Bringing (Contingent) Loss Aversion Down to Earth — A Comment on Gal & Rucker’s Rejection of “Losses Loom Larger Than Gains” Itamar Simonson Ran Kivetz Stanford University Columbia University Accepted by Sharon Shavitt, Associate Editor Although we disagree with some of Gal and Rucker’s (2018 – this issue) specific evidence and with their over- stated conclusion regarding loss aversion, their overarching message makes a worthwhile contribution. In par- ticular, loss aversion is less robust and universal than has been assumed while its most prominent empirical support — the endowment effect and the status quo bias — is susceptible to multiple alternative explanations. Instead of accepting loss aversion as true unless proven otherwise, we should treat it like other decision prop- erties and psychological accounts that are contingent on various moderators and call for an analysis of psy- chological mechanisms. In this commentary, we suggest that gatekeepers, such as reviewers, tend to favor loss aversion and other widely accepted tendencies, while demanding a much higher support-threshold for alternative or newer accounts. Although building on prior theories and concepts is of course important, the bias in favor of incumbent assumptions can impede scientific progress, bar new ideas from the literature, and reinforce well-established but contingent notions that may apply under some conditions but not others. Keywords Behavioral economics; Loss aversion; Judgment; Decision making; Behavioral decision theory; Contingent judgment properties Introduction: the Limitations of the Extant particular, as discussed further below, G&R’s Empirical Support for Loss Aversion claimed support for an endowment effect without loss aversion (using what is referred to as the “re- After reading David Gal and Derek Rucker’s target tention paradigm”) is itself open to rival accounts. article (2018 – this issue), we were skeptical, in More importantly, G&R unconvincingly and super- large part because like most scholars in our field ficially try to explain away the strongest illustration we have accepted loss aversion as unquestionably of loss aversion, namely, the typical extreme risk true and robust as any decision-making property aversion exhibited in choices between a certain gain can be. We still believe that the loss aversion ten- and a mixed, high expected-value gamble. dency plays an important role in many decisions, Further, some of the evidence proffered by G&R but we agree that loss aversion is not omnipresent appears to actually support loss aversion; for exam- or universal, and when relied upon, should be sub- ple, G&R cite their own finding that people enjoy ject to tests like other plausible accounts. Gal and gaining things that are lost (e.g., finding a lost Rucker (G&R) also correctly point out (see their flashlight) more than they enjoy gaining new things table 1) that loss aversion has received “credit” (Gal & Rucker, 2017), which suggests that losses do (although not from Kahneman and Tversky) for loom larger than gains. Still, consistent with G&R’s phenomena to which it does not apply; for exam- general message, treating loss aversion as another ple, framing effects, such as the well-known Asian mortal generalization that requires psychological Disease problem, represent a reflection effect rather analysis and, like other psychological generaliza- than loss aversion. Having said that, we do not find tions, is contingent and is sometimes violated offers persuasive some of the specific evidence that Gal progress and can bring down to earth what has and Rucker rely on to reach their conclusion that been treated as almost an axiomatic, unquestionable losses generally do not loom larger than gains. In (decision making) law of nature. As an aside, Gal and Rucker’s “strong version” of loss aversion is a Received 20 March 2018; accepted 20 March 2018 Available online 30 March 2018 Correspondence concerning this article should be addressed to Itamar Simonson, Graduate School of Business, Stanford Univer- © 2018 Society for Consumer Psychology sity, 655 Knight Center, 94305 Stanford, CA, USA. Electronic All rights reserved. 1057-7408/2018/1532-7663 mail may be sent to [email protected] DOI: 10.1002/jcpy.1046 2 Simonson and Kivetz rhetorical strawman: we are unaware of any empir- not prove that some decision phenomenon is univer- ical phenomenon in the social sciences that always sal, and the observation of seemingly similar JDM holds, and presenting it as a viable option may cre- and choice phenomena among other species (Shafir, ate the wrong impression that the “weaker version” Waite, & Smith, 2002; Wong & Jennions, 2003) is they advocate represents a compromise view. open to various interpretations. Tversky and Kahneman (1991, p. 1047) described As G&R point out, we do not know much about the impact of loss aversion succinctly without elabo- the psychological mechanisms underlying loss aver- rating on the underlying psychology: “The basic sion, which may be due in part to the perceived intuition concerning