The Sunk Cost Fallacy And Why It Controls Most Decisions

A psychologist poses this scenario: You've paid $300 for a concert ticket. The day of the concert, you feel terrible — sick, exhausted. You'd have much more value lying in bed than attending. What do you do?

Most people say they'd go to the concert. The $300 is already spent — it's gone whether you go or not. But emotionally, attending the concert feels like not wasting the $300, while staying home feels like throwing it away. So people drag themselves to concerts they don't enjoy, through conditions that damage their health, because of money they cannot recover regardless of what they do.

The economic decision is trivially clear. The psychological decision is not, which tells you something important about how human cognition actually operates.

The sunk cost fallacy is one of the most thoroughly documented biases in behavioral economics, and it's also one of the most pervasive in real-world decision-making. Understanding why requires going several levels beneath "people make irrational choices."

Daniel Kahneman and Amos Tversky's Prospect Theory (1979) is the formal architecture underlying sunk cost behavior. Their key finding: utility is not a function of absolute outcomes but of gains and losses relative to a reference point. And losses loom larger than equivalent gains.

In their experiments, the pain of losing $100 was roughly twice as intense as the pleasure of gaining $100. This asymmetry — loss aversion — has been replicated across cultures and experimental designs. It appears to be a fundamental feature of how human value-perception operates, likely evolved in environments where losses (of food, shelter, status) had catastrophic consequences while equivalent gains were merely nice.

Applied to sunk costs: continuing a failing investment maintains the possibility, however irrational, that the loss can be "recovered." Abandoning the investment forces the emotional recognition of a loss that is now finalized and cannot be undone. Even when continuing is financially worse than stopping, the psychological cost of stopping — the definitive loss — is often higher than the psychological cost of continuing. So people continue.

Richard Thaler's mental accounting framework (1980) adds another layer. People track gains and losses in separate mental accounts. A sunk cost is sitting in an account that's already in the red. Continuing gives the mental account additional "activity" — a sense that the account is still open, still potentially resolvable. Quitting closes the account at a loss. Mental accounts strongly prefer to close in profit (or at least break-even), and this preference overrides the rational calculation that the account's balance cannot actually be recovered.

Barry Staw's research in the 1970s and 1980s introduced a related concept: escalation of commitment. When people are personally responsible for a failing course of action, they tend to invest more resources in it over time — not less.

This is the sunk cost fallacy in institutional form, and it's catastrophically expensive. Staw documented it first in a business simulation: participants who had made the initial investment decision allocated significantly more resources to a failing division than participants who had not made the initial decision, even when given identical information about the division's performance.

The mechanism is responsibility and justification. If you made the decision that led to the current situation, admitting the decision was wrong requires admitting you erred. Continuing invests you in a narrative where the original decision was correct and has simply been undermined by bad luck or insufficient resources. The additional investment is not a rational bet on future returns — it's a psychological defense of past decisions.

This pattern appears throughout organizational and political history. Vietnam: each administration escalated rather than accept the sunk cost of prior escalation. The Concorde supersonic jet: both the British and French governments continued pouring billions into the program long after it was clear it would never be commercially viable, partly because the prior investment made stopping feel like waste. Construction mega-projects routinely follow the same pattern: once started, the sunk cost of partial construction makes abandonment feel more expensive than completion, even when completion costs have risen far above projections.

The Concorde case is so illustrative that economists use "Concorde fallacy" as a technical synonym for the sunk cost fallacy.

Intelligence doesn't protect against the sunk cost fallacy. In some respects, it makes things worse.

Smart people are better at constructing rationalizations. They can generate sophisticated arguments for why the sunk cost is actually relevant — why the prior investment represents learning that gives this path unique advantages, why network effects from prior commitment make continuing more valuable than stopping, why the analysis of "starting fresh" is naively ignoring real costs of transition. Some of these arguments are sometimes true. But they are also exactly the kind of arguments the sunk cost fallacy would generate.

Research by Jonathan Baron and others shows that intelligent, educated people are as susceptible to sunk cost thinking as others — they simply explain it better. The rationalization is more polished. The underlying distortion is the same.

The emotional drivers — loss aversion, identity investment, fear of social judgment — do not respond to IQ. They respond to entirely different interventions.

One of the most powerful and least discussed drivers of sunk cost behavior is identity.

When you've spent ten years in a career, that career is not just a job. It's part of how you answer the question "who am I?" The story of your past ten years, the credentials you've accumulated, the professional relationships you've built, the way you introduce yourself at parties — all of it is organized around this career. Walking away means reorganizing all of that. Not just changing a job, but changing a story.

