Winner’s Curse, explained.
The winner's curse arises in common-value auctions when winning selects an unusually high estimate, making the winner vulnerable to overpaying.
Why it happens
Several bidders estimate the same underlying value with noise. The largest estimate is disproportionately likely to contain a positive error. Winning is therefore information about your estimate, not just evidence that you found a bargain.
Common-value auctions expose bidders to selection: winning provides information about how optimistic their estimate was. Milgrom and Wilson's work studies auctions with such information problems.
Read the result
Increase estimation noise or the number of bidders and compare the winning bid with the shared value. One profitable win does not remove the selection effect; inspect repeated outcomes and the model's assumptions.
A worked example
Bidding for an uncertain lot
Three bidders estimate the same lot at 90, 100, and 120, while its actual value is 100.
A first-price winner paying 120 loses 20, despite having the most optimistic estimate.
The highest estimate's selection is what creates the problem; the numerical example is illustrative, not a universal adjustment rule.
OPTIONAL DEEPER DETAILGo deeper: inside the model
Inside this model
Each estimate is independently uniform between 100−e and 100+e. Every bidder submits max(0, estimate−reduction). The highest bid wins and pays its bid. The plot compares the running mean winning bid with the true value across 1,000 seeded auctions.
Where this idea is useful
A practical use
A company bidding for an uncertain project can win precisely because its cost estimate was the most optimistic.
A common misconception
“Winning means my estimate was the most accurate.”
Winning may instead mean your estimate had the largest upward error. Accuracy and optimism are different properties.
What this explanation leaves out
- These bidders use a naive shared rule, not equilibrium bidding. A common reduction can improve a winner's margin here without changing who wins; real competition and seller reserves complicate that trade-off.
Is winner's curse the same as paying too much for a private-value item?
No. In a common-value setting, bids provide noisy information about a shared unknown value. With private values, a bidder may legitimately value the same item more than others.
What does the fact that your estimate beat everyone else's tell you about possible estimation error?
Associated thinkers
Further reading
Explore the original research or the teaching reference behind this experiment.