The puzzle

By all accounts, trade institutions appear eminently well suited for facilitating those violations that lead to more aggregate benefit than aggregate harm. As Trachtman (Reference Trachtman2007: 130) readily admits, WTO law ‘does not normatively demand compliance at all costs’. Countries join the institution out of some form of self-interest, they recognize the uncertainty of the future circumstances they will face, and the resulting incompleteness of the contract they enter into. The possibility of renegotiationFootnote ⁴ – ostensibly as a means of making all trading partners better off, as it requires the acquiescence of all parties involved – is explicitly included in the WTO texts.Footnote ⁵ But since renegotiation, in the sense of putting all issues back on the table (especially under most-favored nation constraints, where any concessions offered to one country must then be extended to all Members) is inherently costly, the ex-ante toleration of adequately compensated breach would seem to follow the law's instrumental aspect by making trading partners better off in the aggregate, and leaving every partner at least as well off as under full compliance.

Moreover, while injury in other domains, such as environmental law, is difficult to grasp and quantify, the losses from breach in trade are readily reducible to monetary terms. This is not to say that the exercise of quantifying injury caused by import barriers is straightforward by any means, as exemplified by the complexity of dispute settlement understanding (DSU) Article 22.6 proceedings over retaliation levels, but it remains, at the very least, conceivable. Note that such quantifiability need not extend to the benefit wrought on the violating state as a result of its breach. From the decision-maker's point of view, that benefit is a political one. While this aspect later becomes an important part of the argument, it need not in and of itself be a barrier to the exercise of efficient breach: compensation, or the transfers required to render affected exporters ‘whole’, is usually assessed on the basis of the injury caused, rather than the benefit that balances it.

Accordingly, a number of scholars have advocated several different means of further accommodating breach and compensation of the affected parties at the WTO. For the most part, these recommendations center on filling the function played by escape clauses. In recognition of the fact that the pressure for trade protection can at times reach peak levels, where compliance with international treaties is no longer politically feasible, most, if not all trade institutions, contain some form of escape clause. Efficient breach advocates argue that the decision to exercise an escape clause or not should hinge on the willingness and ability of the violating state to offer compensation to affected parties (Rosendorff and Milner, Reference Rosendorff and Milner2001; Herzing, Reference Herzing2005; Schwartz and Sykes, Reference Schwartz and Sykes2002; Sykes, Reference Sykes1991). Yet, not only is this not the existing WTO criterion for allowing escape, but the institution has moved progressively further away from relying on such willingness to pay (Pelc, Reference Pelc2009). Indeed, the Agreement on Safeguards now restricts compensation for the first three years of a safeguard measure, the maximum allowable period of which is four years. The required criteria for escape, in turn, have become increasingly precise, and can therefore be challenged in dispute settlement if a country's use of the escape clause fails to meet them. Members must demonstrate that a ‘product is being imported in increased quantities’, and ‘under such conditions as to cause or threaten to cause serious injury’.Footnote ⁶ Notably, none of the criteria put on the exercise of safeguards, or any other sanctioned trade remedy, attempts to quantify the injury caused as a test for the efficiency of a breach, and none requires compensation as a condition for exercise.Footnote ⁷ The very existence of enforced criteria, moreover, speaks against viewing safeguards through the lens of efficient breach, as the latter's use is circumscribed only by the ability and willingness to provide compensation, rather than any demonstration of injury.

Importantly, the changes brought during the Uruguay Round to the Agreement of Safeguards, as compared to its predecessor, GATT Article XIX, did not come out of the blue. Indeed, compensation or retaliation following the use of the escape clause was only prevalent in the first decade of the GATT, when a majority (58%) of escape clauses were followed by some form of voluntary compensation or retaliation. This rate dropped sharply in the following decade, and although overall reliance on escape clauses increased in the 1970s and 1980s, the proportion of compensated or retaliated instances of escape plummeted to around 5% (Pelc, Reference Pelc2009: 357). In other words, with the exception of its first decade, compensation as an expected consequence of exercising the escape clause was never a part of GATT. The Agreement of Safeguards as it is today, which enforces the criterion of ‘unforeseen developments’ that was so hard to enforce during GATT,Footnote ⁸ thus constitutes the logical conclusion of a trend that developed in the institution's early years.

