The theater subscription
You buy an annual theater subscription for $800. You go to a few shows. The remaining shows are terrible. You do not want to attend. But you think: I already paid for it. If I do not go, the money is wasted. So you go. You sit through performances you dislike, and you are miserable. The money was spent the moment you bought the subscription. Going to the show does not recover it. But it feels like it does.
This is the sunk cost fallacy in its purest form: the decision to continue investing in a losing proposition because of what has already been invested, rather than because of what the future holds. The money is gone. The decision you face now is whether the remaining experience is worth your time. Not whether the past expenditure was worth it. But the sunk cost fallacy collapses those two questions into one, and you answer the wrong one.
The sunk cost fallacy is one of the most expensive cognitive errors in daily life. It keeps people in bad jobs because they have already given years to the company. It keeps people in bad relationships because they have already invested years in the partnership. It keeps people watching bad movies because they have already watched 45 minutes. The investment does not make the future better. It just makes leaving harder.
Escalation of commitment
The sunk cost fallacy often escalates. When an investment is failing, the sunk cost fallacy does not say "stop." It says "invest more." You have already spent so much. Surely you cannot stop now. You need to double down to justify what you have already spent. This is escalation of commitment: the tendency to increase investment in a failing course of action because of what has already been invested.
Escalation is visible in business strategy constantly. A company launches a product that is failing. The executives know the product is failing. But they invest more in marketing, more in development, more in promotion — because they cannot accept that the initial investment was wasted. The additional investment does not make the product better. It just makes the failure more expensive.
Politicians escalate commitment to failed policies for the same reason. They cannot admit the policy was wrong, because that would mean the initial investment was wasted, and the waste would be political. So they double down. More troops to a failing war. More money to a failing program. The doubling down is rationalized by the sunk cost, not by the evidence.
The endowment effect
The sunk cost fallacy is closely related to the endowment effect: once you own something, you value it more highly than you did before you owned it. You paid $800 for the theater subscription. Now it is yours. The subscription has become part of your identity — you are someone who goes to the theater. Letting it go feels like losing part of yourself.
This is psychological, not economic. The subscription's value to you does not change based on ownership. But your perception of its value changes. You now feel entitled to its benefits, and giving it up feels like a loss, not a rational optimization. This is why people keep gym memberships they do not use. The membership is an identity object. Letting it go would mean accepting that you are not the kind of person who goes to the gym, and that is uncomfortable.
The endowment effect also explains why investors hold losing stocks longer than they should. Once they own a stock, they are reluctant to sell it at a loss, because selling would make the loss real. The paper loss is uncomfortable, but paper losses feel reversible. The actual loss, crystallized by selling, is final. So they wait. The wait is justified by the sunk cost of what they originally paid, not by the likely future performance of the stock.
The gambler's fallacy connection
The sunk cost fallacy often pairs with the gambler's fallacy: the belief that future outcomes are affected by past results. A gambler who has lost money at a casino believes that a win is "due." The more they have lost, the more they believe the win is coming. They double their bet, trying to recoup what they have lost. The doubling is driven by sunk cost and the gambler's fallacy, and it usually leads to greater losses.
In business, escalation driven by sunk cost and gambler's fallacy creates dangerous cycles. The company has invested heavily in a strategy. The strategy is failing. But the executives believe the strategy will eventually work, because they have already invested so much. They cannot accept that the investment was wrong. So they wait, and invest more, waiting for the vindication that the sunk cost demands.
The combination of sunk cost and gambler's fallacy is the engine of many financial disasters. Enron continued to lie about its financial health even as evidence mounted, because the alternative — admitting failure — would have required acknowledging the magnitude of the sunk investment in the fraudulent model. The fraud escalated because the sunk cost of the lie was too large to admit.
Breaking the grip
The correction for sunk cost is simple to state and difficult to execute: you must separate the past from the future. The question is not "have I already invested?" The question is "what is the best action from this point forward, given where I am now?" If the answer involves abandoning an investment, then the rational choice is to abandon it, regardless of how much has already been spent.
This requires what psychologists call "mental accounting" — specifically, the ability to close one mental account and open another. The sunk cost is in the closed account. The future decision is in the new account. Conflating the two accounts is the error. Keeping them separate is the correction.
In practice, this means asking: if I had not already invested, would I invest now? If the answer is no, then the sunk cost is not a reason to continue. It is a reason to recognize the error, close the account, and redirect resources to a better use. The money, time, and effort are gone. The only thing you can control is what happens next. Choose based on that, not on what you cannot change.
Continue exploring