IX Coach

The Sunk Cost Fallacy: Escaping Bad Investments

Why past investment traps future decisions — and the practices that escape the trap

What is the sunk cost fallacy and how do you escape it?

The sunk cost fallacy is the tendency to continue a losing course because of unrecoverable past investment rather than on the basis of future expected value. It is one of the most robustly documented biases in behavioral economics. The corrective is to evaluate forward-only: what will each path deliver from here, regardless of what has already been spent.

Money spent, time invested, emotion poured in — none of it comes back. Yet the human mind treats irrecoverable past costs as reasons to stay in bad situations: finishing a terrible meal because you paid for it, continuing a failing project because the team has been on it for a year, staying in a relationship that no longer works because of shared history. This is the sunk cost fallacy, and it is remarkably expensive. Understanding the mechanism and training a cleaner decision process can recover significant time, money, and wellbeing.

Practices

How these practices connect

Every one of these practices works by relocating the decision's reference point from the past to the future. The sunk cost fallacy operates by making unrecoverable prior investment emotionally present at the moment of forward choice, so each practice either erases that past investment from the calculation, substitutes a person or policy untouched by it, or replaces backward-looking loss with forward-looking loss. None of them argue you should feel differently about the past; they change what the past is allowed to touch.

What's the actual difference between zeroing out the past and the new investor test?

Zeroing out the past is something you do alone, in your own analysis: you set prior investment to zero and evaluate only what each future path offers from here, without adding or subtracting the sunk amount. The new investor test does the same job through a different mechanism — instead of disciplining your own math, it asks whether a rational person with no history in the decision would choose to invest starting today. The new investor test tends to be easier to run under pressure because it externalizes the judgment onto an imagined person rather than requiring you to hold a strict accounting rule in your own head, but it fails the moment you brief that imagined investor on how much you've already spent.

Why doesn't just knowing about the sunk cost fallacy stop me from falling for it?

Zeroing out the past requires deliberately excluding prior investment from a forward-looking calculation, but knowing the rule and applying it aren't the same thing: writing the past investment down and then mentally trying to 'ignore' it leaves it visible, and it keeps influencing judgment anyway. Its own evidence caveat states this plainly — knowing about the fallacy reduces it but does not eliminate it, because the emotional salience of past investment is difficult to fully override even with correct analysis. That's a reason to prefer structural fixes, like a stop-loss policy set before you're invested, over relying on willpower once you already are.

When should I actually set a stop-loss policy — can I do it after I've already started something?

Stop-loss policies work by being set before you are sunk: you define exit criteria at the start, at a time when no sunk costs yet exist to bias the decision, so the pre-committed policy can later act as an anchor against the emotional pull of accumulating investment. Setting one after you're already invested doesn't get you this benefit, because the same emotional salience that distorts the original decision will distort the criteria you write for exiting it. The policy also only works if it's actually enforced — vague stop criteria provide false comfort while investment keeps accumulating, and telling someone else about the policy materially increases the odds you'll follow it.

Is sunk cost fallacy the same thing as escalation of commitment?

No. The sunk cost fallacy is a reasoning error at the level of a single judgment — letting unrecoverable past costs influence a decision that should only weigh future costs and benefits. Escalation of commitment, the behavioral pattern Barry Staw documented starting in 1976, is what plays out over time when a decision-maker personally responsible for a failing course commits further resources to it in successive rounds, more than an uninvolved observer looking at the same facts would. Sunk cost is largely cognitive and explains why leaving feels wasteful; escalation is social as well as cognitive — self-justification, reputational exposure, and the fact that each new round of investment becomes its own sunk cost all intensify it, which is why escalation deepens with every round in a way a single sunk-cost judgment does not.

Why does quitting feel like admitting I'm a failure, and how do I get past that?

This happens because sunk costs get tangled with identity: the more an investment has become part of how you see yourself — 'I'm the one who started this company' — the more exiting feels like self-betrayal rather than an ordinary update to new information. Separating your identity from the investment reframes exit as information processing rather than failure, since the fact that you chose something originally doesn't mean continuing is who you are. This isn't a one-time reframe — it requires ongoing cognitive and emotional work, and deeply identity-linked investments like careers or relationships are especially resistant to it, so expect to have to reapply the distinction rather than settle it once.

How is regret minimization different from just thinking about what I've already lost?

The sunk cost fallacy produces a backward-facing regret — 'I'll regret wasting what I've already put in' — that pulls toward continuing. Regret minimization reorients the same emotional force forward, asking which path a future self with full information would regret more: stopping now, or having continued into a deeper hole. It uses the identical mechanism as sunk cost thinking, just pointed in the opposite direction, but it only works if you genuinely imagine the regret of continuing and not only the regret of stopping — imagining just one side selectively reintroduces sunk cost thinking under a different name.

What does 'the cost of delay' mean and how does it connect to the new investor test?

