Yves here. As this article explains, the finding that loss aversion may not be as prevalent a cognitive bias as behavioral economists had long posited is now raising questions about other findings in that discipline.
The problem I have with attempts to do lab-type studies is that behavioral economics have also found that how subjects respond to various economic propositions that are identical in value very much depends on how the bet or offer is framed. So give how suggestible people are, that would also support the idea that reactions are context-dependent and malleable.
More specifically, if you are familiar with the classic mug experiment, it is obvious why the new version is not equivalent. The original version is “stay as you are or take money to change your state.” The second is, “You have to give up something”.
More generally, most people are irrationally attached to things. I am much more willing to throw things out than most people and even so, when I moved, I spent good money hauling things around that were unlikely to make any sense outside Manhattan, or in a life where I didn’t have to dress like a well-heeled professional to meet clients.
By Michael Schulson, a contributing editor for Undark. His work has also been published by Aeon, NPR, Pacific Standard, Scientific American, Slate, and Wired, among other publications. Originally published at Undark
While most people have likely never heard of loss aversion, the concept — arising in the social sciences some four decades ago — is among the most influential in the behavioral sciences. In a nutshell, it holds that when people make decisions, the impact of losing something carries greater weight than the impact of gaining something of similar value — or that, in the often-quoted words of psychologists Daniel Kahneman and Amos Tversky, “losses loom larger than gains.”
The idea has come to inform empirical efforts to understand everything from investor behavior to insurance markets. Some analysts have even tried to quantify the ratio between the pain of loss and the pleasure of gains. Loss aversion, as one decision-research firm describes it, “is a cognitive bias that describes why, for individuals, the pain of losing is psychologically twice as powerful as the pleasure of gaining.”
And yet, in recent years, many behavioral scientists have begun to question whether loss aversion is quite so ironclad a principle of the human mind. Recent experiments, for example, have suggested that other factors — quite aside from a particular orientation toward losses and gains — might play key roles in quirks of human decision-making once chalked up to loss aversion. A blog post published earlier this year by the behavioral science consultant Jason Hreha went so far as to proclaim the death of the field of behavioral economics, largely because, he wrote, the “core finding of behavioral economics, loss aversion, is on ever more shaky ground.”
Few researchers are ready to make such extreme pronouncements, but if there are indeed cracks in the loss aversion firmament, it might well be because the concept was never meant to be a grand, unifying theory of human behavior. According to Kahneman, one of the two researchers most closely associated with the birth of the idea, the concept was originally not much of an experimental finding at all, but rather an intuition. It was something that he and his longtime collaborator Tversky — who passed away in 1996 — thought was so obvious as to be almost trivial: Losing something has a bigger impact on us than winning something.
“What our grandmothers knew, which I think is an intuition that everybody has,” Kahneman said in a recent call with Undark, “is that somehow ‘bad’ is stronger than ‘good’ in some ways.”
It was some years ago that researchers first handed out university-branded coffee mugs to half the members of an undergraduate law and economics class at Cornell University. What happened next would become an iconic moment in the field of behavioral science: The researchers asked the mug-owners to select a sale price for their new mugs, and the mug-free students to identify the amount they were willing to spend to buy one off a classmate. The mug-owning students wanted, on average, more than $5 for their mugs. The un-mugged undergrads were only willing to pay about half that. (At the university bookstore, the mugs retailed for $6.)
The researchers next tried the same setup, but with ballpoint pens. Same finding: The people who owned the objects valued them much more highly than the people who did not.
This divide between the haves and the have-nots, dubbed the endowment effect, has been replicated many times since. And after studying those 44 Cornell students and another group at Simon Fraser University, the mug researchers — Kahneman alongside economists Jack Knetsch and Richard Thaler — proposed an explanation for the phenomenon. It was, they wrotein a widely-cited 1990 paper, a manifestation of loss aversion, a concept they defined as “the generalization that losses are weighted substantially more than objectively commensurate gains in the evaluation of prospects and trades.”
