What are examples of the sunk cost fallacy?
Eating past full at a buffet because you paid, staying in a long line because of the time already spent, and funding a failing project because of past spending are all examples. In each, money or time that is gone for good steers a choice that only future costs and benefits should decide.
Every example below has the same shape: a cost that is already paid and cannot be recovered under any option gets treated as a reason to keep going. The numbers are illustrations, not data. The buffet. You paid $25 for an all-you-can-eat lunch and you are full, but you go back for a third plate to get your money's worth. The $25 is gone whether you eat more or not, so the only question is whether another plate makes you better off, and when you are full it does not. The line. You have waited 40 minutes for a theme park ride, and a sign now says 60 more. The 40 minutes are spent either way, so compare the next 60 minutes in line with the best other use of that hour.
The college major. A student has finished two years toward an engineering degree and wants to switch to economics, but hesitates because of the tuition already paid. That tuition is sunk. Credits that transfer to the new major are not, because they still shorten the path, so they stay in the comparison. The used car. An owner spent $1,200 on repairs last month, and now the transmission fails with a $2,500 repair quote. A comparable replacement costs $6,000 and the broken car sells for $500, so replacing costs $5,500 net. If the repaired car would serve as well, repairing is $3,000 cheaper from here. The $1,200 plays no part: repair because $2,500 is less than $5,500, not to protect last month's spending.
The restaurant. A restaurant spent $60,000 renovating its dining room last year. It now takes in $45,000 a month, while the costs it could still avoid by closing (staff, food, utilities, and a lease it can exit) run $50,000 a month. Staying open loses $5,000 a month on avoidable costs, so unless the owner expects sales to improve, closing is the better choice. The renovation is sunk, except for fixtures that can be sold, and their resale value counts. The software feature. A company has spent eighteen months building a feature, and user testing shows customers do not want it. Those months are sunk. Launching makes sense only if the remaining cost of finishing and supporting the feature is smaller than the revenue it can still be expected to bring in.
The public project. In a hypothetical case, a city has spent $40 million on a rail extension that is half built. Finishing it would cost $30 million more, and the finished line is expected to deliver benefits worth $25 million. Comparing $30 million with $25 million, stopping is better by $5 million, before counting any penalty for cancelling contracts, which belongs in the comparison because it is still a future cost. The $40 million already spent belongs on neither side. If finishing cost only $20 million, the same rule would say finish. The rule is not a bias toward quitting; it says to ignore past spending in both directions.
To spot the fallacy, ask one question: would this cost come back under any option still open? If not, it is sunk and drops out. If part of it can be recovered, such as resale value, a refundable deposit, or transferable credits, that part stays in. Then compare only the costs and benefits that differ between the remaining choices, which is marginal analysis applied over time. The full decision rule, including a worked case where ignoring sunk costs means finishing a project, is in the sunk cost fallacy explained, and the bias that keeps people stuck is compared in loss aversion vs the sunk cost fallacy.
Go deeper
Related questions
- Is every past cost a sunk cost?
- No. Only the part that cannot be recovered under any remaining option is sunk. A refundable deposit, equipment that can be resold, or a lease that can be ended early is still a live cost, because one choice gets that money back and another does not.
- What is the difference between a sunk cost and a fixed cost?
- A fixed cost does not change with output, but it may still be avoidable, for example a lease the firm can exit by shutting down. A sunk cost cannot be avoided by any choice. That is why the short-run shutdown rule compares revenue with variable cost: fixed costs the firm must pay either way do not affect whether it should keep producing.
- Is it always irrational to stick with something you already paid for?
- No. Continuing is the right call whenever the remaining benefit is larger than the remaining cost, as in the used car example, where repairing beats replacing. The fallacy is letting the past payment be the reason, not the decision to continue itself.