Handicapping under Uncertainty in an All-Pay Auction

Abstract

A fundamental result of contest theory is that evenly matched contests are fought most intensely, implying that a contest designer maximizes effort from each contestant by artificially boosting the chances of the underdog. Such "handicapping" is credited with making sports contests more exciting, improving efficiency in internal labor markets, increasing effort from students competing to enter college, and raising revenues in auctions. We reexamine the handicapping problem in a two-period contest where the only information available on player ability is performance in the first period. When a contest is perfectly discriminating (i.e. an all-pay auction), the player who exerts the most effort wins, but the weaker player will not participate with some probability, resulting in lower total effort. However, we find that in the two-period contest, handicapping the loser of the first period increases total effort for all ability differences. When the objective is to increase accuracy in identifying the better player, handicapping the winner is optimal.

Keywords: all-pay auction, asymmetric information, handicapping

JEL Classification: C72, D44, D82