The Review of International Organizations

, Volume 4, Issue 4, pp 329–359

Political-economic problems in trade capacity building


    • Department of Political ScienceColumbia University

DOI: 10.1007/s11558-009-9068-3

Cite this article as:
Urpelainen, J. Rev Int Organ (2009) 4: 329. doi:10.1007/s11558-009-9068-3


Theoretically, trade capacity building should contribute to export-led growth and support liberal economic policies. Unfortunately, it often fails to meet this ideal due to resource misallocation, misplaced focus on existing obligations, and donor-driven implementation. This article presents a formal theory of political-economic problems in trade capacity building. I analyze trade liberalization as a repeated game with imperfect public monitoring between a developed and developing country. Modeling trade capacity building as an investment by the developed country, I show that it suffers from two problems. First, the need to enforce trade liberalization drives resource misallocation: costly projects are implemented only to build commitment capacity while others are not implemented because they encourage protectionism. Second, donor interests distort trade capacity building. Counterintuitively, if the donor can seek compensation from the recipient when it violates international trade law, it sometimes refuses to invest in low-cost trade capacity building while funding projects that hurt the recipient.


Trade capacity buildingDevelopment assistanceInternational cooperation

JEL Codes


1 Introduction

The triumph of export-led development strategies in rapidly industrializing countries and the widespread frustration with the record of conventional development aid have drawn attention to the opportunities that an open international economy offers to developing countries. On this agenda, liberal trade policies occupy a pride of place for three reasons. First, developing countries have themselves shown willingness to open their economies. Milner (1999, 91) writes, for example, that “[o]ne of the most salient changes in the world economy since 1980 has been the move toward freer trade among countries across the globe.” Second, the vast majority of development economists agree that trade liberalization is a key to robust economic growth (e.g., World Bank 2006). Finally, trade liberalization creates the right set of incentives for political and economic actors in developing countries to guide their efforts from rent-seeking activities to wealth creation (Bhagwati 1982; Krueger 1974).

Yet many developing countries have experienced difficulties in reaping the gains of trade liberalization. On the one hand, they lack the economic infrastructure to exploit their competitive advantage. For countries that have few paved roads, ports, and airports, the cost of exportation is prohibitively high. On the other hand, many developing countries also lack the the institutional capacity to engage in international trade. Without effective customs procedures, legal expertise, and laboratories for quality control, it is difficult to comply with regulations and rules set by major importers. Simply put, the prevailing economic and institutional environment in these countries is not conducive to export-oriented production.

The high economic cost of these impediments has not gone unnoticed. There has been a surge in “trade capacity building,” defined as “the reform of trade policies and the strengthening of trade-related institutions and infrastructure” by the World Bank (2006). Multilaterally and bilaterally, donors now give billions of dollars to fund projects that improve the infrastructure necessary for international trade and enhance legal and administrative capacity in the recipient country. Trade capacity building is highly attractive because it fits perfectly the idea of development aid as “help for self-help.”

The current practice of trade capacity building has been criticized for many reasons. The criticisms range from inadequate resources to misallocation and donor-driven implementation that fails to generate local ownership (Prowse 2002; Shaffer 2005). For instance, examining assistance for implementing the Uruguay Round commitments in developing countries, Finger and Schuler (2000, 522) argue the donors have adopted a “do it my way!” strategy with limited benefits for the recipients. Suwa-Eisenmann and Verdier (2007, 486) highlight the pernicious effect of tying aid to purchasing goods from the donor country. They also note that “no study so far has convincingly evaluated the direct impact of trade assistance on the reduction of the trade transaction costs” (Suwa-Eisenmann and Verdier 2007, 502). Finally, Brazys (2009) shows that the primary determinant of U.S. trade capacity building is donor interest instead of recipient need.

What obstacles exist to effective trade capacity building? This article presents a formal theory of political-economic problems in trade capacity building. I construct a simple repeated game of international trade liberalization in which the traders are a developing country and a developed country. I assume that both countries have an incentive to defect by imposing trade policies that protect domestic interest groups from competition (Bagwell and Staiger 1999; Downs and Rocke 1995; Rosendorff 2005). Before trading begins, the developed country can invest in infrastructural and institutional trade capacity to increase both parties’ payoff from trade. Institutional capacity also reduces the probability that the developing country fails to implement liberal trade policies. This analytical distinction between infrastructural and institutional trade capacity is the first key innovation of this article.

Another key innovation is to study the relationship between trade capacity building and the rules of the international trade agreement that states have chosen to enforce liberalization. The key premise is that trade capacity building occurs in the shadow of the international trade agreement. Many factors shape the design and implementation of any given project “on the ground,” but a self-interested donor chooses the broad contours of trade capacity building so as to promote its strategic aims in trade policy. I compare “symmetric” trade agreements, in which protectionism prompts a temporary increase in tariffs and other barriers to trade on both sides, with “asymmetric” trade agreements, in which the defector must compensate the victim to restore cooperation (Downs and Rocke 1995; Ludema 2001; Rosendorff 2005). This distinction proves to have important consequences for trade capacity building. In symmetric agreements, both countries have incentives to avoid a trade dispute, but in asymmetric agreements the developed country can actually benefit from an opportunity to obtain compensation for a perceived implementation failure.

What problems does the need to enforce trade liberalization cause for trade capacity building? In addition to the straightforward problem of diverging donor and recipient interests, the analysis shows that enforcement problems and perverse incentives to seek compensation systematically skew the allocation of resources for trade capacity building. I present three specific analytical results.

First, trade capacity building could increase the incentive to defect and therefore complicate enforcement. While trade capacity building increases the payoff to liberalization, it could also increase the payoff to protectionist policies. For example, training competent bureaucrats facilitates removal of quality non-tariff barriers to trade, such as discriminatory health standards, but it also allows sophisticated legal strategies of discrimination that help the government cater to organized special interests.

Second, under an asymmetric agreement, a donor could refuse to build institutional trade capacity even at a negligible cost. This surprising possibility stems from the enforcement mechanism that is based on compensation for a perceived implementation failure. If the developing country has limited institutional capacity, it frequently fails to meet is obligations. If the resulting compensation is profitable enough for the donor, it benefits from ensuring that the recipient fails to implement trade policies. For example, if corrupt customs officials discriminate against imports and a dispute resolution mechanism punishes the developing country for this, the donor can request monetary compensation or sanction the recipient. What is more, limited commitment capacity can actually benefit the developing country if the only way to sustain trade liberalization is to reduce the risk of implementation failure through institutional trade capacity building.

Finally, I find the conditions under which infrastructural support and institutional capacity are political complements. With adequate institutional capacity in place, the cost of not providing infrastructural support is highest when trade capacity building also raises the payoff from protectionism. This finding helps understand why investment in institutional capacity to fulfill multilateral contractual obligations is sometimes accompanied by a startling neglect of infrastructural support. If the developed country is self-seeking but must ensure compliance by the developing country, the cost of not investing in infrastructure is particularly striking. For example, adequate training for trade policy officials is futile unless the developing country has a transportation system that allows exportation and importation.

