mutual defection. Here, should another state defect, a punishing state might well benefit economically from the opportunity to defect itself and there may be no free rider problem in enforcement These last observations suggest some further principles. If externality problems can be handled adequately on a bilateral basis, such arrangements will tend to be preferred to multilateral arrangements. The case for agreements involving larger numbers of states maps directly onto the case for international agreements in the first instance- multilateral cooperation is potentially desirable mainly when bilateral cooperation itself creates externalities for third states. The caveat is that if a multitude of bilateral agreements would all look about the same, it may be more economical to create one multilateral agreement that sets out the same principles in a single document A related point is that when bilateral or small numbers agreements create externalities, an expansion of the number of states participating in an agreement will not simply increase the costs of agreement, but also the benefits. Hence, the international community may move toward large scale multilateral agreements despite their higher Finally, not only does the number of states involved affect the costs of achieving and enforcing agreement, but the size distribution of states may also matter. In particular, the presence of a few large states may help at least on the enforcement problem. If punishment is seen as costly to the punishers, the free rider problem will be less acute hen some states are large enough to capture a considerable portion of the joint gains from enforcing the agreement. It may then be in their private interest to act as enforcers even if smaller states will free ride. This possibility suggests a constructive role in some contexts for what political scientists term a"hegemon, a powerful state that enforces its will in an environment populated by other smaller and less powerful states ii. Treaties as Contracts As noted one may model international agreements fort by states to reach their Pareto frontier in the presence of externalities. They do so subject to a participation constraint for each state and perhaps other constraints depending on the context Readers familiar with the literature on optimal contracting will find this type of problem quite familiar. Indeed, from an economic standpoint, a treaty is a contract albeit one between states rather than individuals. As long as one is prepared to reduce the preferences"of the state to an ordering akin to that of an individual rational actor, there is no difference in a general sense between the problem of designing an optimal contract
mutual defection. Here, should another state defect, a punishing state might well benefit economically from the opportunity to defect itself and there may be no free rider problem in enforcement. These last observations suggest some further principles. If externality problems can be handled adequately on a bilateral basis, such arrangements will tend to be preferred to multilateral arrangements. The case for agreements involving larger numbers of states maps directly onto the case for international agreements in the first instance— multilateral cooperation is potentially desirable mainly when bilateral cooperation itself creates externalities for third states. The caveat is that if a multitude of bilateral agreements would all look about the same, it may be more economical to create one multilateral agreement that sets out the same principles in a single document. A related point is that when bilateral or small numbers agreements create externalities, an expansion of the number of states participating in an agreement will not simply increase the costs of agreement, but also the benefits. Hence, the international community may move toward large scale multilateral agreements despite their higher cost. Finally, not only does the number of states involved affect the costs of achieving and enforcing agreement, but the size distribution of states may also matter. In particular, the presence of a few large states may help at least on the enforcement problem. If punishment is seen as costly to the punishers, the free rider problem will be less acute when some states are large enough to capture a considerable portion of the joint gains from enforcing the agreement. It may then be in their private interest to act as enforcers even if smaller states will free ride. This possibility suggests a constructive role in some contexts for what political scientists term a “hegemon,” a powerful state that enforces its will in an environment populated by other smaller and less powerful states. ii. Treaties as Contracts As noted, one may model international agreements as an effort by states to reach their Pareto frontier in the presence of externalities. They do so subject to a participation constraint for each state and perhaps other constraints depending on the context. Readers familiar with the literature on optimal contracting will find this type of problem quite familiar. Indeed, from an economic standpoint, a treaty is a contract albeit one between states rather than individuals. As long as one is prepared to reduce the “preferences” of the state to an ordering akin to that of an individual rational actor, there is no difference in a general sense between the problem of designing an optimal contract
