Confidence and Organizations

Abstract: Miscalibrated beliefs are widely viewed as compromising the quality of employees' decisions. Why, then, might an organization prefer to hire an individual known to be overconfident? This paper develops a theory of organizational demand for employees' levels of confidence when private information interacts with conflicts of interest. I study a model in which an employee uses private information to make decisions on behalf of the organization and analyze the belief design problem, namely, how the organization would like the employee to interpret his observations. I show that organizations prefer employees whose actions reflect a constant expected conflict of interest across observations. A well-calibrated employee is optimal if and only if private information does not affect this conflict. When the conflict varies with information, organizations optimally select employees whose confidence distorts their responses to information. Overconfidence is optimal when the organization seeks stronger adjustments to information than a well-calibrated employee would provide.