This is psychologically expensive in a way that pure financial loss aversion doesn't capture. It's not just that you're losing an investment — you're losing a version of yourself. The self that made all those decisions and built all those things has to be replaced by a self that is starting over, which feels like regression rather than liberation.

The identity dimension of sunk costs explains why career changes are so much harder than they "rationally" should be, why marriages that both parties know are over stay in place for years, why leaders persist with organizational strategies that everyone including them knows have failed. The investment isn't just in the external project. It's in the internal narrative that organized the self around that project.

The corrective here is not just cognitive reframing but genuine identity work: building a sense of self that is grounded in values and capacities rather than specific external commitments. The person whose identity is "someone who learns and grows" can abandon a failing path much more cleanly than the person whose identity is "the VP of Marketing at this company."

The standard corrective for sunk cost thinking is to ask: If I were making this decision fresh today, without any prior investment, what would I choose?

This question is structurally correct. It removes the sunk cost from the decision architecture. It forces evaluation of the current and future situation independent of the past.

It is also psychologically brutal to apply to decisions where identity is heavily invested. Here's what makes it hard:

The hypothetical self is not quite real. "If I were starting fresh today" involves imagining a version of yourself that doesn't have the history you have. But you have the history. The skills, relationships, and knowledge gained through the prior investment are real and have value. A genuinely fresh start would involve forfeiting those. So the question, applied literally, overstates the case for quitting.

The transition costs are real. Stopping and starting something else involves genuine costs — financial, social, opportunity. These costs are legitimately relevant to the decision. The sunk cost fallacy doesn't mean transition costs don't exist.

Distinguishing legitimate persistence from fallacy is genuinely hard. Long-term value creation often requires persistence through periods where stopping would feel rational. The investor who holds through a market correction, the entrepreneur who continues through the inevitable trough of a startup, the writer who finishes the book when it's not working — these people are not necessarily committing the sunk cost fallacy. They may be applying genuine long-term thinking.

The distinction is in what's doing the reasoning. Is the persistence driven by an honest analysis of the future value of continuing? Or is it driven by the emotional discomfort of accepting the loss?

The practical test: Can you articulate a specific, forward-looking reason why continuing is likely to produce value that stopping would not — independent of what you've already put in? If yes, you may have a real case for continuing. If the argument collapses when you remove the sunk cost from it, the sunk cost is the real driver.

The counterargument is worth taking seriously: pure sunk cost calculation, applied universally, produces a person who is constantly starting over and never finishing anything.

There are legitimate reasons to factor in prior commitment:

Reputation and reliability. The person who always finishes what they start has something valuable — a track record of reliability. Others can trust them. This is worth something, and it's partly built by absorbing costs you'd avoid in a purely rational analysis.

Learning curves. Almost every worthwhile endeavor gets harder before it gets easier. The sunk cost framing of "I've invested X and should I continue?" often coincides with the trough of a learning curve where quitting looks most rational — right before the returns to continued investment accelerate.

Precommitment value. Making commitments that are hard to reverse has strategic value in some contexts — it signals genuine intent to counterparties and removes the option of easy exit, which can be a game-theoretically useful move.

These are real. None of them, however, are arguments for staying in situations that are clearly not working. They're arguments for being thoughtful about the distinction between "this is genuinely hard and I should persist" and "this is genuinely wrong and I'm staying because of what I've invested."

The question is always the same: What does a clear-eyed analysis of the future say, independent of the past?

At organizational and political scales, the sunk cost fallacy is among the most expensive cognitive errors humans commit.

Wars are continued past the point of strategic sense because of soldiers already lost — as if killing more soldiers undoes the deaths of the first ones. Government programs persist decades past their demonstrated failure because defunding them would mean admitting the initial investment was wasted. Corporate strategies are maintained through obvious collapse because the leadership team that designed the strategy controls the decision to abandon it.

In each case, the structure is the same: decision-makers responsible for the initial commitment have the strongest psychological incentive to continue, and they're also the ones with decision-making authority. The people who could see most clearly that stopping is correct — people with no prior stake in the decision — are often excluded from the decision.

The structural fix that works in organizations: ensure that decisions to continue or stop are made by people who are not the same people who made the initial decision. Independent review, leadership rotation, sunset clauses that require active re-authorization rather than passive continuation — all of these reduce the identity-investment problem that makes sunk cost thinking so powerful.

The fix that works for individuals is simpler and harder: developing the practice of honest forward-looking analysis as a habit — not just pulling it out when stakes are high, but applying it routinely enough that it becomes a natural cognitive move rather than a painful exception.

The past is information. It is not obligation. The decision you made last year was made by the person you were last year, with the information you had last year. You are not that person anymore. The question is what you, now, should do — and the only relevant inputs are current conditions and future possibilities.

Everything else is the fallacy.