The Winner's Curse

Ten oil companies bid on the same offshore tract. Nobody knows what’s really down there, so each firm’s geologists hand in an estimate — some high, some low, scattered around the truth. The company that bids the most wins the lease. Now ask the uncomfortable question: which company was that? Not the one with the best information. The one whose estimate happened to land highest — the most optimistic guess in a room full of guesses. Winning didn’t reward good judgment. It selected for the biggest overestimate.

That is the winner’s curse: in a competitive auction for a prize of uncertain common value, the winner is systematically the bidder who most overestimated what it was worth, so the act of winning is itself evidence you paid too much. The model was named by three Atlantic Richfield engineers — Capen, Clapp and Campbell (1971) — who noticed their firm kept “winning” oil leases that then underperformed. The culprit wasn’t bad luck. It was a selection effect hiding inside the auction, and once you see it, you find it everywhere people compete to buy something whose true value nobody knows.

This course builds the idea rung by rung. It starts by separating the two kinds of value — common value (an oil field, a company, a spectrum licence, a free-agent athlete), where the curse bites, from private value (how much you personally enjoy a painting), where it doesn’t. It shows the engine: conditioning on winning changes the odds, so that E[value | you won] sits below E[value], and the maximum of many noisy estimates floats far above the truth — pure regression to the mean waiting to happen. It gives you the fix: bid shading, and the counter-intuitive rule that the more rivals you face, the less aggressively you should bid per signal. Then it tours the real arenas — the acquisition premium and post-merger underperformance in M&A, IPOs, spectrum and mineral-rights auctions, sports free agency, competitive hiring, online ad auctions, and housing bidding wars. And it ends where every honest model does: at its limits. The curse is a warning for the naive, not an iron law — sophisticated bidders in equilibrium already correct for it, private-value auctions are immune, well-designed mechanisms blunt it, and over-shading until you never win is its own kind of failure.