ournal of Marketing Research - PDF document

ournal of Marketing Research *Dan Ariely is Luis Alvarez Renta Professor of Management Science,Massachusetts Institute of Technology (e-mail: ariely@mit.edu). JoelHuber is Professor of Marketing, Fuqua School of Business, Duke Univer-sity (e-mail: joel.huber@duke.edu). Klaus Wertenbroch is Associate Pro- Kahneman and TverskyÕs (1979) famous dictum thatlosses loom larger than gains implies that people imputegreater value to a given item when they give it up than when When Do Losses Loom Larger Than Gains135the valuations, the buying or selling only occurs probabilis-tically. Novemsky and Kahneman compare probabilisticselling to putting the good in the pot of a poker game,whereas probabilistic buying is analogous to putting moneyinto a poker pot that contains the good. In the former case,the loss of the good may or may not bring about positivemonetary rewards, whereas in the latter, the good may ormay not be acquired. Novemsky and Kahneman find thatthe addition of (balanced) risk to selling does not affect val-uations, implying that there is no risk aversion beyond thestandard loss aversion for parting with a product under cer-tainty. In contrast, the addition of risk to buying creates riskersion with respect to possibly losing that money.Novemsky and Kahneman qualify this latter result becauseit is somewhat less reliable empirically.o methodological issues arise with the examination ofFirst, the probabilistic commerce used in NKÕs article is dif-ferent from the buying or selling of small items that consti-tute the standard tests of the endowment effect. In additionthe novelty of this procedure raises questions about theays that the cognitive processing of such probabilistictransactions differs from that of nonprobabilistic transac-tions. Do these methods invoke different psychological pro-cesses or different scripts? More important, do they invokea different approach and attitude to risk? The analogy to apoker game provides clarity for the reader, but if the respon-dents developed the same analogy, their responses would becolored by their beliefs about gambling in general and hav-ing prized possessions in a poker pot in particular. Giventhe novelty of the task, the appropriate question is whetherthe expressed willingness to trade would change after expe-rience with probabilistic commerce. To the extent that prob-abilistic transactions are somewhat different from certainould benefit from further investigation before a more gen-Becker, DeGroot, and MarschakÕs (1964; hereinafter BDM)procedure. The BDM procedure ensures that all participantshave a (weakly) dominant strategy to reveal their maximumwillingness to pay (WTP) truthfully. Although BDM hasmany desirable theoretical properties, the extent to which(e.g., Wertenbroch and Skiera 2002). Ariely, Mazar, andšszegi (2005) manipulate the price distribution used in theBDM procedure and show that a right-skewed distribution(uniform distribution between $1 and $14, with an addi-tional peak at $25) produces different values than a left-skewed distribution (uniform distribution between $1 and$14, with an additional peak at $.10). The two distributionsaltered WTP, almost tripling the value of a mug between theleft-skewed distribution (mean = $3.33) and the right-skewed distribution (mean = $8.47).In addition to the general questions about how peopleadds another layer of risk to the decision. For example, inthe certainty equivalence condition, the bidder does notknow whether the trade will occur until the random BDMprocedure is completed. This additional risk means that theindifference levels are not determined with certainty butonly with less risk than in risky buying or selling. Toa comparison requires many more respondents because thedistribution of acceptable levels would be known onlythrough the analysis of choices across respondents, but it(with its increased efficiency) and binary choice.In summary, the advantage of probabilistic buying andselling is that it enables new questions to be answered withrespect to buying and selling. However, the very novelty ofAlthough it is possible that the results would remain largelyunchanged, there is