Sunk cost

A sunk cost is a past expenditure of money, time, or effort that cannot be recovered regardless of future choice. Standard economic theory holds that sunk costs should not influence forward-looking decisions; only expected future costs and benefits matter to the decision at hand. The sunk cost fallacy is the widely observed pattern in which people continue investing in a course of action because of past commitment rather than because of expected future return.

The construct sits at the intersection of behavioral economics and decision psychology. Arkes and Blumer (1985) documented the effect experimentally in Organizational Behavior and Human Decision Processes and gave the contemporary literature its anchor demonstration. Related work on escalation of commitment (Staw 1976, 1981) developed the organizational-behavior side of the same phenomenon; loss aversion in prospect theory (Kahneman & Tversky 1979) supplies one of the leading mechanism candidates. The effect has replicated many times in hypothetical scenarios, including a large multi-site replication (Klein et al., 2014), though evidence from studies with real stakes and from field settings is more mixed (for example, Ashraf, Berry & Shapiro, 2010).

Three points are routinely missed in popular treatments. First, the fallacy is about forward-looking decisions: spending past money is not the error, treating past spending as a reason to continue is. Second, the everyday voice — “I’ve put too much in to walk away now” — captures the pattern in natural language but is not itself a mechanism; what produces the pattern is contested and varies by context. Third, knowing about the fallacy does not guarantee avoiding it, which is why safeguards set up in advance (kill criteria, outside review, conditions for stopping named in advance) are often recommended, although they have rarely been tested outside the lab.

The sunk cost fallacy is consequential precisely because it operates on the high-stakes decisions where the costs of error are largest: major financial investments, career transitions, long-term relationships, organizational projects, and government infrastructure. In each domain, the bias systematically tilts decisions toward continuation past the point at which continuation can be justified on forward-looking grounds.

The patterns are visible at every scale. Personal finance research has linked sunk cost reasoning to underperformance in individual investment decisions — investors hold losing positions longer than is rational because selling crystallizes the loss. Infrastructure research has documented similar institutionalized escalation in major project planning, with cost overruns highly skewed toward catastrophic right-tail outcomes. Business literature catalogs many examples of corporations continuing to invest in failing initiatives until external pressure forces correction.

For individuals, recognizing the fallacy can help with personal decisions. It is also a common consideration in stay-vs-go decisions — the structured choice about whether to remain in or leave a voluntary high-stakes commitment.

The technical concept of sunk cost is foundational in microeconomics, with the principle of "ignore sunk costs" articulated in early 20th-century economic writing. Hal Arkes and Catherine Blumer's 1985 paper "The psychology of sunk cost" in Organizational Behavior and Human Decision Processes brought the concept into experimental psychology, demonstrating that ordinary people consistently honor sunk costs even when explicitly told the prior expenditure was unrecoverable.

Their best-known study was a field experiment: people buying season tickets to a campus theater series were randomly charged full price or given a discount, and those who paid full price attended more plays in the first half of the season. The plays were the same for everyone; the only difference was how much each group had already paid, a sunk cost that should, in theory, have been irrelevant. (The familiar example of driving through a snowstorm to a game because the tickets were paid for comes from Thaler, 1980.)

Barry Staw's foundational 1976 paper "Knee-deep in the big muddy" formalized the related concept of escalation of commitment, the observable behavioral pattern in which decision-makers continue investing in a failing course of action. The two concepts are intimately related but distinct: the sunk cost fallacy is the internal cognitive error; escalation of commitment is the sustained behavioral pattern that often results from it. Escalation can also be driven by social and organizational factors — face-saving, accountability concerns, agency problems — that the cognitive sunk cost fallacy alone does not capture.

Several distinct psychological mechanisms contribute to sunk cost reasoning. Mitigation strategies that target one mechanism without addressing the others tend to fail.

