Monotone methods for equilibrium selection under perfect foresight dynamics

This paper studies a dynamic adjustment process in a large society of forward-looking agents where payoffs are given by a normal form supermodular game. The stationary states of the dynamics correspond to the Nash equilibria of the stage game. It is shown that if the stage game has a monotone potential maximizer, then the corresponding stationary state is uniquely linearly absorbing and globally accessible for any small degree of friction. A simple example of a unanimity game with three players is provided where there are multiple globally accessible states for a small friction.