Skip to content

Common-Value Auction


Definition

In a common-value auction the object has the same value vv for every bidder, but no bidder knows vv. Each bidder ii privately observes a signal θi\theta_i that is informative about vv, and her payoff from winning at price pp is v−pv-p. Bidder ii's expected value therefore depends on the other bidders' signals as well as on her own.

A two-bidder example with this structure has v=θ1+θ2v=\theta_1+\theta_2, where bidder ii observes only her own component θi\theta_i, and θ1,θ2\theta_1,\theta_2 are independent.


Interpretation

The contrast with a private-value auction is about information, not about who values the object more. With private values, learning a rival's signal would not change one's own valuation; with a common value it would. Classic examples are drilling rights to an oil field of unknown size, or a jar of coins whose total is estimated by eye.

In the additive example below, the equilibrium bids are strictly increasing, so a bidder wins when her signal exceeds the rival's. Because the common value increases with the rival's signal, winning lowers its conditional expected value for interior own signals. Common values make the rival's information relevant; the direction of this update depends on the signal model. A rational bidder therefore bids on the basis of the conditional expectation of vv given her signal and given that she wins (or, in a second-price auction, given that she is just tied). Ignoring this produces the winner's curse.

Truthful bidding results for private values do not carry over directly. In a second-price auction the case argument for bidding "one's value" fails, because one's value is not known; the equilibrium bid must instead be computed as a Bayesian Nash equilibrium.


Key Results

  • In the additive uniform example below, for 0<θi<10<\theta_i<1 and increasing equilibrium bids, E[v∣θi, i wins]=3θi/2<θi+1/2=E[v∣θi]\mathbb{E}[v\mid\theta_i,\ i\text{ wins}]=3\theta_i/2<\theta_i+1/2=\mathbb{E}[v\mid\theta_i]. The conditioning event has positive probability in this range.
  • For v=θ1+θ2v=\theta_1+\theta_2 with θi\theta_i i.i.d. Unif⁡(0,1)\operatorname{Unif}(0,1): the symmetric BNE of the second-price auction is b(θ)=2θ=E[v∣θ1=θ2=θ]b(\theta)=2\theta=\mathbb{E}[v\mid\theta_1=\theta_2=\theta], and that of the first-price auction is b(θ)=θb(\theta)=\theta.
  • In that example both formats give expected revenue 23\frac{2}{3}.
  • The Dutch–first-price equivalence holds with common values; the English–second-price equivalence can fail with three or more bidders, because drop-out prices reveal signals.

Examples

Take v=θ1+θ2v=\theta_1+\theta_2 with θ1,θ2\theta_1,\theta_2 independent and uniform on [0,1][0,1].

Second-price. Let bidder 2 bid 2θ22\theta_2. Bidder 1 with signal xx who bids b≤2b\le2 wins when θ2<b/2\theta_2<b/2 and pays 2θ22\theta_2, so her expected payoff E[u]\mathbb{E}[u] (written uu because vv is the common value) is

E[u]=∫0b/2(x+y−2y) dy=∫0b/2(x−y) dy.\mathbb{E}[u]=\int_0^{b/2}\big(x+y-2y\big)\,\mathrm{d}y =\int_0^{b/2}(x-y)\,\mathrm{d}y.

The integrand is positive for y<xy<x and negative for y>xy>x, so the payoff is maximized at b/2=xb/2=x, i.e. b=2xb=2x. Revenue is 2 E[min⁡(θ1,θ2)]=2⋅13=232\,\mathbb{E}[\min(\theta_1,\theta_2)]=2\cdot\frac{1}{3}=\frac{2}{3}.

First-price. Let bidder 2 bid θ2\theta_2. Bidder 1 with signal xx bidding b≤1b\le1 wins when θ2<b\theta_2<b, with expected payoff

E[u]=∫0b(x+y−b) dy=xb−12b2,\mathbb{E}[u]=\int_0^{b}(x+y-b)\,\mathrm{d}y=xb-\tfrac{1}{2}b^2,

maximized at b=xb=x. Revenue is E[max⁡(θ1,θ2)]=23\mathbb{E}[\max(\theta_1,\theta_2)]=\frac{2}{3}.

Numbers. With x=0.6x=0.6, E[v∣θ1=0.6]=1.1\mathbb{E}[v\mid\theta_1=0.6]=1.1, but E[v∣θ1=0.6, θ2<0.6]=0.6+0.3=0.9\mathbb{E}[v\mid\theta_1=0.6,\ \theta_2<0.6]=0.6+0.3=0.9. The first-price bid 0.60.6 leaves an expected margin of 0.30.3 when winning.

Value 0.6 plus theta_2 plotted against bidder 2's signal for a bidder with signal 0.6; the winning region theta_2 below 0.6 is shaded and the levels 1.1, 0.9 and the bid 0.6 are marked.

The figure fixes bidder 1's signal at x=0.6x=0.6, so the value is 0.6+θ20.6+\theta_2. Averaged over all θ2\theta_2 it is 1.11.1, but in the first-price equilibrium bidder 1 wins only in the shaded region θ2<0.6\theta_2<0.6, where the average value is 0.90.9. Against the rival strategy b2=θ2b_2=\theta_2, bidding 1.11.1 wins for sure and pays exactly the unconditional expected value, so its expected profit is 00. This is worse than the equilibrium bid 0.60.6, whose winning probability 0.60.6 and conditional margin 0.30.3 give expected profit 0.180.18. If both bidders instead naively bid θi+0.5\theta_i+0.5, winning with signal 0.60.6 means θ2<0.6\theta_2<0.6, so paying 1.11.1 gives conditional expected profit 0.9−1.1=−0.20.9-1.1=-0.2, a winner's curse.


Prerequisites



Sources

Aliases: Common Value Auction, pure common-value auction, common-values auction Keywords: private signal, interdependent values, wallet game, v = theta_1 + theta_2, conditioning on winning Symbols: integral, theta

Last updated: 27 October 2026 11:52