What Is the Sunk Cost Fallacy — Why Past Spending Should Never Drive Future Decisions

In 1985, behavioural-economics researchers Hal Arkes and Catherine Blumer ran a now-famous experiment. They sold theatre season tickets at full price ($15), with a $2 discount, or with a $7 discount — randomly assigned to buyers. Over the following season, they tracked attendance. The result: buyers who paid full price attended more performances than discount buyers, despite all groups having identical access to the same shows. The full-price buyers were attending performances they didn't actually want to see, driven by an unwillingness to let their higher ticket cost feel "wasted." Arkes and Blumer named the pattern the sunk cost fallacy — letting irrecoverable past spending drive forward-looking decisions where it economically shouldn't. The 1985 paper "The Psychology of Sunk Cost" (Organizational Behavior and Human Decision Processes) remains one of the most-cited works in behavioural economics, with over 10,000 academic citations.

This post covers what the sunk cost fallacy actually is, how it differs from loss aversion (covered in what is loss aversion in finance), the real-world finance scenarios where it costs the most, the decision-rule reframe that breaks the bias, and why time-based sunk costs are often the most psychologically powerful.

What the sunk cost fallacy actually is

The sunk cost fallacy is the tendency to continue investing money, time, or effort into a commitment based on what has already been spent, rather than on whether continuing makes economic sense from this point forward.

The economic principle being violated is straightforward: past spending is irrecoverable. Once money is spent, it's spent — no future action can bring it back. The only economically relevant inputs to any forward-looking decision are:

1. The future costs of continuing
2. The future benefits of continuing
3. The opportunity cost of the resources that continuing requires

Past spending is none of these. It's already gone. The decision to continue or stop should be made entirely on the comparison of future costs to future benefits, with past spending ignored.

In practice, humans systematically violate this principle. The Arkes-Blumer theatre-ticket experiment was the clean lab demonstration, but the pattern shows up everywhere:

• Finishing a film you're not enjoying because you paid for the ticket
• Eating a meal you don't like because you're at a paid buffet
• Holding a losing stock because "you've already lost too much to sell now"
• Continuing a degree you don't want because of years already studied
• Staying in a deteriorating relationship because of years invested
• Renovating a money-pit house instead of selling because of money already spent
• Throwing more money into a failing side business because of the capital already deployed

Each of these decisions feels like loyalty to the past investment. Each is the same economic error: letting irrecoverable past spending dictate forward-looking choices.

The original Arkes-Blumer research

The 1985 study designed a clean test specifically because lab researchers had been struggling to isolate sunk cost effects from confounding variables. The theatre-ticket setup eliminated most confounds:

• All subjects had access to the same performances
• All subjects had paid something (avoiding the "free is different" complication)
• The price variation was random, not self-selected (eliminating reverse causation)
• Attendance was directly observable

The result confirmed what Arkes and Blumer hypothesized: the more someone had paid for a ticket, the more performances they attended, even though the marginal cost of attending each individual performance was zero for all subjects. Full-price buyers were dragging themselves to performances they didn't want to see, specifically because skipping a performance felt like wasting money they'd already spent.

Subsequent research has extended the finding across:

• Different cultures (the effect appears in studies done in the US, UK, Japan, Israel, China, India)
• Different age groups (children as young as 6 show measurable sunk cost effects; the bias appears earlier than most cognitive biases)
• Different decision contexts (project management decisions, military procurement, R&D funding, personal relationships)
• Different stake sizes (the effect appears for small stakes but is amplified for larger ones)

A 2014 meta-analysis of 81 studies on sunk cost effects (Arkes and Hutzel) confirmed the original effect size and showed it replicates reliably across decision types. The bias is not a fringe finding — it's one of the best-documented systematic deviations from rational decision-making in the entire behavioural-economics literature.

