Why Finish a Film You Hate? Money Already Spent Should Not Matter
By the BrainSnail editorial team. How these articles are written and checked, and how to tell us when one is wrong.
Rational decision making considers only future costs and benefits, since what has already been spent cannot be recovered whatever you choose. People reliably fail to do this, and the failure scales from films to wars.
The principle and the failure
A cost is sunk when it has been incurred and cannot be recovered regardless of what happens next, and the standard analysis holds that such costs should be irrelevant to any decision, since the question is always what the available options yield from here. A person who has paid for a ticket to something they are not enjoying faces exactly the same choice as somebody who was given the ticket, because the money is gone either way, and yet the first is far more likely to stay. The tendency to let irrecoverable costs influence the decision is robust, appears across many domains, and is one of the more securely established departures from the standard model of choice.
Why people do it
Several mechanisms contribute and they are not mutually exclusive:
- •Aversion to waste, since abandoning something makes the loss explicit and definite
- •Loss aversion, since continuing preserves the possibility that the investment pays off
- •Self-justification, since abandoning admits the original decision was wrong
- •Reputation, where others observed the commitment and abandoning it is visible
- •Treating the spending as an investment with an account that must be balanced
- •Optimism that additional effort will turn the situation around
The large-scale version
The same pattern operates on projects and policies where the sums are enormous. Development projects, weapons programmes and infrastructure schemes are continued past the point where the remaining costs exceed the remaining benefits, with the money already spent cited explicitly as a reason to continue, which is exactly the error. The phenomenon is sometimes named after a supersonic aircraft programme that two governments continued despite clear commercial projections, and comparable cases are documented across countries and decades. Military commitments show the pattern acutely, since casualties already suffered are invoked as a reason to continue, which is emotionally comprehensible and is logically the same mistake. Escalation of commitment is the term used in organisational research, where the conditions producing it have been studied extensively.
When continuing is correct
Not every decision to persist is an error and distinguishing the cases is the practically useful part. Continuing is right whenever the remaining benefits exceed the remaining costs, which can be true of a project that has already overrun enormously, so the overrun itself proves nothing. Learning changes the calculation, since effort spent may have revealed information that improves the prospects rather than merely consuming resources. Reputation for finishing what one starts has real value in repeated dealings, which makes abandoning genuinely costly in a way the simple analysis omits. Contracts and obligations can make abandonment expensive in cash rather than only in pride. The error is specifically letting the irrecoverable expenditure enter the reasoning, not persisting as such, and conflating the two produces its own bad decisions.
What reduces it
Several interventions have measurable effects. Separating the decision to continue from the people who made the original commitment removes the self-justification component, which is why independent review of ongoing projects is recommended and why organisations that rotate decision-makers show less escalation. Setting explicit abandonment criteria in advance, meaning conditions that will trigger a stop before anyone is invested, converts the judgement into a rule. Framing the question as whether one would begin the project today with the current information is a simple and effective reframing. Accounting practice that separates past expenditure from forward projections helps. And recognising that the tendency is strongest where the original decision was personal and public suggests that reducing the personal stake in being right is the lever that matters most.
The takeaway
A cost that cannot be recovered should not affect a decision, since the options yield the same from here regardless, and people reliably let it anyway. Aversion to visible waste and reluctance to admit the original decision was wrong are the main drivers. Separating continuation decisions from the people who made the commitment and setting abandonment criteria in advance both reduce it.