Kahneman and Tversky had first outlined a version of the concept in a 1979 paper. They highlighted a longstanding finding about gambles: that many people seem to shy away from bets that, for example, offer a 50 percent chance of winning some sum and an equal chance of losing it. (Indeed, even a bet that tweaks those odds to, say, a 50 percent chance of winning $20, and a 50 percent chance of losing $15, often feels unfavorable.)
Fueled by research like the Cornell mug study, loss aversion grew to be axiomatic in the emerging field of behavioral sciences. “The concept of loss aversion was, I believe, our most useful contribution to the study of decision making,” Kahneman wrote in a 2002 autobiography after winning the a Nobel Prize in economics.
Not everyone, though, was convinced the concept was so robust. In 2003, a Stanford graduate student named David Gal ran a version of the classic mug experiment. He found the typical result, he said: People who owned the mugs valued them more than people who didn’t. But, Gal recalled, something bothered him: His subjects mostly just seemed indifferent to the whole thing. “You have this vision — like, you’re picturing people in the study, that they’re really invested in these mugs,” he told Undark. “But for the most part, people just didn’t really care very much.”
Instead, Gal began to wonder if subjects were hesitant to part with their mugs — or to offer money to buy one — mostly from inertia, rather than any strong feeling about losing or gaining a mug.
Other researchers have also raised questions about loss aversion. “Losses appear to loom larger than gains in some settings, but not in others,” wrote a pair of Israeli behavioral scientists, Eyal Ert and Ido Erev, in one 2013 paper. In a series of six experiments, they showed how tweaking certain parameters — such as changing the stakes of a decision, or changing which option allowed the subject to maintain a status quo — could affect how research subjects weighed losses and gains.
Other researchers, including Eldad Yechiam at Technion-Israel Institute of Technology, were raising similar concerns. (“I’m somehow obsessed by it,” Yechiam told Undark of his years-long work on the topic. Loss aversion “means that we are wired to sort of give the negative things in the world this huge weight. And it doesn’t seem to agree with how I view the world, with how I view people.”)
In a 2018 paper, Gal, now a professor at the University of Illinois-Chicago, returned to the mug experiment, along with a colleague, the Northwestern University social psychologist Derek Rucker. This time, though, they tweaked one variable. Instead of asking people how much money they would accept to sell a mug, they indicated that they would take the mug away — and then asked how much subjects would pay to keep it. They also asked people who didn’t have a mug how much they would pay to buy one.
In theory, they argue, if the prospect of losses loomed larger than gains, people would pay more to keep a mug they already had than to buy one they did not. But that didn’t happen: Whether they had the mug or not, people largely assigned it a similar value.
Another set of researchers set up a similar study, and, in results published earlier this year, reported a similar result: By asking people how much they’d pay to keep something they had been given, signs of loss aversion disappeared. While she doesn’t argue against the existence of loss aversion, Wendy Liu, an associate professor of marketing at the University of California San Diego and an author on the paper, said it’s not applicable in every circumstance. “Based on our own evidence, it’s not a universal thing,” she said. The research, she said, suggests that loss aversion may not be a good explanation for the endowment effect, and that researchers should stop equating the two.
Gal, Yechiam, and some other researchers now argue that many phenomena chalked up to loss aversion could, at least in theory, be explained by other causes, such as a bias toward inaction over action.
Gal describes the definition of loss aversion as “fuzzy and loose” — sometimes used to describe an underlying feature of human cognition, and other times used to describe a phenomenon (like the endowment effect) without necessarily making a claim about some deeper cause.
Many of his colleagues, he said, have soured on the principle. After the 2018 paper, Gal said, he received emails from colleagues saying they had come to believe that, in their past work, they had wrongly chalked up specific phenomena to loss aversion. Other researchers, he said, told him, “‘I’ve been trying to find this stuff for years, and couldn’t find any evidence for loss aversion” — but that peer reviewers were often hostile to such findings, and unwilling to publish the results.
