Plate v in print

The Informed

adverse selection in the spread: that the maker always loses to those who know more, and widens his quote to recoup from the rest — the part of the spread that prices his own ignorance.

Whoever quotes a price to all comers will be taken up most eagerly by those who know it is wrong; the spread is what the maker charges the uninformed in order to survive the informed.

— the principle of this plate, after the model of Glosten and Milgrom, 1985

The dealer who quotes a price to everyone faces a danger that no amount of diligence can remove: among those who take his quote will be some who trade precisely because they know something he does not — and against them he is bound to lose. The third component of the spread is the price of that danger.

The danger is a form of adverse selection, the hazard the volume on asymmetric information will pursue in full. The dealer posts a price to buy and a price to sell, good for all comers. But traders are not all alike: some trade for reasons unrelated to the asset's value — to invest savings, to raise cash, to rebalance a portfolio — and these the dealer is glad to face; but others trade because they have information he lacks, and know the asset to be worth more or less than he is quoting. The informed buy from him exactly when his ask is too low, and sell to him exactly when his bid is too high; whenever he trades with them, he is on the wrong side. He cannot tell, at the moment of trading, which kind he faces — the order looks the same — and so every trade carries the suspicion that the counterparty knows better.1

1. Adverse selection in market-making: the dealer quotes to all, but informed traders take his quote only when it is mistaken in their favor — buying when his ask is too low, selling when his bid is too high. He cannot distinguish informed from uninformed orders at the moment of trade, so each trade risks facing someone who knows more. See The Asymmetry.

Against a loss he cannot avoid on any single trade, the dealer protects himself in the only way open to him: by widening his spread, so that the margin he earns from the many uninformed traders covers the losses he suffers to the few informed ones. He sells a little dearer and buys a little cheaper than he otherwise would, taxing every trade to pay for the ones that go against him. The spread, in this component, is not a charge for any service but a charge for the dealer's own ignorance — a levy on all traders to fund his survival against those who know more. The uninformed, in effect, subsidize the dealer's losses to the informed, and the wider the asset's scope for private information, the larger that subsidy must be.2

2. The dealer widens his spread so the margin earned from uninformed traders offsets the losses to informed ones — taxing all trades to survive the few that go against him. This component prices the dealer's ignorance, not a service; the uninformed subsidize his losses to the informed, and the spread rises with the scope for private information.
⁂

There is a striking way to see this component: the spread is the rate at which the market learns. When an informed trader buys, he reveals, by the very act, that he thinks the price too low; the dealer, seeing the buy, revises his estimate upward and raises his quotes — and so the information the trader held leaks, through his trade, into the price. The spread is the dealer's defense during the interval before he has learned, the margin that keeps him solvent while the market digests what the informed know. In this light the bid-ask spread is the visible price of the market's imperfect information: it would vanish if everyone knew the same things, and it widens precisely as knowledge becomes more unequal. The spread measures, in money, the asymmetry of information in the market — a quantity the volume on asymmetry will meet again as the root of far more than dealers' quotes.3

3. The adverse-selection spread is the price of unequal information: informed trades move the dealer's quotes — information leaks into price through trading — and the spread is his margin while he learns. It would vanish under equal information and widens with informational inequality, so the spread measures, in money, the asymmetry of information. Cross-reference The Asymmetry.

The point was put memorably, before it was formalized, by a market practitioner writing under a borrowed name: the dealer, he said, always loses to the informed and must make it back from the liquidity-motivated, the uninformed who trade for reasons of their own; his spread is set so that the second group pays for the first, and if the informed ever come to dominate his order flow, he must widen his spread without limit, and the market fails. This is the dealer's predicament in its starkest form: he is the only game in town, obliged to quote to all, unable to refuse the informed, and able to survive only by charging everyone for the ignorance forced upon him. When private information runs too deep — as in a market where some traders reliably know what others cannot — the spread may widen until trade dries up altogether, the market killed by the very suspicion of the informed.4

4. The practitioner Jack Treynor, writing as "Walter Bagehot" (1971), framed the dealer as always losing to the informed and recouping from the uninformed; if informed order flow dominates, the spread must widen without bound and the market can fail. Extreme information asymmetry can shut a market entirely — a foretaste of Akerlof's market collapse in The Asymmetry.

With this component the volume's vindication of the middleman reaches its limit, and begins to turn. The price of immediacy and the price of inventory risk were payments for services rendered — the abolition of waiting, the warehousing of risk — and the middleman earned them. But the adverse-selection component is not a payment for a service; it is a transfer, levied on the uninformed to cover the dealer's losses to the informed, corresponding to nothing the dealer does for anyone. It is closer to the ancient charge's toll than to a service's fee — though a toll forced on the dealer by his ignorance, not one he chooses to exact. And it points toward the flagship's deeper unease: that once we look closely, the spread is not cleanly a price for a service at all, but a fused thing, in which what is earned for serving and what is merely transferred or extracted cannot be told apart in the single number the market shows.5

5. The turn: unlike immediacy and inventory risk, the adverse-selection component is a transfer — uninformed to dealer, covering losses to the informed — not a payment for a service; it resembles a toll more than a fee, though forced by ignorance rather than chosen. This prefigures the flagship's claim that earned and levied components fuse inseparably in the spread.

The third component of the spread, then, is the price of the dealer's ignorance: the margin he must charge all traders to survive the few who trade against him knowing more, a levy that prices, in money, the inequality of information in the market. With it, the clean picture of the spread as the fee for a service begins to blur, for here is a part of the spread that pays for nothing done, and that shades from fee toward toll. The flagship plate completes the turn. It returns to the market-maker at the center of all this — the one who quotes the two prices — and asks what he really is: parasite or maker, servant or sovereign; and it finds that the spread he charges, fusing the earned and the levied beyond any separating, reveals at the last that the single price we imagined a thing to have was never there at all.