Stationary Markov perfect equilibria in risk sensitive stochastic overlapping generations models

Abstract

In this paper, we study intergenerational stochastic games that can be viewed as a special class of overlapping generations models under uncertainty. Making use of the theorem of Dvoretzky, Wald and Wolfowitz [27] from the statistical decision theory, we obtain new results on stationary Markov perfect equilibria for the aforementioned games, with a general state space, satisfying rather mild continuity and compactness conditions. A novel feature of our approach is the fact that we consider risk averse generations in the sense that they aggregate partial utilities using an exponential function. As a byproduct, we also provide a new existence theorem for intergenerational stochastic game within the standard framework where the aggregator is linear. Our assumptions imposed on the transition probability and utility functions allow to embrace a pretty large class of intergenerational stochastic games analysed recently in macroeconomics. Finally, we formulate a set of assumptions under which the stochastic process induced by the stationary Markov perfect equilibrium possesses an invariant distribution.