Trust on eBay arises from the same source as it does at the corner store and at Wal-Mart: seller reputation. Where brick-and-mortar merchants build reputations over time by selling brand-name products of verifiable quality, however, "No seller has been long in the electronic marketplace," says Ramsey professor of political economy Richard Zeckhauser. Most on-line sellers have had little opportunity to establish reputations, and few can deploy tools like advertising and favorable word-of-mouth.
But through eBay's Feedback Forum, a so-called "reputation system" that lets users rate and comment on transactionsand assigns a score based on this feedbacksellers are able to build not just track records, but positive reputations. The nature of on-line commerce requires impeccable service: as amazon.com CEO Jeff Bezos once explained, the customer who gets poor service at the corner store tells five or 10 friends, but the one who gets poor service on the Internet tells a million friends. Because the Feedback Forum isn't limited to the captious customer, satisfied buyers are encouraged to praise responsible and honest sellers.
In a controlled field experiment, Zeckhauser and colleagues at the University of Michigan were able to measure the impact of reputation on auction prices and sales. Their results suggest that a seller with a longstanding, superior reputation on eBay will sell more merchandise, and at higher prices, than a seller new to the site.
The researchers collaborated with a vintage-postcard dealer with a near-impeccable reputation on eBay (2,000 positive comments and one negative). Alongside his established eBay User ID (the name under which goods are auctioned), the dealer created seven new User IDs that appeared on eBay as unrelated either to him or to each other. Because they had not sold anything, these new "sellers" had no reputation. The dealer then organized matched pairs of vintage-postcard sets of equivalent value, assigning one set from each pair to his established ID, and distributing the other sets randomly among the new IDs. Finally, he auctioned off the postcard sets during the next 12 weeks, using all eight identities.
The selling prices that the established dealer received were, on average, 8.1 percent greater than those of the new sellers. "This translates to about 12 to 13 percent higher profit, depending on your assumptions about the cost of goods," says Zeckhauser. (In general, when expenses are fixed, profit margins rise faster than selling prices when the latter go up.) Additionally, the established dealer sold 63 percent of his postcard listings (124 sets), and the new sellers 56 percent (111 sets). Zeckhauser recognizes that these profit differentials are fairly small (the median selling price for the sets was $14.99), and suggests that reputation may matter more for expensive goods.
Perhaps the most perplexing of Zeckhauser's findings is the willingness of users to leave feedback at all. According to his study, half of all buyers doand, surprisingly, only 1 percent of those responses are negative. Such voluntary and apparently selfless participation violates conventional economic theories, which generally predict self-interested consumer behavior and free-riding. Zeckhauser ascribes the high participation rate to simple courtesy, akin to saying "Thank you." But he admits the altruism is curious. "The Feedback Forum illustrates Yhprum's Law," he says, explaining that Yhprum is Murphy spelled backward. "Systems that shouldn't work sometimes do."
~Catherine Dupree
Richard Zeckhauser e-mail address: richard_zeckhauser@harvard.edu