a need to test the behavioral stability ofprobabilistic commerce with respect to both experience andcontext, as well as the impact of the additional level of riskWHAT MODERATES LOSS AVERSION FOR GOODS?In this section, we note a few topics that we considerfruitful avenues for further investigation. When we standback and examine the literature on the endowment effect, itis difficult not to be impressed by how much attention thetions. At the same time, it is surprising to observe how littleis known about the psychology that underlies the phenome-non. We draw readersÕattention to two different psycholog-ical approaches to the endowment effectÑemotional attach-ment and changes in cognitive perspectiveÑin the hope thatfuture investigations will follow with more robust psycho-logical investigations.Emotional AttachmentA seemingly plausible mechanism to produce the endow-ment effect is emotional attachment. From this affectiveperspective, reluctance to give up items increases as con-sumersÕattachment to the items increases. An endowmenteffect due to attachment is often adaptive in that progressivexperiences of ownership enable people to adjust other pos-sessions as well as their endowment of skills, akin to build-ing up consumption capital (Becker and Stigler 1977). Wespeculate that the instantaneous endowment effects oftenobserved in the literature (e.g., Kahneman, Knetsch, andrelation to possessions, even when such consumption capi-tal may not have been built up yet (e.g., when respondentshave not had a chance to use their mugs).Strahilevitz and Loewenstein (1998) take the attachment-based approach to the endowment effect seriously and pro-pose that consumers adapt to ownership over time. Theyposit a gradual adaptation process as the consumerÕs psy-chological state moves from no ownership, to partial owner-ship, to complete ownership. On the basis of this proposi-tion, they show that prior ownership can increase the valuethat consumers place on an object. Thus, the value increaseswith the duration of ownership, as does the perceived attrac-tiveness of the owned item. Furthermore, they show thathigher valuations of items once owned persist for someperiod. That is, consumers continue to value an item even 136JOURNALOF MARKETING RESEARCH, MAY2005after they have lost it, and that positive valuation diminishesCarmon, Wertenbroch, and ZeelenbergÕs (2003) theory ofoption attachment provides a more direct test of the attach-theory (e.g., Pham et al. 2001; Schwarz 2001), they proposethat losing an item is experienced as unpleasant, yieldingpsychological discomfort, which positively colors the valu-ation of the threatened item. In support of this theory, theymanipulate the extent to which respondents becomeattached to different options in the choice set they faced asthe choice options, or the degree to which respondents elab-orated on the benefits of the options). The attachment to aconsidered option created a sense of prefactual ownershipsuch that consumers experienced not choosing it as a loss.The resultant mental endowment effect does not requireactual possession. As a result, respondents who developedoption attachment evaluated forgone options more posi-tively than did respondents who did not. Important from theperspective of affect as a driver of the endowment effect,aluations of forgone options were mediated by ratings ofIn a similar vein, the notion of endowment as an exten-sion of attachment has also been examined in the domain ofonline auctions. In this environment, Ariely and Simonsoning that they are the leaders of the auction, begin to thinkAriely and Simonson term a Òpseudoendowment effect.