Liberals are interested in how states can cooperate in an anarchic inter- national system, and a fundamental problem in game theory called prisoners’
dilemma explores how states can achieve a better collective outcome through
cooperation. Game theory investigates the interaction of two or more indi- viduals or states, in which each individual or state acts according to rational choice; it seeks to explain how the actors’ decisions are interrelated and how these decisions affect outcomes.33 Prisoners’ dilemma is a “mixed-motive
game,” in which two players can benefit from mutual cooperation but have an incentive to “defect” or cheat on each other and become free riders. The term prisoners’ dilemma derives from the story used to describe the game: The police arrest two individuals, A and B, for committing fraud, and they suspect that A and B have also committed robbery but cannot prove it. To get A and B to confess, the police put them in different cells so they cannot communicate with each other, and question them separately. In Figure 4.1, prisoners A and B “cooperate” with each other if they do not confess to com- mitting robbery, and they “defect” (or cheat on each other) if they confess. The sentences the prisoners receive depend on the decisions they make. The numbers in bold at the top right-hand corners of the squares are A’s years in prison, and the numbers at the bottom left-hand corners are B’s years in prison. The police make a tempting offer to induce A to confess (i.e., defect). They inform A that conviction for fraud is certain and will result in a two-year sentence for both prisoners if they do not confess (square I in
FIGURE 4.1 Prisoners’ Dilemma
Figure 4.1). However, if A confesses to robbery (i.e., defects) and B does not (i.e., cooperates), A will go free and B will get 10 years in prison (square II). If both A and B defect and confess to robbery, they will get a reduced sen- tence of five years (square III). Finally, if A does not confess (i.e., cooperates) but B confesses (i.e., defects), A will get 10 years in prison and B will go free (square IV). The police provide the same offer to B.
What will the prisoners do? According to individual rationality, if B defects, A is better off defecting (5 years in prison) than cooperating (10 years). If B cooperates, A is also better off defecting (goes free) than coop- erating (2 years in prison). Thus, individual rationality pushes A to defect regardless of what B does, and the same reasoning applies to B! Furthermore, A and B mistrust each other, and they both fear that they will receive the worst possible penalty by cooperating (10 years) if the other prisoner defects. As a result, A and B are both likely to defect and end up with five years in prison (square III), even though both would get only two years (square I) if they
Defect Actor A* Actor B Cooperate Cooperate Defect 2 10 0 5 2 0 10 5 I II III IV
Liberalism and Institutions 87 cooperated. Square I is the best collective outcome or the Pareto-optimal out-
come for A and B, because no actor can become better off without making
someone else worse off (i.e., if A confesses and goes free, B will get 10 years in prison). Square III is an inferior collective outcome or Pareto-deficient out-
come, because both actors (A and B) would prefer another outcome (square I).
As is the case for the provision of public goods (see Chapter 3), prisoners’ dilemma presents a collective action problem in rational choice analysis, because rational actors may be “unable to reach a Pareto-optimal solution, despite a certain degree of convergence of interests between them.”34 The
dilemma in both the provision of public goods and the prisoners’ dilemma game is that individual rationality differs from collective rationality: The decision of rational, self-interested states to become “free riders” may inter- fere with the provision of public goods, and the decision of rational, self- interested prisoners to defect may lead to a Pareto-deficient outcome (square III in Figure 4.1) for both prisoners.35 In IPE, we ask how states can move
from a Pareto-deficient (mutual defection or DD) to a Pareto-optimal outcome (mutual cooperation or CC). In the liberal view, “cheating” or free riding by states can inhibit cooperation, and mutual cooperation is possible if cheating can be controlled. A global hegemon can prevent cheating by pro- viding public goods and coercing other states to abide by agreed rules and principles. Institutions such as IOs can also prevent cheating by bringing states together on a regular basis. A state that interacts regularly with others is less likely to cheat because the other states have many opportunities to retaliate. International institutions also enforce principles and rules to ensure that cheaters are punished and they collect information on members’ policies, increasing transparency or confidence that cheaters will be discovered. Furthermore, international institutions contribute to a learning process in which states realize that mutual gains can result from cooperation.36
Realists are more skeptical than liberals that international institutions have an important role in moving states to a Pareto-optimal (CC) outcome for sev- eral reasons. First, realists see international institutions as existing more often in “low politics” socioeconomic areas than in the “high politics” areas of national security and defense. Second, realists often view institutions as having no independent standing, because they serve the interests of the most powerful states. Third, realists see state concerns with relative gains as posing a major obstacle to cooperation. Even if two states have common interests, they may not cooperate because of each state’s concern that the other will receive greater gains. Institutions can promote cooperation, according to realists, only if they can ensure that members’ gains are balanced and equitable; but this is difficult to achieve because gains are rarely equal.37