We consider a battle between two groups D (defenders) and A (attackers). Each group has the same number of agents, but they differ by their abilities. Every agent confronts only one opponent from the other group once in either an all-pay contest or a Tullock contest such that the number of the...

We study a best-of-three all-pay auction. It is shown that with values of winning and without values of losing, this auction is less productive (the players' total expected effort is smaller) than the one-stage all-pay auction. However, with different values of losing over the contest's stages...

We find that two-stage contests could be ineffective, namely, there is a higher chance of low-ability players participating (and winning) than high-ability players. However, imposing a fee on the winner can guarantee that the contest will be effective.

We show that a non-Bayesian learning procedure leads to very permissive implementation results concerning the efficient allocation of resources in a dynamic environment where impatient, privately informed agents arrive over time, and where the designer gradually learns about the distribution of...