How Large Looms the Ghost of the Past? State Dependence versus Heterogeneity in Coordination Games

Abstract

In games with multiple, Pareto-rankable equilibria and repeated play, does a history of playing an inefficient equilibrium make it harder for players to reach the efficient equilibrium? In other words, can people “get stuck” in bad equilibria? Previous studies have found support for this, but they have relied on naturally occurring variation in precedent. I implement randomized control to establish that precedent effects are important, but that naturally occurring variation exaggerates the importance of precedent. I present evidence that some of the endogeneity of naturally occurring precedents is due to variation in risk attitudes. This is because in the coordination games used, the inefficient equilibrium is associated with a safe strategy. Understanding the causal effect of precedent is important since many development problems are viewed as coordination games. Moreover, an appreciation of the way in which potential heterogeneity may interact with the policy is essential when trying to lift groups out of bad precedents.