Learning and Hysteresis in a Dynamic Coordination Game

This paper introduces a dynamic coordination game with incomplete information defined by a state variable that evolves stochastically. Incomplete information enables us to use iterated dominance argument in order to resolve the indeterminacy issues. The key endogenous variable is the belief that each agent holds about the state of the world. We show that as agents update their heterogeneous beliefs through learning sequentially, they adjust their beliefs to justify the status quo. This effect induces equilibrium actions that support the status quo, a property we call hysteresis.