ÓThis analysis, reinforced by other empirical evidence (Hey-man, Orhun, and Ariely 2004), suggests that consumerswho have entered the highest bid and anticipate winning therate it into their psychological, as opposed to real, endow-ment. Thus, when someone else enters a higher bid, thepseudoendowment effect, may increase his or her bidbeyond the initial WTP.In summary, there is evidence that affect-based attach-ment might be a central driver of the endowment effect. Aay to test this proposition would be to manipulate respon-dentsÕaffective attachment (or its salience) to the object of atrade to determine the extent to which it moderates theendowment effect. For example, student respondents shouldbe more emotionally attached to coffee mugs with their ownuniversityÕs logo than to mugs with a different, or perhapsen a competing, schoolÕs logo. Consistent with thisproposition, Dhar and Wertenbroch (2000) find a greaterreluctance among their respondents to give up (affect rich)hedonic than (affect poor) utilitarian items, a result to whichwe return subsequently. A better understanding of attach-ment to objects, over time and as a function of the environ-ment, and reasons for the acquisition would also yield addi-tional insights and clarity to understanding the affectiveunderpinnings of the endowment effect.Changes in Cognitive PerspectiveA cognitive approach to the endowment effect is basedon the difference in tasks for buyers versus sellers and onays that the difference in tasks can alter information pro-cessing. Carmon and Ariely (2000) propose that the buyingthe different perspective on the exchange for buyers versussellers. In particular, they proposed that both sellers anduyers focus on what they stand to give up in a transaction,which means that buyers and sellers focus on differentaspects of the exchange when they provide valuations of anitem that is to be traded. Specifically, they suggest that sell-ers naturally focus on the aspects of the exchange that theymight lose if the exchange occurs (i.e., the item in ques-tion), whereas buyers naturally focus on the aspects of thexchange that they might lose if the exchange occurs (i.e.,the expenditure). This leads sellers to come up with higherprices and buyers to come up with lower prices, thus result-ing in buyerÐseller price gaps. In studies of purchases ofNational Collegiate Athletic Association basketball tickets,Carmon and Ariely find evidence of shifts in cognitive per-spective as measured by verbal protocols, by importanceratings of different aspects, and by implied preferences in aconjoint study. Moreover, they find that manipulation of thesalience of the ticket benefits or of the opportunity costs ofuying the tickets enhances or reduces buyerÐseller priceThe differential perspective account suggests that differ-ent decision-making roles impose a differential focus on theattributes of the transaction (see, e.g., LeBoeuf and Shafir2004), which can moderate the endowment effect. Thisview of the endowment effect leads to other predictionsabout the circumstances in which the endowment effect willand will not be observed. In particular, when the sellersÕgoal is to sell the product to earn revenue, their focus mightbe on the aspects of the transaction that they stand to gainrather than to lose, and in this case, the endowment effectshould not be observed. However, research on the endow-ment effect has typically not examined owners of goodshave induced endowments experimentally, only then to askrespondents to part with their newly acquired possessions.An exception is Simonson and Drolet (2004), who showthat when consumers have decided to sell an item, the pre-ailing market price becomes the primary driver of theirminimum asking price. Further research could uncoveradditional ways that the frame of mind that people bring toa transaction affects the endowment effect.The critical importance of the purpose of the exchange isconsistent with NKÕs P, which states that goods exchangedas intended are not evaluated as losses. We suggest that areason for the proposition to hold is the change in cognitiveperspective with its concomitant emphasis on differentaspects of the trade that change sellersÕevaluations. A casein point is Lerner, Small, and LoewensteinÕs (2004) recentinding that previously induced emotions affect the cogni-tive appraisal objectives with which people approach atransaction. For example, they find that endowment reverseschoosersÕindifference prices exceeded sellersÕwillingnessto accept, thus facilitating the trade (the cognitive appraisalobjective). In support of our view that changes in cognitiveperspective may moderate the endowment effect, it isimportant to note that Lerner, Small, and LoewensteinÕsaffect