Given how international law does not benefit from a central enforcement authority, there remains the option of going ahead and violating the agreement, and facing the consequences of an admittedly weak enforcement system. Such is the claim of scholars that suggest viewing the entire dispute settlement system, where affected states bring claims of violation, as facilitating efficient breach: ‘[t]he use of the DSP [dispute settlement procedures] therefore allows a contracting partner to violate the agreement, compensate the losers, and still remain within the community of cooperating nations’ (Rosendorff, Reference Rosendorff2005: 390). The first implication of such a view is that a state's cooperativeness can be rescued entirely through the offer of compensation; the second that compensation in dispute settlement occurs automatically, and that the role of the dispute settlement understanding (DSU) is reducible to an objective price-setting mechanism that rules on the amount required to be transferred to the injured party. Yet WTO disputes almost never result in retaliation for a host of reasons, chief of which are the cost of retaliation for the retaliator country itself, and its impracticability for smaller countries.Footnote ⁹ In short, retaliation has occurred in less than 1% of cases at the WTO, and was a non-event during 50 years of GATT history: suspension of concessions was only authorized once during the GATT years, in a case pitting the Netherlands against the United States, but the Netherlands never pulled the trigger. As for voluntary compensation (as opposed to retaliation by the affected party), it has only been observed once since the WTO's inception, in the case of Japan–Alcoholic Beverages.Footnote ¹⁰ WTO arbitration, moreover, has offered support for the view that compensation, if it does take place, is meant to induce a return to compliance, rather than act as a rebalancing mechanism.Footnote ¹¹ And as Pauwelyn observes, ‘not a single WTO member, not even the United States, has ever argued that compensation or retaliation can fully replace performance as an equally available option’ (2007: 72). Overall, then, efficient breach is not a feature of observable state behavior at the WTO. The question is, why?

The state of the debate

Few studies have addressed this puzzle from an empirical standpoint. For the most part, scholars have turned to the question in its prescriptive form by arguing over whether efficient breach should occur. These accounts fall under two broad categories. The first emphasizes the transaction costs involved in the necessary act of compensating affected parties; the second has centered on the effects of efficient breach on compliance rates, and on the existing normative principles it would contravene.

Providing compensation through the abatement of trade barriers in another sector is politically tricky. Under the GATT's Article XIX, compensation through additional liberalization was ‘often difficult to identify’ (Bown, Reference Bown2002: 51).Footnote ¹² As a result, compensation effectively takes the form of authorizing a trade partner to raise import barriers (Friedmann, Reference Friedmann1989; Pelc, Reference Pelc2009), resulting in further trade distortion. Indeed, such retaliation often costs the retaliator as much as it does the country being retaliated against. To put it bluntly, ‘in economic terms, the balancing rationale for retaliation is a fiction’ (Hudec, Reference Hudec2000: 22). It amounts to ‘shooting oneself in the foot’, as the retaliating Member will see the targeted imports increase the costs to consumers and industrial users (Bronckers and van den Broek, Reference Bronckers and van den Broek2005). And while this much is certain, explanations for the non-occurrence of efficient breach that rely on the inefficiency of the transfers among states effectively side-step the issue. If efficient breach is undesirable because it leads to high transaction costs, then arguably it is not truly efficient in the sense of resulting in a Pareto improvement. Moreover, the implication of these arguments is that if the transaction costs resulting from transfers of utility could be reduced – for example, through greater institutional sophistication, as with ex-ante agreements on the levels of compensation (Rosendorff and Milner, Reference Rosendorff and Milner2001) – then efficient breach would find favor among Members. Moreover, some have argued that monetary compensation would offer a venue around distortionary transfers through retaliation and allow for true efficient breach. And yet all proposals for monetary compensation that have been floated at the WTO have been promptly rejected by Members.Footnote ¹³

In institutional settings where it is allowed for, efficient breach can be conceived of as exhibiting an appealing ‘self-enforcing character’, where it occurs if, and only if, the total benefits from breach are greater than the sum of the total harm and the transaction costs of compensating affected parties.Footnote ¹⁴ If in spite of this it does not occur, then ostensibly the breach is not efficient. Ultimately, then, the possibility of true efficient breach, net of all costs of transfer, is an empirical question. In this way, explanations relying solely on transaction costs are not fully satisfying, given how they purport to explain the non-occurrence of efficient breach by ruling out its theoretical possibility.

