Why the Auction Winner Often Overpays
The winner of a competitive bid is often the person who most overestimated the prize. Evidence from Alaskan oil leases to eBay shows when that happens, and how far experience helps.
Key takeaways
- In Bazerman and Samuelson's 1983 classroom experiment, jars worth $8 drew an average winning bid of $10.01, and the average winner lost $2.01.
- Kagel and Levin's 1986 experiments found profits with three or four bidders and losses with six or seven, because bidders grew more aggressive as rivals multiplied.
- Malmendier and Lee found in 2011 that 42 per cent of eBay auctions of one board game ended above the fixed price on the same page, a result Schneider argued in 2016 that search costs can explain.
In 1969 oil companies bid for drilling leases on Alaska's North Slope in a sealed-bid auction. Each company put a price on each tract without seeing what anyone else had offered. When the bids were opened, the winning offers added up to $900 million. The second-highest offers, from companies studying the same ground, added up to $370 million. On about a quarter of the tracts the winner had bid at least four times as much as the runner-up, and on more than three-quarters at least twice as much. Two years later, three engineers at Atlantic Richfield published a paper explaining why that gap should worry the winners.
Why did oil winners lose?
The engineers were E. C. Capen, R. V. Clapp and W. M. Campbell, and their 1971 paper in the Journal of Petroleum Technology has the whole idea in the first sentence of its abstract: "If it is true, as common sense tells us, that a lease winner tends to be the bidder who most overestimates reserves potential, it follows that the 'successful' bidders may not have been so successful after all." They pointed to the industry's record in the Gulf of Mexico, where, setting aside the era before 1950, they wrote that "the Gulf has paid off at something less than the local credit union."
The logic is simple once it is laid out. When a prize is worth roughly the same to everyone but nobody knows exactly how much, as with oil under a seabed, each bidder's estimate is a guess with some error in it. The average guess may be close to the truth. The highest guess is, by construction, one of the most optimistic, and the highest bidder wins. Winning is itself a piece of bad news about your estimate. Capen and his colleagues added that the effect was not confined to oil, writing that "the phenomenon we will be talking about pervades all competitive situations."
Winning is itself a piece of bad news about your estimate.
Economists call it the winner's curse. In 1988 Richard Thaler devoted one of his "Anomalies" columns in the Journal of Economic Perspectives to it, crediting the three engineers with its first discussion in the literature. The column also reported the Alaska figures above, with the caution that they did not prove anyone had behaved irrationally, only that they were consistent with the curse.
What fooled the coin bidders?
The cleanest demonstration came from a classroom. In 1983 Max Bazerman and William Samuelson ran 48 sealed-bid auctions among 419 MBA students at Boston University. Each lot was a jar of pennies, nickels or paper clips whose contents were valued at $8, which the students did not know. Asked to estimate the value, they guessed low, at $5.13 on average. Yet the average winning bid was $10.01, and the average winner lost $2.01. Two things widened the loss: more uncertainty about what the jar held, and more people bidding.
That second factor became the centre of a long run of laboratory work by John Kagel and Dan Levin. In their 1986 paper in the American Economic Review, bidders in small groups of three or four made money, about two-thirds of what theory predicted. In groups of six or seven they bid more aggressively and lost money where theory said they should profit. Theory says a bidder should bid more cautiously as the crowd grows, because winning against more rivals means having beaten more estimates. The bidders did the opposite.
The 1971 oil paper had already put the point as a question about auction size: "If one wins a tract against two or three others he may feel fine about his good fortune. But how should he feel if he won against 50 others?"
| Year | Setting | What the winners did |
|---|---|---|
| 1969 | Alaska North Slope oil leases | Winning bids totalled $900 million; second-highest bids $370 million |
| 1983 | Jars worth $8, Boston University MBA classes | Average estimate $5.13; average winning bid $10.01 |
| 1986 | Laboratory auctions with six or seven bidders | A loss of 88 cents per period, where theory predicted a $4.68 profit |
| 2011 | eBay listings of a board game | 42 per cent closed above the fixed price on the same page |
The evidence on experience is mixed. Kagel and Levin reported that their experienced bidders responded to strategic considerations, yet the curse still produced losses in large groups. In 1996 Douglas Dyer and Kagel reported that experienced construction executives also suffered from it in laboratory auctions, while doing well in their real trade. Their explanation was that the industry has its own mechanisms for escaping the curse, and that the executives' learned, trade-specific ways of judging a job do not travel to an unfamiliar setting.
