Dynamic tournament design: Evidence from prediction contests

Online contests have become a prominent form of innovation procurement. Contest platforms often display a real-time public leaderboard to provide performance feedback. The impact of information disclosure on players’ decisions is theoretically ambiguous: some players may get discouraged and quit, while others may decide to keep working to remain competitive. We investigate the impact of a leaderboard on contest outcomes using two complementary approaches. First, we estimate a dynamic model using observational data and compare the equilibria with and without a leaderboard. Second, we present experimental evidence from student competitions. We find that a leaderboard on average improves competition outcomes.