Abstract: I study a principal-agent problem between an infinitely lived firm and finitely lived workers. Each worker's type is initially unknown to both parties, and high effort helps to better identify it. My model is built on Camargo and Pastorino (2016), who show that probationary periods (commitment to employment) can be part of the equilibria only when the workers' effort has a delayed impact, if both the type space and the output space are binary. I show that if the type space and the output space contain at least three elements, and the workers' outside option gradually increases depending on their past performance, probationary periods are part of the equilibria, even when effort has no delayed impact. This formulation helps to explain environments where a worker who produces medium output for more than one period can secure an alternative job matching his skills, even though the firm is better off switching to a new candidate.
Working Papers
Abstract: I study a dynamic principal-agent problem with limited commitment between a firm and a worker. Both parties live for infinitely many periods, and the worker's cost of production is his private information. I show that the firm can learn the worker's type without firing the high-cost worker, when the production of a certain quality for some finitely many consecutive stages leads to an increase in both the worker's outside option and in the firm's future payoff. Delaying the time when the high-cost worker starts enjoying a higher outside option is crucial in learning. These findings contrast with the results of the ratchet effect literature, in particular with Gerardi and Maestri (2020), who show that there exists no PBE where the firm learns the worker's type without a firing menu, when the parties have a high discount factor.