Conditional Markov equilibria in discounted dynamic games

摘要

This paper introduces conditional Markov strategies in discrete-time discounted dynamic games with perfect monitoring. These are strategies in which players follow Markov policies after all histories. Policies induced by conditional Markov equilibria can be supported with the threat of reverting to the policy that yields the smallest expected equilibrium payoff for the deviator. This leads to a set-valued fixed-point characterization of equilibrium payoff functions. The result can be used for the computation of equilibria and for showing the existence in behavior strategies.