The Sunk Cost Fallacy: Why You Keep Wasting Time on Things That Are Already Lost
Why Your Brain Would Rather Waste More Than Admit the Loss
The Invisible Trap That Keeps Bad Decisions Alive
You are halfway through a terrible film. Turning it off now feels as though the ninety minutes already spent would be wasted. Or perhaps you have waited two hours for food, paid £80 for an event you no longer want to attend or spent months on a project that no longer makes sense.
The strange instinct is always the same: somehow, more time, money or discomfort might recover what has already gone. Yet the extra sacrifice does not rescue the earlier sacrifice. It merely gives the past another claim on the future.
What Is a Sunk Cost?
A sunk cost is any cost that has already been incurred and cannot now be recovered. It might be money, but it can also be time, effort, attention, emotional energy or reputation. The £30 paid for a non-refundable ticket is sunk; so are the ten hours spent reading a novel that remains dreadful.
The fallacy occurs when that irrecoverable cost distorts the next decision. A rational choice should normally compare what still lies ahead. The question is not, “How much have I put into this?” but, “Which available choice now produces the best expected outcome?”
That principle is sometimes called the “bygones” rule. HM Treasury’s 2026 Green Book states it plainly: sunk costs should not affect what happens next, while the opportunity cost of continuing to use already-purchased resources still matters. In other words, the past expenditure disappears from the decision, but the future use of your remaining money, time and assets does not.
The £5,000 Project That Should Still Be Stopped
Imagine that someone has already spent £5,000 on a project. Completing it will require another £3,000, but new information suggests the finished result will be worth only £1,000. It is tempting to say, “I cannot waste the £5,000, so I might as well finish.”
That reasoning turns a £5,000 loss into a £7,000 loss after receiving the £1,000 result. Stopping means losing the £5,000 already spent. Continuing means spending £3,000 to obtain £1,000, making the person another £2,000 worse off from this point forward.
The original £5,000 matters when evaluating how the project performed and learning from the mistake. It may matter for accounting, accountability or tax. It does not make an unattractive new £3,000 investment attractive.
What the Research Actually Shows
The modern evidence is strongly associated with a 1985 paper by psychologists Hal Arkes and Catherine Blumer. They defined the sunk-cost effect as a greater tendency to continue once money, effort or time had been invested, and linked it partly to the desire not to appear wasteful. In a field study, people who paid more for a season theatre subscription attended more performances during the following six months than people who had paid less.
Richard Thaler placed such behaviour within the wider framework of mental accounting: the informal accounts people use to organise, label and evaluate money and choices. Thaler, awarded the 2017 Nobel Memorial Prize in Economic Sciences for bringing psychologically realistic assumptions into economic decision-making, has used a simple sporting-ticket example. People may feel more compelled to make a difficult journey through bad weather when they paid full price than when the same tickets were free, although the experience awaiting them is identical.
The evidence should not be exaggerated. A 2021 University of Warwick experiment found that 23 per cent of participants stuck with a dominated lottery they had earned through effort rather than switch to a better one. That demonstrates the effect, not that everybody always honours sunk costs or that persistence is inherently irrational.
Why Quitting Feels Like Failure
Quitting often changes the emotional status of a loss. While a project, relationship or investment continues, the mind can classify the damage as temporary, unfinished or recoverable. Ending it makes the loss feel final.
Loss aversion strengthens that resistance. Prospect theory, developed by Daniel Kahneman and Amos Tversky, showed that people evaluate outcomes relative to reference points and often feel losses more sharply than equivalent gains. Continuing can therefore feel like keeping open a route back to “even”, even when the route is becoming more expensive.
Pride and consistency add another layer. People want to see themselves as sensible judges who finish what they start, while leaders want others to see them as decisive and reliable. Evidence that a choice was poor threatens that identity, so supporting the choice for longer can become a way to defend the earlier self who made it.
Fear of regret often favours delay. If you quit and the project later succeeds, the decision feels painfully visible. If you continue and it deteriorates gradually, responsibility can be spread across circumstances, colleagues and time.
Films, Tickets, Subscriptions and Broken Technology
The smallest examples reveal the mechanism cleanly. Finishing a bad film does not recover the first hour; it sacrifices another hour. Completing a book you neither enjoy nor need may preserve a pleasing completion statistic, but that is useful only if completion itself has genuine future value.
