|
Article Excerpt
1. Introduction: The Problem of Trust in Online Markets
Many online markets rely on electronic reputation or "feedback" systems to promote trust in transactions. The reliance on reputation per se does not distinguish online markets from traditional markets, legal enforcement being in all cases expensive. On balance, however, online markets have more problems with fraud. A recent report by the research group GartnerG2 (2002) concludes that "Internet transaction fraud is 12 times higher than in-store fraud." A U.S. Department of Justice (2002) survey also cites high levels of online fraud, pointing especially at frauds common on auction sites (many with online feedback systems) that "induce their victims to send money for the promised items, but then deliver nothing or only an item far less valuable than what was promised (e.g., counterfeit or altered goods)." (1)
The power of reputation to promote trust in business transactions is closely associated with networked communities, places where there is a good deal of interpersonal communication as well as exchange. Online and traditional markets are networked in different ways. Whether the networking pattern is critical to the amount of trust in the market is a matter of contention. The laboratory experiments we present in this paper investigate how differences in the flow of information--that is, the source of the information and how it disseminates in the market--might differentially influence trust. Our results imply that even if information about reputation is shared and reliable, online feedback systems provide fewer incentives to trust or to be trustworthy than do traditional markets, where long-term relationships play a larger role. The results also suggest some ways in which online incentives might be improved.
In traditional business communities, the patterns of information flow and contact that promote trust often interact in subtle ways. Vietnam's free market reform in the mid-1980s illustrates how effective informal reputational controls can be. At the time, there was little in the way of legal protection against exchange malfeasance, but markets nevertheless flourished. According to McMillan (2002, p. 59), "People in the same line of business would meet each other every day in teahouses and bars ... to discuss the reliability of particular customers.... About half of a sample of entrepreneurs said that they had had no prior connections with the businesses that were to become long-standing trading partners." Thus, in traditionally networked communities, interaction between members promotes trust in two ways. For one, the pattern of interaction promotes long-term relationships; a business partner whose trust has been rewarded is, all things equal, more likely to return to do future business. Second, information about individual reliability is transmitted by word of mouth to third parties, some of whom are prospective future trading partners. (2)
One of the great advantages of online markets is the opportunity to trade with a larger, fluctuating set of partners. This means less reliance on long-term relationships. In data collected from eBay over a five-month period, Resnick and Zeckhauser (2002) found that 89% of all encounters were one-shot. To promote the exchange of information on the reliability of individual traders, many platforms, including Amazon, Cnet, eBay, Half, and Yahoo, have instituted online reputation mechanisms, known as "feedback" systems, to provide the kind of word of mouth available in traditionally networked markets. In fact, online feedback systems have advantages over traditional word of mouth, in that accessing information online does not require personal contacts in the trading community, and in that feedback information from even large numbers of buyers can easily be collected, processed, and disseminated. Recent field studies of online auction platforms find that feedback systems have at least some of the desired economic effect in the sense that reputable sellers are more likely to sell their items (Resnick and Zeckhauser 2002), and can expect price premiums (e.g., Lucking-Reiley et al. 1999). (3)
Amazon's used-books market platform for independent dealers (brick-and-mortar bookstores as well as private individuals) provides a simple but illustrative example of how these systems work. Sellers post the price and a description of the book's condition on Amazon's site. Buyers pay through Amazon, who takes a percentage, but sellers ship directly to buyers. The moral hazard problems inherent in the seller's side of the deal--stipulating the book's condition and the shipping--is addressed through the feedback system in which buyers are invited to post comments on the transaction that future buyers can view when deciding whether to make a purchase. (4)
One way of stating the difference between online and traditional reputation networks is to note that they emphasize different types of reciprocity. Traditional markets rely more on direct reciprocity: "I trust you because you were trustworthy with me before." Online markets rely more on indirect reciprocity: "I trust you because you were trustworthy with others before." In both cases, information about reputation enforces trust by inducing a reciprocal response; past trustworthiness is a prerequisite to future business. The two differ, however, in terms of the flow of information. In particular, in direct reciprocal dealings, traders make decisions based on reputational information culled from their own past transactions, and their present dealings produce information that they themselves will use in the future. However, in indirect reciprocal dealings, reputational information is obtained from others, and others will use information from the present dealings in the future. (5)
Putting our investigation in these terms, we look at how well markets based on indirect reciprocity build and sustain trust in comparison to markets based on direct reciprocity. Game-theoretic models imply that indirect reciprocity can be just as effective as direct reciprocity, so long as traders in indirect reciprocal relationships have access to sufficient information about reputations; that is, game-theoretic models imply what we will call the information hypothesis: It is the information per se, independent of its flow, that matters. Standing in counterpoint to the information hypothesis is an argument that says that the flow of information is, in fact, critical. Granovetter (1985), for example, in discussing market trust, argues that people put more stock in information acquired "from one's own dealings." This suggests that direct reciprocity is a more effective way of developing trust even when sufficient information for indirect reciprocity is present. However, even if so, understanding why the information flows matter might help to improve online trust.
