A Misranking/Masquerading-Proof Mechanism for Online Reputation System∗

In e-commerce based online marketplaces, information asymmetry is a big problem. On one hand, the buyers need to learn more than just the price to make decisions whether to buy from a particular seller; on the other hand, the sellers need to distinguish themselves from sellers selling goods of worse quality. Without a well established mechanism to prevent this kind of moral hazard problem, the market tends to collapse. (Akerlof, 1970) Basically, in an online marketplace, the buyers rely a lot on the reputations of the sellers. The reputation can be a result of word-of-mouth, comparison report on a magazine, or rating from the marketplace itself. For big online entities such as Amazon.com, Dell.com, they can charge a premium above the average price of a homogeneous product than other less famous sellers. Reputation sustains the market; in fact, reputation in the online marketplace context is the one and only means to provide information about a particular product’s quality. There is an even more critical problem for online marketplace providers like eBay.com1. Unlike Amazon.com or Dell.com, the sellers in eBay are all anonymous sellers from all over the world, there is no way to find a comparison report on a magazine or to learn about them by word-of-mouth. When a seller turns out to be cheating, there is also not much can be done to him or her. Without a good reputation mechanism, the sellers tend to take the fly-by-night or hit-and-run strategy. One incentive for this research comes from the (somewhat surprising) empirical evidence that the fraud rate is very low among the astronomically large number of transactions. EBay attributes the success to the rating system. Indeed, Lucking-Reiley et al (2000) found that sellers’ reputation points on eBay have a measurable effect on auction prices, and negative ratings has a bigger effect than positive ones. Another very interesting phenomenon we want to capture is the truth-telling behavior of the buyers. Without buyers giving fair ratings, the system will not post correct ratings to reflect the seller’s behaviors, and the market will not sustain. Some effects of unfair ratings are surveyed in Dellarocas (2001b).