An expensive event ticket creates the same confusion. Once it cannot be sold or refunded, its price is no longer a benefit of attending. The relevant comparison is now between the likely enjoyment and costs of going — travel, fatigue, food and lost alternatives — and the value of staying home or doing something else.
An unused gym membership can survive for months because cancelling seems to admit that the original health plan failed. Yet keeping the direct debit does not create exercise. The forward-looking questions are whether the membership is likely to be used, whether it motivates useful behaviour and whether another option would work better.
Old technology creates a physical version of the trap. Someone keeps repairing an unreliable laptop because it was expensive or has already received three repairs. Those bills are sunk; its expected reliability and future costs should be compared with replacement.
Careers, Businesses and the Price of a Past Identity
Career choices are harder because qualifications are not always pure sunk costs. A person may remain in the wrong profession because they spent years training, built status around the role and fear that changing direction would make the qualification “wasted”. But skills, contacts and credibility may transfer, so the past can create future benefits without dictating the same future job.
The comparison is between the expected futures of staying and switching, including retraining costs, lost income, risk, satisfaction and transferable knowledge. Persistence is sensible when the role is temporarily difficult but improving; it is suspect when the qualification itself is the main reason to remain.
Founders face an even sharper version. A failing business may contain years of labour, personal savings and a public identity. Continuing may be rational if credible new orders, a viable pivot or valuable learning improve its future prospects, but “I have already put everything into it” is not a commercial forecast.
Companies repeat the pattern with unsuccessful products. Teams protect launches they designed, revise sales forecasts and authorise more marketing to justify development spending. A product should survive because its remaining expected value exceeds its remaining cost.
Investments, Gambling and the Desire to Get Back to Even
Investors sometimes hold a falling asset until it returns to the purchase price because selling below that number feels like converting a paper loss into a real one. Research on the “disposition effect” has documented a tendency to hold losing investments too long and sell winners too soon. The purchase price can become a powerful psychological reference point even though the asset’s future prospects do not know what one particular investor paid.
That is not personalised investment advice. Holding or selling may be justified by future returns, taxes, costs, diversification or a long-term plan. “I will wait until I break even” is different because it treats a private historical number as though it changes the asset’s future value.
Gambling makes the danger immediate. After a loss, another stake can feel like a way to erase the first one, but each new bet carries its own odds and can deepen the loss. The Gambling Commission identifies chasing losses as an indicator of increased risk of harm, precisely because the aim can shift from making a considered wager to undoing money that has already gone.
Video Games, Loyalty and Manufactured Progress
Games make accumulated investment visible through levels, collections, streaks, ranks, battle passes and unlocked rewards. Leaving can feel like abandoning a version of yourself built across hundreds of hours. Loyalty programmes use related mechanics: the closer somebody feels to a reward, the harder it becomes to walk away.
There is nothing irrational about enjoying progress, a gaming community or earned rewards. The fallacy appears when past progress justifies future time or money on an experience that no longer provides enough value. A high level does not automatically make the next hundred hours worthwhile.
This sits within the wider way behavioural psychology shapes work, relationships and everyday choices. Products can reward consistency so effectively that stopping feels like losing an asset, even where access to that asset requires more spending.
The Relationship Sunk Cost
“I cannot leave because I have already given this person years” is one of the most painful expressions of sunk-cost thinking. The years cannot be retrieved by adding more years, and duration alone cannot establish whether a relationship’s future will be healthy, safe or fulfilling. But relationships are not investment spreadsheets, and no general article can tell an individual whether to stay or leave.
Love, children, housing, care, health, shared history, financial dependence, safety, change and responsibility all matter. Some of these are future costs and benefits; some are moral commitments; some are practical constraints. Research on relationship commitment also shows that satisfaction, perceived alternatives and investments can all influence commitment, which is more complicated than a single bias.
The useful correction is modest. “We have built something worth repairing” is a forward-looking claim that can be tested against behaviour and prospects. “The past must not be wasted” asks the future to repay a debt it cannot repay.
The Terms People Often Confuse
A sunk cost is irrecoverable. A switching cost is a future price of changing course, such as a cancellation fee, retraining period or data migration. An opportunity cost is the value of the best alternative given up when resources stay where they are.