• Loss aversion. Prospect theory (Kahneman & Tversky, 1979) holds that losses loom larger than equivalent gains; later estimates put losses at roughly twice the weight of gains on average (Tversky & Kahneman, 1992), though estimates vary. Quitting a failing project crystallizes the loss; continuation preserves the possibility, however slim, that the past investment will eventually pay off.
• Self-justification and identity defense. Past commitments become entangled with self-image. Walking away means acknowledging the original decision was wrong, which the mind resists. Staw's foundational research highlighted this in organizational contexts; subsequent work has documented it in personal investment, relationships, and careers.
• Mental accounting and framing. The mind treats money in separate "accounts" rather than as fully fungible (Thaler, 1999). Money already spent on a project is mentally bucketed with that project; abandoning it feels like wasting the money, even though it is already gone.
• Social and reputational concerns. In organizations, continuing a failing project preserves the option of eventual vindication; terminating it accepts a definite cost in reputation. This is not a cognitive error but a rational response to misaligned incentives, and it strengthens the apparent sunk cost effect institutionally.
• Goal-completion drives. Closeness to a goal, particularly when partial progress is visible, intensifies commitment. Projects often see escalation when they are perceived as nearly complete, regardless of whether completion is actually likely.

Developmental evidence is mixed: Arkes and Ayton (1999) argued that young children honor sunk costs less than adults, possibly because adults overapply a learned 'don't waste' rule, but later studies have not settled the question. Some animal species show analogous behavior in foraging decisions (Sweis et al., 2018).

What it can do. Recognizing the sunk cost fallacy provides a vocabulary and a structure for forward-looking decisions. The most reliable practical question is: "If I were starting from today's position, with today's information, would I choose to invest the next dollar/hour/effort in this option?" This reframing severs the link to past investment by asking the decision purely in marginal terms. Pre-mortems and structured decision reviews that document expected outcomes at the point of original commitment also help — they make the original case explicit so that current evidence can be evaluated against it.

What it can't do. Awareness alone may not be enough. Expertise does not reliably protect against the fallacy, although the evidence is mixed. Loss aversion, self-justification, and mental accounting are not eliminated by knowing about them. Structural protections, such as kill criteria set at the start of a project, periodic review by people with no stake in past spending and decision logs that name conditions for stopping in advance, are meant to take the decision away from the person most invested in it rather than relying on willpower; preset limits reduced escalation in one study (Simonson & Staw, 1992), but such safeguards have rarely been tested in the field.

"Ignoring sunk costs means abandoning anything that hasn't paid off yet." No. The principle is that the size of past investment should not, by itself, drive whether to continue. Genuine forward-looking reasons to continue may exist — high expected future returns, low marginal additional cost, important learning value. The principle simply removes the past investment as one of those reasons.

"The sunk cost fallacy and escalation of commitment are the same thing." Related but distinct. The sunk cost fallacy is the internal cognitive error of weighting past investment in forward decisions. Escalation of commitment is the observable sustained behavior of increasing investment in a failing course of action. Escalation is often driven by sunk cost reasoning but also by social, reputational, and agency factors that the purely cognitive fallacy does not capture.

"Smart people don't fall for the sunk cost fallacy." Intelligence and expertise do not reliably protect against it, but the evidence is mixed: brief training in cost-benefit reasoning has improved choices in experiments (Larrick, Morgan & Nisbett, 1990), and effects are often stronger when people were personally responsible for the original decision (Staw, 1976).

"The fallacy is always about money." It is not. Sunk-cost effects have been found for past time and effort as well as money, though some studies find past time weighs less than past money (Soman, 2001). People stay too long in failing relationships, finish bad books, complete graduate programs they should leave, and continue businesses they should close, and more than one mechanism may be involved.

Consider a professional three years into a doctoral program that has stopped progressing. Funding has run out, the dissertation topic has lost momentum, the advisor has become unavailable, and the job market the degree was meant to enter has shifted. The forward-looking analysis is clear: completion would require two more years of unfunded work for a credential whose marginal value has declined. Yet the decision feels excruciating — three years and tens of thousands of dollars are at stake.

