Performance Incentives with Award Constraints

This Paper studies the provision of incentives in a large US training organization, which is divided into about 50 independent pools of training agencies. The number and the size of the agencies within each pool vary greatly. Each pool distributes performance incentive awards to the training agencies it supervises, subject to two constraints: the awards cannot be negative and the sum of the awards cannot exceed an award budget. We characterize the optimal award function and derive simple predictions about how award prizes should depend on the number of agencies, on their sizes, and on their performances. Our results indicate that the constraints on the award distribution bind and reduce the overall efficiency of the incentive system.