loss aversion is that losses (out- status of loss aversion as unquestionably true. As comes below the reference state) loom larger than detailed below, we agree with G&R that loss aver- corresponding gains (outcomes above the reference sion needs to be treated like other empirical gener- state).” The evidence presented in that 1991 article alizations and studied like other contingent revolves primarily around manifestations of the phenomena that are sometimes violated. endowment effect and the status quo bias, although Thus, consistent with a key implication of the other demonstrations of loss aversion are also pre- G&R discussion, the question relevant at the pre- sented. As G&R and other researchers have pointed sent time for our field is not whether loss aversion out, both the endowment effect and the status quo occurs on average (we think it does), but what fac- bias have additional viable explanations, which tors moderate its presence and magnitude, and make these phenomena less than ideal as the flagship relatedly, what are its boundaries? There are many sources of evidence for loss aversion. Furthermore, other tendencies, perhaps not as sweeping, that as they point out, the status quo bias and the endow- may apply more often than not but have been trea- ment effect are violated under certain conditions. ted all along as contingent and as calling for speci- Beyond the endowment effect and the status quo fic psychological explanations. Thus, conformity, bias, there is certainly additional evidence for loss the desire for fairness, preference for compromises aversion (Kahneman & Tversky, 1979); for example, and asymmetrically dominating options, variety the tendency of most people to reject a gamble seeking, and the desire for others’ approval are all involving a 50% chance to win $200 and a 50% observed in many cases, yet questions about their chance to lose $100 is less open to rival accounts. psychological mechanisms, applicability, modera- Unlike G&R, we do not find persuasive their inter- tors, and boundaries came to the fore as soon as pretation of such choices in terms of the status quo they were introduced. bias, which hinges on the presumption that not Conformity, for example, is not a surprising phe- accepting the bet is the status quo option; this nomenon, is perhaps overly intuitive, and it lacks assumption appears ad hoc and is unsupported. But the mystique of loss aversion. Indeed, Asch’s various loss aversion findings do appear contingent famous experiments (and other studies) notwith- and are sometimes violated, as discussed below, standing, known violations of conformity (e.g., which reinforce the need to treat loss aversion as a need for uniqueness) have probably stimulated normal tendency that is sometimes violated and call more research than supportive findings did. Loss for greater attention to moderators, boundary condi- aversion, on the other hand, was quickly accepted tions, and psychological mechanisms. as true and as a major contribution of prospect the- Loss aversion has had, and continues to have, a ory, and it has therefore become the default account special status among the insights offered by the judg- for phenomena that might also be explained by ment and decision making (JDM) field, better known other mechanisms. For example, we agree with today as behavioral economics. In fact, loss aversion G&R that the phenomena that are the primary has been elevated to something akin to a law of nat- sources of support for loss aversion — the endow- ure that might be hard-wired from birth and may ment effect and status quo bias — have additional hold also for some other species. It represents a fun- viable explanations (although loss aversion is prob- damental and all-encompassing driver of judgments ably one contributor). and decisions — the weight of losses compared to gains. Monkeys show loss aversion (Chen, Lakshmi- narayanan, & Santos, 2006), and fMRI studies can Rival Accounts for the Endowment Effect find evidence from brain activity that has been inter- preted as consistent with the presence of loss aver- As indicated, the endowment effect has arguably sion (Tom, Fox, Trepel, & Poldrack, 2007). It is been the most prominent illustration of the conse- noteworthy that “monkey-evidence” by itself does quences of loss aversion. For example, Kahneman, Bringing (Contingent) Loss Aversion Down to Earth 3 Knetsch, and Thaler (1991, p. 197) concluded: “This to isolate the impact of loss aversion. For example, suggests that the main effect of endowment is not is the pain associated with giving up something we to enhance the appeal of the good one owns, only thought we owned a reflection of loss aversion, of the pain of giving it up.” Kahneman et al. (1991) something else, or of some combination of factors? went on to attribute the status quo bias and other The challenge of isolating the impact of loss aver- phenomena to loss aversion: “These observations, sion is certainly a weakness of the loss aversion and many others, can be explained by a notion of explanation.