The paper is a survey written for the Sixth World congress of the Econometric Society. It is devoted largely to a discussion of the progress made in the last decade in understanding the structure of self-enforcing agreements in discounted supergames of complete information. Perfect and imperfect monitoring models are considered in turn, with attention given to the case of substantial impatience as well as to the various “folk theorems.” The emphasis is on the features of constrained-optimal perfect equilibria, causes of inefficiency, and some relationships among different strands of the literature. The remainder of the paper is a critical and comparative consideration of recent work on renegotiation in repeated games.
Keywords: Repeated game, self-enforcing contracts, supergames, folk theorem
Traditional agency theory assumes that the principal has no more information about the agent’s actions than the enforcement authorities have. This is unrealistic in many settings, and in repeated models, additional information possessed by the principal changes the nature of the problem. Such information can be used in implicit, self-enforcing contracts between principal and agent, that supplement the usual explicit contracts. This paper studies the way in which the two kinds of contracts are combined in constrained efficient equilibria of the agency supergame. The agent’s compensation is comprised of both guaranteed payments and voluntary bonuses from the principal. We give a simple characterization of the composition of remuneration in the optimal dynamic scheme.
Cooperation in repeated games relies on the possibility that equilibrium play following some t-period history depends on more than simply the structure of the game remaining after the first t periods, that structure being always the same. In a nondegenerate theory of renegotiation, what a player expects, and the statements he finds credible at the end of period t must be affected by the history that has transpired, and perhaps by the implicit agreement that was in force. The solution concept proposed in this paper acknowledges both these influences, while imposing a certain stationarity on beliefs regarding what renegotiation options are available: renegotiation to an equilibrium sigma will not take place if, after some history h, the continuation equilibrium sigma given h is itself vulnerable to renegotiation to sigma (in the sense that all players prefer sigma to sigma given h).