The order book everyone quotes is measured in dollars that changed size

New orders for durable goods is the closest thing American statistics has to an order book, which is why it gets quoted whenever anyone wants to argue that industrial demand is turning. It is a forward-looking series in a field crowded with backward-looking ones: an order is a commitment to buy something that has not yet been made. I use it. I also think it is one of the most casually misread figures in circulation, and that the misreading runs in a consistent direction.

The readings are 259.3 in 2021, 282.2 in 2022, 292.7 in 2023, 287.8 in 2024 and 311.2 in 2025, in billions of dollars. Four observations rise, one falls, and the last is the highest in the window. My argument is that almost nothing a general reader concludes from that sentence is safe, for two reasons that have nothing to do with the economy and everything to do with the units.

It is a monthly rate

The first is mechanical. The series is stated as a monthly rate. The 311.2 billion dollars attached to 2025 is not the value of a year's orders; it is the average month within that year. A year's worth of durable goods orders is a much larger number, and this series does not print it anywhere. Anyone who reads the figure as an annual total has silently shrunk the American order book by an order of magnitude, and that particular error passes unnoticed precisely because the result still sounds like a plausible national statistic.

This sounds pedantic until you watch the figure travel. A number lifted into a speech, a briefing note or a headline arrives stripped of its unit line, and once the unit is gone nobody can reconstruct it from the digits. The unit is not a footnote to the number. In a series like this one, the unit is half the information.

It is measured in current dollars

The second reason is the one that actually matters for policy. These are current dollars. They are not adjusted for what happened to prices over the same five years, and something did happen. The producer price index for all commodities, on a 1982 base of 100, read 227.3 in 2021 and 260.3 in 2025. The dollars in which the order book is denominated did not hold still while the order book was being filled.

Some portion of the rise from 259.3 to 311.2 is therefore price rather than quantity. I cannot tell you what portion, and I want to be exact about why. I do not have a constant-dollar version of this series, and the all-commodities producer index is the wrong instrument for deflating durable goods orders in any case. Dividing one by the other would produce a number that looked authoritative and meant nothing. So I am not going to produce it.

That is also why there is no chart on this piece. A line running from 259.3 up to 311.2 is visually a claim about industrial demand recovering, and a reader is entitled to take a chart as evidence for the claim the surrounding text is making. This chart would not be evidence for that claim. It would be a picture of a nominal aggregate wearing the costume of a real one. The honest thing is to print the readings, name what they are, and leave the reader's eye out of it.

The objection from the shop floor

The strongest objection is that this is an academic's complaint about a practitioner's number. A firm does not book a real order; it books a dollar order. Nominal orders are what fills a schedule, what a supplier plans against, what a bank looks at when it prices a working capital facility. Prices are not a distortion of business, they are part of it, and a series that captures what firms actually committed to pay is describing the world as the people in it experience it.

That is entirely correct at the level of the firm, and I would not argue with a treasurer about it. It is not correct at the level of the claim being made when this number is quoted in a national argument. When someone says the order book shows industrial demand returning, they are making a claim about things being built: about machines, vehicles, turbines and tooling actually being produced in greater quantity. That is a claim about volume, and a nominal series cannot carry it. If the claim is instead that industrial revenue has risen, that is a real and useful thing to say, and it should be said in those words.

There is one feature of the series that survives all of this intact, and it happens to be the most interesting one. Orders fell in 2024, from 292.7 to 287.8. A decline in a nominal series over a period when the price level was not falling is a harder fact than any of the increases, because the two possible explanations, fewer orders or cheaper ones, both point somewhere worth looking. That single down year is the part of this record I would want explained, and it is the part that never gets quoted.

None of this is an argument for ignoring the order book. It is an argument for quoting it with its unit line attached and its currency named, and for declining to illustrate it when the illustration would say more than the data does. A brief that will not draw the chart it cannot defend is a brief whose charts mean something.