In short
Irrational escalation is the pattern in which two or more parties keep raising their stake in a contest because each new bid looks cheaper than accepting the loss of the last one, until the total committed exceeds anything the prize was worth. It shows up in bidding wars for a house or a contract, in lawsuits pursued past the value of the claim, in price wars between rivals, in arms races, and in arguments kept going because stopping would mean having lost.
How irrational escalation works
The dollar auction is a trap built from reasonable steps. The first bid is a bargain: 5 cents for a chance at a dollar. The second bidder's 10 cents is also a bargain. From then on, whoever is behind faces a choice between a certain loss of their last bid and a small extra payment that might win the dollar, and the small extra payment keeps looking better. Shubik identified two turning points. When the two top bids add to more than a dollar, the auctioneer is guaranteed a profit whatever happens next. When one bid reaches a dollar, the other bidder can pay 1.05 to avoid losing, say, 80 cents outright, and from then on both bidders are losing money while continuing to bid in order to lose less.
What makes the escalation irrational is not any single bid but the sequence. Each decision is taken by comparing the cost of the next step with the cost of stopping, and the money already bid is treated as recoverable if only the bidding continues. Shubik described the position in terms of a damage exchange rate: once a bidder has reached a dollar, a rival can inflict a loss of a dollar on them for the price of 5 cents, and without an upper limit there is no end to how far the exchange can go. The bidders are no longer playing for the prize; they are playing not to be the one who pays for nothing.
Allan Teger studied what people actually feel as this happens and reported it in Too Much Invested to Quit in 1980. Early in the game the motives are economic: win the dollar, limit the loss. As bids pass the value of the prize the motives change character, and bidders describe wanting to save face, to not let the other person win, and to justify what they have already spent. The contest becomes about the opponent rather than the dollar. Brockner and Rubin called this family of situations entrapment: a conflict in which the costs already incurred become the reason to keep incurring more, and reviewed the conditions that deepen it, such as a visible audience and an opponent who seems about to quit.
Game theory treats the auction differently from the people who play it. Shubik noted that an opening bid of a dollar which nobody contests is an equilibrium of the game, and that once two bids are in there is no natural stopping point short of an external limit. Barry O'Neill showed in 1986 that when each bidder has a known, finite budget there is a rational opening bid, worked out backward from the end of the game, that the other bidder should not contest, and he concluded that the auction is not inherently a trap. The escalation seen at parties and in classrooms is therefore a fact about how people play, not a property of the game that no strategy can avoid. In most real contests the budgets are not known and nobody has worked the game backward, which is why deciding a limit before the first bid is the practical substitute for the solution.
The original study
Martin Shubik, an economist and game theorist at Yale, published "The Dollar Auction game: A paradox in noncooperative behavior and escalation" in the Journal of Conflict Resolution in March 1971. The paper is three pages long and describes what he called an extremely simple, highly amusing and instructive parlor game. An auctioneer sells a dollar bill to the highest bidder, with the understanding that both the highest and the second-highest bidder pay. Bids rise in multiples of 5 cents, and the game ends when nobody bids for a set time. Shubik recommended a large crowd and a party in which spirits are high and the propensity to calculate does not settle in until at least two bids have been made.
His empirical claim is a single sentence from experience rather than a controlled experiment: once two bids have been obtained, the paradox of escalation is real, it is possible to sell a dollar bill for considerably more than a dollar, and a total of payments between three and five dollars is not uncommon. The rest of the paper is analysis. Viewed as a cooperative game the auctioneer is at a disadvantage, since the two bidders could agree that one bids 5 cents and the other abstains. Viewed noncooperatively, the disadvantage shifts to the bidders: an opening bid of a dollar that nobody contests is an equilibrium, and once the bidding has started the damage one bidder can do to the other for 5 cents has no limit short of an external one. Shubik located the key to the game in communication: bidders cannot make binding commitments, so the only signal available is the bid itself.
