Knowing the Gap - Intermediate Information in Tournaments ∗

We consider a simple tournament model with two stages and compare two information conditions: Either intermediate information on the subjects’ relative positions after the first stage is available or it is not. According to our theoretical predictions, equilibrium behavior in both stages is not affected by intermediate information. In laboratory experiments, we also find no difference between average first and second stage efforts in both treatments. With intermediate information, however, subjects adjust their effort to a higher extent. Subjects who lead tend to lower their second stage effort, subjects who lag still try to win the tournament. These results do not differ from the respective results with and without intermediate information when we introduce costly intermediate information. Overall, intermediate information does not seem to endanger the effectiveness of rank-order tournaments: incentives do neither break down nor does a rat race arise.