Hal Arkes and Catherine Blumer wanted to know whether people would honor a debt that no longer existed. In 1985 they set up a natural experiment at the Ohio University theater. Customers arriving to buy a season ticket were randomly given one of three prices: the full fifteen dollars, a two-dollar discount, or a seven-dollar discount. The discounts were presented as a promotion, applied at the counter, unrelated to anything about the buyer.
Everyone then held the same ticket to the same ten plays. The only difference was how much they had paid for something that was, from that moment on, entirely spent.
Over the first half of the season, full-price buyers attended significantly more plays than discounted buyers. The money was gone in every case and could not be recovered by attending. It changed behavior anyway.
The sunk cost effect is the tendency to continue an endeavor once an irrecoverable investment of money, effort, or time has been made. Economically it is straightforward error: only future costs and benefits should bear on a decision, because the past cannot be altered by anything you do next. Psychologically it is well-motivated, and understanding the motive is what makes it possible to counter.
Arkes and Blumer's own account emphasizes a desire not to appear wasteful — including to oneself. Abandoning the investment converts an expenditure into an acknowledged loss, and acknowledged losses are painful in the way the previous chapters describe. Continuing keeps the account open and the loss unrealized. Barry Staw's work on escalation of commitment adds the other half: the decision-maker is defending not just the money but their earlier judgment, and the more publicly and personally responsible they were for the original decision, the harder they escalate. This is why sunk cost is worst in the people with the most authority over the project.
At scale it produces some of the most expensive mistakes on record. Concorde, which gave the effect one of its names, was continued long after both governments understood it would never repay development. Military escalations have been justified explicitly on the ground that withdrawal would dishonor the losses already taken — an argument that, stated plainly, proposes spending more of the same thing that has already been spent for nothing. Failing acquisitions, doomed IT programs and unfinishable construction projects all run on it.
At personal scale it is what keeps people in degrees they do not want, businesses that will not work, and relationships they have already decided to leave. And it is what a manipulator reaches for once you have invested anything at all. Every argument of the form think how far you have come, you cannot stop now, after everything you have put into this is an appeal to a cost that is by definition unrecoverable. Investment fraud escalates deliberately for this reason: the pig-butchering scam and the advance-fee fraud both work by making each additional payment feel like the small final step that protects everything already paid.
The correction is a zero-based test, and it has to be phrased carefully to keep the past out. Not is it worth continuing, which invites you to weigh the investment. Instead: if I were offered this position today, with these prospects, at this cost, and none of the history — would I take it?
Everything spent is spent regardless of what you decide. The only question the past can legitimately answer is what it taught you about the future.
Defense: The Money Is Already Gone: Escaping Sunk-Cost Entrapment