The analysis has important implications for international institutions and organizations. First, while states must invest resources in building commitment capacity so as to enable international cooperation, they must also sometimes refrain from developing seemingly useful capacities that complicate enforcement. As a yardstick for successful capacity building in international cooperation, policy effectiveness can be highly misleading unless the effect on compliance is accounted for. This observation is a significant caveat to “managerialist” theories of international cooperation that emphasize the virtues of capacity building for compliance (Chayes and Chayes 1995). While capacity building is often insufficient to remove incentives to defect, the results show that it could even create them.

Second, the model adds a new dimension to the problem of international treaty design (Koremenos et al. 2001). In the context of cooperation among unequals, it is not enough to consider the credibility of the policies designed to support cooperation. If the treaty is designed so that the incentives to build compliance capacity are distorted, the cost could be high for both parties. By designing agreements under which both powerful and weak states are willing to build capacity so as to improve equilibrium policies, states could reap substantial benefits. It is particularly important to avoid the possibility of “legitimate” exploitation of weak states by powerful states. If international law prescribes too lucrative compensation for compliance failure, powerful states could deliberately refuse to build compliance capacity. To my knowledge, this problem of creating incentives for capacity building has not been previously recognized in international cooperation theory.

I first discuss trade capacity building and then introduce the game-theoretic model. I conduct the formal analysis and present the results before offering concluding remarks.

2 Trade Capacity Building

The concept of trade capacity building, sometimes referred to as “aid for trade” (Suwa-Eisenmann and Verdier 2007), comprises services that reduce the transaction costs of international trade (Prowse 2002). Both bilateral and multilateral donors invest in trade capacity building. Trade-related assistance documented in the Joint WTO/OECD Capacity Building Database reached $15 billion in 2005 (WTO/OECD 2007). Developed countries allocate a substantial share of their bilateral foreign aid to trade capacity building. For example, the United States has increased its aid for trade from approximately $296 million to $1,212 million between 1999 and 2007.1 The European Commission has pledged that European aid for trade reach two billion euros by 2010, half of which would be commissioned by the European Union and the rest by the Member States. The World Bank has contributed $38 billion to trade capacity building between 1987 and 2004, most of it recently, and the WTO has a capacity building program that focuses on institutional capacity (World Bank 2006; WTO/OECD 2007).

Trade capacity building as a key component of development aid has its basis in two desiderata. First, trade capacity building is a natural strategy to create export-led growth in developing countries (Hoekman 2002; Prowse 2002; Suwa-Eisenmann and Verdier 2007). Second, trade capacity building accords with the notion that development aid cannot lead to lasting results unless it provides the right set of incentives for the receiving country (Easterly 2002). In principle, trade capacity building combines the virtues of conditionality with a benign approach to development that does not require excessive sacrifices by the domestic population in the short term.

A quick review of the current level of trade capacity in least developed countries also suggests that the economic returns to improving trade capacity could be high. Limão and Venables (2001) estimate the impact of inadequate infrastructure on transport costs and find that adequate infrastructure is the most important explanation for the low levels of trade that poor African economies engage in. Suwa-Eisenmann and Verdier (2007) review this and other recent studies to confirm this finding. Henson and Loader (2001) report that most developing countries find their inability to comply with environmental and health regulations the single most important obstacle to increasing exports.

Trade capacity can be grouped into two categories. First, the most important and expensive form is infrastructural capacity (Blackhurst et al. 2000; Limão and Venables 2001). It covers paved roads for transportation, ports and airports, telecommunication, and facilities such as sewage, electricity, and water distribution. Between 2000 and 2004, roughly one fourth of the $5.1 billion in credit for trade capacity building given by the World Bank was allocated to infrastructural support (World Bank 2006). The United States allocated less than one fifth of its aid for trade to physical infrastructure in 1999 and more than two fifths in 2007.2 Projects documented in the aforementioned WTO/OECD database were valued at more than $12 billion in 2005, although it should be noted that much of this infrastructural support only facilitates trade indirectly (WTO/OECD 2007).

Second, institutional capacity comprises customs operations, compliance with environmental and health regulations in the export markets, the implementation of international trade agreements, and competition policy. Most of the credit not allocated to infrastructural support falls under institutional capacity. Of the donors introduced above, both the World Bank and the United States have allocated more than one half of their resources to institutional capacity (World Bank 2006).3

The difference between infrastructural and institutional capacity is essential for my analysis. Improvements in infrastructural capacity lay the physical and technological foundation for profitable exportation and importation of goods and services. This increases the payoff to international trade, but it does not directly influence the difficulty or cost of implementing various trade policies. Consequently, it only influences the incentive to liberalize through changes in the costs and benefits of international trade. In contrast, institutional capacity also enables effective implementation of trade policies. To be sure, tariff removals do not require institutional capacity, but de facto international trade liberalization requires greatly more than that, including the removal of core and quality non-tariff barriers to trade (Kono 2006; Sykes 1995). Thus, institutional capacity increases the payoff to liberalization both by reducing the transaction costs to economic actors and facilitating policy implementation.

The indirect effect of improved institutional capacity is important because it allows the donor to better monitor the recipient. If the recipient has high institutional capacity, the donor understands that failure to liberalize stems from an intentional defection. But if the recipient has low institutional capacity, the donor cannot be sure if the recipient did not implement a trade policy because the government is protectionist or because it really tried but nevertheless failed. In building institutional capacity, the donor must therefore consider the political-economic implications of improved monitoring.

Although trade capacity building has potential as an engine of development, the record of existing measures highlights several deficiencies that are relevant for my argument. First, trade capacity building has drawn criticism for misallocation of resources, by which I refer to failure to target available resources so that they maximize economic development in the recipient country. This criticism applies both to the broad contours of trade capacity building and specific projects. For example, Finger and Schuler (2000) provide project-specific evidence on this pattern of misallocation, which they relate to a strong emphasis on implementing the Uruguay Round commitments (see also Shaffer 2005). They find that the cost of implementing those three Uruguay agreements that require domestic regulatory reform climbs to more than $150 million, which is more than the annual development budget of eight of the twelve least developed countries for which they found data. Prowse (2002, 1239) criticizes the lack of coordination among international organizations, suggesting that aid for trade “has been delivered frequently randomly, indiscriminately and more often than not on a stand-alone basis.”

A related criticism of trade capacity building is its donor-driven nature and lack of ownership. The heavy emphasis on WTO obligations examined by Shaffer (2005) suggest that donors have often used trade capacity building to ensure that least developed countries are capable of complying with rules set mostly by developed countries (Steinberg 2002). Finger and Schuler (2000) conclude that the “do it my way!” strategy adopted by donors in ensuring that developing countries fulfill their Uruguay Round obligations results in haphazard and forced implementation of projects without considering the recipient’s needs and resources. In general, the conditions attached to trade capacity building that artificially boost exports from the donor to the recipient suffer from similar problems (Suwa-Eisenmann and Verdier 2007, 486).