and the problem of designing an optimal treaty. I briefly note here some of the issues that have received some attention in the international literature, and that are directly related to the literature on optimal contracts Treaties as Incomplete Contracts. Some treaties address simple matters about which optimal behavior changes little over time. But many treaties address complex matters in an environment subject to uncertainty. For familiar reasons, treaties in the latter group are likely to be incomplete as to certain behaviors and contingencies. States may then benefit from various devices to fill the gaps in their agreement. Specialized tribunals may supply useful default rules in some cases, as may overarching treaties on interpretation such as the Vienna Convention on Treaties. As in the literature on default ules for private contracts, one can ask what the guiding principle should be in designing them--should they replicate what the parties would have negotiated expressly if they had addressed the matter, penalize a party that has withheld pertinent information, or something els The Contractability of Desired Behavior. It is a commonplace in the literature racting to suppose that some beh the behavior in question cannot be observed by other parties or at least cannot be verified by a court. The same class of problems arises under many international agreements regarding everything from hidden nontariff trade barriers to cheating on nonproliferation commitments. Treaties may then employ the familiar types of (generally second-best) solutions. Treaty obligations may be conditioned on verifiable behavior that is correlated with the unverifiable behavior, as in the classic principal-agent models where payoffs nust be conditioned on observable outcomes rather than unobservable effort Alternatively, provisions may be designed to encourage decisionmakers to internalize the externalities from their choices, and thereby to eliminate the divergence between privately and socially optimal behavior Renegotiation, Modification and Efficient Breach. Treaties are often negotiated under conditions of uncertainty. a variety of shocks may cause particular commitments to become inefficient, or may leave some signatory worse off than it would be by exiting Some treaties will address the problem simply by providing for the possibility of withdrawal, perhaps after a period of notice(the Salt treaty, for example). But many treaties address a broad range of issues, and changed circumstances may justify only modification of the bargain, not a complete end to it. Accordingly, treaties may contain provisions providing for the renegotiation of parts of the bargain. or may specify contingencies under which states may deviate from their prior commitments. A treaty
and the problem of designing an optimal treaty. I briefly note here some of the issues that have received some attention in the international literature, and that are directly related to the literature on optimal contracts. Treaties as Incomplete Contracts. Some treaties address simple matters about which optimal behavior changes little over time. But many treaties address complex matters in an environment subject to uncertainty. For familiar reasons, treaties in the latter group are likely to be incomplete as to certain behaviors and contingencies. States may then benefit from various devices to fill the gaps in their agreement. Specialized tribunals may supply useful default rules in some cases, as may overarching treaties on interpretation such as the Vienna Convention on Treaties. As in the literature on default rules for private contracts, one can ask what the guiding principle should be in designing them—should they replicate what the parties would have negotiated expressly if they had addressed the matter, penalize a party that has withheld pertinent information, or something else? The Contractability of Desired Behavior. It is a commonplace in the literature on optimal contracting to suppose that some behavior is noncontractable, usually because the behavior in question cannot be observed by other parties, or at least cannot be verified by a court. The same class of problems arises under many international agreements regarding everything from hidden nontariff trade barriers to cheating on nonproliferation commitments. Treaties may then employ the familiar types of (generally second-best) solutions. Treaty obligations may be conditioned on verifiable behavior that is correlated with the unverifiable behavior, as in the classic principal-agent models where payoffs must be conditioned on observable outcomes rather than unobservable effort. Alternatively, provisions may be designed to encourage decisionmakers to internalize the externalities from their choices, and thereby to eliminate the divergence between privately and socially optimal behavior. Renegotiation, Modification and Efficient Breach. Treaties are often negotiated under conditions of uncertainty. A variety of shocks may cause particular commitments to become inefficient, or may leave some signatory worse off than it would be by exiting. Some treaties will address the problem simply by providing for the possibility of withdrawal, perhaps after a period of notice (the SALT treaty, for example). But many treaties address a broad range of issues, and changed circumstances may justify only a modification of the bargain, not a complete end to it. Accordingly, treaties may contain provisions providing for the renegotiation of parts of the bargain. or may specify contingencies under which states may deviate from their prior commitments. A treaty
