012 min
Loss aversion at work: teachers who had to give money back
Nine schools in Chicago Heights, Illinois, offered their teachers a bonus in the 2010–11 school year. Four economists ran it as an experiment: Roland Fryer, Steven Levitt, John List and Sally Sadoff. In all, 150 teachers joined. A teacher could earn $80 for each percentile their students reached, up to $8,000. A student's percentile came from comparing their test results with those of similar students.
The teachers were split into groups. One group was paid at the end of the year, the way bonuses usually work. The loss group got a $4,000 check at the start of the year. Teachers in the loss group signed a promise to hand back the difference between $4,000 and their final bonus if their students finished below average. A teacher whose students did equally well ended up with the same money in either group.
You'd expect the same pay to push both groups the same way. It didn't. The students of teachers in the loss group gained between 0.2 and 0.4 standard deviations in math. A standard deviation is the usual gap between a student's score and the average score. For teachers paid at the end of the year, the researchers found no effect.
The researchers explain the result with loss aversion. A bonus you might earn is something you hope for. A check already in your account is yours, so handing part of it back feels like losing something. A loss counts for more than an equal gain, so the same amount of money moved teachers more when it could be lost.
022 min
Why loss aversion happens
In 1990, three researchers tested what owning something does to its value: Daniel Kahneman, Jack Knetsch and Richard Thaler. They gave every other student in a class of 44 at Cornell University a $6.00 mug from the campus bookstore. Then they asked each mug owner the price at which they would sell, and asked the students without a mug what they would pay.
The two prices did not match. In the mug markets of that paper, the median price owners asked was more than twice the median price buyers offered. Before the mugs, the same students had traded paper tokens that each had a fixed cash value, and there the median buying and selling prices were identical. So the students knew how to trade. The gap came from owning the mug. Repeating the market didn't close it: owners and buyers kept the same split over four or five rounds of trading.
Here is the idea underneath. You judge a trade by what changes from where you start, and where you start includes what you own. Kahneman and the psychologist Amos Tversky called that starting point the reference point. Selling your mug moves you down from it, so it counts as a loss. Buying a mug moves you up, so it counts as a gain. Because a loss counts for more, owners ask for more than buyers will pay. The gap between those two prices is called the endowment effect.
The starting point can move. Kahneman and Tversky wrote in 1979 that how a choice is worded and what a person expects can both change where they start.
Loss aversion may also run deeper than culture. In 2006, Keith Chen, Venkat Lakshminarayanan and Laurie Santos taught capuchin monkeys to trade small disks for food. In one test, two experimenters each handed over one slice of apple. One showed two slices first and took one back. The other showed only one slice. The monkeys traded far more with the experimenter who showed one slice. The authors read this as loss aversion, and because the monkeys had little experience with money, they suggest it may be innate rather than learned. Psychologists still argue about exactly why losses count for more.
031 min
Loss aversion theory: the 1979 questions behind it
In 1979, Daniel Kahneman and Amos Tversky gave students and university faculty a long list of imaginary choices. Each choice was between a money gamble and a sure amount, and people said which they would pick. One result stood out. Most people found a 50-50 bet to win or lose the same amount unattractive, and the bigger the stake, the less they liked it. From many answers like these, the pair drew a curve for how much an outcome is worth to a person. The curve starts at the point you are at now. It falls faster on the loss side than it rises on the gain side. They called the whole account prospect theory and published it in the journal Econometrica.
Were they right? In 2020, a team led by Kai Ruggeri asked 4,098 people the same kinds of questions. They lived in 19 countries and spoke 13 languages. Most results held: 94% of the items showed the same pattern again, though the effects were weaker than in 1979.
041 min
How much more a loss counts than a gain
You'll often hear that a loss feels twice as bad as a gain. The figure is how many times more a loss counts than an equal gain, and it is a rough average. Four researchers led by Alexander Brown collected 607 estimates of it from 150 papers in economics, psychology and neuroscience. Their average came out at about 1.96. Across those studies, the likely range of the average runs from 1.8 to 2.1, and single studies sit far from it in both directions.
