘‘second’’ country.

• Two questions arise about the relevance of this argument. 

1. First, how important is the scope for this kind of information dissemination for trade policies? Some trade policies, such as tariffs and quotas, are fairly transparent, however there are two important dimensions in which trade policies are much less transparent, particularly to third countries:

1. There are a number of non-tariff barriers to trade (such as government practices that favour domestic producers over foreign ones, antidumping actions, technical and safety requirements on imported products, etc.) that can be hard for third countries to perceive.

2. Current trade agreements allow countries to increase tariffs or quotas above their ‘‘baseline’’ levels under certain contingencies. If a country erects a trade barrier by invoking such special circumstances, third countries may not know what the underlying circumstances are, hence whether there has been a violation of the agreement. 

2. The second issue concerns the idea that an offending country can be punished by a loss of cooperation with third countries. What forms can this loss take in reality? 

• There are several ways in which the WTO community can respond to a violation of the agreements:

1. First, WTO members can withdraw trade concessions to the defecting country. For example, it is the interpretation of several scholars that Article XXIII of the GATT agreement provides for the possibility of expelling a repeatedly offending country from GATT (see John H. Jackson, 1969 pp. 186–87; Michael Finger, 1993).

2. Second, WTO members can impose costs on the offending government in more subtle ways, by withdrawing some of their ‘‘goodwill’’ toward that government: 

• They can be less forthcoming with the offending country in subsequent negotiations, in the same or related areas of cooperation

• They can be more reluctant to enter new agreements with the offending country

• If trade liberalization takes place in a gradual fashion, they can slow down the liberalization process vis-a`-vis the offending country

3. Third, countries can reduce their cooperation with the offending country at the level of institutional procedures; for example, if country A does not follow the recommendations of the DSP panel, third countries may feel free to do the same in future disputes against country A. 

• Having discussed the kinds of punishment that the WTO community potentially can inflict on the offender, there are two questions that can be addressed: one normative (what the WTO community should do to the offender), the other positive (what the WTO community does do to the offender, and why).

The theory presented in this paper, which I will preview shortly, can be interpreted as addressing the normative question. Whether multilateral punishment threats have played a role in practice— and thus, whether the theory can be interpreted as explaining the actual role of the WTO—is a more subtle question:

1. First, to paraphrase Thomas Schelling’s expression, the effectiveness of an army sometimes must be judged by how little it is used. While it is true that explicit multilateral sanctions have never been observed in the WTO, it is also true that there have been no cases of blatant and repeated violations of key WTO rules, even by strong countries against vulnerable trading partners; it is reasonable to think that strong countries have been deterred from abusing weaker partners by an implicit threat of multilateral sanctions. 

2. Second, more subtle forms of multilateral punishments, such as the ones discussed in the previous paragraph, are hard to pinpoint in practice, even if they do take place. The best we can hope for is some indirect, anecdotal evidence from the history of GATT disputes. 

In the last section of the paper, I review a number of disputes which share the following theme: a strong country was accused by a weaker country of violating an agreement, and the strong disputant ended up withdrawing the contested policy, in spite of a clear opportunity to abuse the weaker partner. These anecdotes, I will argue, are suggestive of an effective, if subtle, role of multilateral enforcement pressures in improving compliance with international trade agreements. 

With regard to the role of multilateral punishment threats, it is useful to anticipate a result of the model that is important for evaluating the positive relevance of the theory. While the model shows that a multilateral enforcement mechanism is desirable, it also points out that third-party sanctions should be ‘‘minimal,’’ in a sense to be made precise, and should be threatened only for certain violations, namely those that are hard to deter with bilateral sanctions alone. This result is broadly consistent with the fact that, in the GATT experience, multilateral enforcement pressures seem to play a more subtle and selective role than bilateral ones. 

In what follows, I give a brief overview of the model and of the main results. The theoretical analysis is based on a multi country model in which governments repeatedly select import barriers. To examine the potential benefits of an institution that verifies and publicises violations, I compare two extreme scenarios: 

1. A WTO-less world, in which third countries remain unaware of violations

2. A world with WTO, in which all governments are informed of any violations. In this setting, the only possible benefit of information dissemination is that it enables third countries to punish violators. 

Thus, the question becomes whether a multilateral enforcement mechanism can sustain more cooperative outcomes than a bilateral one. A key feature of the model is the presence of bilateral imbalances of power, where the ‘‘more powerful’’ country in a given pair is the one that stands to lose less (or to gain) from a trade war. 

In this situation, a multilateral enforcement mechanism may be beneficial, because it allows a transfer of enforcement power across relationships that is not possible under bilateral enforcement. In particular, the optimal self-enforcing agreement requires stronger countries to make more generous concessions than weaker countries, but this can be implemented only with a multilateral enforcement mechanism. The gains from multilateral enforcement do not arise simply because punishments are more severe, but specifically because of imbalances in power: in the benchmark case where power is balanced, there are no benefits from multilateralism. The analysis also indicates that international monetary transfers, if feasible, can mitigate the effects of bilateral power imbalances, but cannot fully substitute for multilateral enforcement. 

