Mind & Psychology · Thinking & decisions
The sunk-cost trap
Learn to separate money and time you can't get back from what a choice can still change, and see why sticking with something isn't always the mistake it looks like.

What you will learn
- Spot a past cost you can't get back
- Tell sunk costs apart from the future costs and benefits of a choice
- Explain why future value can justify continuing despite a past loss
- Apply a forward-looking comparison to a hypothetical project
Lesson 1 of 4 · 4 min
What counts as a sunk cost, and what doesn't
Say you bought a movie ticket for tonight, and it can't be refunded. Then a friend calls with a better plan. You feel a tug. Skipping the film would waste the ticket.
Look at the ticket money under each plan. If you go to the film, it's spent. If you go with your friend, it's spent. That makes it a sunk cost, a cost you've already paid that no choice can bring back.
So the ticket can't tell the two evenings apart. Everything that differs between them is still ahead.
Now change one detail. The theater will refund the ticket until 6 p.m. Canceling gets you money back, and going doesn't. That money now differs between your options, so it belongs in the comparison.
So a payment isn't sunk just because it's in the past. It's sunk when every option you have leaves it lost. Before you call a cost sunk, check what you could still get back.
With the ticket set aside, compare the two evenings as they'd go. How much would you enjoy each one? What would each still cost you in bus fare and time?
The film can still win. Maybe you've wanted to see it for months. Then you're going for the evening ahead, and the ticket price has nothing to do with it.
When a cost you can't recover pushes you to keep going anyway, that's the sunk-cost effect. Its best-known case is an airplane.
In 1962 Britain and France agreed to build Concorde, a supersonic airliner, at an estimated cost of 70 million pounds. By 1976 the bill was at least 1.5 billion. The governments kept paying after the plane had lost its economic case, and only 20 were ever built. Politics and treaty law also made it hard to pull out. Economists now call the effect the Concorde fallacy as well.
"I have to go because I paid" is sunk-cost reasoning. But seeing someone go to the film doesn't prove they reasoned that way. They may expect to enjoy it.
Try to recallWhen you're deciding whether to go out tonight, why is a nonrefundable movie ticket different from the fare to get to the theater?
The ticket payment stays the same whichever option you choose. You can still avoid the travel fare by not going, so it's a future cost that belongs in the comparison.
Before a decision, sort the costs. A cost that's lost under every option is sunk, so leave it out. A cost or a refund that differs between options stays in. Then choose on what's still ahead.
Lessons in this topic
What counts as a sunk cost, and what doesn't
Learn to spot a cost that no choice can bring back, and see how two governments kept paying for Concorde long after it stopped making sense.
When finishing a costly project still makes sense
The sunk-cost rule doesn't say quit. See how the same lost $30 million fits with stopping one project and finishing another.
Why sticking with something isn't proof of a bias
Mice, rats and people all seemed to fall for sunk costs in one study. See how a perfectly rational coffee drinker produces the same data.
Does wasted time count like wasted money?
A 2001 study said lost hours sting less than lost dollars. See what held up when 821 people repeated it, and what didn't.
Three quick stories from this topic
The Concorde fallacy
Britain and France kept funding Concorde after the plane had lost its economic case
The supersonic airliner was estimated at 70 million pounds in 1962. By 1976 the bill had passed 1.5 billion, and only 20 were ever built. Economists now call spending more because of what you've already spent the Concorde fallacy.
Why the old money shouldn't count
Money already spent is gone whether a project continues or stops. Only the costs and benefits still ahead differ between the two choices.
More than psychology
Political and legal ties also made it hard for either government to pull out. Persistence can have more than one cause.
Sunk costs in mice
Mice and rats seemed to fall for sunk costs, until a rational model produced the same data
In a 2018 study, the longer the animals had waited for food, the less likely they were to quit. A 2022 model of a coffee drinker who ignores past waiting did the same. On days she cares less she quits early, so long waits come from days she cares a lot.
The 2018 finding
Rodents waited for flavored food pellets and people waited for short video clips. In all three species, time already invested seemed to make quitting less likely.
Attrition bias
When the less motivated drop out first, those who remain look committed. Statisticians call this attrition bias. The 2022 team proposed changing the offer at random during the wait to tell it apart from a real sunk-cost effect.
Lost dollars against lost hours
48% chose the show they'd paid more for, but only 13% the one they'd worked longer for
In a 2023 replication, 821 people imagined holding tickets to a play and to a rock concert they'd enjoy more. Those who had paid more money for the play often chose it anyway. Those who had worked more hours for it mostly went to the concert.
The design behind the numbers
Participants imagined putting more of a resource into a theater ticket than into a concert ticket, while preferring the concert. Only the resource differed, money in one condition and time in the other. Choosing the theater performance meant letting the investment they couldn't recover steer the decision.
The scenario matters
In a project-choice scenario, the same replication found sunk-cost effects for both time and money. A passage teaching participants to account for time didn't strengthen the time effect. The gap between money and time is real in some settings and absent in others.
Sources
- Sunk Cost Effects for Time Versus Money: Replication and Extensions Registered Report of Soman, Petrov et al., International Review of Social Psychology; PMC full text
- Apparent sunk cost effect in rational agents, Ott et al., Science Advances; PMC full text
- Sunk cost, Wikipedia, Wikimedia Foundation
- Concorde, Wikipedia, Wikimedia Foundation