In plain numbers, a multiplier of 2 means that a coin flip that could cost you $100 only feels fair if it could also win you $200. The number also differs from person to person.
051 min
Loss aversion vs risk aversion and sunk cost
Loss aversion is often mixed up with risk aversion, the endowment effect and the sunk cost fallacy. Each row below names a neighbour and the one point where it differs.
| Concept | What it describes | How it differs from loss aversion |
|---|---|---|
| This pageLoss Aversion | ||
| Not in the library yetRisk Aversion | ||
| PsychologyEndowment Effect | ||
| PsychologySunk Cost Fallacy | ||
| PsychologyProspect Theory | ||
| PsychologyDisposition Effect |
061 min
Loss aversion in bonuses: a Chinese electronics factory
At Wanlida Group, a company that makes consumer electronics in China, economists Tanjim Hossain and John List tested two ways of wording the same team bonus. In one, teams were told they would earn 80 yuan for each week their average output reached a target. In the other, teams were told at the start that they had a bonus of 320 yuan. They would lose 80 yuan for each week they fell below the target.
Both versions raised output. Teams responded more to the bonus worded as a loss, and the wording alone changed total team output by about 1%. A 1% change sounds small, yet the money and the target were the same in both versions.
071 min
Loss aversion in investing and trading
Why have stocks paid investors so much more than bonds over the last century? Economists had tried to explain the gap with plain risk aversion, but that needs people to dislike risk to an implausible degree. In a paper first circulated in 1993, the economists Shlomo Benartzi and Richard Thaler offered another reason. Investors are loss averse, and they also check their portfolios often, even when they are saving for something decades away. They called this combination myopic loss aversion, where myopic means looking only a short way ahead.
Think about what checking often does. If you open your investment app every day, you see a loss on many of those days, and each one counts for more than a gain of the same size. Stocks then feel less safe than their long record says. Benartzi and Thaler ran simulations of investors who check their portfolios once a year. For those investors, the extra return on stocks came out where earlier studies of loss aversion predict. Investors seem to choose as if they will look again in about twelve months.
The disposition effect, selling shares that have risen while keeping the ones that have fallen, is a related habit that shows up in trading records.
081 min
Loss aversion in an app streak: Duolingo's 847-day widget
Duolingo counts the days in a row that you finish a lesson, and the count is called a streak. One of its phone widgets shows a streak of 847 days in large white type. A flame icon with a red exclamation mark sits beside it, and the line "Save your streak!" sits under it. The widget names what skipping today would cost, which is 847 days, instead of inviting you to learn. A number that large is something you would hate to lose, and losses count for more than gains of the same size. Duolingo's own blog says its streak taps into loss aversion. The app also offers a Streak Freeze, which pauses a streak for a day so that you can skip a day without losing it.
091 min
Loss aversion on a cancel sheet: the Vocabulary app
Cancelling a subscription is a moment when you are about to give something up, and the Vocabulary iPhone app words it that way. A sheet slides up over its Manage subscription page with a heading that tells you not to lose what you have achieved with Premium. Under it, a box lists what you would lose access to if you cancel. The four items are backup and sync between devices, unlimited Premium content, streak progress and settings. The button that keeps you subscribed is about keeping your data safe, and the way out is a plain text link that says "Proceed with cancellation". No price you would save appears anywhere on the sheet, only things you already have. People give up what they own reluctantly, so a list of what you would lose gives you more reason to stay than a list of what you would save.
101 min
Loss aversion in a downgrade: Spotify's cancel page
Spotify is a music streaming service, and its paid plan is called Premium. The web page for cancelling Premium opens with a heading that says how you listen will change. Below it runs a column of large numbers, each with one line of explanation. They include ads about every 15 minutes, 0 songs you can play any time on mobile and 6 skips per hour. A line above them says you will move to the free plan on a date shown and lose your Premium benefits. Each change is turned into a number, so what you would give up by leaving takes up the whole page. The two buttons read "Cancel Premium" and "Back to account".
