Principal–agent models where a principal is only affected by extreme performances

Abstract

We consider an independent agents setting, where only the highest or lowest achievement affects the principal. An example of highest achievement scenario is the output of parallel research and development (R&D) teams, when only the highest/ best achievement matters. An example of lowest achievement could be an assembly system, where the number of product units that can be assembled equals the lowest number of adequate parts of any component. The findings show that when only extreme performances matter, a principal employs weaker incentives for multiple agents. Furthermore, the reward parameter is especially small if only the lowest achievement is relevant. | INTRODUCTIONIt is often assumed that a principal (P, "she") who employs several "independent parallel" agents (As, "he") benefits from the sum of their outputs. P then wishes to create incentives so as to maximize the (expected) sum of outputs net of the rewards paid to the agents (see,