The Trade-Agreement Embarrassment

The dominant academic literature about trade agreements maintains that they are only about national terms-of-trade manipulation and not at all about purely political concerns. Non-academic economists, commentators, and diplomats by contrast think that trade agreements are all about political concerns. There are two substantive and important distinctions between the two views. i) Practitioners maintain that policymakers care virtually not at all about the terms of trade or about trade-tax revenue. ii) Practitioners, unlike academics, maintain that trade-agreement negotiations themselves change the underlying political economy. Observation of actual trade policy measures, though not conclusive, suggests that the practitioners are right and that the academics are wrong.


I. Introduction
The process of multilateral trade liberalization began after World War II with the negotiation of the General Agreement on Tariffs and Trade (GATT), which sponsored a succession of negotiating rounds addressing trade barriers. Active

II. The Basics of Multilateralism
I now describe what I regard as the basic features of multilateralism, which both approaches attempt to explain. Exceptions to these features do exist, and some might be debated, but I think that most of us would agree that they nonetheless give an accurate overall description of multilateralism.
1. Tariffs are the instruments of protection; that is, quantitative restrictions are in general banned. 2. Export subsidies are not allowed. 3. Nondiscrimination characterizes trading relations among the partners; in the old language, each partner is a Most-Favored Nation of every other partner. 4. Negotiated agreements feature reciprocity: the mutual exchange of concessions. 5. The countries of the world multilaterally agree to negotiated tariff reductions. 6. The negotiated liberalization is gradual.

Punishments for violations of agreements have consistently been
commensurate with the violation, as have renegotiations of agreements. The objective is to maintain reciprocity.
Several comments are in order. First, properties 1, 2 and 3 are basic principles and, unlike tariffs, not matters whose degree is to be negotiated about. Second, although, as 5 says, the agreements are multilateral, they typically consist of constituent parts each resulting from negotiations among a small number of interested countries. Furthermore, the dispute-settlement mechanism alluded to in 7 is inherently bilateral in its application. Third, negotiations have moved well beyond tariffs to encompass other trade-related measures. But distinctive characteristics of such measures do not figure in the issues discussed below, so I shall not address them. Fourth, for the most part these features also characterize regional, or preferential, trade agreements, provided they are between members of the WTO, which includes most countries. An exception is 6, since such arrangements are supposed to provide for substantially free trade between participants. Also, regarding 3, the arrangements do of course involve discrimination between members and non-members, though not otherwise.

III. An Analytical Framework
A common theoretical framework helps enormously to facilitate comparisons between the SAM and the PCW. The SAM has usually been expressed in terms of formal models, whereas the PCW very often has not. So a natural approach is to express the SAM formally and use that to discuss how the PCW seems to differ.
The standard expressions of the SAM are Staiger (1999, 2002) and Helpman. (1994, 1995). I use a (simplified) version of Grossman and Helpman. There are two countries (Home and Foreign), two factors (Kapital and Labor), and N + 1 traded goods (0, 1, ... , N). Good 0 is a numeráire good, produced by labor alone. Goods 1 to N are produced by capital and labor, with capital specific to each sector. H imports goods 1 to n and exports goods n + 1 to N.
I follow the preponderant part of the political-economy literature in assuming a sector-specific factor. Specific factors, together with the demand separability discussed below, allows the analysis to employ simple partial-equilibrium techniques.
Ownership of each specific factor i is distributed uniformly over a fraction α i of the population (labor force), with each individual owning some of at most one of the specific factors. Let α = α 1 + ... + α N denote the fraction of the population owning specific factors. An exogenously given subset of the α i are organized into separate pressure groups that lobby the government about trade policy.
Choose units so that a unit of good 0 is produced by a unit of labor. Thus, assuming good 0 is actually produced, the wage w = 1.
In each country individual preferences are summarized by the utility function where c i denotes consumption of good i. This implies individual demand where Q i denotes the domestic relative price of good i in terms of good 0. Residual income is all spent on the numeráire good 0. I assume that endowments in both countries are such that each country both produces and consumes good 0. Then an individual's utility can be expressed in the indirect form , the consumer surplus derived from good i, and y denotes the individual's income. Note that, if D i denotes aggregate Home demand (that is, d i summed over all individuals) and V aggregate Home indirect utility (v summed over all individuals), ∂V/∂Q i = D i . Note also that, if R i denotes the income generated by ownership of specific factor i, ∂R i /∂Q i = x i , where x i denotes Home production of good i. Each country may tax or subsidize either imports or exports. For H, let P i denote the international relative price (in terms of the numeráire) of good i, and τ i one plus the ad-valorem trade tax t i . Thus Q i = τ i P i for i = 1, ... , n, and Q j = P j /τ j for j = n + 1, ... , N. Analogous F variables will be distinguished by asterisks.
Equilibrium in the world market for good i, i = 1, ... , n, is represented by where M i and X i * respectively denote H import demand and F export supply. H's import tax and F's export tax thus determine P i , independently of other sectors.
The equilibria in the world markets for goods n + 1 to N can be represented analogously. H imports of goods 1 ... n need not equal in value H exports of goods n + 1, ... , N: Trade balance is reached with a net exchange of good 0.

