The industrial policy number nobody sets a target for

Industrial policy in this country is measured in jobs created, dollars committed and facilities announced. Those are all counts of inputs. Not one of them tells you whether the thing being built can survive the wage bill of the country it is being built in. The number that answers that question is output per hour, and it is the one number no programme has ever been asked to hit.

I want to argue that this is the central omission, not a technical one. A plant in a high-wage economy is durable only if each hour worked in it produces enough to justify what that hour costs. If it does not, the plant survives on subsidy, and subsidy has a political half-life shorter than a factory's useful life. Every serious argument for making things domestically ends, sooner or later, at productivity. Almost no public argument for it starts there.

The number itself

Output per hour in the nonfarm business sector, indexed so that 2017 equals 100, reads 109.1 in 2020, 111.4 in 2021, 109.8 in 2022, 112.1 in 2023, 115.4 in 2024 and 118 in 2025. The 2025 figure means the average hour worked in that sector produced 18 percent more than the same hour in 2017.

The gain is real, it is recent, and it is unclaimed. There is exactly one decline in the window, from 2021 to 2022, when the index gave back 1.6 points; every observation after that is a climb, and the climb is concentrated at the end of the period rather than spread across it. I raise the shape only as a premise. My argument does not depend on which years did the work, and I have no interest in claiming that a curve nobody set out to move proves anything about the people who did not set out to move it.

Set that beside the industrial production index, which shares the same 2017 base and reads 95.1 in 2020 and 101.2 in 2025. The two series cover different populations, so this is a framing rather than a sum, and I am not going to divide one by the other. But the framing is worth holding in the mind. In a period when the volume of industrial output moved very little, the amount produced per hour worked across the business sector moved a great deal. Whatever has been happening in the American economy over these six years, it is not mainly a story about volume, and a policy apparatus organised entirely around volume will keep missing it.

The objection that productivity means fewer jobs

The strongest objection to making output per hour a target is that it can be hit the wrong way. Output per hour is a fraction, and the hours are underneath it. A firm that sheds a shift and keeps production flat improves the measure. A programme judged on productivity could therefore reward exactly what industrial policy was built to prevent: production that stays and employment that does not. For anyone whose interest in reshoring is an interest in jobs, that is not a quibble. It is the whole objection.

I think the objection is correct as stated and I do not want to talk anyone out of it. A single ratio used as a target will be gamed through its denominator; that is the ordinary fate of single ratios used as targets, in this field and every other. Anyone proposing output per hour as the sole test of an industrial programme is proposing something that will be met by cutting hours.

The answer is not to drop the measure but to refuse to let it travel alone. Employment counts and output per hour are already published side by side by the same statistical system; the failure is one of political habit, not of data. A programme that reports both is legible in a way that a programme reporting either one is not. Jobs alone tell you the plant is staffed. Output per hour alone tells you it is efficient. Only together do they tell you it will still be there in fifteen years, which is the only question a public capital commitment is actually asking.

What a stated target would change

The practical consequence of naming a productivity expectation is that it changes the argument at the moment money is allocated. A programme with no such expectation is free to treat any plant as equivalent to any other plant. A programme carrying one has to say what kind of production it is buying: whether the process being installed is at the technical frontier or a decade behind it, whether the equipment is the equipment a competitor would buy, whether the site will still make sense when the subsidy ends. Those are answerable questions at the point of decision, and they are almost unanswerable afterwards.

I am not arguing that the recent climb in output per hour vindicates any particular programme. This series cannot attribute anything to anything, and I would distrust anyone who claimed otherwise from six annual points. I am arguing something narrower and, I think, harder to dispute: the number that will decide whether domestic production survives contact with domestic wages is currently nobody's responsibility. It is measured, published and ignored. That is a strange way to spend public money on a twenty-year bet.