Markets Know More Than Stories Do

This issue has spent a hundred pages showing that prices are imper­fect. They are late. They depend on size and unit and venue. They can be con­structed by com­mit­tees, frozen by bugs, can­celled by exchanges, and quoted long after they stopped meaning any­thing. All of that is true. None of it touches the strong­est reason we use prices at all.

In 1945 Friedrich Hayek pointed out that the knowl­edge needed to run an economy does not exist in any one place. It is scat­tered among mil­lions of people as knowl­edge of par­tic­u­lar cir­cum­stances: that this ware­house is nearly empty, that this harvest looks poor, that this ship will be late, that this buyer is about to walk away. Nobody can collect it. Much of it cannot even be written down. A price is how it gets used anyway. When a farmer who has seen the rain holds back grain, the price rises for a baker who has not, and the baker economises without ever learn­ing why.

The alter­na­tives to prices are not perfect numbers. They are stories and com­mit­tees. A story is a single person's account of what matters, however well informed, and it is usually wrong about some­thing it did not know to look at. A com­mit­tee aggre­gates more views, slowly, and fil­tered through whoever chairs it. The twen­ti­eth century ran a large exper­i­ment in replac­ing prices with plans, and the plan­ners' con­sis­tent com­plaint was that they could not find out what any­thing was really worth.

Markets are not wise and traders are not espe­cially clever. The claim is nar­rower and harder to escape: that people with money at stake, acting on what they each know, produce a number that incor­po­rates more of what is known than any alter­na­tive method we have found. It is why fore­cast­ers who beat market prices con­sis­tently are so rare, and why pre­dic­tion markets have often matched or beaten expert panels.

The flaws this issue describes are, mostly, flaws that markets reveal about them­selves. The dis­per­sion across five venues was nine basis points and cor­rected con­tin­u­ously by arbi­trage. LIBOR failed because it was not a market price, and was replaced by one. The nickel episode was, among other things, a market telling every­body with enor­mous force that a very large short posi­tion could not be deliv­ered. A stale last trade is a problem pre­cisely because the market that would have updated it is absent.

It is fair to insist that every price comes with its def­i­ni­tion, its age and its size. It is a mistake to move from there to the idea that prices are just one story among many. Stories are cheap to tell. Prices are expen­sive to move, because moving them costs the mover money if they are wrong. That cost is the reason the number deserves more trust than the nar­ra­tive around it.

Nor does the imper­fec­tion of prices rec­om­mend any par­tic­u­lar replace­ment. The honest alter­na­tive to a noisy market price is not a clean number. It is a number with the noise hidden: an appraisal that does not move because nobody has asked again, a model mark that is smooth because the model is. A price that jumps around on a bad day is doing its job. A val­u­a­tion that stays calm on the same day is not being wise. It is being late.

Prices are imper­fect, con­tex­tual and con­structed. They still know more than we do.