IV. The Standard Academic Model
Although the SAM emphasizes the role of the terms of trade, it by no means ignores domestic politics sensitive to income distribution. This concern is directed toward relative domestic prices Q i . In the Grossman-Helpman ⓒ Korea Institute for International Economic Policy framework of the previous section, each lobby cares both about its own Q i , which determines the income of its members, and also about other Q i , which influence the real incomes of its members through the cost of their consumption.
A key point is that a sovereign government can always control these relative prices with its own choice of trade policy.
In the model of the previous section, separability ensures that each Q i is determined independently of the others. So, given P i , a government can control each Q i via its choice of τ i . Therefore, given P i , each government can itself fully address domestic political concerns via its choice of τ i .
That choice, though, will influence world prices P i , as long as the country is not small. If other governments care about world prices, to at least some degree, the choice of trade policy by one government affects other governments: a terms-of-trade externality. In the SAM this is the only international externality and, therefore, the only reason for governments to negotiate trade agreements. Given its assumptions, this conclusion of the SAM is unassailable.
Since the SAM implies that a terms-of-trade externality is the only reason for a trade agreement, it does not logically require that governments actually care a lot about the terms of trade. Just that they care to at least some degree. But if that concern is slight, so is the practical importance of trade agreements, raising the question of why they involve so much effort and attract such concern. So, in practice, the SAM does require that governments be much concerned about the terms of trade and about trade-tax revenue.
In the SAM, the basic political economy is the same regardless of whether a government negotiates trade agreements or not. Staiger (1999, 2002) utilize a reduced-form government objective function, with arguments only in Q i and P i , which they assume is invariant to whether a trade agreement is being negotiated. Grossman and Helpman (1995) give a complete analytical model of the politics. Central to their model is a distinction between those interests that are assumed to be politically organized and those that are not. The crucial point, made in the previous section, is that they assume that this distinction is invariant to whether a trade agreement is being negotiated or not.
One might indeed argue that a concern for the terms of trade, or for trade-tax revenue (which is where a tariff-induced terms-of-trade improvement appears) is central not only to the SAM's view of trade agreements, but also to its implicit view of the political economy of protection. A marginal increase in the domestic price of a traded good from its free-trade value will increase the reward to the relevant specific factor by x i and reduce consumer surplus by D i . Then, ⓒ 2013 Journal of East Asian Economic Integration depending on the details of the political economy, if x i is large enough relative to D i a tariff will be forthcoming. This will also raise some tariff revenue. But the tariff will cause an increase in x i and a decrease in D i , increasing even further the political case for protection. What prevents the political process from producing a prohibitive tariff is the fact that tariff increases will eventually cause tariff revenue to fall, and there is presumably sufficient concern about that to prevent a prohibitive tariff.
This logic is clear in Helpman (2004, 2005), whereas Staiger (1999, 2002) employ a reduced-form government objective function that, while consistent with this logic, does not mandate it. In any event, government concern about the terms of trade and about trade-tax revenue is absolutely central to the SAM.

