Optimal Information Disclosure in Contests with Communication

We study optimal information disclosure in static contests where players do not know their own values of winning but can learn them, publicly or privately, from the designer. The designer chooses a disclosure policy that maximizes the total expected effort, and commits to it before observing the realized value profile. A distinct feature of our model is that conditional on receiving private information from the designer, contestants are allowed to communicate with each other by sending informative (truthful) or uninformative (empty) messages. As our results show, the contestants have incentives to share their private information with each other if and only if the values of winning are positively correlated. At the same time, learning is rarely perfect because mixing between the two types of messages leads to higher expected payoff in the communication game. Since with a positive probability communication results in an asymmetric contest associated with lower expected effort, the designer prefers concealment to any other disclosure policy available. This result is in a stark contrast with the no communication benchmark where private disclosure is best when the values of winning are sufficiently positively correlated