A 500% error, then a 90% cut: the food market is losing its shared yardstick

Two counts of the same field

In the third week of August a bus of crop scouts drove across Illinois, stopped at roadside rows and pulled ears of corn apart by hand. They came out at 184 bushels an acre. The government’s figure for the same state, built from farmer surveys and models, was 212.

Both were made by serious people. Neither is obviously lying. Within two days corn crossed five dollars a bushel for the first time in a year and a half, priced off the scouts.

What a crop number is actually doing

It is easy to think a harvest estimate is a description of a field. It is not, or not mainly. It is what a grain elevator writes into a contract in September. It is what a bank writes into a loan in October, and what a miller in Egypt buys against in November.

None of those people can go and look. They are not buying accuracy. They are buying a common reference: one figure the person on the other side of the deal is also using. A price is an agreement, and an agreement needs something both sides can point at.

So the value of a public count is a strange one. It is not correctness. It is sameness. A shared number that runs eight bushels light still lets a thousand contracts close, because everyone is eight bushels light together. A perfectly accurate number that only you hold closes nothing.

Doubt is not the same as error

This is why doubt does something worse than a mistake. An error is a number in the wrong place. It gets revised, and the market moves on.

Doubt takes away the pointing. That is the real cost of a beef export figure overstated fivefold and then cut by ninety percent. Not the wrong number. The next number, and the one after that, now arrive carrying a question. Corn acreage revised far enough that the agency opened a review of itself does the same thing. Six senators writing to ask what staff losses had to do with it is not an accounting query. It is someone checking whether the floor everybody stands on is still there.

Who can afford to count for themselves

Here is the part that is not natural, although it looks it.

When a shared count is doubted, the obvious answer is to count yourself. That answer costs money. A trading house with satellites, agronomists and its own scouts can buy a private picture of Illinois. A farmer with sixty thousand bushels cannot. Neither can the elevator in the next town, and neither can the miller in Egypt.

A public count is therefore not merely convenient. It is the only count most people in the market will ever have. When it weakens, the information does not disappear. It moves behind a price only the large can pay. Everyone else keeps trading, but now against people who can see the field. None of that was announced or decided. It arrives as a run of small technical failures.

The repair now under way, moving the national crop survey off paper clipboards and onto satellite and machine data, is real and probably overdue. It also arrives after this harvest.

You are standing on several of these already

You have never weighed the flour you buy. You trust the number on the bag because an inspector somewhere defines a kilogram, and the shop does not.

You have never surveyed your own street. The line between your house and the one next door sits where a public register says it sits, not where your tape measure says.

Railways could not run to a timetable until towns gave up their own perfectly good clocks for one shared, arbitrary time. Every town’s clock was accurate. That was the problem.

Almost everything anyone relies on daily rests on a number somebody else counted and everybody agreed to use. The shopper cannot check it. Nor can the farmer, nor the trader who prices the grain. The arrangement holds because it is boring and because nobody has recently had a reason to look.

A shared number can be shared and wrong for a very long time. And the same fact that makes it useful, that everyone has to use it, is exactly what makes it worth someone’s while to capture.