V. Practitioners' Conventional Wisdom
It's much more difficult to give a concise description of the PCW than of the SAM. Adherents of the former are policymakers, negotiators, commentators, and applied economists. People who do not often express themselves through formal models. To be sure, there are formal treatments reflecting to some degree the PCW: Ethier (2004Ethier ( , 2007, Hillman (1982Hillman ( , 1989Hillman ( , 1990, Hillman, Long and Moser (1995), Hillman and Moser (1996). But it is not clear how representative these are of the PCW in general.
One consequence of this is that adherents of the SAM claim that the PCW is really just the SAM in different language. For example, Bagwell and Staiger argue (2002, pp 28, 29) that, "[w]e may interpret 'cost shifting,' 'terms-of-trade gain,' and 'market-access restriction' as three phrases that describe the single economic experience that occurs when the domestic government raises its import tariff and restricts foreign access to its market." These are indeed three effects of a single act, but it is clear that proponents of the PCW do not regard them as synonyms.
To my mind, there are two key distinctions between the SAM and the PCW. The first is that the PCW gives little or no weight to governments' concerns about the terms of trade or about trade-tax revenue 3 . As Regan (2013)  period to the present, there is not one word to suggest that tariff revenue [where a terms-of-trade improvement shows up] played even the slightest role in motivating tariffs. Which means the United States did not engage in terms-of-trade manipulation during the past eighty years." This is a real distinction between the two views. But it does not itself imply that the SAM's conclusion that nothing except a terms-of-trade externality can justify a trade agreement is erroneous from a PCW perspective. This is where the second distinction comes in.
That second distinction is that, at least implicitly, the PCW maintains that the commencement of trade-agreement negotiations changes the nature of the political economy in each country. In my (2004) language, each government conveys a "political externality" on the other when they undertake to reach a trade agreement. In this sense, the Bagwell-Staiger assumption that the government objective function is invariant to whether an agreement is being negotiated or not, and the Grossman-Helpman assumption (mentioned in Section III) that the line-up of organized lobbies is similarly invariant, distinguishes the SAM from the PCW.
If that is so, one must ask why the PCW maintains that negotiations produce this change in the political economy. Here the scarcity of an agreed formal modeling of the PCW is again a problem. Regan (2013) advances a number of informal arguments, which he tellingly labels "squishy," that may in practice explain this. But there is also a more formal argument.
The SAM argument that a government can unilaterally use trade policy to control all domestic relative prices, and therefore address purely political domestic concerns, rests crucially on the ability of the government to subsidize the exports of politically significant export sectors. This is explicit in Helpman (2004, 2005) and implicit in Staiger (1999, 2002).
Basic multilateralism feature 6 (Export subsidies are not allowed) in fact rules this out. But that is not by itself convincing, since it is a property of the multilateralism that we are trying to explain. What does matter here is that countervailing-duty (CVD) laws, providing for tariffs to neutralize export subsidies by trading partners, were in existence before the advent of multilateralism -the original US CVD law was enacted in the 19 th century.
(CVDs are discussed in more detail in section VIII below). The implication is that, in many cases, countries were unable to do anything, with trade policy, to benefit many politically influential exporters. This point was raised by Hillman and Moser (1996), subsequently by Ethier (2008), and argued to be endogenous ⓒ 2013 Journal of East Asian Economic Integration in the political-support approach to protection in Ethier (2011Ethier ( , 2012.
Whether one finds such arguments persuasive or not, they do clearly delineate a distinction between the PCW and the SAM. Once multilateral trade agreements were reached, they included feature 6. This powerfully reinforced, from the PCW perspective, the motivation for further trade agreements. That is not to say that that was the motivation for 6. There are other reasonable explanations: a desire by industrial countries to limit competition between themselves in third countries, for example. But, regardless of the motivation, the effect on the incentive to reach trade agreements for domestic political purposes is clear. At least from the PCW perspective.

VI. Does It Matter?
I have thus far argued that the contrast between the SAM and the PCW is not just a failure to speak the same language, but a substantive disagreement about what the world is really like. One might still ask whether that disagreement actually matters.
The reason for this is that each approach can in fact imply the basic features of multilateralism that I have summarized in Section II above. See Bagwell and Staiger (1999) and Ethier (2004). Each feature is either directly implied by each approach or implied by an argument inconsistent with neither approach. (I abstain here from detail). So, does it matter? I believe that it does. For three reasons. First, Regan (2013) argues that, irrespective of the basic features I have summarized, "There are many particular features of the WTO that the SAM does not explain but that make perfect sense if trade agreements are about reducing protectionism." For example, the failure of the WTO to constrain export taxes. Regan (2006) gives a further discussion. I shall not describe this in detail, but just point out that, regardless of how much one agrees with Regan, he does raise real issues.
A second point is that an understanding of what trade agreements are all about can be critical for their implementation. For example, Regan (2013) points out, "[C]onsider the effect if the WTO Appellate Body accepted the SAM. If they were reviewing some measure for WTO-consistency in a context where the meaning of the treaty was unclear, they could quite properly argue: 'We think this measure was motivated by protectionism. Therefore (as the SAM tells us) it is not the sort of behavior the treaty was designed to restrain. Therefore But it is the third point that most interests me. Trade agreements are agreements between sovereign states: They must be self-enforcing. They are also necessarily incomplete in constraining any country's trade policy in an uncertain future. So individual countries will still be conducting trade policy, either consistently with trade agreements or perhaps not. A multilateral order requires a way of dealing with such actions. But how individual countries so behave is critically sensitive to what the trade agreements in fact try to constrain. Understanding that is absolutely central to having a multilateral order that is in fact feasible.
It really does matter.