3 The Model

The model represents trade between countries i = 1,2 that sign an international agreement for trade liberalization. Country 1 is a developing country with limited trade capacity and country 2 is a developed country that has adequate trade capacity. Following Downs and Rocke (1995) and Rosendorff (2005), I consider a repeated game in which the governments can opportunistically impose “politically optimal trade policies” such as tariffs and quotas (Bagwell and Staiger 1999). During the game, the countries can enforce trade liberalization through reciprocity: if a country defects, the other country suspends cooperation (Keohane 1986). Most importantly, I add a pre-game stage at which the developed country can invest in trade capacity building to help the developing country.

I do not model the trading economy in any detail (see Bagwell and Staiger 2005; Rosendorff 2005). Instead, I impose a series of plausible assumptions on the parameters of the repeated game. This approach has three advantages. First, it is general because the results do not depend on any particular model of international trade (see Feenstra 2004). Second, most conventional models reduce trade capacity to transport costs, so it is not obvious how to model it. Finally, the approach is simple and saves notation.4

To begin with, suppose each country i can implement trade policies for all traded goods at each time t. These policies can be tariffs, quotas, or discriminatory regulation. I normalize the payoff from the one-stage Nash equilibrium to 0. The countries aim for (possibly partial) trade liberalization, so let αi > 0 denote the payoff to country i when both countries comply with the international agreement. To capture the possibility of defection, let βi > αi denote the payoff from politically optimal policies when the other country complies. Similarly, let γi < 0 denote the payoff from liberalization when the other country imposes its politically optimal policies. Denoting liberalization by C (cooperate) and the optimal deviation by D (defect), one can construct a simple two-player Prisoners’ Dilemma. Figure 1 illustrates.
Fig. 1

Trade liberalization as a stage game

These payoffs are compatible with conventional terms-of-trade tariffs for large economies (Johnson 1953), the “protection for sale” model in Grossman and Helpman (1994), and a wide variety of other political-economic models based on relative prices (Bagwell and Staiger 1999). This generality is important because many least developed countries have no market power and thus cannot change the terms of trade to their advantage. Politically, however, it is plausible that governments seek domestic redistribution through trade policies even in a small economy.

For notational convenience, let δ ∈ (0,1) represent a common discount factor. The payoff from the repeated game is
$$ U_{i}=\sum\limits_{t=0}^{\infty}\delta^{t}u_{i}\left(a_{1}^{t},a_{2}^{t}\right), $$
where \(a_{i}^{t}\) is the action by player i at time t and ui is the one-stage payoff. Where convenient, I use the average per-period payoff (1 − δUi instead. Below, I introduce uncertainty so that the payoff is probabilistic.

An important element of the model is information. If a country fails to liberalize, for example, it could be due to defection or implementation failure (Downs and Rocke 1995). Similarly, a country could fail to defect, despite attempts to do so, because of bureaucratic inertia. Such problems are particularly acute in international trade with developing countries because they are vulnerable to internal and external exogenous shocks and possess limited trade capacity (Hoekman 2002; Krasner 1985).

For present purposes, it is useful to model this signal structure as a game of “imperfect public monitoring” (Mailath and Samuelson 2006). In games of imperfect public monitoring, the players receive a common signal that is incorrect with some probability. For example, if both payers cooperate (C,C) it is nevertheless possible that they “see” the developing country defecting (D,C). Formally, I introduce two probability measures. First, pD|C is the probability of signal (D,C) when they play (C,C) or (C,D). Substantively, this is the probability that the developing country fails to liberalize despite an attempt to do so. Second, pD|D is the probability of signal (D,C) when they indeed play (D,C) or (D,D). This is the probability that the developing country is “caught” by the developed country upon defection. Clearly, one has to assume pD|C < pD|D so that defection increases the likelihood of a “bad” signal. Figure 2 summarizes the information structure.
Fig. 2

The information structure. The first column “Play” indicates the true actions chosen by players i = 1,2 and the other columns show the probability Pr(·) of different public signals conditional on this play. The zero-probability cells are hypothetical incorrect signals on the action chosen by the developed country, which is not possible in the model because the developed country has adequate trade capacity

Note that the incorrect signals pD|C (cooperation incorrectly seen as defection) and (1 − pD|D) (defection incorrectly seen as cooperation) pertain to the action chosen by the developing country. This asymmetric modeling technique, which resembles that adopted by Svolik (2006), is natural when the developed country has adequate trade capacity. It does not capture the possibility of domestic hardship in the developed country (Downs and Rocke 1995; Rosendorff 2005), so it allows a sharp focus on trade capacity. Moreover, for notational convenience I have assumed that the probability of an incorrect signal is independent of the trade policies enacted by the developed country. As I show below, this assumption is not required to obtain the key results.

Given imperfect public monitoring, the players cannot condition their future behavior on the true history of the game. Instead, they must use the signals that they observe.5 A strategy for player i thus consists of a selection rule for each time t that depends on the t − 1 signals that they have seen. Empirically, the signals correspond to perceived (possibly incorrect) cooperation and defection in the past by both countries, while the strategies si correspond to the choice between cooperation and defection at any given time.

A note on the binary {C,D} action space is in order. It omits the complexity of choosing trade policies when the signal varies continuously with the distance between the chosen policy and the provisions of the international agreement. A more realistic representation in continuous strategies would enable “trigger” strategies studied by Green and Porter (1984). Unfortunately, continuous strategies complicate the analysis because the effect of trade capacity building would have to be defined on functions, whereas it can be defined on scalars in the binary model.

3.1 Trade Capacity Building

I incorporate trade capacity building in the model by introducing a pre-game decision-theoretic problem, in which the developed country can invest to improve trade capacity in the developing country. The analysis of trade capacity building thus amounts to a formal derivation of the “willingness to pay” for trade capacity by the donor.

As I have argued above, trade capacity can be roughly divided into two categories. First, I model infrastructural capacity building as a vector \(x^{I}=(x_{1}^{I},...,x_{m}^{I})\) of possible projects and assume αi and βi are increasing and strictly concave in each \(x_{j}^{I}\). Infrastructural capacity increases the payoff to trade liberalization and the politically optimal policies, but the benefits are decreasing on the margin. Importantly, the developed country can fund the projects separately. Moreover, improved infrastructure is useful for both countries whether they cooperate or defect. Improved infrastructure facilitates imports and exports, and no a priori reason exists to assume that this capacity would not increase the effectiveness of politically optimal policies. The capacity also helps the developed country because it can block imports and simultaneously exploit the developing country infrastructure to reduce the price of its exports.

Second, improved institutional capacity both increases the payoff from trade and reduces the probability of incorrect signals. As with infrastructure, the government needs institutional capacity whether it cooperates or defects. But as discussed above, improved institutional capacity also implies that the developing country cannot use lacking capacity as an excuse for discriminatory policies. Suppose the developed country observes an illegal trade policy, such as unnecessarily complex customs procedures or discriminatory health standards. If the developing country has high institutional capacity, the developed country correctly infers that the failure to liberalize almost certainly results from opportunism (low pD|C), as the government could have easily removed these policies. With low institutional capacity, these policies could result from incompetent or corrupt bureaucrats that the government cannot control. Similarly, if the developed country observes compliance, it assigns a low likelihood on failure to defect despite willingness to do so (high pD|D).

