We find that a number of problematic cases in public policy can be traced to inefficient Nash equilibria with the games considered in strategic normal form. In this sense, we may say that noncooperative game theory is a powerful diagnostic tool for public policy. For now we reserve judgment as to whether it may also be helpful in prescription. Nash equilibrium models have proved a powerful tool of problem identification for public policy. While the Prisoner’s Dilemma has commanded the central position in this perspective, a wide variety of other Nash equilibrium models may result in inefficient equilibria in the absence of some public action. In conflict situations, and some others, it may be rational for agents to be unpredictable, and to randomize their strategies; but in other circumstances, coordination and anticoordination games, randomization may be something to be avoided, and avoiding it may require some information not within the game itself. The Nash equilibrium in pure strategies has a property of consistent beliefs that helps to explain how a Nash equilibrium, once established, can be persistent even when other Nash equilibria may be more efficient. Extensions of these models, with large numbers of players, trial-and-error adaptive learning in place of ideal rationality, and lack of symmetry among players, may seem more “realistic” than the simpler Nash equilibrium models. Where there are two or more Nash equilibria, refinements may eliminate some as less plausible. Pareto-dominance, evolutionary stability, and resistance to disruption by coalitional shifts of strategy provide criteria for refinement in particular cases.
- 1. For a case study see McCain et al. (2011).
- 2. I am indebted to my colleague Richard Hamilton for discussions that contributed to this example.