What is the sunk cost fallacy?
Money already spent should not steer the next choice, and in a 1982 theatre field experiment it did. A definition, the evidence and the size of the effect.
Answer to the question
What is the sunk cost fallacy?
The sunk cost fallacy is the habit of letting money, time or effort already spent, and not recoverable, steer what you do next. In Arkes and Blumer's theatre field experiment, full-price buyers attended more plays in the first half of the season. Large replications find the effect real but small.
Key takeaways
- Arkes and Blumer (1985) sold season tickets at random for $15, $2 off or $7 off. Over the first five plays the full-price group used 4.11 tickets against 3.32 and 3.29, and over the last five the groups did not differ.
- Thaler (1980) named the sunk cost effect. In Arkes and Blumer's radar-blank plane scenario, 41 of 48 respondents told that $10 million was already spent said yes to continuing, against 10 of 60 when nothing had been spent.
- Kahneman (2011) routes the fallacy through mental accounting and the top-right cell of the fourfold pattern: abandoning a project closes the account at a loss, and a likely loss makes people risk seeking.
- The original Many Labs project (Klein and colleagues, 2014) found a small effect in 36 samples of 6,344 people, and only half the samples reached significance alone. Beknazar-Yuzbashev and colleagues (2025) found about 1.3 extra minutes of play in an online task, and no significant overall effect across 11,328 YouTube videos.
- Staw (1976) separates escalation of commitment from the sunk cost fallacy: authorship of the failing decision, not the money already spent, drove the doubling down.
The ticket stubs in the Ohio University Theater's 1982 season booklets were colour coded, so that two psychologists could count which buyers turned up to which play. They were testing the sunk cost fallacy.
The psychologists were Hal Arkes and Catherine Blumer, and the fallacy is the habit of letting money already spent, and gone for good, steer what you do next. Every booklet gave the same seats at the same plays. Only the price each buyer had paid differed, and chance had set it.
Three Prices Offered to Buyers of One Theatre Season
The first 60 people to buy a season ticket were sold one of three versions at random: the normal $15, a $2 discount, or a $7 discount. The discount was explained as a departmental promotion. Six buyers were dropped for purchasing as couples, which left eighteen at full price, nineteen at $2 off and seventeen at $7 off.
Over the first five plays, full-price buyers used a mean of 4.11 tickets. The $2 group used 3.32 and the $7 group 3.29. Over the last five plays the three groups did not differ, at 2.28, 1.84 and 2.18. Here the sunk cost fallacy had a measurable footprint: a price assigned by chance, and unrecoverable by anyone, predicted attendance for about half a season and then stopped predicting it.
Daniel Kahneman reports the same field experiment in the endnotes of Thinking, Fast and Slow (2011), the book behind this piece. Full-price buyers were more likely to attend, he writes, especially in the first half of the season. He adds that skipping a show one has paid for feels like closing an account at a loss.
Kahneman links not using a ticket one has paid for to the sting of a loss on a mental account.
What the Definition Says and the Motive Behind the Behaviour
Kahneman gives the sunk cost fallacy a plain definition: continuing to fund a losing account when better uses of the money exist. His example is a company that has spent $50 million on a project that is behind schedule, with forecasts worse than at the start. Finishing needs $60 million more. The same $60 million could go into a new project with better prospects. The $50 million is gone either way, so it can say nothing about which project to back.
The sunk cost fallacy is putting more money into a losing project when better uses for it exist.
People do not reason that way, and the sunk cost fallacy is the name for the gap. In the blizzard item Kahneman retells from Richard Thaler, two fans plan to drive 40 miles to a game, one holding a ticket he paid for and one holding a free ticket. Asked which of them is more likely to brave the storm, almost everyone answers at once: the one who paid.
Kahneman files the pull under [mental accounting](https://thewhyweproject.com/articles/what-is-mental-accounting). Spending opens an account, and abandoning the project closes it at a loss. He places the choice in the top-right cell of his fourfold pattern, where a likely loss makes people risk seeking, so that continuing becomes the gamble that postpones booking the loss.
Thaler named the effect in 1980. His hypothesis was that paying for the right to use a good raises how much it is used, all else equal. Arkes and Blumer's 1985 paper turned the idea into a research programme of ten experiments, and their own first experiment is footnoted as adapted from Thaler. In their radar-blank plane scenario, 41 of 48 respondents told that $10 million was already spent said yes to continuing, against 10 of 60 when nothing had been spent.
