Sunk Cost Fallacy

The sunk cost fallacy occurs when a person or organization continues, expands, or retains a commitment because of unrecoverable past investment rather than because the choice has the best expected future value. The past investment can be money, time, effort, reputation, or organizational attention.

A large historical loss does not prove the fallacy occurred, and continuing is not automatically irrational. The diagnosis requires evidence that Sunk Costs influenced the decision after relevant future costs, benefits, risks, exit effects, and alternatives were considered.

Key Takeaways

• A sunk cost is a cost classification; the sunk cost fallacy is a decision error.
• The error is not incurring a loss. It is allowing an unrecoverable past amount to distort a current forward-looking choice.
• Continuing can be rational when updated expected future benefits exceed future costs or when stopping creates larger exit, legal, strategic, or reputational costs.
• Past experience can provide new information even when past spending is excluded from the financial comparison.
• Break-even targets, original purchase prices, prior budgets, and years of effort can become psychological anchors rather than relevant forecasts.
• Escalation is more difficult to challenge when the same decision-maker sponsored the original commitment or fears blame for stopping.
• Opportunity Cost remains relevant because continuing uses resources that could support another feasible alternative.
• Stage gates, independent review, precommitted stop criteria, updated base rates, and separate accountability reviews can improve decisions.
• A decision process can be biased even if the project later succeeds, and a disciplined decision can still produce a loss.
• The concept should not be used to diagnose an individual, infer misconduct, or recommend a specific investment.

How the Fallacy Can Develop

Past commitment can create pressure to justify the original choice, avoid appearing wasteful, preserve identity or status, and postpone recognition of a loss. New evidence may then be interpreted more favorably than the same evidence would be for a new proposal.

    flowchart TD
	    A["Past investment becomes unrecoverable"] --> B["Pressure to justify the commitment"]
	    B --> C["Weak evidence receives optimistic weight"]
	    C --> D["More resources are committed"]
	    D --> E["The accumulated sunk cost grows"]
	    E --> B

The loop is possible, not inevitable. Additional funding may be rational when a new test succeeds, market evidence improves, completion costs fall, or abandonment would destroy a valuable option. The analysis must compare updated alternatives rather than assume motive from sequence alone.

Worked Example: Escalating a Weak Project

Assume a company has already spent $8 million on a manufacturing system. That amount is unrecoverable. Management now chooses between continuing and stopping.

The forward-looking calculations are:

Stopping is expected to lose $200,000 from this point, but continuing is expected to lose $1.5 million. Stopping therefore preserves an estimated $1.3 million relative to continuing:

If management says, “We must continue because otherwise the $8 million will be wasted,” that reasoning displays the fallacy under the stated assumptions. The $8 million cannot be recovered under either option.

The recommendation could change if the forecast omits a credible completion contract, customer penalty, strategic dependency, regulatory obligation, resale option, or other future consequence. The example identifies the logic; it does not establish what an actual company should do.

Sunk Cost Fallacy Versus Rational Continuation

The phrase should be applied to reasoning, not merely to the choice. Two organizations can both continue the same type of project: one because updated future value is positive, the other because leaders refuse to recognize an unrecoverable loss.

Opportunity Cost Still Matters

Even if equipment, research, or licenses have already been paid for, continuing to use them can forgo sale, lease, redeployment, or alternative operating value. Staff time, borrowing capacity, collateral, and management attention also have alternative uses.

HM Treasury’s Green Book 2026 distinguishes sunk spending from the opportunity cost of continuing to use resources already paid for. This prevents “already owned” from being confused with “economically free.”

Where the Fallacy Appears in Finance

Capital projects

A company can continue a negative-Net Present Value project to justify prior design, permitting, construction, or acquisition spending. Forecast revisions may be delayed, completion probabilities overstated, or the original sponsor allowed to control the review.

Not every cost overrun justifies cancellation. Near-complete projects can have low incremental cost and high future value even if their total historical return is poor. The correct analysis compares remaining alternatives at the review date.

Corporate acquisitions

After paying an acquisition premium, management may keep funding a weak business to defend the original deal. Integration spending, retained capital, executive attention, and delayed disposal can expand the loss.

Selling immediately is not always superior. Markets may be temporarily illiquid, separation can be costly, and operational changes can create future value. A credible review should compare hold, invest, restructure, partner, and sell alternatives using current forecasts.

Lending and restructuring

A lender may provide additional funding because a large amount has already been advanced. That can be rational if new money improves expected recovery more than its cost, has protected priority, or supports a value-preserving restructuring. It can be a sunk-cost error if prior exposure substitutes for analysis of incremental recovery.

The existing claim, collateral, legal position, and restructuring alternatives remain relevant. Prior funding is not simply ignored; it is separated from the economics of advancing the next dollar.

Investment portfolios

An investor may retain a security solely to “get back to break-even” at the purchase price. The market does not know the investor’s personal break-even target, and the historical price does not determine future return from today’s value.

A hold-or-sell decision still requires expected return, risk, diversification, liquidity, taxes, transaction costs, mandate constraints, and alternatives. Tax basis can change future after-tax cash flows, so it should not be dismissed merely because the purchase price is historical. This article does not recommend holding or selling any security.

Product and technology development

Teams can continue a product because of years of code, research, branding, or internal advocacy. User evidence, technical feasibility, switching cost, shared infrastructure, customer commitments, and strategic options should be updated independently of the desire to preserve prior effort.

Public programs and infrastructure

Large programs can develop political, contractual, employment, and service dependencies. Those future consequences belong in appraisal, while unrecoverable historical spending does not. Public decisions may also have distributional, legal, social, and policy objectives beyond financial return.