may provide that a party can breach its obligations at a price, which if set correctly can facilitate"efficient breach The design of such provisions can raise a number of interesting questions. In a multilateral treaty, for example, how does a renegotiation avoid the holdout problem? Does that problem argue for allowing breach at acompensatory price? "How is the price for breach set, and when is it preferable to employ stiffer sanctions that move the regime toward a"property rule"and away from a"liability rule? These latter questions presuppose that the legal regime has the capacity to employ meaningful sanctions for breach and to calibrate them-a subject that will receive further attention in a moment In addition to the problem of facilitating efficient adjustments to changed circumstances, treaties must confront the fact that efforts to modify the bargain may be opportunistic. In the event that a party deviates from its commitments and then offers to renegotiate, will it be in the interest of the other parties at that point in time to hold it to the original bargain and will they have the ability to do so, or will they capitulate to new terms that extract some of their surplus? In many important contexts, therefore, a tre designer must worry whether the agreement is renegotiation proof. This problem is usually related to the presence of sunk costs. After states have made sunk investments in eliance on an agreement, the returns to those investments may be vulnerable to expropriation during opportunistic renegotiation This is but a partial listing of the issues that international agreements must confront, and that have direct parallels in optimal contracting problems. Economically oriented scholars will find many fruitful applications to international law of the ideas developed in the contract literature iii Enforcement and Dispute resolution States contemplating an international agreement often confront another set of problems that is absent for parties to conventional contracts. Private actors commonly rely on state enforcers to hold them to their commitments. An award of damages contract can be enforced through a seizure of the promisor's assets or an injunction backed by threat of imprisonment. by contrast, although the use of military force and the seizure of assets is surely seen in international relations, such devices are not part of the enforcement arsenal for many international agreements. Indeed, the observation that much of international law is not backed by a credible threat of military force or other strong coercive measures has led many commentators to question whether international law is“law” at all
may provide that a party can breach its obligations at a price, which if set correctly can facilitate “efficient breach.” The design of such provisions can raise a number of interesting questions. In a multilateral treaty, for example, how does a renegotiation avoid the holdout problem? Does that problem argue for allowing breach at a “compensatory price?” How is the price for breach set, and when is it preferable to employ stiffer sanctions that move the regime toward a “property rule” and away from a “liability rule?” These latter questions presuppose that the legal regime has the capacity to employ meaningful sanctions for breach and to calibrate them—a subject that will receive further attention in a moment. In addition to the problem of facilitating efficient adjustments to changed circumstances, treaties must confront the fact that efforts to modify the bargain may be opportunistic. In the event that a party deviates from its commitments and then offers to renegotiate, will it be in the interest of the other parties at that point in time to hold it to the original bargain and will they have the ability to do so, or will they capitulate to new terms that extract some of their surplus? In many important contexts, therefore, a treaty designer must worry whether the agreement is renegotiation proof. This problem is usually related to the presence of sunk costs. After states have made sunk investments in reliance on an agreement, the returns to those investments may be vulnerable to expropriation during opportunistic renegotiation. This is but a partial listing of the issues that international agreements must confront, and that have direct parallels in optimal contracting problems. Economicallyoriented scholars will find many fruitful applications to international law of the ideas developed in the contract literature. iii. Enforcement and Dispute Resolution States contemplating an international agreement often confront another set of problems that is absent for parties to conventional contracts. Private actors commonly rely on state enforcers to hold them to their commitments. An award of damages in contract can be enforced through a seizure of the promisor’s assets or an injunction backed by threat of imprisonment. By contrast, although the use of military force and the seizure of assets is surely seen in international relations, such devices are not part of the enforcement arsenal for many international agreements. Indeed, the observation that much of international law is not backed by a credible threat of military force or other strong coercive measures has led many commentators to question whether international law is “law” at all
