The Knowledge in a Price

In 1945 Friedrich Hayek published a short paper asking a question that sounds like a technicality and is not: what is the economic problem society actually has to solve?

The answer everyone assumed was allocation. Given resources, given technology, given the wants of the population, work out the best arrangement. Put that way it is a very large sum, and the natural inference, drawn by clever people on all sides of the argument, is that it will be solved better by a central authority with good statistics than by millions of people bumping into each other.

Hayek's move was to deny the premise. The data required for that calculation, he argued, is never given to anybody in its totality. It exists as fragments: this machine is running warm, that supplier is unreliable in August, this customer will accept a substitute, there is a man in the next town who will do it for less. Much of it is local, most of it is perishable, and a great deal of it is tacit: the kind of knowledge a skilled person uses and cannot write down.

If that is the situation, the problem is not computation. It is transmission, and it cannot be solved by collecting the information centrally, because the information is not in a form that can be collected.

What a market does, on this account, is bypass the problem entirely. Hayek's illustration is tin. Suppose tin becomes scarcer: a mine floods, or someone invents a new use for it a continent away. What has to happen is that everyone who uses tin economises a little, and everyone who can supply it produces a little more. What does a user of tin need to know in order to do the right thing?

Nothing. They need to know that tin has become dearer. The cause is irrelevant to the correct response, which is why the price works: it carries the conclusion without the argument, in a form that fits in one number and can be acted on by someone who has never heard of the flooded mine. Hayek called the price system a mechanism for communicating information, and noted that its economy is the point; the marvel is how little each participant has to know.

This is an argument about knowledge, and it is worth separating from the arguments it usually travels with. It does not say prices are fair; a price reflects willingness to pay, which reflects the existing distribution of money, and a system that allocates by willingness will allocate badly whenever need and money come apart. It does not say prices are always right; they omit costs borne by third parties, they can be manipulated, and asset prices in particular can wander a long way from anything real for a long time. It says that a market solves a specific informational problem remarkably well. That claim is compatible with a long list of other failures, and the strongest versions of both the pro-market and anti-market cases tend to be built by ignoring the distinction.

The practical payoff is in understanding what happens when a price is prevented from moving. The good still has to be rationed (the scarcity does not care about the rule) and if the price is not doing the rationing, something else will. Queues. Waiting lists. Knowing someone. Quietly reduced quality at the same headline price. Rent control is the standard case, and the standard finding is that the controlled units end up held longest by the people with the longest tenure rather than the greatest need, while new supply stops appearing because the return has been capped. None of which establishes that caps are always wrong. Distributional aims are real, and sometimes worth paying for. It establishes that the rationing does not vanish; it changes hands, usually to a mechanism nobody designed and nobody is accountable for.

Which brings us to the uncomfortable case, because it is the one where both sides of the argument are correct.

After a hurricane, a shop raises the price of bottled water from one pound to eight. The economic case for permitting this is not that shops deserve profits. It is that at one pound the first arrivals buy forty cases and the shelf is empty within the hour, and that at eight pounds it is worth someone's while to load a lorry and drive through the night from two hundred miles away. The price is doing both jobs prices do: rationing the existing stock, and summoning new stock.

The objection is equally serious, and it is not sentimental. In a disaster nearly everybody's need is identical, and a price allocates by ability to pay. Efficiency arguments have nothing to say about that, because they take the distribution of money as given, and the distribution of money is exactly what is in question.

What is worth noticing is that the strongest response is neither slogan. If you want the supply signal and not the distribution, you can have both: ration by quantity (two cases each) or put cash in the hands of the affected, and leave the price to do its other job. The interesting policy is almost always the one that separates the two functions rather than the one that argues about whether prices are good.