I study a dynamic principal-agent model in which the agent continuously works on a project which may yield a success. The principal cannot observe the success, but she observes imperfect signals over time after the agent stops working. Making payments at a later time increases the principal's informativeness, but is also more costly due to the agent's relative impatience. I derive optimal contracts in two different settings. First, if success is observed by the agent, he is induced to exert full effort until success and report it truthfully. The principal makes deferred payments after the agent's report of success. Before the report, the principal makes an increasing flow payment starting from a certain positive time. Secondly, if success is unobserved by the agent, the principal sets a deadline and makes a deferred payment after the deadline. To reduce the agent's procrastination rent, the principal either terminates the project randomly or makes payments before the deadline, depending on the information structure.
I study a discrete time principal-agent model where the agent's effort and ability are both private information. The wage is exogenously fixed and the principal designs a firing policy to incentivize the agent to work. In each period, the agent works on a project with binary outcomes. The high type has a higher probability of getting a good outcome than the low type conditional on high effort. The outcome in each period is publicly observed. In the optimal contract, the principal hires the high type for sure and hires the low type with some probability. Conditional on being hired, the high type faces a higher standard of performance.
I study a dynamic model of delegated decision making with adverse selection and imperfect monitoring. In each period, a principal may delegate to a biased agent who has better information. The quality of the agent's information depends on his ability. In the optimal mechanism where the agent's ability is publicly observable, the principal delegates to the agent at the beginning of their relationship and the agent behaves in the principal's interest. Depending on the history, the principal either commits to delegating forever or stops delegating eventually. When the agent's ability is private information, the optimal mechanism features pooling at the top. The principal offers the same mechanism to the agent if his ability is known to be above a cutoff.