Gal has sometimes taken a strong stand — he described loss aversion as a “essentially a fallacy” in a 2018 essay for Scientific American — but not everyone is convinced.
“There are a lot of phenomena that are explained by loss aversion,” said Eric Johnson, a decision science expert at Columbia University. Johnson points to a recent metanalysis, published as a working paper, which analyzed 150 articles and found, on average, strong empirical evidence of people weighting losses more strongly than gains in their decision-making. While critics of loss aversion, he acknowledged, may be able to come up with alternate explanations for any specific application — such as the endowment effect — loss aversion offers the simplest explanation for all those diverse cases. Otherwise, he said, you “end up with this sort of zoo of partial explanations.”
Giving serious credence to loss aversion skeptics, he suggested in a follow-up email, was drawing a false equivalency — similar to giving credence to climate change deniers.
For his part, though, Gal said he was unimpressed with the meta-analysis. “A meta-analysis can calculate an effect size, but in most cases doesn’t tell us much about what causes the effect,” he wrote in an email. “In this case, they are aggregating different effects likely caused by different processes and claiming to have found some general loss aversion coefficient.”
Now 87 and an emeritus professor of psychology at Princeton University, Kahneman has not responded publicly to Gal, Rucker, or other recent challengers of the loss aversion consensus. But during a recent Zoom conversation from his apartment in Manhattan, he made it clear he had been following the controversy closely.
He agreed with Gal and others, he said, that evidence of loss aversion only appears in certain situations. “It’s not a law of human nature that you have to find it in every context,” Kahneman said.
“There are experiments where people don’t find loss aversion,” he added later. “And, again, there’s an explanation for every one of them. That doesn’t violate loss aversion, because there are exceptions to loss aversion.”
What, then, could ever disprove the idea? Is loss aversion falsifiable? Probably not, said Kahneman. “I don’t know, maybe it’s falsifiable, it’s hard to imagine. There are alternative explanations for just about any experiment,” he said. But, he suggested, that didn’t matter: In the science of decision making, the theory had established its place.
“Having a principle that helps understand a wide body of phenomena — that’s considered useful,” Kahneman said. “That doesn’t mean that loss aversion’s true. It means that it’s useful.”
15 comments
Comments are closed.
Well, I just have an aversion to having stuff around that I don’t use or no longer want. Extra coffee mug – I’ll pay YOU to take it. Taking up space – even though I have nothing else to put in the space, just seems like a waste to me. If you have too much stuff, than you can’t find what you really want in that mountain of stuff. I don’t sell extra stuff – that is too much work. Even the work of calling up Goodwill to pick it up is more effort than I am willing to expend. Fortunately, the street in front of my house is busy enough that everything I put out there gets scarfed up – clothes, paint, excess glass tiles, broken office chairs, etcetera.
I mean, stuff I still have the receipt for that I regretted buying a day or two later – too lazy or embarrassed or opened the package, etcetera to go back and get a refund – put out in front for whatever scavenger wants it.
So when loss aversion meets me, I just go all Marie Kondo on its A$$
There’s a Free Table near the Arizona Slim Ranch. I take unwanted stuff over there on the regular.
What is really crazy is that the stupid social science career-making game of identifying and then defending “concepts” like loss aversion – the game of conceptual “discovery” or “identification” which is only central to academic social science because careers have to be based on something and this apparently provides some type of “impartial” means by which people legitimate an employment system that otherwise revolves around hiring, publishing, and promoting their friends and their friends’ friends – which are “proven” to exist on the basis of moronic “experiments” with college undergrads and five dollar bills are then used by other “serious” academics and policy makers to design actual public policies that have (mostly bad) real world impacts on real people’s lives.
