Economists accept a positive level of fraud as a result of information costs (Darby and Karni1973). If a consumer possessed suﬃcient knowledge of seller characteristics and the qualities of goods and services, he could not be defrauded. However, it is costly to discover these data andthe consumer must make choices based on incomplete information. Because of this, misleadingseller claims or advertising could fraudulently persuade the consumer to buy something heotherwise would not.The key questions are what information is costly, what contributes to the informationcosts, and what institutions are more eﬀective in reducing the information costs. Most literatureinvolving uninformed buyers limits the buyer’s uncertainty to information regarding speciﬁcsellers’ characteristics. For example, though buyers may not know the quality of goods producedby a speciﬁc seller, they are assumed to so sophisticated that they know the distribution of qualityof all goods for sale in the market.
This paper presents evidence for a more fundamental lackof information: some consumers may not even possess correct beliefs on the distributions in themarket.To distinguish these naive consumers from the traditional sophisticated but uninformedbuyers, we watched the online market for baseball cards to identify consumer behavior patternsand designed an experiment to test if those patterns corresponded to rational behavior. Aseven-month eBay observation revealed 35-52% higher prices were paid to sellers claiming to beselling high quality cards (as opposed to modest or no claims). This would be perfectly rationalif sellers making grand claims actually delivered better cards or provided more reliable service.Such was not the case.Immediately following the observation period, we purchased the same types of ungradedcards and had them professionally evaluated. These cards were systematically purchased sohalf of the sample came from high claims sellers and the other half from modest or no-claimsellers. In a reversal of the rational expectation story, sellers making the best claims were morelikely to default (no delivery) or send counterfeits. Even conditional on actual deliveries of authentic cards (as validated by the professional grading service), the average quality from highclaims sellers was no better than the average for the group with more modest claims. Further
This assumption is explicit in Akerlof (1970), Shapiro (1983), and many follow-up papers summarized in Tirole(1988). To be precise, buyers in Shapiro (1983) are even over-sophisticated: they anticipate inﬁnite supply of hit-and-run sellers, and therefore refuse to pay non-zero price for non-reputable sellers, although some non-reputablesellers are new honest ones. See footnote 7 for more citations on informed and uninformed buyers.