111 min
Loss aversion counted from a team's own use: Canva
Canva is a design tool. While a team is in a free trial, a window over its billing page lists the features the team would lose. On the left, a warning says the team has 18 days left in its trial. On the right, a dark card says the team will lose features it puts to good use. Beside three of them is a count of the team's own use: Brand Kit 394, Magic Write 20 and Background Remover 8. The numbers describe things the team already uses, so leaving reads as giving them up. Three buttons sit on the left, among them one that switches the team to an individual Canva Pro plan at US$15 a month, and one marked "Continue".
121 min
Loss aversion as a column of lost features: Codecademy
Codecademy's cancel page opens by saying it is sorry to see you leave and then splits into two columns. The left column is headed "No longer available". It lists ten items, each marked with an X. They include all courses, quizzes, skill paths and certificates of completion. The right column is headed "Features still available". It lists four. One of them, practice on the mobile app, is marked "Limited". A line above the columns says the changes apply immediately when you cancel. The losses form a longer list than what you keep, so the cost of leaving takes up most of the page.
131 min
Loss aversion with a deadline: Swiggy's expiring DineCash
Swiggy is an Indian food delivery app, and DineCash is credit it gives you to spend at restaurants. Swiggy's iPhone app shows a DineCash balance of ₹137 at the top of one screen, in large type. Right under it, a line says the ₹137 will expire in 5 days. It adds that you can redeem the cash by paying a bill through Dineout or Scenes, two Swiggy services. Further down, the history lists past entries called "Dineout cash expired", including "- ₹157" and "- ₹86". The balance is money you have until a set day, and the history shows what has already gone.
141 min
Loss aversion when you quit a lesson: Uxcel Go
When you try to leave a quiz in Uxcel Go, a phone app for learning design, a sheet slides over the question. It praises your work so far and tells you that quitting now means losing all your progress. Two buttons sit side by side. The outlined one on the left reads "Exit lesson" and the filled one on the right reads "Keep learning". The sheet describes quitting by what it costs, and the filled button is the one that keeps your progress.
151 min
Loss aversion in a button label: Ladder's exit sheet
In the workout app Ladder, trying to leave a workout opens a dark window that asks "Save Progress?" The window says you can resume later to earn a badge. You must do so before the workout expires. Two buttons are stacked below. The first, outlined in red, reads "EXIT & LOSE DATA". The second, solid white, reads "SAVE FOR LATER". A small line under them gives the time left, 23h 58m 20s. The loss is part of the name of the button you would press to leave, so you read it at the moment you decide.
171 min
A quit sheet that names no loss: Speak
Quitting a lesson in the language-learning app Speak brings up a sheet over the paused lesson that asks whether you are sure you want to quit. It has a wide filled "Keep Learning" button and a plain "Quit" link. Nothing on the sheet says what you would lose. The quit sheet in Uxcel Go, a phone app for learning design, appears at the same moment. It adds that you'll lose all your progress. The two sheets ask the same question, one with a named loss and one without. This is one screen, so it doesn't tell you how Speak words losses elsewhere.
181 min
Is loss aversion real? Where the evidence is contested
Researchers still disagree about how far loss aversion holds, though it appears in many studies. In 2018, David Gal and Derek Rucker critiqued the research on loss aversion. One of their points is about how it is measured. In some tests, refusing a gamble is also the way of doing nothing, so a person who simply prefers to leave things as they are can look loss averse.
Other researchers tested whether loss aversion holds for people who know an area well. In 2020, Kellen Mrkva and colleagues asked people who had recently bought a car about four features. The features were safety, comfort, fuel economy and a navigation system. For each one, the researchers measured how loss averse a buyer was and how much the buyer knew about it. Buyers who knew more about fuel economy were less loss averse about fuel economy, but not about comfort. Across five samples, people with more knowledge of an area were less loss averse, though people at every level of knowledge and experience still were.
