What Is Auction Theory? Designing the Rules So Bidding Reveals Value
By the BrainSnail editorial team. How these articles are written and checked, and how to tell us when one is wrong.
A seller who knew what each bidder was willing to pay would simply charge it. An auction exists because the seller does not know, and the rules of the auction determine whether bidders reveal what they know or conceal it. Getting those rules right is worth billions in spectrum sales and electricity markets, and getting them wrong has produced some expensive public failures.
The four basic formats
Almost every auction is a variant of four arrangements, and they behave differently despite selling the same thing:
- •The English or ascending auction, where the price rises until one bidder remains, which is familiar and lets bidders learn from each other as the bidding proceeds
- •The Dutch or descending auction, where the price falls until someone accepts, used for flowers and fish where speed matters because each lot takes seconds
- •The first-price sealed bid, where bids are submitted privately and the highest wins and pays their bid, which forces bidders to shade their bids below their true value and therefore to guess about others
- •The second-price sealed bid, or Vickrey auction, where the highest bid wins and pays the second highest, which has the remarkable property that bidding your true value is the best strategy regardless of what anyone else does
- •Hybrid and multi-unit formats, including simultaneous ascending auctions used for spectrum, where many related items are sold at once because their values depend on which others a bidder wins
- •Reserve prices and entry fees, which are not details but instruments that substantially change both revenue and who bothers to participate
The winner's curse
When bidders are competing for something whose value is the same for everyone but unknown, such as the oil under a tract of land or the revenue from a licence, the winner is by definition the bidder who estimated highest, which means the winner is systematically the one who overestimated most. Bidding your honest estimate therefore loses money on average, and the correct strategy is to shade the bid downward by an amount that increases with the number of competitors, because more bidders means the highest estimate is further above the truth. The phenomenon was identified by petroleum engineers analysing offshore leases in the 1970s, who found that returns on won tracts were persistently disappointing, and it has since been documented in construction contracting, corporate takeovers, sports transfers and publishing advances. Experienced bidders correct for it and inexperienced ones frequently do not, which is one reason auctions opened to new entrants can produce prices the winners regret.
Why design matters in practice
Theory became consequential when governments started auctioning public assets. Radio spectrum auctions, beginning in the United States in 1994 and designed with heavy input from economists, replaced administrative allocation and beauty contests with a mechanism that both raised revenue and put licences in the hands of those who valued them most. The design problems were real: bidders in early auctions used the trailing digits of their bids to signal which markets they wanted and to collude tacitly, which was fixed by rounding bids; some auctions allowed bidders to win a scattering of licences that were worthless without complementary ones, an exposure problem addressed by allowing package bids. The 2000 British third generation spectrum auction, designed by Ken Binmore and Paul Klemperer, raised an extraordinary sum by encouraging entry and preventing incumbents from carving up the market, while comparable auctions elsewhere raised far less because their rules allowed exactly that. Robert Wilson and Paul Milgrom received the 2020 Nobel prize for the underlying theory and for the formats they invented.
Where it goes wrong
The recurring failures are instructive. Collusion is the central threat, since a small number of repeat bidders who expect to meet again have every incentive to divide the market rather than compete, and auction rules can either make that easy or hard, with transparency about who bid what being helpful for accountability and helpful for enforcing a cartel. Entry is the other key variable, since an auction with two serious bidders raises little regardless of design, which is why encouraging weaker bidders to participate frequently matters more than the format. Complexity can defeat the purpose when bidders cannot compute a sensible strategy, and several elaborate formats have performed worse than simple ones. And revenue is not always the objective, since a government auctioning spectrum may prefer a market structure with more competitors over the highest immediate payment, which means the design question is what outcome is wanted before it is how to extract the most money.
The takeaway
Auction formats change behaviour even when selling the same thing: ascending auctions let bidders learn from each other, first-price sealed bids force strategic shading, and second-price auctions make honest bidding optimal. The winner's curse means the winner is the bidder who overestimated most, so bids must be shaded further as competitors increase. Spectrum auction design produced large revenue differences between countries, and collusion and lack of entry cause the main failures.