Their explanation for the sunk cost fallacy is moral rather than arithmetical. People follow a rule against appearing wasteful and apply it where it does not belong. Christopher Olivola's eight experiments, published in 2018 with 6,076 participants, fit that reading. According to the paper's abstract, people honoured costs that other people had paid, whether or not they felt close to those people and whether or not anyone was watching.
How Large It Is, and What It Does Not Show
The original Many Labs project tested the sunk cost fallacy with one item in 36 samples with 6,344 participants. The scenario was a football-ticket vignette descended from Thaler's, not the theatre experiment. The effect appeared in the expected direction and was a little larger than in the original study. It was also small: only half of the 36 samples reached significance on their own.
Laboratory games with real stakes shrink the sunk cost fallacy further. Friedman and colleagues built a treasure-digging game around eleven treatment factors and reported a surprisingly small sunk cost effect, largely unaffected by the psychological drivers proposed. Beknazar-Yuzbashev and colleagues (2025) varied a ticket price by $2 for a fourteen-minute online task with 1,806 people, and the higher price bought about 1.3 extra minutes of play. Their YouTube field experiment, across 11,328 videos, found no significant overall effect on engagement.
The sunk cost fallacy is often merged with escalation of commitment, and the two come apart. In Barry Staw's 1976 study, participants who had personally chosen the failing course invested the most, while those handed the same failing division by someone else escalated far less. Authorship of the first decision did the work there. A manager who carries on with a project because a predecessor spent the money is committing the sunk cost fallacy without escalating anything.
Arkes and Blumer concluded that their effect cannot be fully reduced to any one social psychological theory, dissonance included. Why You Defend What You Just Bought follows the dissonance route, and Why You Finish the Series You Stopped Enjoying applies the sunk cost fallacy to time spent watching television. Loss Aversion and the Bet You Turn Down covers the loss side of the same account.
Set aside what has been spent, and ask what the next hour buys.
Common questions
What is the sunk cost fallacy?
It is treating money, time or effort already spent, and not recoverable, as a reason to continue. Richard Thaler named the effect in 1980, and Arkes and Blumer ran the first systematic experiments in 1985.
Who discovered the sunk cost fallacy?
Richard Thaler named it in 1980. Arkes and Blumer ran the first systematic experiments in 1985, and credit is shared in that order.
Is the sunk cost fallacy the same as escalation of commitment?
No. The sunk cost fallacy turns on money, time or effort already spent. Escalation of commitment, from Staw's 1976 study, turns on who made the original decision.
Does the sunk cost fallacy always change behaviour?
No. It appeared in the theatre field experiment for about half a season, and large replications find a small effect that is weaker in incentivised and field settings.
Sources
Kahneman, D. (2011). Thinking, Fast and Slow. New York: Farrar, Straus and Giroux (sunk cost via mental accounting; the fourfold pattern; the theatre field experiment in the endnotes). Seed book; vault hub Books/Thinking Fast and Slow.
Thaler, R. H. (1980). "Toward a positive theory of consumer choice." Journal of Economic Behavior and Organization, 1(1), 39–60.
Arkes, H. R. & Blumer, C. (1985). "The Psychology of Sunk Cost." Organizational Behavior and Human Decision Processes, 35(1), 124–140.
Staw, B. M. (1976). "Knee-Deep in the Big Muddy: A Study of Escalating Commitment to a Chosen Course of Action." Organizational Behavior and Human Performance, 16, 27–44.
Klein, R. A., et al. (2014). "Investigating variation in replicability: a Many Labs replication project." Social Psychology, 45(3), 142-152.
Friedman, D., Pommerenke, K., Lukose, R., Milam, G. & Huberman, B. (2007). "Searching for the sunk cost fallacy." Experimental Economics, 10(1), 79–104.
Beknazar-Yuzbashev, G., Ichiba, S. & Stalinski, M. (2025). "To the depths of the sunk cost: Experiments revisiting the elusive effect." Journal of Economic Behavior & Organization, 241, 107352.
Olivola, C. Y. (2018). "The Interpersonal Sunk-Cost Effect." Psychological Science. Eight experiments, N = 6,076. Effects not moderated by social closeness or by whether the sunk costs were observed. Abstract and reported design only.