So give how suggestible people are, that would also support the idea that reactions are context-dependent and malleable. The Hawthorne experiments in the 1930s – which were at least workplace experiments done with actual workers doing their actual work jobs – demonstrated quite clearly that it is impossible even in that setting to disentangle substantive causation (does improved lighting increase worker productivity) from experimental effects. For those not familiar, the researchers found that improved lighting increased productivity, but so did worse lighting and pretty much ever other intentional manipulation they tried, concluding it was the existence of the experiment itself that raised productivity and not any of the specific interventions.
The notion that loss aversion could be attributed, or not, to a coffee mug that an undergrad had owned for a matter of minutes is so patently ridiculous… Also, it has become virtually impossible to write commentary without liberal use of scare-quotes. We are all in the fun-house.
I appreciate your ability to articulate scorn. This kind of stuff leaves me sputtering. Bringing in the Hawthorne experiments is spot on.
I’d just quibble with the emphasis on suggestibility because it seems to emphasize manipulation. The sorts of big questions, or scenarios, that we’d want loss aversion research to help us with, like how populations react to sacrifices necessary to carry out a war or to mitigate climate change, seem utterly beyond it, and those involve not so much manipulation as a shared understanding of necessary sacrifice, or not. I often kvetch about studies not being open to drawing subjects into a reflection on the forces the experiment is trying to gauge and that really goes here.
I think behavioral economics has a great value as a foil to the patently false rationality assumed by the Chicago school of economics. Understanding the ways in which rationality breaks down is the only path to building economic models that are more humane.
Even though “…the pain of losing is psychologically twice as powerful as the pleasure of gaining” is clearly true among horse bettors, card players, and roulette wheelers, in my experience, that is irrelevant to their behavior. Gamblers need money to get into the action, not to make money, per se. If they win – great – more money to play with. If they loose – crap – have to scrape up some money somewhere to play with. So-called “consumers” are often addicted too and just need to consume something now and then. Insurance buyers are not addicted in the usual sense, but they are compelled to buy if they have assets to protect, so prices are secondary to the need to “get into the action.”
To me the main question in handling money is always: Am I going to have enough to do what I have decided to do before it runs out? If not, I usually decide to behave differently. When I didn’t do that, it usually hurt real bad and I felt stupid. It’s mostly about judgement, not fear or pleasure.
Yes, if you lose 50% you then have to make 100% just to get back to where you were. Lose 50% again and your task is that much bigger. Pretty soon you may be out of the game altogether, or having spam for dinner instead of a nice steak, and that hurts.
I can’t recall if I have heard of the “Mug” experiment but I certainly have heard of the loss aversion bias principle referenced in popular articles to explain human behavior. It is disturbing to think how easily I could accept such explanations. Contrast the “Mug” experiment to a more complex game I have participated in, the White Elephant gift exchange. At the last company I worked, we did this for Christmas where the gifts were of actual value. This changed the character of what was supposed to be a “fun” event, to what seemed like some strange experiment in gain and loss. Example of the gifts were movie tickets, wine, toaster ovens, gift cards and better. One could see some people really valued certain gifts and may have experienced real loss when taken. Since this was supposed to be “fun”, it was usually hard to determine true feelings because people would try to hide them. So, I guess my thinking here is this event starts by giving everyone something of value and then options for loss and exchange. However there are more options, a context,( i.e., “fun event”) and some element of risk. In this game compared to the “Mug game”, I don’t see loss aversion bias having much impact on the decision making process. The “Mug” experiment, clearly seems too simple. What other experiments show that loss aversion would be highly determinative in more complex environments with more choices?
