Over-workers and Drop-Outs in Competitions : Contests with Expectation-Based Loss-Averse Agents

Abstract

Competition is often presumed to enhance performance, however, empirical evidence points out bi-modal effort provision in competitive environments: over-workers and drop-outs. To address this discrepancy, I analyze a multiple-prize contest with expectations-based loss-averse contestants. The predictions of the model are able to align the observed behavior on effort provision, which is hard to reconcile under classical preferences. In particular, I show that high-ability players, holding high expectations, exert effort aggressively while low-ability players, holding low expectations, exert little or no effort in comparison to the standard theoretical predictions. Furthermore, the optimal prize allocation differs markedly in the presence of loss-averse contestants. I show that awarding multiple prizes can become optimal in the cases where a single prize is predicted to be optimal under standard preferences. Intuitively, awarding an additional prize motivates drop-outs to exert effort, while it de-motivates over-workers due to lowered competition. Which of these two countervailing forces dominates, and thus the optimal allocation of prizes, hinges on the interplay between the number of competitors, the ability heterogeneity of competitors, and the degree of loss-aversion. These results have significant implications for the optimal design of organizational environments: muting competition by increasing the number of prizes and decreasing prize inequality can increase total output.