Escalation of commitment is the broader pattern of increasing support for a chosen course despite negative evidence; sunk-cost thinking may help cause it. Loss aversion describes the disproportionate psychological weight of losses. Throwing good money after bad is the everyday description of adding recoverable resources to an unattractive position mainly because of what is already lost.
Persistence means continuing despite difficulty; it can be admirable and rational. Loyalty can carry genuine relational or reputational value. Strategic commitment deliberately restricts future options to build trust or influence others, while the empty claim “I have invested too much to stop” supplies no evidence about what happens next.
Why Governments and Companies Cannot Let Go
Organisations intensify the problem because a project can become attached to careers, budgets, public promises and institutional prestige. Executives who sponsored a product may fear that cancellation will expose their judgement. Ministers who defended a project may face accusations of waste, while teams learn that optimistic updates are rewarded and dissent delays promotion.
Groups do not automatically correct individual bias. Research on escalation of commitment has found that responsibility for an initial choice, self-justification and impression management can encourage further investment. Groupthink can then filter warning signs, while each approval stage creates another constituency whose status or funding depends on continuation.
Concorde is frequently called the “Concorde fallacy” because its development costs rose far beyond early estimates while its commercial case remained contested. Parliamentary debate recorded an estimate of £75.85 million in 1962 and a cost of £795 million by 1979. Yet treating the entire decision as a laboratory example is disputed: governments also pursued technological capability, skilled employment, national prestige and Anglo-French cooperation, while Concorde became a genuine engineering achievement.
That ambiguity matters. A cost overrun does not prove a sunk-cost fallacy, and cancellation is not automatically wise. Claimed strategic benefits must be tested as future benefits, not dignified protection for the past.
Modern public appraisal tries to enforce that separation. The Treasury Green Book excludes sunk costs, includes opportunity costs and requires attention to risk and optimism bias. Staged business cases, independent assurance and periodic reappraisal exist partly because an organisation needs formal opportunities to reconsider a promise before the promise becomes its identity.
When Continuing Is Rational
Quitting is not the default rational answer. Continue when the remaining benefits genuinely exceed the remaining costs, even if the project began badly. A difficult degree, early-stage company or infrastructure programme can still have a strong future case.
Abandonment may also trigger contractual penalties, redundancy costs or legal liabilities. Switching may require expensive retraining, migration or replacement. Learning, reputation and relationships created by earlier work may have future value, while new evidence, improved technology or changed demand may make the original objective more achievable.
State those reasons without relying on the past sacrifice. “Another £100,000 should produce £180,000 of expected benefit” is forward-looking. “We have already spent £2 million” is history.
The Seven-Question Decision Test
When a choice feels trapped by its past, ask:
If I had not already invested anything, would I choose this today?
What will continuing cost from this point forward?
What future benefit can reasonably be expected?
What else could the remaining time or money achieve?
Am I improving the outcome or merely avoiding admitting the loss?
What evidence would persuade me to stop?
If a friend described this situation, what would I advise them?
The first question creates a mental reset, but a newcomer may lack your knowledge, relationships or contractual obligations. The others restore those forward-looking factors without letting pain about the past return in disguise.
Build the Exit Before You Need It
Predetermined exit rules reduce the temptation to rewrite standards after disappointment. A business can define the sales, cost or delivery thresholds that trigger review. A project can release funding in stages, requiring fresh evidence before the next tranche rather than treating the total budget as one irreversible promise.
Independent reviews help because the original decision-maker may be invested in vindication. Research has examined external oversight and separating initial and later decisions, although safeguards consume time and can lose context. Independent reviewers therefore need access to the original evidence.
Stop-loss limits, spending caps and time limits are forms of pre-commitment, not universal instructions. Cooling-off periods create distance from the urgency of winning back money, defending a choice or completing a streak. Reviews should also specify what evidence would justify continuation, so “stop” and “go” face the same burden of proof.
The Past Cannot Be Repaid With the Future
Recognising a sunk cost does not mean becoming someone who abandons every difficult book, relationship, career or project. Achievement often requires patience through periods when the rewards are delayed and the evidence is incomplete. The purpose is to make sure persistence still serves an achievable future.
Time, effort and money already spent may teach, shape and inform you. They cannot be recovered by sacrificing more simply to avoid calling them lost. The strongest decision is sometimes to continue and sometimes to stop, but it should be made facing forwards: the past can provide evidence, never a refund.
