A contest architecture specifies how the prize sum is split among several prizes, and how the contestants (who are privately informed about their abilities) are split among several sub-contests. We compare the performance of such schemes to that of grand winner-take-all contest from the point of view of designers that maximize either the expected total effort, or the expected highest effort. An important explanatory variable is the form of the agents� cost functions. The analysis is based on simple but powerful results about various stochastic dominance relations among order statistics and functions thereof.
© 2008-2024 Fundación Dialnet · Todos los derechos reservados