We apply the average cost optimality equation to zero-sum Markov games, by considering a simple game with one-sided incomplete information that generalizes an example of Aumann and Maschler (1995). We determine the value and identify the optimal strategies for a range of parameters.

Recently, Askoura et al. (2013) proved the nonemptiness of the α-core of a finite Bayesian game GR with Young measure strategies and nonatomic type spaces, without requiring that the expected payoffs be concave. Under the same hypotheses as theirs, we demonstrate that Scarf’s method (1971)...