There seems to be a bit of a backlash at the moment against Khaneman, largely because of the extreme artificiality of many of his experiments, and the sweeping conclusions that some have drawn from them. Artificial experiments with small numbers of college students can only get you so far, after all. But that said, “loss aversion” is a real thing, when you know what you have, but you are uncertain about what you might gain. Anyone who’s had children knows how human beings will cling to their possessions and refuse too change them for something else. But what this implies is actually just the common-sense observation that change is difficult when you can’t clearly show that the new state will be as good, if not better than the old. I’m not sure you need experiments for that.
isn’t this all dependent upon how much the receiver likes the mug and needs a mug like that in their life?
if we own something and it represents a hassle to us, as fresno dan above says “i’ll pay you to take it off my hands”. but if one is given something for free and then approached to sell it, of course they might try to get the most of out the transaction–because the money is actually useful to them and the mug is just a nonfactor (most people can afford to take or leave another free mug). same with the “buyers” in the experiment–imagine being asked “how much will you pay for this thing that you never asked for and probably don’t really want or need?”
the mugs are virtually valueless to both the buyers and the sellers in the initial experiment, but they are both trying to get a “better deal” out of the game that has been set up. initial experiment explains nothing except some kind of game theory imaginary outcome because the token item has virtually zero value to either. which means it doesn’t prove loss aversion exists in that scenario, but it also doesn’t discount that it may exist in another.
this is America. they should have done this with automobiles or something valuable and easily subsumed into personal identities & lifestyles for the mass of people instead, and tried to see where that would take them. and even then, you would have to match your trading partners along socioeconomic lines. a car means a lot more to a poor wretch like me than someone who can go out tomorrow and buy one if they need to do so.
The mug experiment as described doesn’t feel like loss aversion to me at all. It seems more like people trying to sell high to other people trying to buy low. I.E. the beginning of haggling.
And if I were setting up an experiment around loss aversion I would start with the hypothesis that some people have it some people don’t. As in: some people are obsessed with keeping their stuff from thieves, moochers and the gubmint; while other people happily pay their taxes, donate to food banks and give their surplus stuff to Goodwill rather than trying to sell it on eBay.
In other words, some people but not all have a libertarian mindset.
Exactly! I find it odd that people would even interpret the mug experiment in such an awkward way. The mug-holders were given the mug for free. Of course, they will try to maximize the profit they could get by selling it, while being cheaper than the university gift shop. The ones trying to buy factor in that the other side got it for free. Therefore, they offer much less. Additionally, they might also hold the opinion that the gift shop sells this junk for too much money anyway. A simple explanation that does not require to invoke a concept, loss aversion, that does not seem falsifiable.
Those of us who worked in mathematical psychology and associated fields – NOT behavioural economics – knew all this to be true since mid 80s. If you already own something your knowledge and crucially degree of confidence involving its value is higher (lower variance).
Something new is something you are less sure about (higher variance). Result? Choice (frequencies – estimates of probabilities used – incorrectly – to infer only VALUE) probabilities look more skewed towards loss aversion than gains. It’s there in the Likelihood function! This was solved by Yatchew & Grillisches in mid 80s and siloed academic fields means this zombie idea refuses to die.
A new addition for behavioural economics.
Short term, self interest is part of the human condition.
We can now see 21st century bankers aren’t really any different to 1970’s union leaders.
They both work in their own short term, self interest with no thought to the consequences.
I don’t understand what the issue is. OF COURSE, losing $1000 is more harmful than gaining $1000 is beneficial. Descending a notch on the ladder is more painful than rising a monetarily-equivalent notch would be pleasant. For me losing $1000 / year would wipe out a substantial proportion of my discretionary spending and put me in a different world, whereas gaining $1000 would merely increase my discretionary spending somewhat but leave me in a nicer version of the same place.
Same for things I have the things I have because they are the things I want and the things I’m habituated to. New things of equal cash value may or may not be right for me, and even if I think they will be right, I might find out that I’d guessed wrong. The choice is seldom between a good thing and a bigger good thing. It’s almost always the choice between a good actual thing and a bigger possible thing. Facing choices like this, my level of trust is low. A more optimistic and more trusting person might respond differently.