Cost-of-delay framing calculates the forward cost of not stopping: the value of a better alternative that keeps accumulating for as long as you stay committed to the worse one, which turns 'stopping equals loss' into 'continuing equals ongoing loss.' It shares its forward-only orientation with the new investor test — both ask what a decision looks like judged only by what happens from here — but the new investor test asks whether you'd start at all, while cost-of-delay quantifies what continuing is actively costing you against a specific alternative. That specificity is also its limit: without a concrete alternative to measure against, the cost-of-delay calculation becomes guesswork rather than a number.

How do you choose between these?

If you're setting up a new project, relationship, or commitment and haven't invested anything yet, this is the moment for a stop-loss policy — write the exit criteria now, while no sunk cost exists to bias them, and tell someone else so it's enforced later. If you're already invested and trying to evaluate a decision in the moment, zeroing out the past or running the new investor test are the direct tools — use whichever is easier to hold onto under pressure, since they do the same job through different mechanisms. If what's stopping you isn't the math but the feeling that quitting means admitting failure, separating identity from the investment or reframing through regret minimization address that emotional layer directly rather than trying to out-argue it. And if you notice each new investment is only making the next exit harder, that's a sign to check for escalation of commitment specifically — a pattern that's easier to catch early, before successive small investments add up to a large one, than to interrupt once it's underway.

What these approaches do not settle

None of these practices claims to eliminate the bias; each evidence caveat says the same thing in a different way — awareness reduces but doesn't remove the pull of sunk costs, perspective-taking and reframing techniques are most useful as checks rather than definitive verdicts, and identity separation takes ongoing work rather than a single insight. The practices also don't agree on where the real leverage is: some treat this as a cognitive accounting error best fixed by better math (zeroing out the past, the new investor test), while escalation-of-commitment research suggests that in organizational settings, personal responsibility and social visibility are doing at least as much work as the cost-accounting confusion itself — meaning a purely individual, private fix may address a smaller share of the problem than it would in a public or organizational one.

Practice this with IX Coach

Practice this with IX Coach

Frequently asked questions

What is the sunk cost fallacy and how do you escape it?

The sunk cost fallacy is the tendency to continue a losing course because of unrecoverable past investment rather than on the basis of future expected value. It is one of the most robustly documented biases in behavioral economics. The corrective is to evaluate forward-only: what will each path deliver from here, regardless of what has already been spent.

How do you zero out past investment before evaluating the forward decision?

The sunk cost fallacy works by making past investment emotionally salient at the moment of forward decision. Deliberately zeroing it out in the analysis creates a clean forward-only frame. The prior investment is neither added to nor subtracted from the value of future paths — it simply does not appear in the calculation, because it cannot affect outcomes.

How do you apply the new investor test?

The new investor test removes the asymmetry between you (who carry the sunk cost’s emotional weight) and a fresh observer. If a well-informed outside investor would not put money or time into this project starting today, that is strong evidence that only sunk cost psychology is keeping you in. The test makes the counterfactual concrete and gives it a person — which is more cognitively tractable than abstract principle.

How do you set stop-loss policies before starting projects?

The sunk cost fallacy is hardest to resist when you are already invested. Stop-loss policies are a precommitment device: by specifying exit criteria in advance, you make the decision at a time when no sunk costs exist to bias it. The pre-committed policy then serves as an anchor that resists the emotional pull of accumulating investment.

How do you separate your identity from the investment?

Part of what makes sunk costs sticky is identity: abandoning a course feels like admitting a mistake, which threatens self-image. The more the investment has become part of one’s identity ("I’m the one who started this company / relationship / project"), the more exit feels like self-betrayal rather than rational updating. Separating the decision from identity reframes exit as information processing rather than failure.

What is Escalation of commitment vs sunk cost fallacy: catch the spiral early?

These two terms get used interchangeably and are not the same thing. The sunk cost fallacy is a reasoning error at the level of a single judgment: you let unrecoverable past costs influence a forward-looking decision, when rational choice says only future costs and benefits should count. Escalation of commitment, the phenomenon Barry Staw documented starting in 1976, is the behavioural pattern that plays out over time and largely in organisations: a decision-maker who is personally responsible for a failing course of action commits further resources to it, in successive rounds, more than an uninvolved decision-maker looking at the same facts would. The practical distinctions matter. Sunk cost is one bias among several drivers of escalation; Staw’s experiments isolated personal responsibility as a separate force, because self-justification — the need to prove the original decision was sound — is doing work that pure cost-accounting confusion does not explain. Escalation is also social where sunk cost is largely cognitive: public visibility, reputational exposure, and the cost of admitting error to others all intensify it. And escalation is dynamic — each additional investment becomes its own sunk cost, so the trap deepens with every round, which is why recognising the pattern early is substantially easier than stopping later. In short: sunk cost explains why leaving feels wasteful; escalation explains why you keep paying to stay.

How do you use regret minimization as a forward-looking check?