Formally, let \(x^{A}=(x_{1}^{A},...,x_{n}^{A})\) represent projects that enhance institutional capacity. The payoffs αi and βi are increasing and strictly concave in each \(x_{j}^{A}\). Probability pD|C is decreasing and strictly convex in each \(x_{j}^{A}\) and probability pD|D is increasing and strictly concave in each \(x_{j}^{A}\). These assumptions imply that the probability of an incorrect signal decreases as institutional capacity increases.

Infrastructural and institutional capacity interact. It is useful to exclude the possibility that the two capacities were substitutes, because it is implausible that improved infrastructure would reduce the need to build institutional capacity, and vice versa. For example, even the most advanced transportation systems are virtually useless in promoting agricultural exports unless the developing country can comply with the sanitary and phytosanitary standards set by the developed country (Henson and Loader 2001). Although domestic producers could transport goods to the border, they could not market them without permission to do so by the officials of the developed country. I therefore assume that αi and βi exhibit weakly increasing differences with respect to any two coordinates (xj,xk). To save notation, I write x = (xA,xI).

Note that trade capacity building does not affect the one-stage Nash equilibrium payoff (0) or the payoff when the other country defects (γi). It is natural to assume that when the states limit trading, improved trade capacity does not affect the payoff. When the other country defects, it is also plausible that the benefits from improved trade capacity are minimal.

I have also omitted the analysis of litigation by the developing country (Bown 2004; Bown 2005; Busch and Reinhardt 2003). This form of trade capacity is important, but it is not analytically comparable to trade capacity building because it has no direct bearing on the payoff from liberalization per se. In this article, I focus on trade capacity that is directly related to trade.

Finally, if trade capacity building was free, the developed country might invest any amount of resources. Instead, I assume that the developed country must pay a “fee” \(z=z\left(\sum x_{j}\right)\), where z is strictly increasing and convex, so that the cost is weakly increasing on the margin. Even though this fee usually corresponds to the monetary cost of trade capacity building, the formal analysis can be extended to other costs. For example, one could consider the political difficulty of securing domestic support for implementing institutional reform in a developing country.

3.2 International Trade Agreements

I analyze two possible agreements designed to sustain trade liberalization (C,C) in equilibrium.6 The key assumption is that the countries must commit to the form of the agreement before trade capacity building. For example, the countries cannot rewrite international trade law after trade capacity building or redesign the WTO dispute resolution mechanism. However, within these constraints they can modify the details of the agreement, which in this article pertain to the severity and likelihood of punishments upon defection. The model has only two players, so it is intuitive to think of the resulting equilibria as preferential trading agreements. However, the key substantive implications of the analysis also offer insight into the effect of the GATT/WTO multilateral trade regime.

First, a symmetric agreement involves a temporary or permanent reversion to the Nash equilibrium upon any other signal than (C,C). Specifically, both players cooperate as long as the signal has been (C,C) in the past. If the developing country defects, they revert to the one-stage Nash equilibrium (D,D) for T periods (Bagwell and Staiger 2005; Downs and Rocke 1995; Rosendorff 2005). Afterwards, they clear past deviations and start anew with (C,C). This agreement could either have a dispute resolution mechanism that authorizes retaliatory tariffs but no compensation, or simply contain a clause that permits a temporary suspension of cooperation upon defection. Notably, this suspension need not be a total trade war that effects a reversion to autarky. It is equally possible to apply the model to a single industry, or assume that the politically optimal tariffs are not close to being prohibitive.

Second, an asymmetric agreement contains a “penal code” requiring that a defector compensate the victim (Abreu 1988). In particular, both players cooperate as long as the signal has been (C,C) in the past. If the developing country defects, they play (C,D) for T periods. This asymmetric punishment is costly for the defector but profitable for the victim. Any deviation during the punishment period restarts it. Intuitively, an asymmetric agreement contains a compensation scheme according to which the victim is allowed to retaliate while the defector must comply to restore cooperation. If the developed country defects, both players permanently revert to the one-stage Nash equilibrium.

In both agreements, if the developed country defects, both countries revert to the one-stage Nash equilibrium permanently. This is but an innocuous technical simplification: given that the behavior of the developed country is perfectly observable, it never defects in equilibrium so the severity of the punishment does not affect its payoff. The key difference between the two agreements is that in the asymmetric agreement, incorrect signals are not unambiguously detrimental for the developed country. If the compensation scheme overvalues the damage caused by defection, the developed country could gain from an incorrect signal.

The two international agreements that I consider are ideal types chosen to shed light on the relationship between enforcement and trade capacity building. Real international trade agreements are more complex. On the one hand, they often deliberately limit retaliation or compensation and contain escape clauses that states can use under domestic hardship (Rosendorff 2005). Such agreements fall somewhere between symmetry and asymmetry, but the analysis of limited compensation or flexibility does not yield important insights into political-economic problems in trade capacity building, so I omit it. On the other hand, to build enforcement power, states could retaliate defections by suspending cooperation in other issue areas as well. This extension is not germane to trade capacity building, so I omit it too.

Please note that the argument does not require explicit provisions for trade capacity building in the agreement, or even implicit expectations thereof in the negotiations. The analysis therefore applies both to international trade negotiations in which trade capacity building is discussed, such as the Doha Round, and those in which it is ignored. What is important is that trade capacity building occurs in the shadow of the international trade agreement. Many factors determine what goes into trade capacity building, but it should not undermine a formal treaty between the two governments. While the implementation of any given project “on the ground” might not have a direct link to the international trade agreement, the self-interested donor does not continue aid for trade if it consistently contradicts its primary strategic goal of trade cooperation.

The argument also does not require equal or even comparable retaliation capabilities between the donor and the recipient. My formulation is not intended to literally represent a “trade war” but to highlight the contrast between enforcement strategies under which both parties prefer to avoid trade disputes (symmetric) and those under which one of the parties benefits from compensation (asymmetric). The recipient is a least developed country that cannot unilaterally affect the terms of trade, so it suffers much more from a trade dispute than the donor does. Empirically, it is most plausible that donor-recipient trade suspensions are quite limited and serve to discipline the developing rather than the developed country. Asymmetric retaliation capabilities also imply that the international trade agreement must closely reflect donor interests because it simply refuses to comply otherwise. This power asymmetry explains why the recipient must often accept an unequal agreement that prescribes compensation to the donor upon perceived compliance failure.

I ignore the possibility that trade capacity building directly influences bargaining over the economic terms of trade liberalization. In reality, it is possible that trade capacity building has significant implications for optimal tariff rates or other trade policies. But to rigorously examine such interactions, it would be necessary to explicitly model the two economies. To keep the model as simple as possible, I only consider the effect of trade capacity building on how the agreement is enforced.