How sunk cost shows up in real-world finance

Three patterns produce the most measurable financial harm:

1. Holding losing investments "until they recover"

A retail investor buys a stock at ₹500, watches it fall to ₹300. The rational forward-looking question is: "Given everything I now know about this company, would I buy this stock today at ₹300?" If the answer is no, the rational action is to sell at ₹300 and redeploy the capital into a better investment.

Most retail investors do not ask the forward-looking question. They ask: "Can I wait for it to come back to ₹500 so I don't have to lock in a loss?" That question is sunk cost reasoning — the ₹500 entry price has become a psychological anchor that makes the ₹200 paper loss feel like a debt that needs to be "made back" before exiting.

The cost: holding a deteriorating position eats opportunity cost. The ₹300 stuck in a losing stock could have been compounding in a diversified index fund at 10-12% annually. Across a typical 3-5 year hold-until-recovery cycle, the foregone return often exceeds the recoverable loss even if the original stock eventually does recover.

2. Continuing to fund a failing project or business

Indian small-business owners and US founders both report the same pattern: it's easier to commit ₹10 lakh of new capital to a struggling business than to walk away after ₹20 lakh has already been deployed. The new ₹10 lakh decision should be evaluated on its own forward-looking merits — does this additional capital have a positive expected return given the current state of the business? — but it usually gets evaluated through the lens of "we've already put in ₹20 lakh, we can't walk away now."

The "we can't walk away now" reasoning is structurally identical to the theatre-ticket study. The ₹20 lakh is gone whether you walk away or continue. The only relevant question is whether the new ₹10 lakh has higher expected return deployed in the failing business or deployed elsewhere. In most cases where this scenario arises, the answer is "deployed elsewhere" — but the sunk cost lens pushes the decision toward continued deployment.

3. Renovating instead of selling a money-pit house

A buyer purchases a property for ₹80 lakh, discovers ₹15 lakh of structural issues that need fixing. After fixing, they discover another ₹10 lakh of problems. After another year, another ₹8 lakh. At every decision point, the right question is: "Is the next ₹X of repair cost less than the value it adds to the property?" — not "We've already put ₹33 lakh into repairs, we can't walk away now."

This sunk cost trap is common enough in real estate that the term "money pit" exists as shorthand for it. The financially correct decision is usually to sell at market value, take the loss, and redeploy capital. The emotionally common decision is to keep pouring in money, hoping each new round of repairs will finally make the property "worth it."

How sunk cost differs from loss aversion

Loss aversion and sunk cost are related but distinct biases. The relationship:

• Loss aversion = the asymmetry where losing ₹1,000 hurts about 2× as much as gaining ₹1,000 feels good (covered in what is loss aversion in finance)
• Sunk cost fallacy = the specific behavioural consequence where loss aversion pushes you to avoid recognizing a loss by continuing the failed commitment

The link is this: walking away from a failed commitment forces the loss to be explicitly recognized ("I lost ₹50,000 in that stock"). Continuing the commitment lets the loss remain notional and not-yet-realized ("it's still down but it might recover"). Loss aversion's pain response makes the notional-loss path feel better than the recognized-loss path, even when the notional-loss path produces worse forward-looking outcomes.

So sunk cost behaviour is loss-aversion-induced denial — refusing to convert a paper loss into a realized loss, at the cost of continued exposure to further downside.

The two biases reinforce each other and often appear together in the same decision.

The fresh-start decision rule

The most effective behavioural-economics-research-backed technique for breaking sunk cost reasoning is the fresh-start rule:

If I were starting from today with nothing committed — no money already spent, no time already invested, no relationship history — would I choose to enter this commitment now, given what I currently know?

If the answer is no, the right forward-looking decision is to exit, regardless of how much has been spent so far.

The fresh-start framing works because it explicitly removes the past-spending anchor that triggers sunk cost reasoning. Instead of "should I continue this thing I've already invested in?", the question becomes "if I were making this decision today from scratch, would I choose this option?"