Such skepticism is misplaced in my view for at least two reasons. First, states that join international agreements can and do create international regimes where strong coercive measures are possible-witness the deployment of troops after UN authorization through the years, or the various episodes of u N. sanctions that have affected the economic vitality of rogue states. International law should not be viewed as hopelessly weak because of the limited enforcement regime that is imposed on it exogenously. Instead, one must recognize that the choice of enforcement regime is endogenous. There is much that states can do, if they wish, to make their commitments more credible. One must therefore ask why some international legal commitments carry much greater punishments for breach than others whether weaker enforcement are inadequate to their task or are instead chosen precisely because nothing more Is necessary Second, although parties to private contracts can often appeal to state enforcers, many settings they cannot. Litigation costs may swamp the benefits of an appeal to the courts when the monetary stakes are modest. Transnational contracts can confront difficult issues of securing personal jurisdiction in the event of a dispute, enforcing awards, and securing unbiased decisionmakers. Such problems have led economic scholars to recognize that there are valuable mechanisms for making contractual commitments credible that do not rely on the existence of a third-party enforcer with coercive powers. (See, e.g., Telser 1980) Even when these mechanisms cannot achieve the first-best, they can often accomplish a great deal So too it is at the international level I thus offer a brief catalog of the mechanisms that are available for the enforcement of international agreements that do not rely on the coercive powers of third party enforcers. Some have been mentioned in passing already but deserve more detailed comment Hostage Exchange and Bond Posting. The paradigm example of the hostage exchange involves a hypothetical peace treaty between warring kings. Each king sends a son to live in the other kingdom with the understanding that if he attacks that kingdom his son will be killed. Both kings value their sons' lives more than any possible gains from aggression, and so peace is sust The exchange of human hostages threatened with death is rather unseemly by modern standards, of course and we do not observe it in international agreements. But analogous situations may arise, as when foreign nationals or their assets travel and become subject to the jurisdiction of other states. When such movements are reciprocal a situation akin to a hostage exchange may arise implicitly--recall the suggestion by
Such skepticism is misplaced in my view for at least two reasons. First, states that join international agreements can and do create international regimes where strong coercive measures are possible—witness the deployment of troops after U.N. authorization through the years, or the various episodes of U.N. sanctions that have affected the economic vitality of rogue states. International law should not be viewed as hopelessly weak because of the limited enforcement regime that is imposed on it exogenously. Instead, one must recognize that the choice of enforcement regime is endogenous. There is much that states can do, if they wish, to make their commitments more credible. One must therefore ask why some international legal commitments carry much greater punishments for breach than others, and whether weaker regimes of enforcement are inadequate to their task or are instead chosen precisely because nothing more is necessary. Second, although parties to private contracts can often appeal to state enforcers, in many settings they cannot. Litigation costs may swamp the benefits of an appeal to the courts when the monetary stakes are modest. Transnational contracts can confront difficult issues of securing personal jurisdiction in the event of a dispute, enforcing awards, and securing unbiased decisionmakers. Such problems have led economic scholars to recognize that there are valuable mechanisms for making contractual commitments credible that do not rely on the existence of a third-party enforcer with coercive powers. (See, e.g., Telser 1980) Even when these mechanisms cannot achieve the “first-best,” they can often accomplish a great deal. So too it is at the international level. I thus offer a brief catalog of the mechanisms that are available for the enforcement of international agreements that do not rely on the coercive powers of third party enforcers. Some have been mentioned in passing already but deserve more detailed comment. Hostage Exchange and Bond Posting. The paradigm example of the hostage exchange involves a hypothetical peace treaty between warring kings. Each king sends a son to live in the other kingdom, with the understanding that if he attacks that kingdom his son will be killed. Both kings value their sons’ lives more than any possible gains from aggression, and so peace is sustained. The exchange of human hostages threatened with death is rather unseemly by modern standards, of course, and we do not observe it in international agreements. But analogous situations may arise, as when foreign nationals or their assets travel and become subject to the jurisdiction of other states. When such movements are reciprocal a situation akin to a hostage exchange may arise implicitly—recall the suggestion by