Section 2 develops the information hypothesis, and discusses the counterargument. Section 3 details the design of the experiment. Section 4 lays out results. Section 5 discusses the implications for improving online markets. The appendices include the instructions for our experiments and robustness checks for our statistical analyses.
2. Three Markets and the Information Hypothesis
To put matters on a tangible footing, we describe the fundamental trust problem in the context of a simple market platform. Using this platform as a base, we characterize three specific markets that differ by the flow of information. We describe the information hypothesis in the context of these markets.
2.1. The Basic Trading Platform for Three Markets
In each market, transactions take place over a series of rounds. (In the experiments, of course, there will be a finite number of rounds, but the number of rounds--finite or infinite--is not critical to our basic argument.) At the beginning of each round, a potential buyer is matched with a potential seller. The buyer then chooses whether to purchase an item at a fixed price. If the purchase order is sent, the seller decides whether to ship or simply keep the buyer's money. The moral hazard is that, on receiving the money from the buyer, the seller has no immediate pecuniary incentive to ship the item. Thus, a transaction that is in both parties' interests may be impeded either because the seller proves untrustworthy, or because the buyer, anticipating this risk, chooses not to trust.
Figure 1 illustrates the exact moves in the buyer-seller encounter. Both the seller and the buyer are endowed with 35, which is the payoff when no trade takes place. The seller offers an item for sale at a price of 35, which has a value of 50 to the buyer. The seller's cost of providing the buyer with the item--costs associated with executing the trade, shipping, handling etc., as well as production costs--is 20. (6) Therefore, each successfully completed trade increases efficiency by creating a consumer surplus of 15 and a net profit of 15 for the seller. If the buyer chooses to buy the item, he sends his endowment of 35 to the seller, who then has to decide whether to ship the item. If the seller does not ship, he receives the price plus his endowment of 35 for a total of 70. If he ships, he receives the price minus the costs plus his endowment for a total of 50. If the buyer chooses not to buy the item, no trade occurs. In this sense, the buyers can choose with whom to trade and with whom not to trade.
The three markets are characterized as follows: In the strangers market, individual buyers and sellers meet no more than once and the buyer has no information about the seller's transaction history. Here the moral hazard has full force, because the actions of the seller are not conveyed to future prospective customers. In the feedback market, an online feedback system tracks seller histories of shipping decisions and provides this information to prospective buyers. This affords the type of indirect reciprocity associated with online markets. In the partners market, the same buyer-seller pairs interact repeatedly, in every round. This affords the type of direct reciprocity associated with more traditional markets.
[FIGURE 1 OMITTED]
2.2. The Information Hypothesis
The critical question we investigate is whether the flow of information, in addition to the information content, affects trust and trustworthy behavior. If not, then feedback and partners markets should, in theory, produce just as much trustworthy exchange. If so, then the partners market may have an advantage.
A robust finding of game-theoretic investigations of reputation building is that a market networked for indirect reciprocity can support just as much trustworthy exchange as a market networked for direct reciprocity. That is, standard game theory implies the information hypothesis (our null hypothesis), that information about reputation determines trusting and trustworthy behavior independent of the pattern of the flow of information.
To illustrate the reasoning behind the hypothesis, suppose...
|