The sunk cost fallacy produces regret-anticipation that is backward-facing: "I’ll regret wasting what I’ve put in." Regret minimization reorients regret forward: which path will produce more regret from the perspective of a future self with full information? Jeff Bezos popularized this frame. It uses the same emotional force as sunk cost thinking but points in the correct direction — toward future consequences rather than past investment.

How do you calculate the ongoing cost of delay?

Sunk cost thinking focuses on the backward cost of stopping. Cost-of-delay thinking focuses on the forward cost of not stopping: the value of the better alternative accumulating while you remain committed to the inferior course. This reframes "stopping = loss" as "continuing = ongoing loss," which is both more accurate and motivating. The switch in framing makes exit feel like gain rather than abandonment.

What's the actual difference between zeroing out the past and the new investor test?

Zeroing out the past is something you do alone, in your own analysis: you set prior investment to zero and evaluate only what each future path offers from here, without adding or subtracting the sunk amount. The new investor test does the same job through a different mechanism — instead of disciplining your own math, it asks whether a rational person with no history in the decision would choose to invest starting today. The new investor test tends to be easier to run under pressure because it externalizes the judgment onto an imagined person rather than requiring you to hold a strict accounting rule in your own head, but it fails the moment you brief that imagined investor on how much you've already spent.

Why doesn't just knowing about the sunk cost fallacy stop me from falling for it?

Zeroing out the past requires deliberately excluding prior investment from a forward-looking calculation, but knowing the rule and applying it aren't the same thing: writing the past investment down and then mentally trying to 'ignore' it leaves it visible, and it keeps influencing judgment anyway. Its own evidence caveat states this plainly — knowing about the fallacy reduces it but does not eliminate it, because the emotional salience of past investment is difficult to fully override even with correct analysis. That's a reason to prefer structural fixes, like a stop-loss policy set before you're invested, over relying on willpower once you already are.

When should I actually set a stop-loss policy — can I do it after I've already started something?

Stop-loss policies work by being set before you are sunk: you define exit criteria at the start, at a time when no sunk costs yet exist to bias the decision, so the pre-committed policy can later act as an anchor against the emotional pull of accumulating investment. Setting one after you're already invested doesn't get you this benefit, because the same emotional salience that distorts the original decision will distort the criteria you write for exiting it. The policy also only works if it's actually enforced — vague stop criteria provide false comfort while investment keeps accumulating, and telling someone else about the policy materially increases the odds you'll follow it.

Is sunk cost fallacy the same thing as escalation of commitment?

No. The sunk cost fallacy is a reasoning error at the level of a single judgment — letting unrecoverable past costs influence a decision that should only weigh future costs and benefits. Escalation of commitment, the behavioral pattern Barry Staw documented starting in 1976, is what plays out over time when a decision-maker personally responsible for a failing course commits further resources to it in successive rounds, more than an uninvolved observer looking at the same facts would. Sunk cost is largely cognitive and explains why leaving feels wasteful; escalation is social as well as cognitive — self-justification, reputational exposure, and the fact that each new round of investment becomes its own sunk cost all intensify it, which is why escalation deepens with every round in a way a single sunk-cost judgment does not.

Why does quitting feel like admitting I'm a failure, and how do I get past that?

This happens because sunk costs get tangled with identity: the more an investment has become part of how you see yourself — 'I'm the one who started this company' — the more exiting feels like self-betrayal rather than an ordinary update to new information. Separating your identity from the investment reframes exit as information processing rather than failure, since the fact that you chose something originally doesn't mean continuing is who you are. This isn't a one-time reframe — it requires ongoing cognitive and emotional work, and deeply identity-linked investments like careers or relationships are especially resistant to it, so expect to have to reapply the distinction rather than settle it once.

How is regret minimization different from just thinking about what I've already lost?

The sunk cost fallacy produces a backward-facing regret — 'I'll regret wasting what I've already put in' — that pulls toward continuing. Regret minimization reorients the same emotional force forward, asking which path a future self with full information would regret more: stopping now, or having continued into a deeper hole. It uses the identical mechanism as sunk cost thinking, just pointed in the opposite direction, but it only works if you genuinely imagine the regret of continuing and not only the regret of stopping — imagining just one side selectively reintroduces sunk cost thinking under a different name.

What does 'the cost of delay' mean and how does it connect to the new investor test?

Cost-of-delay framing calculates the forward cost of not stopping: the value of a better alternative that keeps accumulating for as long as you stay committed to the worse one, which turns 'stopping equals loss' into 'continuing equals ongoing loss.' It shares its forward-only orientation with the new investor test — both ask what a decision looks like judged only by what happens from here — but the new investor test asks whether you'd start at all, while cost-of-delay quantifies what continuing is actively costing you against a specific alternative. That specificity is also its limit: without a concrete alternative to measure against, the cost-of-delay calculation becomes guesswork rather than a number.

Research

Related concepts

Articles that cover this