To build intuition, consider some examples. The Dispute Settlement Understanding of the WTO is an illustrative example because it falls somewhere between the two ideal agreement types. On the one hand, it contains important elements of asymmetry. A state that has violated international trade law can restore cooperation by compensating the victims (Rosendorff 2005). Empirical research shows that developing countries with limited market power and legal capacity are at a disadvantage in WTO dispute resolution (Bown 2004; Busch and Reinhardt 2003). On the other hand, WTO dispute resolution is based on the legal principle that compensation should be proportional to the damage caused. This principle constrains the magnitude of compensation, and in practice the prescribed compensation usually falls short of this ideal, as Spamann (2006) shows.

Many preferential trading agreements contain elements of asymmetry. The North American Free Trade Agreement has an arbitration tribunal that can prescribe compensation for perceived violations of international trade law. McCall Smith (2000, 157) shows that many trade agreements, such as the 1969 Andean Pact and the 1960 Central American Common Market, have binding dispute resolution mechanisms and standing tribunals that can prescribe compensation for non-compliance. Unique among these agreements is the European Community, in which dispute resolution under the rules of the single market is fully legalized and reciprocal suspension of cooperation is rarely necessary to enforce compliance.

In contrast, the various preferential trading agreements that the European Free Trade Association negotiated with various Eastern European countries and Turkey after the collapse of the Soviet bloc have elements of symmetry. As data collected by McCall Smith (2000, 156) show, these agreements with no dispute resolution mechanism or provisions for sanctions are quite common even when the trade partners are “economically asymmetric” based on their gross domestic products. Under these agreements, states can enforce compliance by threatening a perceived defector with suspension of cooperation either in trade or other issue areas.

4 Analysis

The analysis proceeds in two parts. First, I find the set of payoffs that are sustainable as a subgame-perfect Nash equilibrium of the repeated game. Second, I analyze the pre-game decision-making problem. The payoffs from the repeated game now depend on the parameters, which are in turn determined by trade capacity building. Consequently, the developed country can use trade capacity building to shape the Pareto-frontier of the repeated game. I assume that the developed country maximizes its own payoff (e.g. Bueno de Mesquita et al. 2003).7

4.1 Symmetric Trade Agreements

Consider first a symmetric trade agreement. Given that the punishment—the one-stage Nash equilibrium—is trivially self-enforcing, it is enough to ensure that neither player can profitably deviate when they are supposed to cooperate. To do this, let \(V_{G}^{i}\) denote the average future payoff to player i at any time t when they obtain the signal (C,C). Also let \(V_{B}^{i}\) denote the average future payoff to player i when they obtain signal (D,C). For any other signal, the average payoff is clearly 0. Substantively, these payoffs correspond to the expected gains from trade liberalization in the future as a function of whether the players perceived a violation.

To enforce trade liberalization, neither player should have an incentive to defect. If the developing country defects, the immediate payoff increases from αi to βi but the likelihood of a trade war increases. Defection is thus not profitable if and only if
$$ \begin{array}{lll} V_{G}^{1}&=&(1-\delta)\alpha_{1}+\delta\left((1-p_{D\mid C})V_{G}^{1}+p_{D\mid C}V_{B}^{1}\right)\\[3pt] &\geq&(1-\delta)\beta_{1}+\delta\left((1-p_{D\mid D})V_{G}^{1}+p_{D\mid D}V_{B}^{1}\right).\label{eq:2} \end{array} $$
To the contrary, if the developed country defects, it triggers a permanent trade war with certainty:
$$ V_{G}^{2}=(1-\delta)\alpha_{2}+\delta\left((1-p_{D\mid C})V_{G}^{2}+p_{D\mid C}V_{B}^{2}\right)\geq(1-\delta)\beta_{2}.\label{eq:3} $$
If these conditions hold, neither player expects to gain from defection because the threat of a future trade war is enough to deter opportunism. The conditions reduce to
$$ V_{G}^{1}-V_{B}^{1}\geq\frac{1-\delta}{\delta}\frac{1}{p_{D\mid D}-p_{D\mid C}}(\beta_{1}-\alpha_{1})\label{eq:4} $$
for the developing country and
$$ (1-p_{D\mid C})V_{G}^{2}+p_{D\mid C}V_{B}^{2}\geq\frac{1-\delta}{\delta}(\beta_{2}-\alpha_{2})\label{eq:5} $$
for the developed country. Intuitively, the difference between the expected gains from trade liberalization and the trade war must be large enough to outweigh the immediate payoff from politically optimal policies.
To maximize the payoff from the repeated game, the players need to use the least costly punishment that renders trade liberalization self-enforcing. Under imperfect public monitoring a trade war occurs with positive probability each period, so the cost of an incorrect signal should be as low as possible. In the Appendix, I derive the maximal payoffs for each state, assuming that trade liberalization is feasible:
$$ V_{G}^{1max}=(1-\lambda)\alpha_{1} $$
$$ V_{G}^{2max}=(1-\lambda)\alpha_{2}, $$
where \(\lambda=\frac{p_{D\mid C}}{p_{D\mid D}-p_{D\mid C}}\frac{\beta_{1}-\alpha_{1}}{\alpha_{1}}\). Importantly, a symmetric agreement aligns the countries’ interests. In each period, they either obtain the expected payoff from trade liberalization (α1,α2) or the one-stage Nash equilibrium payoffs (0,0). Both therefore prefer exactly the same enforcement mechanism, which minimizes the likelihood of a trade war over time.

Some straightforward comparative statics are readily available. First, the discount factor does not directly affect the maximum per-period payoff. Instead, it only determines whether trade liberalization (C,C) is possible in equilibrium. Second, the maximum payoff for both players i increases with the value of cooperation (α1,α2) and decreases with the value of defection (β1,β2). The former effect is a direct consequence of cooperation in equilibrium, while the latter effect follows from the need to choose more severe punishments when the incentive to defect increases. Third, an increase in the probability of an incorrect signal upon trade liberalization (pD|C) reduces the maximum payoff, because states erroneously engage in trade wars more often. Similarly, an increase in the probability pD|D increases the maximum payoff, because less ruthless punishments are necessary to achieve incentive compatibility.

4.2 Asymmetric Trade Agreements

In principle, an asymmetric trade agreement is more complicated than a symmetric trade agreement because one also has to verify that neither player has an incentive to defect when being punished. The problem can be simplified considerably, however, because I have assumed that the behavior of the developed country is fully observable. In equilibrium, the signal never indicates that the developed country has defected, so one can assume without loss of generality that if it did, the punishment would be a permanent trade war. Let \(W_{G}^{i}\) denote the average future payoff to player i upon signal (C,C) when the countries are supposed to cooperate. Similarly, let \(W_{B}^{i}\) denote the average future payoff to player i upon signal (D,C) so that the developing country will be punished. Finally, let \(W_{T}^{i}\) denote the average payoff to player i upon signal (D,C) in the first period that the developing country is punished. This payoff differs from the average future payoff \(W_{B}^{i}\) upon signal (D,C) because one period of the punishment has already elapsed.