Does eBay repeat the mistake?
Later work turned from information to emotion. In 1999 the fibreglass cows from Chicago's Cows on Parade public art exhibition were sold for charity, live and online, and prices ran to roughly seven times their estimates. Gillian Ku, Deepak Malhotra and J. Keith Murnighan drew on those sales in a 2005 paper proposing a competitive arousal model of what bidders call auction fever: rivalry, time pressure and an audience raise arousal, and arousal raises bids. A summary of the work for Harvard Business School's Working Knowledge reports that live bidding produced larger overbids than internet bidding, and that frequent bidders overbid no more than novices.
Online auctions gave economists a cleaner test. Ulrike Malmendier and Young Han Lee, in a 2011 paper titled "The Bidder's Curse", compared eBay prices for an item with the fixed buy-it-now price for the same item on the same page. In detailed data on a board game, 42 per cent of these auctions ended above the fixed price, and 48 per cent did across a broad cross-section of goods. A small fraction of bidders, 17 per cent, was enough to generate the overbidding. The authors' explanation was limited attention: bidders were not looking at the cheaper option beside them.
The finding has been challenged. In a 2016 comment in the same journal, Henry Schneider used their data and new eBay data and argued that a key condition for identifying nonstandard behaviour may not have been met. Once the cost of searching is counted, he concluded, the overbidding is consistent with standard behaviour.
What should you check first?
Richard Shotton's The Choice Factory brings the idea into advertising, late in the book. Bias 19 (pages 131 to 136) argues that the winner of any competitive auction tends to overpay, and that programmatic media, now bought through real-time bidding, has imported the problem. It traces the term to oil-lease bidding in Alaska and credits Thaler for the explanation. The more bidders chase the same broad audience, the book argues, the worse the curse. Its remedies are to bid where others are not, on unusual audiences and at unusual moments, to convert better than rivals, and to shave bids by around a tenth.
The research supports the direction of that advice and complicates the arithmetic. The curse is clearest when the prize is worth much the same to everyone and its value is uncertain, as with oil leases and jars of coins and clips. It grows with the number of rivals, and theory says the right discount grows with them, so a fixed tenth is likely to be too much in a thin market and too little in a crowded one. The eBay evidence is contested, and experience in a familiar setting appears to protect professionals. The curse is conditional, and largest for the bidder with the most rivals and the least information.
Most of the clean evidence comes from laboratories, where nothing is at stake beyond a few dollars, and the field cases, oil leases and eBay, have more than one explanation. That is why the check below is framed as questions rather than rules.
Before you bid in any auction, work through four. Is the prize worth much the same to every bidder, and is its value uncertain? If so, your estimate is the one most likely to be too high when you win. How many people are bidding? More rivals call for a lower bid. Is there a fixed price for the same thing nearby? The eBay data suggest some winners never looked. Did you write your limit down before the bidding started, and are rivalry or a countdown now pushing you past it? If you cannot answer all four questions with confidence, it may be wiser not to bid at all and to let the prize go to somebody else.
Common questions
What is the winner's curse?
It is the tendency for the winner of a competitive bid to overpay when the prize's value is uncertain, because the highest bid usually comes from the most optimistic estimate. Three Atlantic Richfield engineers, Capen, Clapp and Campbell, described it in 1971, pointing to poor returns on oil leases in the Gulf of Mexico.
Do experienced bidders avoid the winner's curse?
Partly. John Kagel and Dan Levin found in 1986 that experienced laboratory bidders still lost money in groups of six or seven. In 1996 Douglas Dyer and Kagel found that construction executives suffered the curse in the laboratory, while industry practices protected them in their real trade.
Do people overpay in online auctions?
Some do. Ulrike Malmendier and Young Han Lee found in 2011 that 42 per cent of eBay auctions of one board game ended above the fixed price on the same page. Henry Schneider argued in 2016 that search costs can explain this without assuming irrational bidding.