The sunk cost fallacy is the voice that says "I've put too much in to walk away now." The protective reframing is: "If I were applying to this program today, with today's funding situation and today's job market, would I enroll?" If the answer is no, the past three years are providing cognitive cover for a forward decision no neutral observer would make. This does not mean abandoning is automatically right — there may be genuine forward reasons (proximity to defense, alternate uses of the credential, value of the writing experience) — but past investment is not one of them. The structural protection is to consult someone without sunk-cost exposure: a mentor outside the field, a former student who left the program, a financial advisor. They can see the marginal calculation the person inside cannot.

• Opportunity cost — the value of the best alternative not chosen. The forward-looking counterpart to sunk cost: opportunity cost should drive decisions; sunk cost should not.
• Stay-vs-go decision — the structured class of choices about whether to remain in or leave a voluntary commitment. Sunk cost is one of the central considerations.
• Cognitive bias — the broader category of systematic deviations from rational judgment. The sunk cost fallacy is one of the most common.
• Decision hygiene — the structural practices (kill criteria, decision logs, periodic outside review) that mitigate sunk cost reasoning before it occurs.
• Escalation of commitment — the observable behavioral pattern of increasing investment in a failing course of action, often driven by sunk cost reasoning combined with social and organizational factors.
• Loss aversion — the prospect-theory idea that losses loom larger than equivalent gains, often estimated at about twice as much (Tversky & Kahneman, 1992). One proposed mechanism behind the sunk cost fallacy.
• Mental accounting — the tendency to treat money in separate cognitive "accounts" rather than as fully fungible. Past investment becomes mentally bucketed with the project, reinforcing the sunk cost effect.
• Bounded rationality — Simon 1955/1956 framework within which sunk-cost reasoning can be understood as a cognitive shortcut, such as a 'don't waste' rule applied too broadly, even though honoring sunk costs departs from expected-utility theory.
• Risk aversion — risk aversion in expected-utility theory; loss-aversion and sunk-cost reasoning are both prospect-theory features that depart from EU-theory predictions.
• Nudge theory — choice-architecture interventions can be designed to counteract sunk-cost reasoning, for example by setting stopping rules in advance; preset minimum targets reduced escalation in one study (Simonson & Staw, 1992).

The LifeByLogic Should I Quit? tool is about leaving a job. It scores burnout, meaning at work and intent to leave with 26 LifeByLogic-written questions, then runs 10,000 simulated versions of staying and of leaving built from forward-looking figures such as pay, expected salary, raises, expenses and search difficulty. Time or money already put into the job is not an input, but the tool does not measure sunk-cost thinking, and its questions have not been validated and have no norms. The full methodology is documented on the tool methodology page.

Take the Should I Quit? test →

This entry is intended as a citable scholarly reference. Choose the format that matches your context. The retrieval date should reflect when you accessed the page, which may differ from the entry's last-reviewed date shown above.

LifeByLogic. (2026). Sunk Cost Fallacy: Definition and Everyday Examples. https://lifebylogic.com/glossary/sunk-cost/

LifeByLogic. "Sunk Cost Fallacy: Definition and Everyday Examples." LifeByLogic, 25 Oct. 2026, https://lifebylogic.com/glossary/sunk-cost/.

LifeByLogic. 2026. "Sunk Cost Fallacy: Definition and Everyday Examples." October 25. https://lifebylogic.com/glossary/sunk-cost/.

@misc{lblsunkcost2026,
  author = {{LifeByLogic}},
  title = {Sunk Cost Fallacy: Definition and Everyday Examples},
  year = {2026},
  month = {oct},
  publisher = {LifeByLogic},
  url = {https://lifebylogic.com/glossary/sunk-cost/},
  note = {Accessed: 2026-10-25}
}