Sunk Cost Fallacy
Resources already spent, and unrecoverable no matter what happens next, still shape decisions about what to do next, even though a purely forward-looking choice shouldn’t weigh them at all.
What Is It?
Behavioral economists Hal Arkes and Catherine Blumer demonstrated the pattern experimentally in a 1985 paper, showing that people were more likely to continue with a chosen course of action, a trip, a project, a purchase, specifically because they’d already invested money or effort in it, even when a fresh, unbiased look at the same choice would favor stopping. The rational standard is straightforward: a past cost that can’t be recovered regardless of what happens next shouldn’t factor into a forward-looking decision, only future costs and benefits should. What people actually do is different, treating the size of a past investment as a reason to continue, as though abandoning the effort would mean the earlier spending was wasted, when in fact that spending is already gone either way, and continuing can waste more on top of it.
Why Does It Matter?
The bias shows up in small, everyday decisions as readily as large ones: finishing a bad meal because it’s already been paid for, sitting through a disappointing conference session because travel was booked, continuing to use a tool or vendor because of the time already spent learning it, rather than migrating to something better. In each case, the honest question, what’s the best choice from this point forward, gets quietly replaced by a different one, how do we avoid feeling like the earlier spending was wasted, and those aren’t the same question, even though they can feel identical from the inside.
Organizations are exposed to this at scale because so many decisions involve visible, quantifiable past investment: budget already allocated, months already spent, a system already built. The clearer and more countable the sunk cost, the easier it is to let it quietly substitute for a genuine forward-looking judgment, particularly when no one has explicitly separated the two questions in the room.
What Changes Once You See It?
You start asking the past investment to leave the room entirely when evaluating a choice, treating the decision as though the resources already spent were zero, and reasoning only from here forward.
You start noticing the specific language that signals sunk-cost reasoning is doing the work, “we’ve already put so much into this,” “it would be a waste to stop now,” since both describe the past, not the future, dressed up as a reason to continue.
You also get more comfortable killing something specifically because it no longer makes sense, independent of how much has already gone into it, since the discomfort of “wasting” a past investment is a poor guide to what’s actually worth doing next.
Common Misunderstandings
- It isn’t the same as Escalation of Commitment, though the two are closely related and often confused. Sunk Cost Fallacy is the narrower decision error of allowing irrecoverable past investment to influence a forward-looking choice, Escalation of Commitment is the broader organizational and psychological process, identity, reputation, self-justification, responsibility, that explains why sunk costs are so specifically hard to walk away from once other people and public commitments are involved. A third, related concept, Path Dependence, is different still: sunk cost asks whether we’re counting money that’s already gone, Path Dependence asks what real dependencies and switching costs the past has actually created that legitimately affect today’s choice.
- It doesn’t mean everything created by a past investment should be ignored going forward. The money already spent is sunk and irrelevant, but capabilities, infrastructure, contracts, and accumulated knowledge that investment created can still be real assets affecting future costs and benefits, ignoring those isn’t correcting for sunk-cost bias, it’s a different mistake.
- It isn’t a claim that persistence is always wrong. Sometimes continuing really is the better choice, on its own forward-looking merits, the fallacy is specifically about letting past, unrecoverable spending tilt that judgment rather than sitting outside it.
- It doesn’t require large amounts of money to operate. The bias can show up on a five-dollar meal or an afternoon’s wasted effort as readily as on a multi-year investment, the mechanism doesn’t require the cost to be large.
- It isn’t fixed by simply knowing about the bias. Awareness reduces but doesn’t eliminate it, structural habits, explicitly asking “if we hadn’t already spent this, would we start now,” tend to work better than willpower alone.
Diagnostic Question
If none of what we’ve already spent on this existed, would we choose to spend what it will cost from here forward?
Explore Further
Field Notes
- None yet.
Related Field Guide
Origin
Hal R. Arkes and Catherine Blumer, “The Psychology of Sunk Cost,” Organizational Behavior and Human Decision Processes (1985), the first systematic experimental demonstration of the effect; the underlying concept has roots in earlier economic writing on the irrelevance of sunk costs to rational forward-looking decisions.