Selection Incentives and Accountability Traps: A Laboratory Experiment

Abstract

In a laboratory experiment, I evaluate subject behavior in the context of a two-period pure adverse selection model of electoral accountability. The equilibrium predictions associated with standard specifications here require minimal effort investment in both periods and voter retaining the representative if and only if the posterior on representative's type exceeds the expected value of the replacement distribution. I find that subjects systematically deviate from these predictions, but do so in ways that differ significantly across different institutional specifications: (a) with vs. without noise in the outcome function, (b) with smaller vs. greater ranges of representatives’ types, and (c) with smaller vs. greater marginal returns on effort in the outcome function. I construct a model of subject behavior that is consistent with these observations. Jointly with the analyzed behavior, the model sheds doubt on the received wisdom that analytically prioritizes adverse selection over moral hazard considerations in environments with agent type heterogeneity.