Equilibrium Contracts and Boundedly Rational Expectations

Abstract: We study a principal-agent framework in which the agent has a misspecified model of the principal's project. She fits this model to the objective probability distribution in order to predict outcomes under alternative actions. Under mild restrictions, the agent has correct beliefs on the equilibrium path, but may incorrectly extrapolate how off-equilibrium actions map into outcomes. The agent's misspecification may relax incentive compatibility or even eliminate the incentive problem. We examine optimal workplace design when the principal has discretion over the agent's subjective model, and obtain new results on technology choice, transparency, narcissistic leadership, and relative performance evaluation.