7JAMES E. HEYMAN ANDYESIM ORHUNare Ph.D. students at the Haas Schoolof Business, University of California atBerkeley.DAN ARIELYis the Louis Alvarez Renta Professor ofManagement Science at MIT’s SloanSchool; e-mail: [email protected] authors thank Itamar Simonson,Uri Simonsohn, Stacy Woods, DinaMayzlin, Ali Hortacsu, and RobertZeithammer for their stimulatingdiscussions concerning this project.JAMES E. HEYMAN, YESIM ORHUN, AND DAN ARIELY© 2004 Wiley Periodicals, Inc. and Direct Marketing Educational Foundation, Inc.JOURNAL OF INTERACTIVE MARKETING VOLUME 18 / NUMBER 4 / AUTUMN 2004Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/dir.20020AUCTION FEVER:THE EFFECT OF OPPONENTSAND QUASI-ENDOWMENTON PRODUCT VALUATIONShe wide adoption of dynamic second-price auctions as the format ofchoice for Internet-based (online) transactions has created an interest inunderstanding how individuals behave in such environments. The currentwork concentrates on two dynamic effects, which we call quasi-endowmentand opponent effect, and finds that these effects may result in over-bidding.The results of two experimental auctions—one involving hypothetical bidsand the other real-money bids—demonstrate that bids reflect valuations thatinclude the nonnormative influences of the two factors. Quasi-endowmentand opponent effects could lead to the behaviors of repeated bidding andsniping commonly observed in second-price online auctions such as eBay.T8 JOURNAL OF INTERACTIVE MARKETINGINTRODUCTIONAt 4 minutes to 7:00 on January 5, a 7-day eBay auc-tion for an Olympus SLR IS-3 DLX camera ended.This was a typical auction with 10 different bidders,each placing between one and five bids. The personwith the user name Belab submitted five proxy bidsstarting with one for $50.00 on the second day of theauction and culminating with a final bid of $306.00placed 15 seconds before the end of the auction (a pat-tern of bidding called sniping). The example of Belabis interesting because it demonstrates the complexityand richness of behaviors in online auctions such asthose hosted by eBay.Belab’s behavior is not an individual anomaly. In fact,multiple bidding and sniping are more the rule thanthe exception (Ariely & Simonson, 2003; Bajari &Hortacsu, 2003, 2004; Hossain, 2003; Ockenfels &Roth, forthcoming; Rasmusen, 2003; Roth & Ockenfels,2002). The fact that this behavior is common raisesthe following questions: Why do so many bidders sub-mit multiple proxy bids over the course of a single(second price) auction? Was Belab reacting rationallyto new information gained from other bidders’ actions?Was she over-weighting the information she receivedfrom other bidders in the auction compared with theextensive amount of information available commer-cially about this particular camera? Was she trying toconceal her true valuation in her early bids? Was shetrying to fool some of the other bidders? Does this pat-tern of bidding have anything to do with the observa-tion that more people join auctions toward their end?Was her understanding of her valuation for the cam-era changing during the auction? Was the prospect ofnot getting the camera once she had started biddingweighing heavily in her mind, and increasing herwillingness to pay as seen in the cartoon (Exhibit 1)?It has been argued by Ariely and Simonson (2003),among others, that for the average consumer, biddingin an auction is fundamentally different from buying aproduct for a fixed price either on- or off-line. Auctionsare dynamic and competitive. Making decisions insuch an environment could influence the process bywhich bidders come to understand their own valuationof the items in question and their willingness to payfor them. Moreover, auctions present their partici-pants with the task of setting the maximum dollaramount they are willing to pay, which is very differentfrom the more common decision of whether to buy ornot to buy at a fixed and known price.EXHIBIT 1The Joy and Agony of Online Auctions: Competition and AttachmentEvidence demonstrating the influences of theseunique features of online auctions on the prices con-sumers are willing to pay has shown that final win-ning prices are positively related to the total numberof bids and to the total number of bidders (Ariely &Simonson, 2003). It has also been shown that biddersplace too much value on information gathered withinthe particular auction in which they are participating,and are thus unduly influenced by that auction’sreserve price (Häubl & Popkowski-Leszczyc, 2001).Finally, it has been noted that bidders sometimesexhibit herd-like behavior in their pursuit of particu-lar goods (Djolakia & Soltysinski, 2001; Häubl &Popkowski-Leszczyc, 2004). It seems that there is sig-nificant evidence suggesting that bidders’ valuationsfor goods offered at auction change during the lifetimeof the auction, leading to the repeated bidding andsniping behavior that is often observed in online auc-tions (Lucking-Reiley & Reeves, 1999).In the current work we propose two possible causesfor an increased valuation for goods as the auctionprogresses: quasi-endowment and the “opponenteffect.” Quasi-endowment is a sense of ownershipthat bidders develop during an auction, even thoughthey are not the owners in any common or reasonablesense of the word. Opponent effect, on the otherhand, involves an increase in the subjective value ofwinning the auction when the behavior of the otherbidders in the auction is perceived to be competitive.To investigate these two effects, we conducted anexperiment to test people’s intuitions of theirresponses to auction conditions, and conducted a setof real auctions where participants were able to bidwith their own money for a variety of products. Bothexperiments demonstrated quasi-endowment andthe “opponent effect.” To the extent that the experi-ments resemble conditions similar to real auctions,these effects may have significant managerialimplications for the optimal design of auctionmechanisms.The rest of this paper is organized as follows: We pro-vide an overview of second-price auctions, since bothof our experiments use variants of this structure; adetailed conceptualization of the two proposed effectsfollows, along with an overall description of the exper-iments; we then detail the experiments and results,and conclude with