ANN ARBOR—Paul Resnick, an associate professor in the School of Information at the University of Michigan, collaborated with Harvard’s Kennedy School of Government in a new study affirming the value of reputation for people and entities selling items on eBay. It is the first known controlled study of eBay sales to quantitatively measure the monetary advantage for sellers who establish and maintain high marks on the auction website’s online feedback system. Buyers are willing to spend 7.6 percent more for items offered by a seller with a solidly positive eBay reputation, according to the results of a controlled experiment involving sales of vintage post cards. Richard Zeckhauser, the Frank Plumpton Ramsey Professor of Political Economy at Harvard’s Kennedy School of Government, conducted the study in conjunction with Kate Lockwood, Resnick, and John Swanson, vintage postcard seller, registered on eBay. The study found that buyers reward sellers who have built an established positive eBay identity. But these buyers are willing to discount a small amount of negative feedback among newcomers, especially when the feedback is brief or non-specific. Previous studies by Zeckhauser and Resnick have found that half the buyers on eBay provided feedback following a transaction, and sellers received negative feedback only 1 percent of the time. "eBay sellers value their reputations and eBay buyers value working with those who have an established track record of fair dealing and timely delivery with their customers," said Zeckhauser. "In the online world, many of the traditional ways to establish a reputation are impossible. eBay has created a system where traditional word-of-mouth recommendations can be broadcast to thousands of potential buyers. This study indicates that their system works." "For a reputation system to create incentives for good behavior, buyers need to reward sellers who have better reputations," said Resnick. "Our study showed that buyers were in fact more likely to buy from a seller with an established reputation, and to pay slightly more." In the first stage of the controlled experiment, 200 matched pairs of vintage postcard collections were sold under an extremely high reputation identity and seven other newcomer identities with little to no established reputation. Each seller identity offered the postcards at the same selling price, tax, shipping cost, payment methods and with comparable text descriptions. Following each purchase, bids placed, final selling price of the items, and feedback of both the buyer and seller at the time of the auction were analyzed. The findings illustrate that the established identity (with high reputation) experienced an increase of 7.6 percent of revenue compared to newcomers with little reputation. The second phase of the experiment compared sales of newcomer identities with and without negative feedback. Researchers bought items from three of the seven newcomer identities and gave each one to two negative comments, thereby reducing their overall score. Thirty-five matching items were then sold under the newcomer identities to compare those with and without negative comments. In phase two, researchers found that negative feedback in the brief reputation of new sellers did not affect their prices. Additional information on the study may be access through the Kennedy School of Government Website at: http://ksgnotes1.harvard.edu/research/wpaper.nsf/rwp/RWP03-007?OpenDocument For more information about Paul Resnick, see www.si.umich.edu/~presnick. To learn more about the U-M School of Information, visit www.si.umich.edu/

E-mail: [email protected]

www.si.umich.edu/~presnick

www.si.umich.edu/[email protected]