Incentive Pay, Fraud, and Self-Reported Performance Measures

Abstract

Executives generally receive incentive pay based on accounting metrics reported by the executives themselves and not otherwise directly observable. The recent prevalence of accounting scandals has led many researchers and practitioners to question the wisdom of these incentive packages. In this paper, I explore the impact of self-reported performance measures by analyzing a principal-agent model in which the principal cannot observe performance directly but instead relies on the agent to report results. In the basic model, I show that optimal incentive pay is lower when self-reporting is used, thereby providing a novel reason why incentive pay is typically lower than other principal-agent models predict. The model also provides a justification for the common practice of making bonus pay proportional to salary. In an extended model with multiple agent types, I show that under some assumptions, the optimal contract provides incentive for almost all agents to falsify results. Under other assumptions, one can replace any such contract with one that relies solely on externally verifiable data, without loss of profit. This model contributes new insights into the advantages of highly leveraged firms such as LBOs. Finally, an extended model with a private benefit for truth telling predicts the observed positive correlations between incentive pay, fraud, and industry or growth rate. Overall, the model contributes towards the current debate over whether incentive pay is too high by offering the following implications. First, incentive pay might be too high if firms overestimate the reliability of performance metrics. Second, fraud might be higher than necessary if firms overlook alternative directly observable metrics. Third, total realized pay might be too high if firms overlook the possibility of imposing fraud fines. Finally, the prevalence of fraud and the positive correlation between fraud and incentive pay can be consistent with an equilibrium model in which firms maximize profits. Thus, these observations alone do not imply that incentive pay is too high.