Goldsmith and Posner that the rules of ambassadorial immunity may be sustained through such a mechanism. a possible difficulty with the hostage exchange mechanism, of course, is that the reciprocal threats to the hostages may not be credible. It may not be in the interest of the king who is attacked to kill the attackers son, for example, if he knows that his own son will then be killed A related mechanism is bond posting. Each party posts a bond that is large enough to exceed its potential gains from cheating on the agreement. Should cheating occur, the bond is forfeit. Compliance with the agreement then becomes an equilibrium as long as cheating will be detected with sufficient probability in relation to the size o the bond. Bond posting differs from hostage exchange in that it generally relies on a third-party arbiter to determine whether breach of agreement exists, but the arbiter need not possess any coercive powers beyond the capacity to declare that one party has forfeited its bond to another. If the arbiter is trusted by all parties, and they cannot interfere with the exercise of the arbiter's authority, then a bond posting mechanism can do quite well in encouraging ly, formal bond posting arrangements seem quite rare in international law, although the reason why is by no means obvious Mutual Threats of Defection. International agreements that address external itie often have many of the qualities of a Repeated Prisoner's Dilemma. Each party makes commitments that it would prefer not to make, other things being equal, in exchange for valuable commitments from others that result in a Pareto improvement. States will be tempted to defect if they think they can do so without retaliation Repeated Prisoner's dilemmas have been studied extensively, both theoretically and empirically. On the theoretical side, the well known Folk Theorem holds that long term cooperation is a possible equilibrium, enforced by mutual threats to defect from cooperation should another party defect. Cooperation requires that the game has no fixed ending(defection would become a dominant strategy in the last period and cooperation would unravel from there), and that the parties have low enough discount rates that the current gains from defection do not loom too large in relation to the long term gains from cooperation. Rasmusen(1989). A requirement that the equilibrium be renegotiation proof makes the equilibrium rather complicated, but such equilibria do exist. ( Fudenberg Tirole 1991). Empirical research suggests that the tit for tat" strategy, whereby a period of defection by one party is met by a subsequent period of defection successful enforcement device, even if the theoretical literature notes that this strategy not subgame perfect. Axelrod (1984). Although much of the discussion in the literature is
Goldsmith and Posner that the rules of ambassadorial immunity may be sustained through such a mechanism. A possible difficulty with the hostage exchange mechanism, of course, is that the reciprocal threats to the hostages may not be credible. It may not be in the interest of the king who is attacked to kill the attacker’s son, for example, if he knows that his own son will then be killed. A related mechanism is bond posting. Each party posts a bond that is large enough to exceed its potential gains from cheating on the agreement. Should cheating occur, the bond is forfeit. Compliance with the agreement then becomes an equilibrium as long as cheating will be detected with sufficient probability in relation to the size of the bond. Bond posting differs from hostage exchange in that it generally relies on a third-party arbiter to determine whether breach of agreement exists, but the arbiter need not possess any coercive powers beyond the capacity to declare that one party has forfeited its bond to another. If the arbiter is trusted by all parties, and they cannot interfere with the exercise of the arbiter’s authority, then a bond posting mechanism can do quite well in encouraging compliance. Interestingly, formal bond posting arrangements seem quite rare in international law, although the reason why is by no means obvious. Mutual Threats of Defection. International agreements that address externalities often have many of the qualities of a Repeated Prisoner’s Dilemma. Each party makes commitments that it would prefer not to make, other things being equal, in exchange for valuable commitments from others that result in a Pareto improvement. States will be tempted to defect if they think they can do so without retaliation. Repeated Prisoner’s Dilemmas have been studied extensively, both theoretically and empirically. On the theoretical side, the well known Folk Theorem holds that long term cooperation is a possible equilibrium, enforced by mutual threats to defect from cooperation should another party defect. Cooperation requires that the game has no fixed ending (defection would become a dominant strategy in the last period and cooperation would unravel from there), and that the parties have low enough discount rates that the current gains from defection do not loom too large in relation to the long term gains from cooperation. Rasmusen (1989). A requirement that the equilibrium be renegotiation proof makes the equilibrium rather complicated, but such equilibria do exist. (Fudenberg & Tirole 1991). Empirical research suggests that the “tit for tat” strategy, whereby a period of defection by one party is met by a subsequent period of defection by others, can be a successful enforcement device, even if the theoretical literature notes that this strategy is not subgame perfect. Axelrod (1984). Although much of the discussion in the literature is