VII. Empirical Work
To my knowledge, there is no empirical work testing the SAM against the PCW, perhaps because much of the expression of the latter is not at all formal, as discussed above. But there has been empirical work of potential relevance. Broda, Limão and Weinstein (2008) show that, for a number of non-WTO countries, tariffs are negatively related to foreign export-price elasticities. This is consistent with an optimum-tariff policy, and they conclude that these countries were indeed pursuing such policies. This would give support to the SAM. But the correlation they establish is consistent with other motives, including purely politically-protectionist motives of the sort that concerns the PCW [Ethier (2011[Ethier ( , 2012, Regan (2013)].
Establishing the correlation is indeed valuable, but it tells us nothing at all about either the motivation for actual tariffs or the relative relevance of the SAM and the PCW.
First, these papers focus only on the tariff formula of the model and ignore ⓒ 2013 Journal of East Asian Economic Integration it's clearly counter-factual implications of export subsidies for politically-organized export sectors, export taxes for politically-unorganized export sectors, and import subsidies for politically-unorganized import sectors. The first of these can be justified by the existing constraints from the GATT and from CVD laws, but import subsidies and export taxes, unlike import taxes, are basically unconstrained either by trade agreements or by national laws of trading partners. Furthermore, these predictions directly reflect the key role of trade-tax revenue in the Grossman-Helpman model. For both reasons these predictions furnish the clearest test of the relevance of the Grossman-Helpman model, a test that is dramatically failed. Second, because national tariffs (unlike export taxes) are bound by trade agreements, these studies use calculated tariff-equivalents of administered protection measures. Such measures very often involved importing countries completely surrendering trade rents to exporters. (This is discussed more fully in the next section). 4 But the Grossman-Helpman result hinges crucially on the assumption that importing countries both capture and value highly such rents. If these countries are constrained from doing this, the implications of the theory change greatly.
Thus the theory and the data used to test it are inconsistent. Third, the part of the Grossman-Helpman theory that is apparently supported empirically is the part that it shares in common with the general political-support view of protection that minimizes the role of trade-tax revenue (and so, by implication, of the terms of trade). [Hillman (1989) and Ethier (2011Ethier ( , 2012].
But this literature supplies a basis for the PCW, not the SAM.
In sum, the empirical literature cannot resolve the disconnect between the SAM and the PCW.
But, to the extent that it does contribute, it offers more support for the latter than for the former.
ⓒ Korea Institute for International Economic Policy