A worked example. An investor holds a stock bought for ₹500/share, currently at ₹280/share, paper loss ₹2 lakh on a ₹5 lakh position. The fresh-start question: "If I had ₹2.8 lakh cash today and was choosing where to deploy it, would I buy this stock at ₹280 over the alternatives available?" If the honest answer is "no — I'd put it in an index fund," then the forward-looking action is to sell, recognize the ₹2 lakh loss, and redeploy ₹2.8 lakh into the index fund. The ₹2 lakh is already gone either way; the only question is whether the remaining ₹2.8 lakh continues to underperform in the original stock or compounds in the better alternative.

The hard part is being honest. Many people who think they're applying the fresh-start rule are still subtly weighting past spending in their reasoning ("well, it could recover" — but would you buy it fresh today? probably not). Genuine fresh-start reasoning requires noticing the moment your mind starts arguing for continuation and treating that argument with suspicion.

Time-based sunk costs are often the strongest

Money-based sunk costs are bad enough. Time-based sunk costs are often worse, because time cannot be recovered under any circumstance.

Common time-based sunk cost traps:

• Education: "I've already done 4 years of medical school, I can't quit now" — except quitting before year 5 saves another year of opportunity cost on top of the 4 already invested
• Career: "I've put 8 years into this company / industry / specialty, I can't pivot now" — except continuing in a misaligned career path keeps compounding the misalignment
• Side projects: "I've spent 18 months on this app, I can't abandon it" — except finishing an app no one wants is more 18 months than walking away
• Relationships: outside the scope of this finance-focused post, but the same logic applies

The honest reframe: every additional hour you spend on a failed direction is a new sunk cost in the making. The 18 months already spent on the dead-end app are gone whether you continue or walk away. The next 18 months are either compounding on the same dead-end or available to redeploy on something better. The right question is forward-looking, not backward-looking.

What to actually do with this

Three practical applications:

Set explicit "exit criteria" before entering big commitments. When starting a new investment, project, or commitment, define in advance what would constitute "this isn't working — time to exit." Examples: "If this stock loses 30% relative to the index over a 12-month period, I sell" or "If this side project has no users after 6 months of consistent effort, I shut it down." Pre-committed exit criteria are easier to follow than in-the-moment decisions because they were made before sunk-cost emotions could distort the reasoning.

Audit your current commitments through the fresh-start lens. Pick one financial position, project, or recurring spending item you're not sure about. Ask: "If I had today's cash / time / energy and was choosing from scratch, would I pick this?" If the answer is no, you've identified a sunk cost trap. The next decision is whether the cost of exiting (transaction costs, social cost, emotional cost) exceeds the cost of continued misalignment.

Treat time-based sunk costs with extra suspicion. Because time can't be recovered, the temptation to "just finish what I started" is strongest for time investments. Combine this with the fresh-start rule for any multi-year commitment — education, career, large side projects, long-term relationships with friction. Sometimes the answer is to continue; the point of the exercise is to make sure that answer is forward-looking, not loss-aversion-driven retroactive justification.

Sources

• Hal R. Arkes and Catherine Blumer, The Psychology of Sunk Cost, Organizational Behavior and Human Decision Processes, Vol. 35, No. 1 (February 1985), pp. 124-140
• Daniel Kahneman, Thinking, Fast and Slow (Farrar, Straus and Giroux, 2011) — chapter 32 covers sunk costs as a regret-avoidance bias
• Richard Thaler, Misbehaving: The Making of Behavioral Economics (W.W. Norton, 2015) — chapter 5 on sunk costs and mental accounting
• Hal Arkes and Peter Ayton, The Sunk Cost and Concorde Effects: Are Humans Less Rational Than Lower Animals?, Psychological Bulletin, Vol. 125, No. 5 (1999) — the famous "Concorde fallacy" comparison
• Securities and Exchange Board of India (SEBI), Investor Education Material on Common Biases — sebi.gov.in
• US Securities and Exchange Commission, Behavioral Biases in Investing — investor.gov