PsychologyCounterintuitive resultExplained★ Famous
Sunk Cost Fallacy
Also known as Concorde Fallacy, Sunk Cost Effect, Escalation of Commitment
You paid for a concert ticket and now feel terrible. If you hadn't paid, you'd stay home. You paid, so you go. Why does a sunk cost change anything?
Counterintuitive result. Nothing is wrong with the reasoning. The result is simply correct and human intuition is simply unreliable here. Not a contradiction at all.
Discover
Arkes and Blumer's canonical experiment: You have spent $100 on a ticket to a ski resort. You then buy a $50 ticket to a different ski resort — one you expect to enjoy more. The weekend arrives, and the two trips are on the same day. You can only go to one.
Rationally, the $100 and $50 are both spent and unrecoverable. They should have no bearing on which trip is more enjoyable. The decision should be simple: go to the better resort.
Most people choose the $100 resort. The larger past expenditure pulls them toward it, even though the enjoyment difference favours the other.
This is the sunk cost fallacy: past investment — already lost and irrecoverable — distorts decisions about the future.
Understand
Economics defines a sunk cost as a cost already incurred that cannot be recovered regardless of future decisions. The normative rule is clear: sunk costs should have zero influence on decisions about what to do next. Only expected future costs and benefits are relevant.
The psychological reality is that people systematically violate this rule. Several mechanisms drive the fallacy:
Waste aversion. Failing to use what you paid for feels like waste. Leaving early from a bad film, or abandoning a project after spending two years on it, is experienced as "throwing away" the past investment. The drive to avoid apparent waste pushes people to continue — even when continuing is the waste.
Loss framing. Kahneman and Tversky's prospect theory shows that losses feel roughly twice as painful as equivalent gains feel pleasurable. Sunk costs feel like losses waiting to be confirmed — stopping feels like acknowledging the loss permanently, while continuing holds out the possibility of recouping.
Consistency and self-justification. Committing publicly to a project and then abandoning it signals that the original commitment was a mistake. Continuing, even against evidence, maintains the appearance of consistency and competence.
Escalation of commitment. In organisations, sunk cost reasoning often escalates into commitment trap: each new investment in a failing project makes it harder to write off the previous investments, leading to more investment, and so on. The project grows too large and too politically committed to be cancelled, regardless of its prospects.
The rational alternative
The correct question is always: "Given where I am now, what is the best action going forward?" The past investment is gone. The only question is whether the expected future benefits of continuing exceed the expected future costs.
Applying this rule is psychologically difficult, partly because it requires reframing past investment as already gone — accepting a loss rather than hoping to recoup it.
The shape of it
Same shape, different costume:
Examples
| Scenario | The sunk cost | The rational choice | The fallacious choice |
|---|---|---|---|
| Bad concert attended because you paid | Ticket price | Enjoy evening at home | Attend miserable concert |
| Project 3 years in, clearly failing | 3 years of work | Cut losses, redirect resources | Continue because "we've come so far" |
| Concorde aircraft funding | Development costs to date | Cancel unviable aircraft | Continue for political/sunk cost reasons |
| Continuing medical treatment with low efficacy | Previous treatment costs | Switch to more promising treatment | Continue because patient has come this far |
| Finishing a bad book | Time already spent reading | Stop and read something better | Finish because started |
The Concorde case is the origin of the "Concorde fallacy" as a name for the effect. Both the British and French governments continued funding a commercially unviable aircraft long past the point where private investors would have stopped, partly because the amount already spent made cancellation feel too painful.
Explore
Loss aversion and the sunk cost. Kahneman and Tversky's prospect theory provides the deepest theoretical account: losses are more…
Loss aversion and the sunk cost. Kahneman and Tversky's prospect theory provides the deepest theoretical account: losses are more painful than equivalent gains are pleasant. Stopping a project means converting a potential loss (the sunk cost) into an actual, acknowledged loss. Continuing holds out hope — however irrational — of recoupment. The loss-aversion asymmetry explains why the sunk cost effect is so resistant to rational correction.
When sunk costs should matter — barely. There are edge cases where past investment is weakly relevant. Expertise acquired during a project may be an asset for future decisions. A reputation for completing projects may have value. But these considerations are about future effects of the past investment, not the investment itself. They are almost never large enough to justify continuing a project that is expected to lose money — and they are often invoked as rationalisations for the sunk cost fallacy.
