Updating Validations and Stochastic Dynamics on Coordination

This work studies an equilibrium selection of infinitely repeated symmetric 2x2 coordination games that show a tension between Pareto efficiency and risk dominance, in which bounded rational agents adopt the following simple behavior rule: each agent has a valuation of actions, and chooses the highest one. Valuations are updated according to the sign of the difference between the current valuation and the realized payoff. By applying techniques from stochastic stable states (Kandori et al. 1993 and Young 1993), it is shown that the risk dominant outcome is selected; that is, it is realized more frequently in the long run.