Having specified these payoffs, I first solve for conditions under which defection is not profitable when the countries are supposed to cooperate:
$$ W_{G}^{1}-W_{B}^{1}\geq\frac{(1-\delta)}{\delta}\frac{1}{p_{D\mid D}-p_{D\mid C}}(\beta_{1}-\alpha_{1}); $$
$$ (1-p_{D\mid C})W_{G}^{2}+p_{D\mid C}W_{B}^{2}\geq\frac{(1-\delta)}{\delta}(\beta_{2}-\alpha_{2}). $$
Moreover, the developing country should have no incentive to deviate during the punishment period:
$$ \frac{1-\delta}{\delta}\gamma_{1}\geq(p_{D\mid C}-p_{D\mid D})\left(W_{T}^{1}-W_{B}^{1}\right) $$
This condition can sometimes be difficult to meet, because the developing country must accept the minimum payoff γ1 during it. The developed country obviously has no incentive to deviate when it is punishing the developing country, because it obtains its maximum payoff β2.
Using Eq. 8 and assuming Eq. 9 holds, one obtains the maximum payoff for the developing country,
$$ W_{G}^{1max}=\alpha_{1}-\frac{p_{D\mid C}}{p_{D\mid D}-p_{D\mid C}}(\beta_{1}-\alpha_{1}). $$
It is not possible to characterize the maximum payoff for the developed country explicitly, but note that
$$ W_{G}^{2max}=\alpha_{2}+\delta\cdot\left((1-p_{D\mid C})W_{G}^{2max}+W_{B}^{2max}\right), $$
where \(W_{B}^{2max}\) is the payoff when the optimal enforcement mechanism is used. This payoff is implicitly determined by constraints (8) and (10).

Notably, the developing country has an identical maximal payoff in the asymmetric and symmetric agreements. While surprising, the reason is clear: a developing country rarely has the opportunity to punish a developed country through sanctions, so it benefits from “fair play” according to the agreement with minimal punishments. To the contrary, the developed country has a maximum payoff that is implicitly defined by the incentive constraints for both countries. It actually benefits from the punishment period, so it prefers incorrect signals to the extent that they are not accompanied with costly loss of trade.

This perverse incentive can be seen by considering the comparative statics for the developed country. To begin with, the discount factor does not vanish because the maximum payoff is determined by the most stringent punishment that the developing country finds acceptable. Moreover, an increase in the probability of an error by the developing country (pD|C) is sometimes beneficial because the difference \(W_{B}^{2max}-W_{G}^{2max}\) is positive. One can also immediately observe that an increase in β2 must be beneficial because it increases the value of \(W_{B}^{2}\) while leaving everything else in the game unaffected.

4.3 Trade Capacity Building

In what follows, I write βi(x) instead of βi and so on to highlight the dependence of the parameters of the liberalization game on trade capacity building. Assuming that capacity building occurs after the states have negotiated the international agreement, the decision-making problem can be formulated as follows. First, consider a symmetric agreement:
$$ \begin{array}{lll} \underset{x}{max} & V_{G}^{2max}-z\left(\sum x_{j}\right) & s.t.\\ & x\geq0; & ;\\ & V_{G}^{1}-V_{B}^{1}\geq\displaystyle\frac{1-\delta}{\delta}\displaystyle\frac{1}{p_{D\mid D}\left(x^A\right)-p_{D\mid C}\left(x^A\right)}\left(\beta_{1}(x)-\alpha_{1}(x)\right) & ;\\[4pt] & \left(1-p_{D\mid C}\left(x^A\right)\right)V_{G}^{2}+p_{D\mid C}(x^A)V_{B}^{2}\geq\displaystyle\frac{1-\delta}{\delta}\left(\beta_{2}\left(x^A\right)-\alpha_{2}\left(x^A\right)\right) & .\end{array} $$
Second, consider the asymmetric agreement:
$$ \begin{array}{lll} \underset{x}{max} & W_{G}^{2max}-z\left(\sum x_{j}\right) & s.t.\\ & x\geq0; & ;\\ & W_{G}^{1}-W_{B}^{1}\geq\displaystyle\frac{(1-\delta)}{\delta}\displaystyle\frac{1}{p_{D\mid D}\left(x^A\right)-p_{D\mid C}\left(x^A\right)}(\beta_{1}(x)-\alpha_{1}(x)); & ;\\[4pt] & \left(1-p_{D\mid C}\left(x^A\right)\right)W_{G}^{2}+p_{D\mid C}\left(x^A\right)W_{B}^{2}\geq\displaystyle\frac{(1-\delta)}{\delta}\left(\beta_{2}(x)-\alpha_{2}(x)\right) & ;\\[4pt] & \frac{1-\delta}{\delta}\gamma_{1}\geq\left(p_{D\mid C}\left(x^A\right)-p_{D\mid D}\left(x^A\right)\right)\left(W_{T}^{1}-W_{B}^{1}\right) & ,\end{array} $$
and the restriction that upon signal (C,C), a punishment period is never initiated.

In both cases, the developed country maximizes its expected payoff subject to the enforcement constraints. In the symmetric agreement, the countries’ interests are aligned so the payoff-maximizing agreement under the equilibrium constraints is identical for both players. In the asymmetric agreement, the design that maximizes \(W_{G}^{2}\) is not necessarily in the best interest of the developing country. Given the unequal power relationship, it seems plausible to focus on the design that is in the best interest of the developed country. This is particularly useful when the developed country interacts with least developed countries, which is often the case in trade capacity building.

5 Results

I now turn to the results. First, I show that trade capacity building can create a commitment problem. Second, I show that in an asymmetric agreement, the developed country often prefers not to build institutional capacity even if it is costless. Finally, I show that adequate infrastructural capacity amplifies the damage caused by failure to improve institutional capacity, and vice versa.

Throughout, I assume the optimization problems are solvable so that trade liberalization is feasible. The relevant benchmark for the analysis is the vector of projects that maximizes the cost-effectiveness of trade capacity building in the developing country, defined as \(V_{G}^{1max}-z(x)\) or \(W_{G}^{1max}-z(x)\). Comparing the equilibria of the game with this benchmark, which is denoted by x*, and the no-investment baseline, which is denoted by x0, uncovers the reasons why trade capacity building fails to deliver.

To begin with, observe that the number of potential projects x can be high. Since, the project-specific outcomes can vary, it is useful to categorize them. First, some projects are not implemented (xj = 0). They could be prohibitively costly, violate the constraints outlined above, or go against the interests of the developed country. Second, some projects are implemented but greater investment would foster faster economic development (\(0<x_{j}<x_{j}^{*}\)). This is possible if full implementation of the project violates the equilibrium constraints or is not in the interest of the developed country. Third, some projects could be close to the first-best developmental investment (\(x_{j}\approx x_{j}^{*}\)). Finally, some projects are not cost-effective for development so that investment should be reduced (\(x_{j}>x_{j}^{*}\)).

Consider first the possibility of a commitment problem:

Proposition 1

(commitment problem).

  1. 1.

    In a symmetric agreement, investment to achieve the developmental optimum x* prevents trade liberalization if and only if the increase in the payoff from optimal defection \(\beta_{i}(x^{*})-\beta_{i}(x^{0})\) is too high for some i = 1,2.