manipulation is independent of any emotional attach- When Do Losses Loom Larger Than Gains137be traded. Rather, the appraisal objectives induced by affectthat is unrelated to the trade Òpersist beyond the (affect-)ing subsequent situationsÓ (Lerner, Small, and Loewensteinthat direct respondentsÕattention to different aspects andobjectives of the trade could have major moderating effectson loss aversion. For example, different decision frames,ord cues, or differential salience of information about sell-ersÕgoals could all lead to changes in the cognitive perspec-tive from which respondents view the trade.CHMENT, PERSPECTIVE CHANGE, AND LOSSAnother direction for further research is the examinationof what types of experiences or goods will not show theendowment effect, or show less of it. The two possible mod-erators of loss aversion and the endowment effect that wepropose, attachment and perspective change, can be broughtto bear on this question in general and on NKÕs finding ofno loss aversion for money in particular.Ithasbeenarguedthatthelackoflossaversionformoneyarisesbecausemoneyisnotusefulinandofitself,anditsonlyusefulnessisitsabilitytopurchaseothergoodsandexperiences.Thisanalysissuggeststhattherearemostlikelyothercasesinwhichtheendowmenteffectisnotobserved,thatis,whenthetradedgoodsareviewedasmerecurrenciesofexchange.Forexample,inabartersociety,inwhichitemsareexchangedonlytobetradedagainlater,afeelingofendowmentmaybemoderatedbecausethegoodisdesignatedfortransactionandnotforconsumption.Aninterestingaspectofmoneyasacurrencyofexchangeisthewaymentalbudgetsprescribeanintendeduse(HeathandSoll1996;Thaler1985)andhowthiscaninteractwiththeendowmenteffect.More pre-cisely,peopletendtostronglyresistspendingmoneythatisnotdesignatedforagivenuse,andthereforeitsusecanbeinterpretedasaloss.Incontrast,moneythatisintendedtobespentbutisnotÒburnsaholeinoneÕspocket,Óandnotspendingitcanbeinterpretedasaloss.ThisviewofauyerÕsperspectiveonmoneyisthemirrorimageoftheviewthatwepreviouslynotedinwhichasellerwhohasdecidedtosellanitemchangeshisorherperspective,resultinginagreateremphasisoncompletingthetransac-tionandarelativelylowerevaluationoftheitem(e.g.,DroletandSimonson2004;Lerner,Small,andLoewen-stein2004).Novemsky and KahnemanÕs finding that there is no lossersion for money that is held for transaction purposes isalso consistent with recent findings that suggest that themagnitude of the endowment effect is different betweenhedonic and utilitarian goods. Hedonic goods and attributesare those that are primarily consumed for the positive affec-tive experience and pleasure that they provide. Utilitariangoods and attributes are those that are primarily consumedfor instrumental purposes. Dhar and Wertenbroch (2000)show that consumers weigh hedonic attributes more heavilywhen they decide which one of two options (a hedonic anda utilitarian item) in their possession to give up (a forfeiturechoice) than which one of the same two options to obtain(an acquisition choice). They propose that this is becausehedonic goods are affectively richer sources of prefactualthinking about what not having an item would be like. Animplication of the findings is that there is less referencedependence for money because money that is held to begoods that Dhar and WertenbrochÕs (2000) findings implysuggests that the affective content of the traded or chosengood enhances loss aversion. In turn, the affect associatedwith the traded item may be linked to how attached con-sumers feel about the item and may mediate the effect ofattachment on loss aversion. Carmon, Wertenbroch, andZeelenbergÕs (2003) and Ariely and SimonsonÕs (2003)indings of mental or pseudoendowment effects as a func-comfort from giving it up thus provide another clue thatemotional attachment, in addition to cognitive perspectivechange, is a key driver of the endowment effect.conclude, NK have done the field an excellent serviceby specifying some boundaries of loss aversion. Their worksuggests that there is no loss aversion for money when con-sumers intend to spend the money. By carefully defining thelimits of loss aversion, they caution the field not to overuseloss aversion as a theoretical account. We propose