2009 Poster Sessions : Reputation Mechanisms for Electronic Marketplaces

Student Name : Christina Aperjis
Advisor : Ramesh Johari
Research Areas: Information Systems
Online marketplaces, such as eBay and the Amazon Marketplace, are successful because they are able to engender trust among buyers and sellers who never meet face-to-face. These marketplaces rely on reputation mechanisms to improve the efficiency of trade among buyers and sellers. Reputation mechanisms collect ratings on past transactions of the sellers and provide aggregate information to potential buyers. We consider a fundamental question: how do we design a reputation mechanism to achieve efficient trade? We focus on reputation mechanisms that successfully deter misleading advertisements by sellers. Our main findings are the following. First, weighting all past ratings equally gives sellers an incentive to falsely advertise their items; importantly, this insight supports eBay's recent decision to base the Positive Feedback percentage on the past 12 months of feedback, rather than the entire lifetime of the seller. Second, weighting recent ratings more encourages sellers to advertise honestly when there are increasing marginal benefits to higher reputation for the seller, as is suggested by some recent empirical analyses of online marketplaces. We establish this fact by studying a family of mechanisms that weight recent ratings more heavily than those in the distant past. Finally, we discuss the dependence of the optimal mechanism on the seller's attributes: increasing the number of ratings that are aggregated in the seller's reputation statistic is more likely to make patient sellers honest, while it is less likely to make high quality sellers honest. These results provide qualitative guidance for the improvement of reputation mechanisms in online marketplaces. (Joint work with Ramesh Johari.)

Christina Aperjis is a PhD student in Operations Research at Stanford University. Her research focuses on information economics and incentives. She has worked on electronic marketplaces and peer assisted content distribution.