Sources
Capen, E. C., Clapp, R. V., & Campbell, W. M. (1971). “Competitive bidding in high-risk situations.” Journal of Petroleum Technology, 23(6), 641-653. doi:10.2118/2993-PA. Abstract and opening pages read (Crossref deposit): abstract sentence quoted; 'the Gulf has paid off at something less than the local credit union'; 'the phenomenon we will be talking about pervades all competitive situations'. Authors' affiliation Atlantic Richfield Co. The p. 645 quotation about winning against 50 others is as quoted in Thaler (1988), full text read.
Thaler, R. H. (1988). “Anomalies: the winner’s curse.” Journal of Economic Perspectives, 2(1), 191-202. doi:10.1257/jep.2.1.191. Full text read: credits Capen, Clapp and Campbell with first discussion; 1969 Alaska North Slope sale, winning bids $900 million against $370 million second-highest, winner at least 4 times second bid on 26 per cent of tracts and at least twice on 77 per cent; Kagel and Levin small groups 65.1 per cent of predicted profit, large groups loss of $0.88 per period against predicted $4.68 profit.
Bazerman, M. H., & Samuelson, W. F. (1983). “I won the auction but don’t want the prize.” Journal of Conflict Resolution, 27(4), 618-634. doi:10.1177/0022002783027004003. Full text read: 419 MBA students at Boston University, 48 sealed-bid auctions, jars worth $8.00; mean estimate $5.13; mean winning bid $10.01; average loss $2.01; uncertainty and number of bidders widen the curse.
Kagel, J. H., & Levin, D. (1986). “The winner’s curse and public information in common value auctions.” American Economic Review, 76(5), 894-920. Abstract read: experienced bidders respond to strategic considerations; 3 to 4 bidders fit Nash predictions, 6 to 7 bidders bid more aggressively, producing a winner's curse with negative profits.
Dyer, D., & Kagel, J. H. (1996). “Bidding in common value auctions: how the commercial construction industry corrects for the winner’s curse.” Management Science, 42(10), 1463-1475. doi:10.1287/mnsc.42.10.1463. Abstract read: experienced construction executives suffer from a winner's curse in laboratory markets; industry-specific mechanisms, learned evaluation processes and private value elements explain their field success.
Ku, G., Malhotra, D., & Murnighan, J. K. (2005). “Towards a competitive arousal model of decision-making: a study of auction fever in live and Internet auctions.” Organizational Behavior and Human Decision Processes, 96(2), 89-103. doi:10.1016/j.obhdp.2004.10.001. Abstract not reachable (publisher blocked). Model as described in Adam, M. T. P., Ku, G., & Lux, E. (2019), 'Auction fever: the unrecognized effects of incidental arousal', Journal of Experimental Social Psychology, 80, 52-58, doi:10.1016/j.jesp.2018.07.009, full text read; Cows on Parade prices roughly seven times estimates, live bidding larger overbids than internet, frequent bidders no more prone than novices, as reported in Malhotra and Ku's summary 'How to harness auction fever', Harvard Business School Working Knowledge, full text read; Chicago, 1999, and live and internet sales confirmed in Ku, G. (2000), Kellogg Journal of Organization Behavior, full text read.
Malmendier, U., & Lee, Y. H. (2011). “The bidder’s curse.” American Economic Review, 101(2), 749-787. doi:10.1257/aer.101.2.749. Abstract read: 42 per cent of board-game auctions exceed the simultaneous fixed price on the same page; 48 per cent in a broad cross-section; 17 per cent of bidders suffice to generate the overbidding; limited attention most consistent. Byline order on the AER page is Malmendier, then Lee.
Schneider, H. S. (2016). “The bidder’s curse: comment.” American Economic Review, 106(4), 1182-1194. doi:10.1257/aer.20120767. Abstract read: using the original and new eBay data, argues a key condition for identifying nonstandard behaviour may not have been met; overbidding consistent with standard behaviour once search costs are included.
Shotton, R. (2018). The Choice Factory: 25 behavioural biases that influence what we buy. Harriman House. Read in full by this publication, with a reliability audit. Bias 19, Winner's Curse, pp. 131 to 136 (oil-lease origin and Thaler credited pp. 131 to 132).