possibleadditions to these limits by exploring two additional factorsthat both mediate and moderate loss aversion, emotionalattachment and cognitive perspective change. Neither ofthese provides a comprehensive explanation of loss aver-sion. However, we hope that they help focus future investi-gations of the limits and origins of loss aversion.Finally, it is important to observe that research on theendowment effect has tended to focus on small and rathermeaningless goods. The appeal of these is obviousÑsim-plicity, cost, and manipulability. At the same time, the pos-sibility of a better theoretical understanding of the endow-ment effect should encourage the examination of products,ollowing this path, it might also be useful to examine adramatic endowment effectÑfor example, loss aversionwith respect to a personÕs child. On the one hand, it is diffi-cult to imagine an offer that would be sufficiently attractivethat a person would be willing to give up his or her child.that particular childÕs parent, the amount he or she would bewilling to pay for the privilege of parenting the child islikely to be very small (if at all positive). Given the strengthof a child-based endowment effect, the examination of rela-tionships and traded goods that are not common marketgoods might help clarify the endowment effect and its psy-Ariely, Dan, Nina Mazar, and Botond Kšszegi (2005), ÒPrice-Sensitive Preferences,Ó working paper, Sloan School of Man-agement, Massachusetts Institute of Technology.ÑÑÑandItamarSimonson(2003),ÒBuying,Bidding,Playing,orCompeting?ValueAssessmentandDecisionDynamicsinOnlineAuctions,ÓurnalofConsumerPsychology13,113Ð23.Becker, Gary and George Stigler (1977), ÒDe Gustibus non EstDisputandum,Ó American Economic ReviewBecker, Gordon M., Morris H. DeGroot, and Jacob MarschakMethod,Ó Behavioral Science 138JOURNALOF MARKETING RESEARCH, MAY2005Carmon, Ziv and Dan Ariely (2000), ÒFocusing on the Forgone:Why Value Can Appear So Different to Buyers and Sellers,Óournal of Consumer ResearchÑÑÑ, Klaus Wertenbroch, and Marcel Zeelenberg (2003),ÒOption Attachment: When Deliberating Makes Choosing FeelLike Losing,Ó ournal of Consumer ResearchCoombs, Clyde H. and George S. Avrunin (1977), ÒSingle PeakedFunctions and the Theory of Preference,Ó Psychological ReviewDhar, Ravi and Klaus Wertenbroch (2000), ÒConsumer ChoiceBetween Hedonic and Utilitarian Goods,Ó ournal of MarketingResearchsumer Decisions,Ó ournal of Consumer ResearchHeyman, James, Yesim Orhun, and Dan Ariely (2004), ÒAuctionFever: The Effect of Opponents and Quasi-Endowment onProduct Valuations,Ó ournal of Interactive MarketingKahneman, Daniel, Jack L. Knetsch, and Richard H. Thaler(1990), ÒExperimental Tests of the Endowment Effect and theCoase Theorem,Ó ournal of Political EconomyÑÑÑ and Amos Tversky (1979), ÒProspect Theory: An Analysisof Decisions Under Risk,Ó LeBoeuf, Robyn A. and Eldar Shafir (2004), ÒAlternating SelvesInconsistency,Ó working paper, Department of Psychology,Princeton University.Lerner, Jennifer S., Deborah A. Small, and George Loewenstein(2004), ÒHeart Strings and Purse Strings: Carryover Effects ofEmotions on Economic Decisions,Ó Psychological ScienceNovemsky, Nathan and Daniel Kahneman (2005), ÒThe Bound-aries of Loss Aversion,Ó ournal of Marketing ResearchPham,MichelTuan,JoelCohen,JohnPracejus,andDavidHughes(2001),ÒAffectMonitoringandthePrimacyofFeelingsinJudg-ment,ÓurnalofConsumerResearchSeptember),167Ð88.Schwarz, Norbert (2001), ÒFeelings as Information: Implicationsfor Affective Influences on Information ProcessingÓ in of Mood and Cognition: A UserÕs Guidebookand Gerald L. Clore, eds. Mahwah, NJ: Lawrence ErlbaumSimonson, Itamar and Aimee Drolet (2004), ÒAnchoring Effectson ConsumersÕWillingness-to-Pay and Willingness-to-Accept,Óournal of Consumer ResearchStrahilevitz, Michal A. and George F. Loewenstein (1998), ÒTheEffect of Ownership History on the Valuation of Objects,Ó nal of Consumer ResearchThaler, Richard (1985), ÒMental Accounting and ConsumerChoice,Ó Marketing ScienceTversky, Amos and Daniel Kahneman (1991), ÒLoss Aversion inRiskless Choice: A Reference-Dependent Model,Ó sumersÕWillingness to Pay at the Point of Purchase,Ó ournal ofMarketing Research