  2. 2.

    In an asymmetric agreement, investment to achieve the developmental optimum x* prevents trade liberalization if the increase in the payoff from optimal defection to the developing country \(\beta_{1}(x^{*})-\beta_{1}(x^{0})\) is too high; and only if the increase in the payoff from optimal defection \(\beta_{i}(x^{*})-\beta_{i}(x^{0})\) for some i = 1,2 or the decrease in the probability of an incorrect signal \(p_{D\mid C}(x^{A0})-p_{D\mid C}(x^{A*})\) is too high.

This result provides two reasons why trade capacity building might fail to deliver even when the countries’ interests are perfectly aligned. First, trade capacity building increases the returns to defection (βi(x)). Even though the developed country is sincerely interested in promoting economic development, it must sometimes abandon profitable projects to ensure that liberalization is enforceable. This possibility covers symmetric and asymmetric agreements, and both countries. However, in an asymmetric agreement, for the developed country, an increase in the returns to defection (β2(x)) does not necessarily prompt protectionism. This is so because the dispute resolution mechanism permits occasional defections over time. If the value of these defections is high enough for the developed country, it prefers forgoing politically optimal tariffs now and reaping the gains from occasional defections over time. Thus, asymmetric agreements have the interesting property that, for the developed country, as the value of defection increases, so does the value of cooperation.

Second, improving the accuracy of the signal through institutional capacity building can paradoxically result in the collapse of the international agreement. This happens only in the asymmetric agreement, because it is possible that profitable punishments upon incorrect signals are the reason why the developed country complies. Again, this is not necessarily so. It is sometimes possible to increase the severity of the punishment to “compensate” the developed country for their reduced frequency. If it is very difficult to secure compliance by the developing country during the punishment phase, however, this strategy does not work.

The result falls under the “theory of the second best” (Krishna and Panagariya 2000). Even when donors and recipients agree on the need for trade capacity building, they must consider its impact on the incentive to defect. Political-economic considerations thus impose constraints that prompt second-best a investment schedule. Empirically, this result implies that the failure of trade capacity building per se is not evidence for the donor’s self-seeking motivations.

The converse also holds. Even if a self-seeking donor prefers not to invest in trade capacity building, it might have to do so to induce compliance by the developing country. The commitment problem cuts both ways. It sometimes prevents growth-enhancing projects, but it is equally possible that it forces the donor to help the recipient. For example, the criticisms against narrow-minded trade capacity building to enable implementation of previous commitments lose their punch when one considers the possibility that there would be no trade capacity building whatsoever.

Proposition 1 offers a potential explanation for the findings of Finger and Schuler (2000) that trade capacity building to implement the Uruguay Round obligations has not benefited the recipients (see also Shaffer 2005). Donors have funded institutional trade capacity building to promote compliance by developing countries (higher \(p_{D\mid D}(x^A)\)), such as customs procedures and standards enforcement, while neglecting investments that reduce the vulnerability of developing countries to litigation by developed countries (lower \(p_{D\mid C}(x^A)\)), such as investment in good governance or legal expertise in the domain of trade. In an asymmetric agreement that permits compensation, the donors have a stronger incentive to invest in capacity to perceive defections (higher \(p_{D\mid D}(x^A)\)) than to build institutional capacity to avoid accidental defections in the first place (lower \(p_{D\mid C}(x^A)\)).

Importantly, the commitment problem that I have identified is quite different from the conventional time-inconsistency issues examined in the political-economic literature on development (Easterly 2002; Svensson 2000). These works highlight the pernicious incentives of recipient governments under limited domestic accountability. The commitment problem that I have identified, however, is created exactly when the recipient government faithfully implements the projects funded by the developed country. Furthermore, they could even be commitment problems by the developed not the developing country.

Consider next the consequences of imperfect preference alignment, so that the donor is self-seeking:

Proposition 2

(resource misallocation).

  1. 1.

    In a symmetric agreement, increased investment in any project xj never reduces the payoff to either country as long as trade liberalization is feasible.

  2. 2.

    In an asymmetric agreement, increased investment in some project xj that has an effect on the payoff from defection to the developing country \(\beta_{1}(x^{*})\) or the likelihood of an incorrect signal upon cooperation \(p_{D\mid C}(x^{A})\), can reduce the payoff to the developed country \(W_{G}^{2max}\) even if trade liberalization is feasible.

  3. 3.

    In an asymmetric agreement, increased investment in any project xj that has a large enough effect on the correct signal upon defection \(p_{D\mid D}(x^{A})\), can reduce the payoff to the developing country \(W_{G}^{1}\) even if trade liberalization is feasible.

  4. 4.

    In either agreement, increased investment in some project xj can increase the net payoff to the developed country (\(V_{G}^{2max}-z(x)\) or \(W_{G}^{2max}-z(x)\)) without being cost-effective for development (\(V_{G}^{1}-z(x)\) or \(W_{G}^{1}-z(x)\) decreases).

These results provide insight into the dilemma of trade capacity building. First, in a symmetric agreement, trade capacity building is never directly harmful to either country unless the international agreement collapses. This intuitive finding sets an upper bound for the extent to which trade capacity building can be considered failure: the developing country is never worse off than without it.

Second, in an asymmetric agreement, this need not be true. On the one hand, trade capacity building can decrease or increase the value of the “outside option” that the developing country has through defection. In this case, the countries must adjust the details of the international trade agreement to redistribute the gains from cooperation to the developing country. Consequently, the developed country has an incentive not to implement these projects at all.

On the other hand, changes in the information structure also affect the equilibrium payoffs. When the incorrect signal upon cooperation is likely with high \(p_{D\mid C}(x^A)\), the developing country has less reason to cooperate, but if it does, the developed country can punish it more frequently and thus gain a higher payoff. Thus, improved ability to cooperate through a decrease in \(p_{D\mid C}(x^A)\) can paradoxically hurt the developed country.

When the correct signal upon defection is likely with high \(p_{D\mid D}(x^A)\), the developed country can tighten its grip on the developing country, because it detects defections detected more often and hence less valuable as an outside option. Thus, in some cases, the developed country could even implement projects for trade capacity building that directly hurt the developing country. The empirical plausibility of this scenario is unclear, however, as one expects that the developing country could refuse to implement such projects.

Finally, it is intuitive that the developed country sometimes implements projects that are not cost-effective from a developmental perspective. Since some projects allow many gains for the developed country but not for the developing country, the developed country uses resources in ways that seem to fail the developmental cost-benefit analysis.

The proposition is empirically testable. It predicts that, in a symmetric agreement, donors never fail to implement sufficiently cost-effective projects for trade-capacity building. Since both countries unambiguously benefit from avoiding trade wars, the donor uses all opportunities to improve trade capacity in the recipient country. However, in an asymmetric agreement, the donor may fail to implement both infrastructural and institutional projects for trade capacity building. These projects can increase the value of defection to the developing country, either because of changes in payoffs or in the information structure, which is not in the interests of the developed country. Thus, it would not fund these projects even if they were costless! The empirical prediction is unambiguous: symmetric trade agreements facilitate trade capacity building by aligning donor and recipient interests. This finding is important important, because symmetric enforcement mechanisms that trigger trade wars, seem unnecessary and even unfair: why should the victim suffer from the defection in the future? The model provides a rationale for such designs in unequal North–South trading relationships, in which trade capacity building is potentially useful.

