Markov Perfection And Cooperation In Repeated Games PDF Download

This paper examines the issue of multiplicity of equilibria in alternating move repeated games with two players. Such games are canonical models of environments with repeated, asynchronous choices due to inertia or replacement. We focus our attention on Markov Perfect equilibria (MPE). These are Perfect equilibria in which individuals condition their actions on payoff-relevant state variables. Our main result is that the number of Markov Perfect equilibria is generically finite with respect to stage game payoffs. This holds despite the fact that the stochastic game representation of the alternating move repeated game is "non-generic" in the larger space of state dependent payoffs. We also compare the MPE to non-Markovian equilibria and to the (trivial) MPE of standard repeated games. Unlike the latter, it is often true when moves are asynchronous that Pareto inferior stage game equilibrium payoffs cannot be supported in MPE. Also, MPE can be constructed to support cooperation in a Prisoner's Dilemma despite limited possibilities for constructing punishments.