Why the Effect Is Difficult to Identify

The foundational Arkes and Blumer study, The Psychology of Sunk Cost, reported greater willingness to continue after investments of money, effort, or time in experimental and field settings. That research supports the existence of a sunk-cost effect; it does not establish that every observed continuation decision has the same cause or magnitude.

In real finance decisions, analysts rarely observe the counterfactual. They may not know:

• what management believed at the decision date;
• which exit costs or strategic dependencies were considered;
• whether new private information improved the forecast;
• whether stopping was contractually or operationally feasible;
• how probabilities and discount rates were estimated; or
• whether the decision-maker personally sponsored the original commitment.

The diagnosis should therefore rely on contemporaneous decision records, alternative analysis, forecast changes, governance evidence, and stated reasons rather than hindsight alone.

Warning Signs in Decision Materials

• The amount already spent appears as the main reason to continue.
• The proposal uses “percent complete” without comparing remaining cost with remaining benefit.
• The original sponsor controls assumptions, approval, and post-investment review.
• Stop, pause, sell, redesign, or partner alternatives are absent.
• Forecasts reset repeatedly without reconciling prior errors.
• Completion is treated as success even if expected operating value is negative.
• Break-even at the historical purchase price becomes the decision target.
• Future funding is described as protecting prior investment rather than creating incremental value.
• Exit costs are exaggerated while continuing costs or opportunity costs are omitted.
• Accountability for the original decision is mixed with the choice that preserves the most value now.

No single warning sign proves bias. Each points to evidence that should be requested and tested.

Decision Controls That Can Help

The U.S. GAO Cost Estimating and Assessment Guide emphasizes scope, technical baselines, documented assumptions, risk analysis, and updates using actual costs. Those practices make escalation easier to detect, but a well-documented forecast can still be biased or wrong.

A Practical Decision Test

1. State the decision in present tense. Ask what should be done now, not whether the original choice was correct.
2. Remove unrecoverable past amounts from the comparison. Keep them in a separate historical-performance record.
3. List feasible alternatives. Continue, stop, pause, redesign, sell, partner, or redeploy may have different future values.
4. Estimate future incremental cash flows. Include exit costs, salvage, taxes, working capital, and implementation risk.
5. Include opportunity costs. Value alternative uses of capital, assets, staff, capacity, and time.
6. Reconcile forecast changes. Explain what new evidence justifies each revision.
7. Use comparable base rates. Contrast inside-view forecasts with outcomes from similar decisions where available.
8. Separate outcome from process. A later success does not validate biased reasoning, and a later loss does not prove it.
9. Identify incentives and ownership. Note who sponsored the original decision and who bears the next loss.
10. Use independent challenge. Give reviewers access to assumptions, alternatives, and authority to recommend stopping.
11. Check legal and nonfinancial obligations. Safety, contract, customer, employee, environmental, and policy effects can be future-relevant.
12. Document the trigger for review. Define which evidence, threshold, or date will cause another decision.

Risks and Limitations

• Hindsight bias: later failure can make a reasonable earlier continuation decision look irrational.
• Hidden information: outsiders may not know the strategic option, contract, or technical evidence supporting continuation.
• Forecast uncertainty: stop and continue values can both be highly uncertain.
• Binary framing: pause, stage, redesign, insure, partner, or sell may dominate both full continuation and immediate abandonment.
• Tax and accounting effects: historical amounts can influence future taxes, covenants, impairment, or reporting.
• Reputation effects: stopping can create future customer, employee, supplier, or policy consequences.
• Career incentives: decision-makers may favor continuation or cancellation for personal reasons.
• Overcorrection: fear of sunk costs can cause premature abandonment of long-duration or experimental investments.
• Scale differences: consumer examples do not automatically predict organizational capital decisions.
• Diagnostic misuse: the label can become rhetoric used to dismiss an opponent rather than analyze cash flows and evidence.

Common Mistakes

• Treating every continued losing project as proof of the sunk cost fallacy.
• Assuming stopping has no cost or that all project assets are worthless.
• Ignoring valid new information learned from prior spending.
• Using total historical loss instead of future incremental value to rank alternatives.
• Treating a return to the purchase price as a financially required target.
• Confusing persistence through temporary volatility with irrational escalation.
• Assuming a disciplined process guarantees a favorable outcome.
• Canceling solely to demonstrate decisiveness without evaluating future value.
• Mixing accountability for the past with value preservation today.
• Applying a behavioral label as personalized investment, legal, or governance advice.

Authoritative Sources

• Arkes and Blumer: The Psychology of Sunk Cost
• HM Treasury: The Green Book 2026
• U.S. Office of Management and Budget: Circular A-4
• U.S. Government Accountability Office: Cost Estimating and Assessment Guide

These sources provide behavioral research, appraisal principles, and estimation practices. They do not diagnose a particular person, project, investment, or institution.

Related Terms

• Sunk Cost: Past cost that has already been incurred and cannot be recovered through the current choice.
• Opportunity Cost: Value of the best feasible alternative forgone by continuing with the current choice.
• Net Present Value: Present value of expected cash inflows minus present value of cash outflows.
• Capital Allocation: Distribution of resources among investment, financing, liquidity, and payout choices.
• Cost-Benefit Analysis: Structured comparison of expected benefits and costs across alternatives.
• Behavioral Finance: Study of how psychology and decision processes affect financial behavior and markets.
• Agency Cost: Monitoring, bonding, and residual loss associated with delegated decisions.

FAQs

This article provides general economic and financial education. It is not a behavioral diagnosis, project recommendation, valuation conclusion, or individualized investment, tax, legal, accounting, or regulatory advice.