The second set of explanations for the undesirability of efficient breach from the Members' standpoint relies on normative grounds and legal interpretation of the current WTO texts and their drafting history. Most clearly, Article 3.4 of the DSU explicitly mentions ‘settlement’, rather than any form of compensation, as the aim of dispute settlement. And while compensation could in principle be included under ‘mutually agreeable solutions’ that form the outcome of most disputes (a majority of which never make it to a ruling), Article 3.5 of the DSU explicitly requires that all mutually agreed solutions be ‘consistent’ with countries' WTO obligations under the agreements. Hence, settlements prior to a panel ruling cannot result in any measures that would continue to flout WTO rules. In short, the DSU views any measures short of a return to compliance as ‘fallback measures’ (Jackson, Reference Jackson2004). Jackson goes on to demonstrate how GATT jurisprudence, the WTO Charter, and the DSU texts all point to the withdrawal of measures in contravention of WTO law as the only satisfactory conclusion to a disagreement among parties.

Along similar lines, scholars have looked to the importance of ‘security and predictability’ as one of the central tenets of WTO law, embodied most clearly in Article 3.2 of the DSU.Footnote ¹⁵ The possibility of efficient breach, by virtue of not being contingent on any circumstantial criteria, leads to unpredictable shifts in countries' trade policies. Such unpredictability has been shown to negatively affect trade flows (Mansfield and Reinhardt, Reference Mansfield and Reinhardt2008), since it is perceived by exporters and investors as a tax on imports. Efficient breach, then, is said to fly in the face of the spirit of GATT/WTO law by the way it affects expectations.

Finally, scholars have forcefully argued against efficient breach mechanisms on the basis of their likely impact on compliance rates at the WTO. These scholars claim that allowing for monetary compensation, or facilitating trade compensation, would likely increase the rate of non-compliance, which would be detrimental to the system (Mercurio, Reference Mercurio2009). Pauwelyn uses the clever analogy of an Israeli daycare center to illustrate the effects of monetary compensation on compliance (Pauwelyn, Reference Pauwelyn2006: 92–93). Faced with parents coming late to pick up their children, the daycare center instituted a system of fines for lateness. The unexpected result was that the number of late parents rose, rather than fall. The social stigma associated with lateness had faded; parents now felt they were released from their obligation to be on time, as long as they accepted to pay a fine.

In reality, the link between compliance records and the existence of compensation likely hinges on the amount of compensation required (Rosendorff and Milner, Reference Rosendorff and Milner2001). More importantly, the argument that efficient breach is objectionable because it would lead to greater non-compliance misses the point. The reasoning rests on the assumption that compliance is the objective of the trade system, which is precisely the belief that efficient trade advocates are arguing against. The latter would likely view a rise in non-compliance as, if anything, a demonstration of the prior inefficiency of the system. Once we view efficient breach as Pareto improving, then the more of it there is, the better. If the daycare center had put the fine at an amount greater or equal to the true total cost of keeping the children longer, then a rise in the number of parents coming late would not have be seen as a negative development. It would have, in the view of efficient breach advocates, left all parties better off.

In sum, evidence for Members' rejection of increased reliance on compensation in GATT/WTO negotiations for reform, and in the texts themselves, sharpens the underlying puzzle, but does not provide an explanation for it. Indeed, the existence of rules sharply opposed to efficient breach only defers the causal question. We are left asking, why have these come about? And in the case of norms, why have proposals aimed at facilitating efficient breach not been accepted on the basis of competing norms, such as those invoked by developing countries calling for an end to what ‘amounts to disenfranchisement’ by making compensation mandatory?Footnote ¹⁶ Why, in short, have normative considerations against efficient breach prevailed over existing normative considerations for it?