Read the diagram and evidence as text
An illustrative outward spiral alternates round and square auction paddles. Each paddle is tethered to a banknote bearing the same bidder mark, partly swallowed by a one-way slot above a locked receptacle: the earlier stake would be lost on quitting. The spiral exits at a balance whose combined-payment purse hangs lower than the unchanged prize banknote. Both bidder marks appear on the purse. The tilt shows ordering only; paddle repetitions, banknotes, purse size and tilt angle encode no amounts or round counts. Shubik described informal party play, not a controlled experiment.
Each bid answers the last
Shubik's report from party games, not a controlled experiment
Opening bids: A few cents for a chance at a dollar: each early bid looks like a bargain
Outbid: Quitting now means paying your last bid and getting nothing
One more bid: A small extra payment looks cheaper than the certain loss, so the loser raises
Both pay past the prize: The dollar sold for considerably more than a dollar; total payments between three and five dollars were not uncommon
Stop and accept the loss: The exit each bidder keeps refusing
Opening bids → the rival answers → Outbid
Outbid → the loss looks avoidable → One more bid (error transition)
One more bid → repeats, roles swapping, until both are losing → Both pay past the prize
| Condition | Value and unit | Evidence quote |
|---|---|---|
| A dollar bill at auction | considerably more than a dollar sale price, party games | a dollar bill could be sold for considerably more than a dollar |
| Both bidders together, not uncommon | between three and five dollars total payments, party games | total payments between three and five dollars not uncommon |
Allan Teger's 1980 book Too Much Invested to Quit reported a program of laboratory studies in which pairs of participants played the dollar auction and related games, with the bidding regularly passing the value of the prize. Teger's contribution was to track the shift in motives as the auction went on, from the economic to the interpersonal, and to connect it with Staw's work on escalation of commitment in organizations. Keith Murnighan's 2002 account of classroom auctions, in which bidding for a bill climbed far beyond its face value, has kept the game in use as a teaching demonstration, and Brockner and Rubin's 1985 book Entrapment in Escalating Conflicts placed it within a wider social psychology of being caught by one's own prior investment.
Everyday examples
The structure of the auction, in which stopping means losing what has been spent and continuing costs only a little more, is common outside parlor games. The rows are illustrations, not findings from the cited studies.
| Situation | What the bias does | What helps |
|---|---|---|
| Two buyers bid for a house at auction | Each raise is made to avoid losing the bidding already done, and the final price passes what either would have paid in a private sale | Fix a maximum in writing before the auction and leave when it is reached |
| A lawsuit over a modest claim | Legal costs on both sides climb past the amount in dispute because withdrawing would make the costs so far a pure loss | Compare the cost of the next stage with the best realistic outcome, ignoring what has been spent |
| Two companies cut prices to win a market | Each cut is a response to the rival's last, and both end up selling below cost | Decide in advance what share of the market is worth and at what margin, and stop there |
| A standoff between two governments over a border or a tariff | Each side escalates to avoid being seen to back down after the last move | Build in a route to stop that neither side has to call a defeat |
| An argument between partners that runs past midnight | Conceding now would mean the last hour was for nothing, so neither stops | Agree beforehand on a signal that ends arguments without a winner |
How to counter it
The trap closes after the second bid, so most of the counter happens before the first.
- Set a limit before you start. Decide the most the prize is worth to you, write it down, and treat reaching it as the end of the contest rather than as a point to reconsider.
- Ignore the money already in. What has been bid is spent whether you continue or stop. The only comparison that matters is the cost of the next step against what the next step can win.
- Notice when the motive changes. When you find yourself bidding to beat the other person or to avoid looking foolish rather than to get the prize, the game has changed and the prize no longer justifies it.
- Watch for the structure. Any contest in which the loser also pays, where exits are costly and each step is small, has the shape of the dollar auction. Recognizing the shape early is most of the defense.
- Do not expect the other side to stop first. Teger's bidders kept going partly because the opponent seemed about to quit. The opponent is thinking the same thing.
What it is not
Irrational escalation is the competitive form of a family of patterns in which prior investment drives further investment. The angle on this page is the rival: each bid answers another bidder's. The nearest relatives lack that feature or share only part of it.