Until the debate settles, treat loss aversion as a strong tendency that varies, not as a fixed constant. If you design a screen or an incentive, test it with your own users. Don't assume that a message about a loss will beat a message about a gain by a factor of two.
192 min
How to overcome loss aversion in your own decisions
You can't switch off loss aversion, but you can change the situation it reacts to. Four habits work on different causes.
Write both options as final totals
If you are deciding whether to sell a share that has fallen, don't ask whether you will lose money. Write down what you would hold in each case and compare the two numbers. This removes the starting point, which is the point you measure from and is usually what you have now. Without it, one outcome no longer counts as a loss and the other as a gain. The framing effect is related: the wording of a choice can change which outcome counts as a loss.
Look less often
If you check an investment every day, you meet many more small losses than if you check once a year. Each one counts for more than a gain. A yearly look shows fewer of them.
Ask whether you would choose it if you did not own it
In a 1990 experiment, Cornell University students who were given a mug asked more than twice what other students offered for it. Owning it was part of the reason. If you would not pay today's price for what you hold, your wish to keep it may come from owning it, not from its worth.
Learn the area you are deciding in
A 2020 study of five samples found that people who knew an area well were less loss averse, though not free of it.
201 min
How to use loss aversion in marketing
Teams in product and marketing use loss aversion because the wording of a loss has been measured to matter. In a Chinese electronics factory, a team bonus worded as a loss raised output more than the same bonus worded as a gain. The same tool can mislead, though. A countdown that resets, a stock warning that isn't true and a decline button worded to make you feel bad are dark patterns, not honest uses.
Before you ship a message that talks about a loss, check that:
?3 questions
Questions people ask
Is loss aversion a cognitive bias?
Is loss aversion irrational?
How do you measure loss aversion?
▶3 videos
Watch
Understanding Loss Aversion
CRITICAL THINKING - Cognitive Biases: Reference Dependence and Loss Aversion [HD]
Wireless Philosophy
The psychology behind irrational decisions - Sara Garofalo
TED-Ed
§10 sources
Sources
Show all 10 sourcesHide sources
Fryer, R. G., Levitt, S. D., List, J. and Sadoff, S. (2012). Enhancing the Efficacy of Teacher Incentives through Loss Aversion: A Field Experiment. NBER Working Paper 18237
Kahneman, D. and Tversky, A. (1979). Prospect Theory: An Analysis of Decision under Risk. Econometrica 47(2), 263-292
Kahneman, D., Knetsch, J. L. and Thaler, R. H. (1990). Experimental Tests of the Endowment Effect and the Coase Theorem. Journal of Political Economy 98(6), 1325-1348
Brown, A. L., Imai, T., Vieider, F. M. and Camerer, C. F. (2021). Meta-Analysis of Empirical Estimates of Loss-Aversion. CRC TRR 190 Discussion Paper 275 (the discussion-paper version of the study)
Ruggeri, K. et al. (2020). Replicating patterns of prospect theory for decision under risk. Nature Human Behaviour (record: )
Benartzi, S. and Thaler, R. H. (1995). Myopic Loss Aversion and the Equity Premium Puzzle. Quarterly Journal of Economics 110(1), 73-92. Working paper version
Hossain, T. and List, J. A. (2009). The Behavioralist Visits the Factory: Increasing Productivity Using Simple Framing Manipulations. NBER Working Paper 15623
Chen, M. K., Lakshminarayanan, V. and Santos, L. R. (2006). How Basic Are Behavioral Biases? Evidence from Capuchin Monkey Trading Behavior. Journal of Political Economy 114(3), 517-537
Mrkva, K., Johnson, E. J., Gächter, S. and Herrmann, A. (2020). Moderating loss aversion: Loss aversion has moderators, but reports of its death are greatly exaggerated. Journal of Consumer Psychology 30(3), 407-428
Duolingo (2022). The Duolingo Streak Uses Habit Research to Keep You Motivated (company blog)