Domestic politics and efficient breach

This paper argues that the possibility of efficient breach negates some of the main benefits for which countries enter international trade institutions in the first place. This is the reason for which we observe WTO members resisting efficient breach, or mechanisms reminiscent of it.

There is broad consensus over the fact that countries join preferential trade agreements (PTAs) and multilateral trade agreements such as the WTO for two main reasons (Goldstein et al., Reference Goldstein, Rivers and Tomz2007). The first is that through the enforceable reciprocal commitments they lead to, trade agreements alleviate the terms-of-trade prisoner's dilemma that arises if governments set their trade policy unilaterally (Mayer, Reference Mayer1981); the second is that commitments made at the international level allow governments to tie their hands domestically in a way that alleviates the problem of time inconsistency arising from socially optimal trade policy (Staiger and Tabellini, Reference Staiger and Tabellini1987).

The observation that even small countries, with no world market power and thus no means of improving their terms of trade through import duties, nonetheless pursue protectionist policies has led scholars to take a closer look at domestic politics, and at the role institutions may have in helping leaders deal with domestic political pressure from import-competing lobbies. According to this view, even leaders that recognize how free trade is in the interest of the economy will subsequently have reason to give into demands from import-competing groups that can help them retain office.Footnote ¹⁷ Hence leaders' ‘time-inconsistent’ preferences: without some external commitment device, leaders cannot credibly commit to maintaining an open trading regime which they would subsequently profit from changing. In this way, international agreements let leaders ‘tie their hands’ vis-à-vis powerful domestic interests (Reinhardt, Reference Reinhardt and Drezner2001; Hudec, Reference Hudec1987; Staiger and Tabellini, Reference Staiger and Tabellini1987; Maggi and Rodriguez-Clare, Reference Maggi and Rodriguez-Clare2007).

Along similar lines, political scientists have pointed to trade agreements as a solution to a parallel commitment problem vis-à-vis voters. Individual citizens who fear that their leaders will cede to private industry pressure, at the cost of the economy as a whole, will tend to blame these leaders for economic downturns, whether or not their behavior in fact contributed to it (Mansfield et al., Reference Mansfield, Milner and Rosendorff2002). In the event of economic downturns, leaders who have entered into reciprocal commitments over their trade policy through international agreements become less vulnerable to attacks from the opposition about their mismanagement of the economy. In sum, states enter into international agreements, in great part, for domestic reasons.

Understanding the motivations which lead countries to delegate power to institutions in the first place helps us explain the effect that the existence of efficient breach mechanisms would have on the trade system. Key to this argument is the fact that the decision to violate a country's WTO obligations ultimately resides with state leaders, specifically with the legislature or the executive. From the point of view of these decision-makers, moreover, the benefit from offering protection is of an entirely political nature. Whether through voting power (the standard example in the American context being textile and apparel industries), campaign donations, or appeals to national security (the American steel industry comes to mind),Footnote ¹⁸ the benefits that industries confer upon leaders ultimately help these leaders retain power. Conversely, denying protection to interest groups can lead to domestic political costs for policymakers. Especially following exogenous shocks, such as import surges, technological changes, or price disruptions, private industries have a higher incentive to surmount collective action problems and increase their demands for protection. In this way, exogenous shocks act as ‘triggers for self-interested action’ (Baldwin, Reference Baldwin1989: 123), and they can thus make it politically unfeasible for political leaders to keep denying affected industries temporary import relief.

The significance of this issue is reflected in the pervasiveness of different flexibility enhancing devices in trade agreements. Such contingency measures, ranging from safeguards to antidumping and countervailing duties, are a partial solution to the ‘incomplete contract’ nature of these agreements. Exogenous shocks that lead to surges in the demand for protection are precisely the type of event that cannot be foreseen by the drafters of the agreement. But as Downs and Rocke (Reference Downs and Rocke1995: 88) make clear, from the point of view of state leaders, the uncertainty about future circumstances that renders the contract incomplete is ‘uncertainty about the future demands of interest groups’. In other words, political leaders experience the vagaries of the market through changes in the pressure for protection by interest groups.