While the model examined in this paper focuses on imbalances in power, these are not the only factor that gives rise to gains from multilateral enforcement. In Section IV, I will discuss other factors that can make multilateral sanctions desirable. 

The next part of the analysis takes a closer look at the optimal enforcement mechanism. The key finding is that third-party sanctions should be kept within limits. In particular, the analysis reveals that: 

• Past a point, increasing the severity of third-party sanctions in response to a bilateral violation of the agreement will not enhance cooperation; in this sense, ‘‘maximal’’ third-party sanctions are not necessary. Also, the threat of third-party sanctions is necessary only for certain violations, namely those by stronger countries against weaker countries. 

• If the monitoring of trade policies is slightly imperfect, and if inflicting a punishment is slightly more costly for third parties than for second parties, then third-party sanctions should be minimal, in the sense of being the least severe that achieve all gains from multilateral enforcement.

The model is then extended to examine a situation in which the Dispute Settlement Procedure (DSP) again is called upon to verify and publicise violations, but where now a government can block this process by, say, refusing to cooperate with the investigation. In this situation, governments must be induced to comply not only with substantive rules, but also with procedural rules (e.g., facilitating the DSP operations). The interesting question is whether the governments’ power to block the transmission of information undermines the multilateral enforcement mechanism. 

The analysis suggests that the non-enforceability of procedural rules does not impose a binding constraint on the enforcement system, as long as the DSP can publicise procedural violations. Aside from enforcement issues, the model indicates that a multilateral approach may be important also at the level of trade negotiations. 

Trade negotiations are modelled as a process of Nash bargaining over the set of self enforcing trade policy configurations. It is found that, in the presence of power imbalances, multilateral bargaining enables countries to achieve deeper trade liberalization than a web of bilateral negotiations. 

The interesting aspect of this result is that each bilateral negotiation is locally efficient (i.e., it is efficient conditional on the outcome of other bilateral negotiations), and there are no externalities across bilateral relationships, yet a system of bilateral negotiations is globally inefficient. 

The inefficiency of bilateral bargaining is similar to the market failure caused by incomplete markets; if one regards trade negotiations as a market where countries exchange trade concessions, bilateral bargaining is inefficient because the market is segmented. At a broad level, this result suggests a further potential role of trade institutions such as the WTO: to the extent that they can promote multilateral rule-making procedures, they can prevent distortions arising from bilateral imbalances of power. The paper is structured as follows. 

• Section I presents the basic model. 

• Section II examines the implications of multilateralism at the enforcement level. 

• Section III discusses the efficiency gains from multilateral trade negotiations. 

• Section IV discusses some extensions of the model. 

• Section V offers concluding remarks.

The Basic Model

I consider a three-country trading system in which bilateral relationships are separable, in the sense that trade policies in a given bilateral relationship do not affect trade flows in the other two bilateral relationships. 

In this setting, trade policies do not cause trade diversion. I will argue later that the possibility of trade diversion tends to strengthen the paper’s main results on the gains from multilateral enforcement and rule-making. 

A simple way to rule out trade diversion is to assume that each pair of countries trade two distinct goods that are neither supplied nor demanded by the third country.

More formally, let 

• N = {A, B, C } denote the set of countries

• N\i the pair of countries that does not include country i. 

• The numeraire good is indicated by a ‘‘0’’ subscript. 

• Each non-numeraire good is indexed by an ordered pair (i, j), with i ∈ N;  j ∈ N; i ≠ j

Country i’s representative consumer has the following utility: 

• where xij denotes consumption of good (i, j). 

Country i is endowed with z0 units of the numeraire good and with zij units of good (i, j), j ∈ N\i.

Given this structure of preferences and endowments, good (i, j) is exported from country i to country j under free trade. 

To be more concrete, consider for example country A: 

• This country is endowed with goods (A, B) and (A, C), which it exports respectively to countries B and C

• And imports goods (B, A) and (C, A) respectively from country B and C

• In country A there is no demand for good (B, C) or good (C, B). 

• Let p[ij x] and p[ij m] denote the price of good (i, j) respectively in country i (the exporting country), and in country j (the importing country). 

p[ij x] and p[ij m] can be seen in their true form in the equation below labelled (1)

• Governments choose specific import tariffs on the non-numeraire goods (the numeraire good is assumed to be freely traded). 

• I focus on tariffs because they are particularly simple to model; however, the qualitative results would be the same if governments chose non-tariff import barriers, as long as the game has the structure of a prisoner’s dilemma. 

• An import tariff creates a wedge between the local price and the offshore price, Thus, if 𝝉ij denotes the import tariff on good (i, j), the following relationship holds, provided the tariff is not prohibitive:

Export taxes and import subsidies are not allowed; I will discuss the role of this assumption later in this section. The population in each country is a continuum of individuals whose total size is normalized to one. Country i’s local consumers demand good (i, j) in amount [where x d(pij) d(·) is the inverse of u*(·)] and good ( j, i) in amount Consumer surplus in coun- m d(p ). ji try i is given by (j √ N "i / m x [s(p ) s(p )], ji ij where s(p) Å u[d(p)] 0 pd(p). The function u(·) is assumed to be quadratic, so that d(·) is linear. Market clearing for good (i, j) requires: