Three Essays in Auctions, Contests, and Leaderboards.

Abstract

In this dissertation, I study three resource allocation problems. In the first study, I present joint work with Yan Chen, Thomas Finholt, and Kan Takeuchi on a laboratory study evaluating three different package auction mechanisms for allocating access to large scientific equipments. I found that mechanisms that allow the bidders to express their intensity of preferences do equally well in maximizing the efficiency of the allocation. This is true regardless if the mechanism allows for multiple rounds of bidding before the final allocation.

In the second study, we evaluate the predictions of a theory of contest in the laboratory. The research question is given a fixed amount of prize money, how can we best partition it to encourage the highest level of total group effort. A theory of contests developed by Moldovanu et al. (2007) predicts that the optimal structure of prizes depend on the distribution of the ability factors of the contestants. Roughly speaking, if the contestants’ ability factors are distributed uniformly, then it is better to combine the prize money into one grand prize. However, if a few contestants have high ability factors but the rest have low ability factors, then it is better to have as many prizes as there are contestants. In our laboratory study, we found that one of the predictions holds but the other fails because when people are unsure of the optimal strategy, they tend to exert more effort.

In the third study, we present an empirical study evaluating the effectiveness of social information in encouraging people to contribute product reviews on Amazon.com. We found empirical evidence that shows that people respond to social information when their action can more readily improve their ranking on a leaderboard.