Payoff Information and Self-Confirming Equilibrium

Eddie Dekel, Drew Fudenberg, and David K. Levine

June 10, 1996

Revised: January 26, 1999

Abstract: In a self-confirming equilibrium, each player correctly forecasts the actions that opponents will take along the equilibrium path, but may be mistaken about the way that opponents would respond to deviations. This models a steady state of a learning process in which players observe actions played by their opponents, rather than a complete specification of their strategies. Consequently, players need not receive evidence that their forecasts of off-path play are incorrect. In practice, players understand that opponents are rational and have some information about their opponents’ payoffs. This paper develops a refinement of self-confirming equilibrium that incorporates the effects of such information. We show that this concept is robust. We also discuss its relationship to other concepts. In particular, we show that it is closely connected to assuming