Strategic Economics

The Bid-ask Spread as the Price of Adverse Selection

A spread is not a fee, a margin, or a convention. It is the exact amount a quoter must charge everyone in order to survive the minority who know more — and it stays strictly positive even when competition has driven profit to zero. Once you can compute it, you can predict which markets have wide spreads before you look at one.

  • Expert
  • 14 min total
  • 13 chapters

What decision this helps you make: Whether the gap between the price you will buy at and the price you will sell at is wide enough to survive the informed side of your own market — and whether the cheaper fix is to widen it or to spend money making yourself less ignorant.

What this topic is

The bid-ask spread is the gap between the price you can sell something for right now and the price you would have to pay to buy it right now. It is the round-trip cost of impatience, and everybody who has ever exchanged currency at an airport has paid one. The interesting question is where it comes from. The obvious answers — the dealer's costs, the dealer's profit — turn out to explain only part of it. The deep answer is that whoever quotes a two-sided price is offering to trade with anyone, including the people who know the price is wrong, and the spread is exactly what it costs to keep doing that.

Why it matters

Glosten and Milgrom proved in 1985 that a spread exists even when quoters are perfectly competitive and earn zero expected profit, because the quoter loses on every trade against an informed counterparty and must recover it from the uninformed ones. That single result reframes an enormous amount of business. It says the spread on any asset is roughly the probability that your counterparty knows something times the size of what they know — which tells you why small-cap spreads are wide, why quotes vanish before an earnings release, why insurers demand a medical, and why an instant-offer business that ignores the composition of who accepts its offers loses money in exactly one category and cannot work out why.

Who should learn it

Anyone who quotes a two-sided price — market makers, dealers, instant-buyout and trade-in services, secondary marketplaces, insurers, lenders — plus investors sizing the real cost of trading, and readers who want the operating version of the lemons problem rather than the diagram.

What you will understand

  • The Glosten-Milgrom result: why a spread is strictly positive even at zero profit, and the one-line formula it produces
  • How to compute a quote as a conditional expectation, and why every trade is a Bayesian update rather than an event
  • The three components of a real spread — order processing, inventory, adverse selection — and how to separate them
  • When widening the spread is the wrong answer and buying information is the cheaper one

Prerequisites

Common misconception

"The spread is what the dealer makes." In the model worked through in this lesson, the dealer makes exactly zero on average — and the spread is six dollars on a fifty-dollar asset. The spread is not profit; it is the transfer that flows from uninformed traders to informed ones, collected and passed through by whoever stands in the middle. That is why competition does not eliminate it. Ten dealers competing to quote the same asset all face the same informed counterparties, so competition drives the spread down to the adverse-selection cost and no further. If you want a narrower spread you have to make the market less asymmetric, not more competitive.