Confidence and Organizations

Miscalibrated beliefs generally compromise the quality of workers' decisions. Why might a firm prefer to hire an individual known to be overconfident? In this paper, I explore the role of such biases when members of the organization disagree about the right course of action. I present a model in which an agent uses his private information to make a choice on behalf of a principal. In this setting, I consider what I call the belief design problem: how would the principal like the agent to interpret his observations? I provide conditions under which the solution indicates a preference for a well-calibrated, an underconfident, or an overconfident agent. A well-calibrated agent is preferred if and only if his information does not affect the expected difference in the players' preferred actions. Overconfidence is optimal when the principal seeks to adjust actions beyond what a well-calibrated agent would do.