Dynamic contracting with long-term consequences: Optimal CEO compensation and turnover

We examine optimal managerial compensation and turnover policy in a principal-agent model in which the firm output is serially correlated over time. The model captures a learning-by-doing feature: higher effort by the manager increases the quality of the match between the firm and the manager in the future. The optimal incentive scheme entails an inefficiently high turnover rate in the early stages of the employment relationship. The optimal turnover probability depends on the past performance and the likelihood of turnover decreases gradually with superior performance. With good enough past performance, the turnover policy reaches efficiency; the manager is never retained if it is inefficient to do so. The manager's compensation depends on the firm value and the optimal performance-compensation relation increases with past performance.