Stochastic Strategy Adjustment in Coordination Games

We explore a model of equilibrium selection in coordination games, where agents stochastically adjust their strategies to changes in their local environment. Instead of playing perturbed best-response, we assume that agents follow a rule of "switching to better strategies more likely". We relate this behavior to work of Rosenthal (1989) and Schlag (1998). Our main results are that both strict Nash equilibria of the coordination game correspond to stationary distributions of the process, hence evolution of play is not ergodic, but instead depends on initial conditions. However, coordination on the risk-dominant equilibrium occurs with probability one whenever the initial share of agents playing the risk-dominant strategy has at least some positive measure, how ever small, within the whole population.