In this dissertation, we explore the optimal design of group contests, with a primary application to organizational economics. A group contest is a strategic situation in which individuals belonging to a team expend costly effort with the hopes of increasing their team's chance of receiving a valuable prize, such as a bonus or award. Managers often employ such contests with the goal of incentivizing high effort from workers. All chapters of this dissertation are concerned with addressing the following questions: When workers differ in their ability, how should a manager sort workers into competing groups with the objective of maximizing firm productivity? Should the manager promote competitive balance, i.e., form teams that are evenly matched or balanced? Or, alternatively, form teams by grouping workers who are similar in abilities, thereby creating a situation where some teams have an inherent advantage over others, i.e., form teams that are unbalanced? .;In the second chapter, we attempt to address these questions by developing a theoretical model which allows for an arbitrary degree of complementarity in the team production process and an arbitrary convex cost of effort. We find that the optimal sorting is not universal, and depends crucially on the degree of complementarity between team members efforts as well as the shape of the individual cost of effort function.;In the third and fourth chapters, we conduct a series of laboratory experiments designed to test the comparative static predictions regarding the optimal sorting of workers under two special cases of our model. Namely, assuming a moderately steep individual cost of effort, (i) when there is no degree of complementarity in team production, a balanced sorting of workers produces higher output than an unbalanced sorting, and (ii) when the degree of complementarity between team members is extreme, an unbalanced sorting is optimal.;The results of our experiment agree with the theoretical predictions, but with various degrees of success. We find strong support of the theory for case (i), as average total output under the balanced sorting was 24.7% higher than the unbalanced sorting, a result close to the theoretically predicted difference of 27.6%. As for case (ii), we find that average total output is higher under the unbalanced sorting than the balanced sorting as predicted; however, the difference is not statistically significant and only half as large as theory predicted. Thus, we only find weak support of the theory for case (ii).;Because we felt that communication plays an important role in the team production process, we ran an additional set of experiments in the same environment as case (ii), with the exception that participants now had a chance to chat with their team member prior to making their effort decisions. Interestingly, we find that average total output under the balanced sorting is 33% higher than under the unbalanced sorting, a large reversal of the theoretical prediction!;In light of our experimental findings, we conclude that a manager who wishes to maximizes firm productivity should use a balanced sorting. This is in line with the "common wisdom" that promoting competitive balance can bolster competition between competing teams, thereby leading to a high level of overall contest output.
展开▼