- It is not escalation of commitment, which Staw studied in people defending their own earlier decisions with more money and time. There is no opponent in Staw's case; the person is escalating against their own past choice, and the driver is the wish for the first decision to have been right.
- It is not the sunk cost fallacy, which is a single decision made badly because money already spent is treated as recoverable. Sunk cost reasoning is what makes each bid in the dollar auction feel sensible; irrational escalation is the sequence of such bids locked together by a rival's responses.
- It is not loss aversion, the general finding that losses weigh more than equal gains. Loss aversion is why the certain loss of the last bid is so hard to accept, but it is a property of how outcomes are valued, not a pattern of contest.
Tests and tools that use this
Related biases
Irrational Escalation FAQ
What is irrational escalation?
It is the pattern in which parties to a contest keep raising their stakes because each new bid looks cheaper than accepting the loss of the last one, until the total committed exceeds the prize. The standard demonstration is Shubik's dollar auction, in which a dollar bill is sold to the highest bidder and the second-highest bidder also pays. Shubik reported in 1971 that total payments of three to five dollars for the bill were not uncommon.
Why does the dollar auction go past a dollar?
Because once one bidder has bid a dollar, the other faces a certain loss of their last bid or a bid of 1.05 that might recover it. Paying 5 cents more to avoid losing 80 or 95 cents looks sensible, and the same logic applies at every step after that. Both bidders are then losing money and bidding to lose less. Shubik noted that without an external limit there is no point at which this reasoning tells either of them to stop.
What did Teger find about people playing the dollar auction?
Allan Teger's 1980 book Too Much Invested to Quit reported laboratory studies in which pairs of participants played the dollar auction and bidding regularly passed the value of the prize. His main finding was a shift in motives as the game went on: early bids were about winning the dollar or limiting a loss, while later bids were about saving face, not letting the other person win, and justifying what had already been spent. The contest became about the opponent rather than the prize.
How does the demo on this page work?
You bid against a scripted opponent for a one-dollar bill in 5-cent steps, with the rule that the runner-up pays their bid and gets nothing. You can stop at any point. The opponent re-raises until the bidding reaches 1.50 dollars and then drops out. The result reports your final bid and your net gain or loss, and the outcome records only whether your bid went past a dollar. No money is involved, and the opponent's limit is fixed so that the demo always ends.
Is irrational escalation the same as the sunk cost fallacy?
They are related but not the same. The sunk cost fallacy is a single decision distorted by treating money already spent as if it could be recovered. Irrational escalation is a sequence of such decisions locked together by an opponent, so that each party's bid is a response to the other's. Sunk cost reasoning makes each bid feel sensible; the rival's reply makes the next bid necessary. Escalation of commitment, which Staw studied, is the version without an opponent.
How do I avoid getting trapped in an escalating contest?
Decide the most the prize is worth before the first bid and treat reaching that limit as the end. Compare the cost of the next step only with what the next step can win, ignoring what has already been spent. Notice when the motive shifts from the prize to the opponent or to face; that is the sign the contest has stopped being worth it. Recognize the structure, a contest where the loser also pays and each step is small, and do not count on the other side quitting first.
References
- Shubik, M. (1971). The Dollar Auction game: A paradox in noncooperative behavior and escalation. Journal of Conflict Resolution, 15(1), 109-111. doi.org
- Teger, A. I. (1980). Too much invested to quit. Pergamon Press. doi.org
- Brockner, J., & Rubin, J. Z. (1985). Entrapment in escalating conflicts: A social psychological analysis. Springer-Verlag. doi.org
- O'Neill, B. (1986). International escalation and the dollar auction. Journal of Conflict Resolution, 30(1), 33-50. doi.org
- Murnighan, J. K. (2002). A very extreme case of the dollar auction. Journal of Management Education, 26(1), 56-69. doi.org
- Staw, B. M. (1976). Knee-deep in the big muddy: A study of escalating commitment to a chosen course of action. Organizational Behavior and Human Performance, 16(1), 27-44. doi.org
Cite this source
Psychology.com. (2026, October 24). Irrational Escalation. Psychology.com. https://psychology.com/biases/irrational-escalation