What the WTO Secretariat calls the institution's ‘architectural challenge’Footnote ¹⁹ comes from the difficult task of differentiating between true exigency, caused by exogenous shocks such as market disruptions, and opportunistic behavior on the part of private industries. The possibility of efficient breach is seen by some as a solution to this problem. It becomes unnecessary to try and assess the ‘legitimacy’ of a country's violation by examining the circumstances that led to it. Instead, the willingness to pay for the breach acts as a selection device, and absolves the country of its violation by leaving all affected parties as well off as they would otherwise have been. But what determines a country's willingness to pay? In making a decision to breach or not, state leaders weigh the costs of doing so (as a measure of the expected cost of dispute settlement action by aggrieved Members, plus any additional informal consequences such as reputational costs) against the benefit they get from offering protection to a domestic industry, or, conversely, the cost they incur from not doing so. Crucially, the advocates of efficient breach make the mistake of treating the components of this calculation as exogenous – that is, as being caused by factors unrelated to the design of the rules. The assumption is that leaders simply deny protection until the costs from doing so grow forbidding, and then they breach and offer compensation. Facilitating breach simply allows this to happen. But in fact, the behavior of domestic interest groups, and thus the costs and benefits involved in leaders' decisions, are themselves strongly determined by whether or not the rules tolerate efficient breach.

Since leaders' costs and benefits flowing from the decision over whether or not to grant protection are of a political nature, they are prone to manipulation. By surmounting collective action problems and increasing lobbying efforts, domestic interest groups can increase the benefits from leaders' points of view of granting protection. Indeed, domestic groups act strategically. And petitioning for protection, whether it is successful or not, always involves costs. If these groups know that formal criteria for legal temporary violations make it unlikely for their demands to be granted, they are not likely to pay the costs of petitioning for protection. Conversely, the possibility of efficient breach effectively means that the odds of getting protection increase with the costs spent in lobbying for it. From the standpoint of domestic industry, then, obtaining protection hinges on driving the political benefits to leaders to outweigh the costs of the violation at the international level. Under an efficient breach scheme, the costs of a violation are fixed, and determined by the injury incurred by trade partners. The benefits of violation, conversely, are variable, and depend on action by domestic groups, which is itself determined by the projected odds of success. Accordingly, the potential for efficient breach is likely to increase domestic pressure for trade barriers.

The end result is thus a reinsertion of power politics at the domestic level. Efficient breach schemes, in other words, have distributional effects: they empower those import-competing interest groups with the ability to muster resources and exert pressure on policymakers, at the cost of smaller, less organized industries. This means both that more powerful industries are likely to succeed in raising trade barriers even when the market conditions they face do not ‘justify’ it (by the standards of legitimate escape as allowed for in WTO law), and that less powerful industries are likely to fail in getting temporary import relief even when market conditions would in fact ‘justify’ it.

The above described ‘architectural’ challenge, of discerning valid state violations from the opportunistic ones, does not therefore go away by virtue of invoking the possibility of compensation, and making temporary breaches contingent on it. The test for validity is simply passed on from the international level (where trade partners and the DSU must determine the validity of trade measures) to the domestic level (where it is now state leaders who must adjudicate between motivated and opportunistic petitions for protection).

Efficient breach thus negates one of the reasons for which countries join international trade institutions such as the WTO. It effectively ‘unties’ the hands of state leaders, thus increasing the returns to domestic industries from lobbying for protection, and empowering the very groups with preferences against trade liberalization.

Interestingly, no part of this argument denies that in any individual instance, efficient breach is the optimal solution. This, in part, is what renders this longstanding debate such a complex one. As its advocates claim, efficient breach does make all actors better off. If it were exercised only once, it would constitute the best course of action for all states involved. That is, if a state values breach more than the negative effects of the breach on affected parties, then that state could indeed transfer some of the utility to its trading partners, leaving them whole, and be left better off than if it had remained in compliance. The problem is in the externalities resulting from the constant possibility of efficient breach. If we endogenize the likelihood of domestic demands for protection, and if we allow that domestic groups strategic behave strategically, then the possibility of efficient breach has a dynamic effect. Powerful domestic groups adjust their expected likelihood of being granted protection, and behave accordingly, by surmounting collective action problems, paying the costs of lobbying, and ramping up their demands. Overall, then, efficient breach increases pressure for protection.