Equilibrium Selection under Different Learning Modes in Supermodular Games

We apply the dynamic stochastic framework proposed by recent evolutionaryliterature to the class of strict supermodular games when two simplebehavior rules coexist in the population, imitation and myopic optimization.We assume that myopic optimizers are able to see how well their payoff doesrelative to what they can get in the stage game and therefore experimentmore in low payoff states. A clear-cut equilibrium selection result isobtained: the payoff dominant equilibrium emerges as the unique long runequilibrium. Furthermore, the expected waiting time until the payoffdominant equilibrium is reached is relatively short, even in the limit asthe population size grows large.