VIII. Actual Practice
Countries bound by a trade agreement will frequently be tempted to alter behavior, either by using a recognized opt-out or by deviating from the agreement (it often being a matter of dispute as to which is which).
The SAM implies that a country would like to capture a terms of-trade gain, if it can, and the PCW that it would like to deny market access to foreign firms, if it can.
A conventional tariff will both protect import-competing firms by denying home market access to foreign firms and improve the terms of trade of a country with market power. This is consistent with both the SAM and the PCW and so cannot distinguish between them.
However, with conventional tariffs bound by trade agreements, countries have used other means. And these other means do very often allow a distinction between the alternative explanations.
Consider in detail the means by which parties to trade agreements have undertaken protectionist initiatives. This is most easily done in terms of the following commonly used UNCTAD classification of non-tariff barriers, slightly modified for the present discussion.
1 voluntary export restraints 2 import embargoes/prohibitions 3 minimum import price requirements/undertakings 4 threats 5 anti-dumping investigations 6 countervailing-duty investigations With a voluntary export restraint (VER) a country desiring protection persuaded the exporters of a trading partner to limit exports to it. This was the dominant form of unilaterally-induced protection between 1975 and 2005. VERs violated the GATT stricture against quantitative restrictions, but were tolerated, because no one formally complained, until in 1995 the Uruguay Round moved to phase them out by 2005.
A VER was equivalent to levying a tariff and then turning the tariff revenue over to the exporting nation. The importing nation in effect "purchased" a denial of market access by allowing its terms of trade to worsen. Indeed, the usual effect of such a restraint was to reduce trade-tax revenue, since the imports ⓒ 2013 Journal of East Asian Economic Integration were usually subject to a conventional tariff. Obviously this is much in the spirit of the PCW and very contrary to that of the SAM.
The exporting nation did receive a terms-of-trade improvement. But this was typically not realized in an increase in trade-tax revenue, but instead in the profit margins of exporting forms. And, crucially, the exporting nation of course did not offer access to its own markets in return for the terms-of-trade gain.
These comments apply equally well to measures 2 -6. These also involve surrendering all trade rents to the exporting country to purchase a denial of market access.
7 import authorization requirements 8 health and safety restrictions 9 licence requirements 10 compulsory inspection 11 labeling/marketing/packaging requirements 12 compulsory product characteristics/standards 13 testing requirements 14 quality standards These eight measures all work by imposing real costs on foreign exporters. Thus there are no rents, or, in other words, they are absorbed by real costs. Thus the importing country captures no trade-tax revenue from these measures. The protectionist aspect of these measures is equivalent to the home country imposing a tariff and then throwing the tariff revenue away. (There may well be alternative, non-protectionist, reasons for such measures, but they are not relevant here). And, again, to the extent that the measures are effective, they also reduce the revenue generated by conventional tariffs.
15 advance payment requirements 16 import quotas These two measures generate trade rents that (except in the unlikely event of quota rights being auctioned off) are captured by private agents rather than by trade-tax revenue, as in the Grossman-Helpman model. But neither of these measures has been significant in US trade policy, at least since the ending of oil quotas in the 1960s. An antidumping law provides for a government to levy a tariff on imports from a country whose firms have been "dumping" exports into the home country, that is, selling them at a price below the price charged in the country of origin or below the cost of production. The duty is set equal to the dumping margin.
Unlike VERs, AD is embodied in national laws and is consistent with the GATT and the WTO, provided certain standards are met. AD has in recent decades become the most common form of unilaterally-induced protection.
If an antidumping action is initiated (which also involves 5 above), a duty can often be avoided if exporting firms raise their prices: the basic purpose of the exercise (this might be a case of 2 above). This is a frequent outcome, and the effect of the action, like with a VER, is to allow a denial of market access at the cost of a terms-of-trade deterioration.
Essentially the same result follows if an AD duty is levied that turns out to be prohibitive, which is also often the case. World prices of the imported good may indeed fall, but the importing country gains nothing by that and loses whatever tariff revenue would have otherwise been obtained.
A non-prohibitive AD duty, like any tariff, may improve the terms of trade of a country with market power and allow that country to buy imported goods cheaper. But if exporters do lower their prices, that increases the dumping margin and therefore the AD duty. This often prevents exporters from charging a lower price than before the commencement of the AD action.
The overall result is that an AD action, like a VER, usually (but not always) involves an importing nation "purchasing" a denial of market access with a terms-of-trade deterioration.
Countervailing duties (CVDs) work just like AD, with the exception that the justifying exporter "sin" is not dumping but rather an export subsidy (implicit or explicit) by the exporting country's government. As with AD, individual countries have established their own CVD laws, and these are consistent with the GATT and the WTO if certain restrictions are followed. Conclusions similar to those reached with regard to AD apply here as well.
The general conclusion is that unilaterally-induced protection usually involves purchasing a denial of market access with a loss of trade-tax revenue. Obviously this is very much in the spirit of the PCW and very much contrary to that of the SAM. But caution is called for. One possibility is that the PCW does indeed misrepresent reality, but that it well describes actual behavior simply because policy-makers, mistakenly, believe in the PCW. This would suggest that the PCW is the better positive theory, but a bad normative one.
Another possibility is that the SAM does explain the negotiation of trade agreements, but that subsequent unilateral actions reflect not a desire to "cheat" but rather exogenous surprise shocks in specific countries causing "regrets" that induce those countries to purchase market-access denial with terms-of-trade deteriorations.
But then one faces the embarrassing question of why the exogenous shocks seem to be almost all in one direction rather than the other.

IX. Concluding Remarks
The dominant academic literature maintains that trade agreements are only about dealing with national terms-of-trade manipulation and cannot properly address purely political concerns. Practicising economists, commentators, negotiators, and diplomats by contrast regard terms-of-trade manipulation as of no practical significance and think that trade agreements are all about political concerns. There is no real dialogue between the two groups, and, indeed, academics often seem to think that the practitioners simply do not understand that they are just saying the same thing as the academics in different language.
But there are in fact two substantive and important distinctions between the two views. i Practitioners, unlike the academics, maintain that policymakers care virtually not at all about the terms of trade or about trade-tax revenue. ii Practitioners, again unlike the academics, maintain that trade-agreement negotiations themselves change the underlying political economy.
Empirical work on the political economy of trade policy, though itself of high quality, is both widely misinterpreted and unable to say much about the comparative practical relevance of the two views. Observation of actual trade policy measures, though not conclusive, is much more suggestive. That suggestion is that the practitioners are right and that the academics are wrong.