Why Smart People Stay in Bad Movies and Worse Wars

In a 1985 paper that became the canonical citation, Hal Arkes and Catherine Blumer ran a simple experiment. They told subjects to imagine they had paid $100 for a ski trip to Michigan, then later paid $50 for a better ski trip to Wisconsin on the same weekend. The trips cannot both be taken. Which do you go on? More than half the subjects chose Michigan — the worse trip — because they had paid more for it. The money was already gone in both cases. The decision should turn entirely on which weekend they would enjoy more.

The sunk-cost fallacy is the tendency to let unrecoverable past spending influence a forward-looking choice. The standard economic prescription is to treat sunk costs as irrelevant: only future costs and benefits matter. The standard human practice is to do the opposite, and the gap is robust across cultures, ages, and stakes.

It shows up in places that hurt. People sit through bad movies because they paid for the ticket. Investors hold losing positions long past the point of cutting them. Couples stay in failing relationships because they have invested years. The British and French governments kept funding the Concorde supersonic program through the 1970s after it became clear it would never be commercially viable; economists later renamed the bias the "Concorde fallacy" in its honor.

The interesting wrinkle is that sunk-cost reasoning is not entirely irrational. Outside the laboratory, walking away from a project signals to others — and to yourself — something about your judgment, your reliability, your ability to commit. Honoring sunk costs may be a costly but credible way to maintain a reputation for follow-through. Animals show analogous behavior: starlings and mice will work harder for a reward they have already invested effort in, even when an easier option is available. The fallacy may be the wrong calculation but the right strategy in a social, repeated game.