An interesting application is the Generalized System of Preferences, which creates market access for developing countries in a limited number of goods and services without reciprocal concessions (Çaglar Oz̈den and Reinhardt 2005). This system reduces the ability of developed countries to litigate trade policy violations by developing countries, which by Proposition 2 creates incentives for the donors to invest in institutional capacity.

The final theoretical result that I present pertains to the complementarity of adequate infrastructure and institutions:

Proposition 3

(political complementarity of infrastructure and institutions). Consider an investment schedule \(\hat{x}\geq0\).

  1. 1.

    If β1(x) increases slowly with \(x_{j}^{I}\) relative to α1(x) at \(\hat{x}\), the marginal effect of increasing \(x_{j}^{I}\) on \(V_{G}^{1max}\) or \(W_{G}^{1max}\) increases with \(p_{D\mid C}(x^A)\) and decreases with \(p_{D\mid D}(x^A)\).

  2. 2.

    If β1(x) increases rapidly with \(x_{j}^{I}\) relative to α1(x) at \(\hat{x}\), the marginal effect of increasing \(x_{j}^{I}\) on \(V_{G}^{1max}\) or \(W_{G}^{1max}\) decreases with \(p_{D\mid C}(x^A)\) and increases with \(p_{D\mid D}(x^A)\).

This proposition has an intuitive interpretation. If one considers the impact of trade capacity building on the maximum payoff for the developing country (\(V_{G}^{1max}\) or \(W_{G}^{1max}\)), a failure to invest in institutional capacity raises the marginal cost of not investing in infrastructure when such investment also raises the payoff to defection (βi(x) grows rapidly with \(x_{j}^{I}\) relative to α1(x)). This is so because improved monitoring (low \(p_{D\mid C}(x^A)\) and high \(p_{D\mid D}(x^A)\)) counters the detrimental effect of an increased payoff from defection, which implies that infrastructure and institutions are political complements. However, if investment in infrastructure has little impact on the payoff to defection (βi(x) grows slowly with \(x_{j}^{I}\) relative to α1(x)), the absence of institutional capacity paradoxically increases the returns to infrastructural support, because the latter has the additional effect of deterring opportunism when monitoring is incomplete.

These findings are crucial for understanding the frustration with the aid for trade that donors have provided to help developing countries comply with extant international trade law. The emphasis on compliance suggests that opportunism is indeed empirically a major problem, so Proposition 3 predicts that investment in institutional support should be accompanied by startling neglect of economically profitable infrastructural capacity. This is exactly what Finger and Schuler (2000) find.

In summary, these findings imply that, in addition to diverging donor and recipient interests, enforcement problems and incentives to seek compensation systematically skew the allocation of resources for trade capacity building. If protection is attractive to the governments, the political complementarity between infrastructural support and institutional capacity implies that failure in one and success in the other prompts what seems to be startling neglect.

6 Conclusion

I have investigated political-economic problems in trade capacity building. The analysis has two principal elements. First, trade capacity building is chosen by the developed country to shape the payoffs from trade liberalization as well as protectionism. Its institutional variant also decreases the probability that the developing country fails to implement the policies that it seeks to enact. Second, I compare symmetric agreements, in which a mutually costly trade dispute occurs upon illegal protectionism, with asymmetric agreements, in which the defector must compensate the victim to restore cooperation.

The first key result concerns the problem of credible commitment. As the literature review suggests, the developmental effect of trade capacity building is mixed. The need to enforce trade liberalization constrains the set of projects that can be implemented. On the one hand, some projects are often necessary to enforce trade liberalization even if their economic viability is questionable. On the other hand, and more counterintuitively, some projects could even increase the returns to protectionism. Unfortunately, these projects could also increase the payoff from trade liberalization substantially.

The second key result pertains to the donor-driven nature of aid for trade. Political economists often explain the dismal performance of development aid with reference to self-seeking donors (Smith and Bueno de Mesquita 2007). I provide a theoretical basis for the divergence between donor and recipient motivations. On the one hand, the donor often invests too much in some projects and too little in others. Notably, however, these projects never actually hurt the recipient in a “symmetric” agreement. Instead, the resulting failure is only one of misallocation. On the other hand, if the international agreement is “asymmetric” in that it permits profitable punishment upon deviation, the donor might not implement costless projects while funding costly projects that hurt the recipient.

Finally, I examine the complementary nature of infrastructural support and institutional capacity. If donors invest in institutional capacity to enable compliance, the marginal returns to infrastructural support from a developmental perspective increase when trade capacity building also raises the incentive to defect. The converse also holds. Consequently, the model correctly predicts a pattern of asymmetric misallocation: when opportunism is a potential problem, large-scale investment in institutional support is often accompanied by striking neglect of infrastructural capacity.

The analysis has a number of limitations, two of which are particularly important. First, I have not explicitly modeled the role of international organizations, such as the World Bank or regional development agencies. International organizations have their own preferences, so they pose additional complications. But to the extent that major donors, such as the United States, can discipline the international bureaucrats, my simple model should provide insight into multilateral trade capacity building. The incentives that my theoretical analysis has uncovered are broadly applicable, and even in multilateral development agencies, the donors ultimately decide how much funding is made available for trade capacity building.

Second, I have not examined another important source of problems, namely coordination failure between different domestic agencies (Prowse 2002). My analysis has focused on strategic problems between states, so it has more to tell about how donors strategically sketch the broad contours of trade capacity building than about details of any particular project. Both issues are important given the large financial flows that trade capacity building generates and the importance of implementation details. A particularly interesting avenue for future research are the interactions between political-economic and other implementation problems. For example, if trade officials are responsible for compliance with international trade agreements while development officials manage trade capacity building, coordination problems could amplify the commitment and enforcement problems that I have identified.


“USAID Trade Capacity Building Database.” Available at


“USAID Trade Capacity Building Database.” Available at


“USAID Trade Capacity Building Database.” Available at


In the Appendix, I show that the relevant assumptions hold even for a small economy in the Grossman and Helpman (1994) model of endogenous trade policy.


The one-stage payoffs should therefore be interpreted as expected values. If the payoffs were certain, the players could ignore the signals and instead simply observe the payoffs. It seems more plausible that an incorrect signal is accompanied by a surprisingly high or low payoff (Green and Porter 1984).


See the Appendix for a full characterization.


I incorporate the possibility of “altruistic” trade capacity building by considering the special case in which the countries’ interests are perfectly aligned.



I thank Todd Allee, Leonardo Baccini, Sam Brazys, Daina Chiba, Hyeran Jo, Jennifer Kavanagh, Daniel Kono, Paul Poast, the editor of the Review of International Organizations, and the anonymous reviewers for comments and advice.

Copyright information

© Springer Science + Business Media, LLC 2009