Animals and sunk costs. Studies of foraging animals — including rats, pigeons, and starlings — find that they do not show the sunk cost fallacy: they abandon patches based on future expected yield, not on how long they have already foraged. This suggests that sunk cost reasoning may be unique to humans or closely related to human-specific features like self-justification, social reputation, and narrative identity.
The endowment effect. A closely related bias: people value things they own more than identical things they do not own. This is partly because giving up something owned is experienced as a loss, triggering loss aversion. The sunk cost fallacy and the endowment effect are both expressions of the same asymmetric treatment of losses and gains.
Historical context
1970s. Economists note the sunk cost principle as a normative rule for rational decision-making. The rule is clear theoretically;…
1970s. Economists note the sunk cost principle as a normative rule for rational decision-making. The rule is clear theoretically; whether people follow it is not studied systematically.
1985. Arkes and Blumer publish the foundational experimental paper, documenting the bias across multiple scenarios with controlled experimental evidence. They identify waste aversion as the primary psychological mechanism.
1990s. Barry Staw's work on escalation of commitment shows how the bias operates at organisational scale: decision-makers who are personally responsible for a project's inception show stronger sunk cost bias when evaluating whether to continue it.
2002. Kahneman's Nobel Prize brings the connected concepts of prospect theory and loss aversion to wider attention, providing the theoretical context for sunk cost behaviour.
2010s. Experimental economists document the Concorde fallacy in a wide range of contexts, including healthcare decision-making, government procurement, and financial investment.
Why it matters
Personal decisions. The sunk cost fallacy affects decisions about relationships, careers, and investments. Staying in a bad…
Personal decisions. The sunk cost fallacy affects decisions about relationships, careers, and investments. Staying in a bad relationship because "we've been together for ten years." Staying in a poor career because of the years invested in training. Holding a losing stock because "I can't sell at a loss." These are all sunk cost fallacies, and they cause significant personal harm.
Organisational decision-making. Failing projects at organisations are almost never killed early enough. The sunk cost fallacy, combined with political incentives (no one wants to admit their project failed), creates systematic under-cancellation of bad initiatives. This wastes resources that could be redirected to better uses.
Policy and government. Large government projects — weapons systems, public infrastructure, technology procurement — are notoriously resistant to cancellation once initiated, partly because of sunk cost reasoning. The political cost of cancelling a project grows with every pound spent on it, creating exactly the wrong incentive structure.
Modern relevance
Startup culture. The "pivot or persevere" decision in startups is structurally a sunk cost problem: founders who have spent two…
Startup culture. The "pivot or persevere" decision in startups is structurally a sunk cost problem: founders who have spent two years on a business model have strong psychological incentives to continue rather than pivot, even when market evidence clearly favours a pivot. Experienced investors explicitly try to pre-commit founders to evidence-based decision rules to prevent sunk cost reasoning from blocking necessary pivots.
AI development. Large AI training runs cost millions or tens of millions of dollars. Once a training run begins, there are strong sunk cost incentives to continue to completion, even if early results suggest the architecture is underperforming. Organisations that have pre-committed to stopping rules (abort if loss exceeds X) resist sunk cost reasoning better than those making the decision ad hoc.
Medical treatment decisions. End-of-life treatment decisions are among the most painful contexts for the sunk cost fallacy: families and physicians may continue aggressive treatment partly because of how much has already been done, rather than purely on the patient's expected quality of life going forward. Advance directives and palliative care conversations are partly attempts to pre-commit to forward-looking decision rules.
Sources & further reading
Primary source
(1985)The Psychology of Sunk CostOrganizational Behavior and Human Decision Processes, 35(1), pp. 124–140.
The foundational experimental paper establishing sunk cost bias as a systematic psychological phenomenon with controlled scenarios.
Elsewhere on the web
How this connects
- Related to
- Anchoring EffectBoth are biases in which past information contaminates current decision-making. The anchoring effect is about numerical anchors; the sunk cost fallacy is about past investment distorting future choices.
- The Paradox of ChoiceBoth concern how cognitive factors corrupt rational decision-making. Paradox of Choice shows that more options produce worse decisions; Sunk Cost shows that past decisions corrupt current ones.
- The Prisoner's DilemmaBoth show that individually rational-seeming reasoning can lead to collectively or personally harmful outcomes. The sunk cost fallacy is individually irrational despite feeling rational.