Auction theory is a branch of economics that examines how auctions allocate goods, services, or rights through competitive bidding. It uses game theory to analyze bidders’ strategic choices and mechanism design to construct allocation and payment rules. Its central questions concern who wins, how much participants pay, what information bidding reveals, and how outcomes change under alternative rules. Applications include procurement, securities, mineral rights, and radio-frequency licenses. (nobelprize.org)
Valuations, information, and objectives
An auction model specifies the objects offered, eligible participants, available information, permissible bids, and rules determining allocation and payments. Information asymmetry is fundamental: bidders generally know something about their willingness to pay that the seller and competing bidders do not. In a basic single-item model, a winning bidder’s payoff equals their valuation minus their payment; a losing bidder pays nothing and receives zero payoff. (nobelprize.org)
In a private-value auction, each bidder knows their own valuation, which does not depend on competitors’ information. Private values need not be statistically independent; independence is an additional modeling assumption. In a common-value auction, the object has the same underlying value to all bidders, but participants possess different estimates. Mineral deposits provide a standard example. More general interdependent-value models combine individual preferences with information held by others. (nobelprize.org)
Design objectives also differ. Allocative efficiency means assigning resources to participants who value them most, subject to relevant constraints. Revenue maximization instead seeks the highest expected receipts for the seller. These objectives can conflict: withholding an item unless bidding exceeds a minimum price may increase expected revenue while preventing some mutually beneficial transactions. (nobelprize.org)
Principal auction formats
Four single-item formats are especially important:
- English auction: bids or an announced price rise until one bidder remains.
- Dutch auction: an announced price falls until a bidder accepts it.
- First-price sealed-bid auction: participants submit confidential bids; the highest bidder wins and pays their own bid.
- Second-price sealed-bid auction, or Vickrey auction: the highest bidder wins but pays the second-highest bid. (theory.stanford.edu)
Under standard private-value assumptions, the Dutch and first-price formats present equivalent strategic choices: a bidder chooses the price at which to buy. An idealized ascending English auction and a second-price auction produce the same allocation and payment when bidders remain active until the price reaches their valuations. With common or interdependent values, however, observing others’ behavior during an open auction can convey information unavailable in sealed bidding. (nobelprize.org)
Strategic bidding and equilibrium
In a single-item second-price auction with private values and the standard payoff structure, bidding one’s valuation is a weakly dominant strategy. Overbidding can cause a purchase at a price above the bidder’s valuation; underbidding can sacrifice a profitable purchase. Truthful bidding therefore implements incentive compatibility without requiring accurate predictions of competitors’ bids. (theory.stanford.edu)
First-price auctions instead generally induce bid shading: bidding below valuation balances a larger surplus upon winning against a lower probability of winning. They are often modeled as a Bayesian game, in which strategies depend on private information and beliefs about competitors. The relevant equilibrium is a Bayesian Nash equilibrium: each bidder’s strategy maximizes expected payoff given the others’ strategies. (nobelprize.org)
For example, with (n\geq2) risk-neutral bidders whose valuations are independently and uniformly distributed on ([0,1]), and no reserve price, a symmetric first-price equilibrium is
[ b(v)=\frac{n-1}{n}v. ]
With two bidders, each bids half their valuation. As the number of bidders increases, equilibrium bids approach valuations. This formula depends on the specified assumptions rather than describing first-price auctions universally. (theory.stanford.edu)
Revenue equivalence and optimal auctions
The revenue equivalence theorem explains why different payment rules can yield identical expected seller revenue. In the standard benchmark with risk-neutral, symmetric bidders and independent private values, the four principal formats allocate the object efficiently and generate the same expected revenue. Equality concerns averages across possible valuations, not necessarily payments in an individual auction. Changing information structures, risk preferences, or bidder symmetry can change the comparison. (nobelprize.org)
Optimal-auction analysis goes beyond comparing familiar formats. Roger Myerson’s 1981 framework characterizes revenue-maximizing mechanisms using virtual valuations, transformations incorporating bidders’ valuation distributions. For independent, identically distributed private values satisfying a regularity condition, the optimal single-item mechanism is a second-price auction with an appropriately chosen reserve price. The item remains unsold if no qualifying bid reaches that threshold. (theory.stanford.edu)
The winner’s curse
In common-value settings, winning conveys information: the winner may have made the most optimistic estimate. Failing to account for this selection effect can produce the winner’s curse, in which the successful bidder overpays relative to the object’s underlying value. Rational bidding incorporates the information implied by winning; losses are not an unavoidable feature of common-value auctions. (nobelprize.org)
Robert Wilson developed foundational common-value models. Paul Milgrom extended analysis to environments combining private and common components, clarifying conditions under which information disclosure and learning during bidding affect expected revenue. Their contributions received the 2020 Nobel Memorial Prize in Economic Sciences. (nobelprize.org)
Multiple objects and practical design
With multiple objects, values may depend on combinations: adjoining spectrum licenses can be worth more together than separately. A combinatorial auction permits bids on bundles, but selecting winners can require substantial mathematical optimization and raise questions of computational complexity. (theory.stanford.edu)
Milgrom and Wilson contributed to the simultaneous multiple-round format first used for United States spectrum sales in 1994. Related licenses are offered concurrently over successive rounds, allowing bidders to adjust their choices as prices evolve. Multi-object design must accommodate complementarities, substitutes, and interactions among participants that simpler single-item models omit. (nobelprize.org)