Contracting with Informed Agents

University Of Chicago, Chicago IL

Investigators

Abstract

Empirical work on pay for performance shows little support for the available theories. This proposal reasons why this is so. First, agency theory predicts that pay for performance should be less common in uncertain environments. However, the data do not seem to support this assertion. I argue that when workers operate in certain settings, the activities that they should engage in are well known, and firms are content to assign tasks to workers and monitor their inputs. By contrast, when the situation is more uncertain, firms know less about how workers should be spending their time. As a result, they delegate responsibility to workers but, to constrain their discretion, base compensation on observed output. Second, agency theory has had little to say about the control of bureaucratic corruption. I argue that standard means of relating pay to bureaucratic performance are unlikely to be successful and that instead, oversight occurs through independent investigations. This project outlines a series of behavioral responses to the use of investigations that limit the use of incentives for bureaucrats.