Foundationsbehavioral
The Sunk Cost Fallacy: Throwing Good Money After Bad
Overview
'We've spent so much, we can't quit now' — the sunk cost fallacy escalates commitment to failing projects, relationships, and wars because past investments feel like reasons to continue. Rationally, only future costs and benefits matter; sunk costs are gone either way. This lesson explains the psychology (loss aversion, waste aversion, self-justification), the classic experiments, the real-world catastrophes (Concorde!), and the decision discipline that defeats it.
Learning objectives
- Define sunk costs and the fallacy
- Reproduce the classic experiments
- Explain mechanisms: loss aversion, waste aversion, self-justification
- Identify real cases: Concorde, wars, failing projects
- Apply zero-based decision discipline
The fallacy
A sunk cost is unrecoverable past expenditure — the 2 billion already spent on the project, the three years in the degree. The fallacy: letting it influence the continue-or-quit decision. Rational rule: decide on prospective costs vs. prospective benefits only. The $500 ticket is gone whether you attend sick or not — the only question is whether tonight's enjoyment beats rest. Yet most people go ('waste not'), most committees continue failing projects ('we've come so far'), most generals reinforce losing wars. Past investment hypnotizes forward judgment — the most expensive cognitive bias per incident.
Ask: 'If I hadn't spent anything yet, would I start (or continue) this today given remaining costs and benefits?' If no, quit. Sunk costs are analytically invisible — train yourself not to see them.
Classic experiments
Arkes and Blumer (1985): 'You've paid 50 Wisconsin trip the same weekend. Wisconsin is the better trip. Which do you take?' Most choose Michigan — honoring the larger sunk cost over the better experience. The radar-blank-plane experiment: subjects allocating budgets continue funding the original project over a superior alternative once money is sunk. Theater tickets: season subscribers attend more in the first half (sunk cost motivates use) — then the effect fades, showing it's psychological, not rational. The pattern is robust: the more sunk, the harder to quit — escalation scales with irrecoverable investment.
Dawkins's name for it: Britain and France kept funding the supersonic jet long after economics doomed it, partly to 'justify' past spending. The fallacy's monument flies in museums.
Mechanisms
Why do vanished costs grip us? Waste aversion: quitting feels like admitting the spending was wasted — we continue to 'get our money's worth,' converting waste into (illusory) value. Loss aversion: quitting realizes the loss; continuing keeps hope alive — the disposition effect applied to projects. Self-justification: decision-makers escalate to defend their earlier judgment — admitting failure threatens identity (Staw's escalation research: personal responsibility for the initial decision increases escalation). And prospect-theory framing: the sunk cost sets a reference point; quitting books a sure loss, continuing is a gamble to break even — and we're risk-seeking in losses. Four mechanisms, one trap.
Staw showed personal responsibility for a failing course deepens commitment to it. Rotating decision-makers and separating initial approvers from continuation reviews are structural fixes.
Catastrophes
Concorde: billions spent, economics hopeless, continued for national prestige and sunk justification. Vietnam: 'we can't abandon the sacrifice' — past lives as sunk cost, the grimmest escalation. Corporate: failing IT projects, acquisitions, and product lines sustained years past viability (the average failed corporate project burns far more in escalation than in its initial mistake). Personal: degrees pursued despite better alternatives, relationships and careers held by tenure invested. Public: infrastructure white elephants, wars, and subsidy programs that persist because ending them 'wastes' past spending. The fallacy's toll is measured in billions and lives — it's the bias with the highest stakes per occurrence.
The initial bad investment is usually smaller than the escalation it triggers. Killing projects early is the highest-leverage decision skill in organizations.
The discipline: zero-based decisions
Personal: the 'would I start today?' test; pre-commit to kill criteria before investing (if X milestone missed by Y date, exit — decided in advance, cold); mental accounting hygiene — close the books on sunk spending. Organizational: separate initial approvers from continuation reviewers; stage-gate funding (tranches contingent on milestones, not lump sums); premortems and red teams; celebrate smart kills (reward managers who terminate failing projects — invert the stigma). Investment: the stock you own is cash at market price — 'would I buy it today?' is the only question. And distinguish: reputation, learning, and relationships built along the way are real (if intangible) prospective assets — count genuine ones, not the spending itself.
Key points
- Sunk cost fallacy: letting unrecoverable past spending influence continue-or-quit decisions — rationally, only prospective costs/benefits matter.
- Classic proof: the $100-vs-$50 ski trip; escalation scales with sunk amounts; Concorde is the monument.
- Mechanisms: waste aversion, loss aversion (quitting realizes the loss), self-justification, and risk-seeking to break even.
- Catastrophes: Concorde, Vietnam escalation, corporate zombie projects — escalation usually exceeds the original mistake.
- Discipline: 'would I start today?' test, pre-committed kill criteria, stage-gating, separated reviewers, celebrated kills.
Common misconceptions
✓ Truth: The spending is gone either way — 'waste' is determined by future choices, not past ones. Continuing a bad project wastes more; quitting stops the waste.
✓ Truth: Grit is persisting toward worthy goals; the fallacy persists toward unworthy ones. Distinguishing them — via kill criteria set in advance — is the actual skill.
✓ Truth: They do it constantly — Staw's research shows personal responsibility deepens escalation, and corporate history is full of zombie projects. Structure (stage-gates, fresh reviewers) beats sophistication.
Practice questions
Q1. The sunk cost fallacy is best defined as:
- Avoiding all risks
- Letting unrecoverable past expenditures influence whether to continue — when rationally only prospective costs and benefits matter
- Spending too little
- Ignoring the future
Sunk costs are gone either way; the continue-or-quit decision should compare remaining costs to remaining benefits. Honoring past spending is the fallacy.
Q2. In Arkes and Blumer's ski-trip experiment, most people chose the inferior Michigan trip over the better Wisconsin one because:
- Michigan has better snow
- They had paid more ($100 vs. free) for Michigan — honoring the larger sunk cost over the better experience
- They disliked Wisconsin
- The question was confusing
The $100 paid for Michigan hypnotized choice: subjects sacrificed the better trip to 'not waste' the ticket — the fallacy in its purest experimental form.
Q3. Staw's escalation research found that escalation of commitment deepens when:
- Outsiders decide
- The decision-maker was personally responsible for the initial investment — self-justification fuels continuation
- Costs are low
- Information is perfect
Personal responsibility for the original decision threatens identity if the project fails, so decision-makers double down. Separating approvers from reviewers is the structural fix.
Click a question to reveal the answer and explanation.
Frequently asked questions
Are there cases where past investment rationally matters?
Yes — when it built genuine prospective assets: skills, relationships, reputation, infrastructure. Count those real assets; don't count the spending itself. The fallacy is honoring the expenditure, not the assets.
What's the difference between sunk costs and fixed costs?
Fixed costs don't vary with output but may be recoverable (resell the machine); sunk costs are unrecoverable. Only sunk costs should be ignored in forward decisions — fixed-but-recoverable costs still matter.
How do stage-gates help?
Funding in tranches contingent on milestones forces periodic zero-based review — each gate asks 'continue given what we now know?' rather than defaulting to escalation.
Why is it called the Concorde fallacy?
Dawkins popularized the term for continuing doomed investments to justify past spending — after the Anglo-French supersonic jet sustained by prestige and sunk billions long after economics failed.