Knowledge Engine

Why the winner of an auction often overpays

·no read data

The idea

Some auctions sell items that are worth roughly the same to anyone who wins them: an oil lease, a jar of coins, the rights to a book. Nobody knows the true value, so each bidder forms an estimate. Even if those estimates are right on average, they are scattered, and the highest bidder tends to be the one whose estimate was most optimistic. Winning therefore tells you something unwelcome: you were probably the one who overestimated. Economists call this setting a common-value auctioncommon-value auction An auction for something that is worth about the same to every bidder, though no bidder knows exactly how much., and the systematic loss the winner’s curse.

The name comes from three petroleum engineers, Capen, Clapp and Campbell, who wrote in the Journal of Petroleum Technology in 1971 that companies bidding for Gulf of Mexico leases were earning surprisingly low returns. As Richard Thaler later reported in his 1988 survey, they said the Gulf had paid off “at something less than the local credit union,” and that bids from serious competitors on the same tract often differed by a factor of five to ten.

A concrete version: Max Bazerman and William Samuelson auctioned jars of coins and paper clips, each worth $8.00, to MBA students at Boston University in 1983. Students’ average estimate of the contents was $5.13, well below the truth. Yet across 48 auctions the average winning bid was $10.01, so winners lost about $2 on average.

Why it might change a decision

The theory makes a specific prediction that people can be tested against. A rational bidder in a common-value auction should bid below their own estimate, and should shade the bid further when there are more rivals or when the item is harder to value, because winning against a large field is stronger evidence of having guessed too high. If people suffer from the curse, the observed pattern should be the reverse: bids that track raw estimates, and losses that grow as the number of bidders rises.

How solid is this?

The laboratory evidence is consistent across several research groups. Bazerman and Samuelson’s study used 419 students in 12 classes. They found the average winning bid rose with how uncertain the item was and, less strongly, with the number of bidders, while individual bids followed estimates and did not adjust much for either factor. That is the pattern the curse predicts.

John Kagel and Dan Levin, in the American Economic Review in 1986, ran sealed-bid common-value auctions in the laboratory, including with experienced subjects. With three or four bidders, a standard game-theoretic model (Nash equilibriumNash equilibrium A situation in a game where no player can do better by changing their own choice while everyone else keeps theirs.) described behaviour fairly well. With six or seven, subjects bid more aggressively, the opposite of what theory recommends, and winners made losses. Thaler’s summary gives the figures: small groups earned 65.1 percent of the profit theory predicted, while large groups lost $0.88 per auction against a predicted profit of $4.68. Kagel and Levin’s later summaries of this line of work report that subjects who had learned to avoid the curse with four bidders fell back into it with seven.

The field evidence is harder to read. Thaler cites a study of 1,223 Gulf of Mexico leases sold between 1954 and 1969 that found an average loss in present value at a 12.5 percent discount ratediscount rate The yearly rate used to shrink money received in the future to what it is worth today.. Another analysis, by Hendricks, Porter and Boudreau, found firms did profit at a 5 percent real discount rate, though 12 of 18 major bidders would have done better by bidding less. Whether oil companies were cursed thus depends partly on the discount rate assumed and on how oil prices moved afterwards. These two lease studies are reported here only through Thaler’s summary; their original papers were not checked for this post. Thaler also mentions book advances, baseball free agents and company takeovers, but he notes that in publishing, a few hugely profitable books could explain unearned advances without any curse.

The strongest competing explanations for apparent field losses are luck, skewed payoffs and hindsight: a portfolio of leases can lose money on average in one period because prices fell, not because of bidding errors. In the lab, where values are known exactly, those explanations do not apply, which is why the experiments carry most of the weight. The popular claim that winners of any auction overpay outruns the data: the curse is specific to items whose value is shared and uncertain, and it is clearest among inexperienced bidders facing many rivals.

Sources

  • Capen, E. C., Clapp, R. V., & Campbell, W. M. (1971). Competitive bidding in high-risk situations. Journal of Petroleum Technology, 23, 641-653. doi:10.2118/2993-PA
  • 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
  • Kagel, J. H., & Levin, D. (1986). The winner’s curse and public information in common value auctions. American Economic Review, 76(5), 894-920. IDEAS/RePEc record
  • Kagel, J. H., & Levin, D. (n.d.). Winner’s curse [Encyclopedia entry]. Ohio State University. asc.ohio-state.edu
  • Thaler, R. H. (1988). Anomalies: The winner’s curse. Journal of Economic Perspectives, 2(1), 191-202. doi:10.1257/jep.2.1.191

Check yourself

Answers are checked in your browser. Nothing is recorded or sent anywhere.

  1. Why does winning a common-value auction carry bad news for the winner?

  2. In Bazerman and Samuelson's jar experiment, each jar was worth $8.00. What happened?

  3. What did Kagel and Levin find when they increased the number of bidders from three or four to six or seven?

  4. Which statement about the field evidence for the winner's curse is most accurate, given this post?

Questions without answers

  1. Hiring a sought-after candidate, buying a house in a bidding war and acquiring a company are all partly common-value contests. In which of them would you expect the curse to be strongest, and what information would you want before judging?
  2. Laboratory subjects bid on jars of coins for small stakes. What features of real auctions, such as repeated play, expert staff or the chance to drop out, might make the curse weaker or stronger than in the lab?

Back to the archive