This paper provides a rationale for the use of up-or-out contracts. We consider a learning game in which a firm and a worker commonly observe noisy performance signals correlated with the worker’s ability. A worker’s choice of effort affects the accuracy of the inference process about ability. The greater the effort, the more informative the output signals are. Output is observable but nonverifiable, while effort can be either observable but nonverifiable, or unobservable. In particular, retention decisions based on performance cannot be part of a formal contract. We model contracts as a commitment, on the part of the firm, to a sequence of retention decisions and to a wage schedule conditional on employment. In this setting, contracts act as mechanisms to induce workers to generate nonverifiable information about their ability. We identify conditions under which an up-or-out contract is offered in equilibrium. |
