We analyze a firm's job-assignment and worker-monitoring decisions when workers face occasional crises. Firms prefer to assign good workers to a difficult task and to not employ bad workers. Firms observe failures but only observe successfully resolved crises if they monitor the worker. If monitoring costs are positive but sufficiently small, for a range of probabilities that the worker is good, the firm assigns the worker to a low task (less sensitive to crises) and monitors her. At probabilities below this range and not too much above it, she is assigned to the low task and not monitored. At high probabilities of being good, she is assigned to the di¢ cult task. We analyze the implications for internal labor markets of the case where a worker has the same ex ante probability of being good at all firms and learning is about ability at this particular firm.
