Entropified Berk-Nash Equilibrium
22 Pages Posted: 30 Oct 2019 Last revised: 1 Nov 2019
Date Written: October 31, 2019
Esponda and Pouzo (2016) propose Berk-Nash equilibrium as a solution concept for games that are misspecified in that it is impossible for players to learn the true probability distribution over outcomes. In general, the beliefs that support Berk-Nash equilibrium are not stable: there may exist a profitable deviation to alternative beliefs in the player’s support that lead to higher payoffs. From an evolutionary perspective, this renders the beliefs that support Berk-Nash vulnerable to invasion. Drawing on the machine learning literature, we propose robust Berk-Nash equilibrium, which is immune to this critique.
Keywords: misspecified learning, evolutionary models, Berk-Nash Equilibrium
JEL Classification: D8, C7, C4
Suggested Citation: Suggested Citation