Contest Design with Stochastic Performance

This paper studies optimal contest design in contests with noisy performance. Here, contest design is a team moral hazard problem that endogenizes the assignment rule that maps performance profiles into winning probabilities. The optimal design features endogenous standards for eligibility, and the number of prizes that are awarded may be stochastic. Generally, one group of agents is identified as “first claimants” of prizes, contingent on performance exceeding a threshold of excellence. However, which group wins prizes more often depends on the designer’s objective function and the performance technologies. Finally, the approach derives endogenous, microfounded, and fully optimal contest success functions. (JEL D44, D82)