The labour-force series do not record who arrived, and participation is not a substitute

My position is that the contribution of migration to productive capacity cannot be estimated from the labour-force data this brief holds, and that the right response is to say so rather than to approximate it. No file on this site records a worker's place of birth, date of arrival or immigration status. A search of the generated data files for a nativity, citizenship or visa field returns nothing. The attribution question is not difficult here. The variable is absent.

An absence of that kind is not a gap to be closed by inference. A labour-force series counts people in work, out of work or looking; it does not classify them. The national series used here are participation, prime-age participation, payroll employment, job openings, the quits rate and the U-6 measure, and the state file adds a civilian labour force count for fifty-one jurisdictions. Every one of them is a count or a rate. Not one is a composition.

The proxy measures age

The series usually offered as a stand-in is the headline participation rate. It fell from 62.6 percent of the civilian population in 2023 to 62.4 percent in 2025, a fall of 0.2 points, and that small fall does a great deal of work in arguments about labour supply. Prime-age participation moved the other way over the same two years, from 83.3 to 83.6 percent of the 25-54 population, a rise of 0.3 points, and 83.6 is the highest reading in its six-year window, held in 2024 as well.

They disagree because they are computed over different populations. One divides by the whole civilian population. The other divides by the 25-54 band alone. A denominator holding the whole civilian population moves as that population ages, with nobody arriving and nobody leaving. Assigning the headline movement to migration assigns to arrivals a change the denominator can produce on its own.

Set the two windows side by side and the point is harder to avoid. Across 2020 to 2025 the headline rate ran 61.7, 61.7, 62.2, 62.6, 62.6 and 62.4 percent, a total range of 0.9 points. Prime-age participation ran 81.4, 81.6, 82.5, 83.3, 83.6 and 83.6 percent, a range of 2.2 points, which is 2.44 times as wide. The series quoted in the argument is the one with less movement in it and the broader population behind it.

One month of a state count is worse still

State figures get pressed into the same service and they carry less. The state-index file holds a civilian labour force count for all fifty-one jurisdictions dated 1 July 2026, with the previous month, 1 June 2026, beside it. Between those two dates the count fell in 39 jurisdictions and rose in 12. California's fell by 70,227, from 19,620,270 to 19,550,043. Texas's rose by 10,866, from 15,904,759 to 15,915,625.

Two observations are the whole state record. Nothing sits behind them, so nothing built on them is a trend. And a month-on-month change in a labour force count cannot separate a person who arrived from one who entered the labour force, left it, moved between states, or was counted differently between the two months. Thirty-nine jurisdictions moving the same way in a single month is a fact about a month.

The answerable question is being displaced

What the pursuit of the unanswerable question costs is the answerable one, which these same files do settle: whether the economy over this window was short of workers. Payroll employment rose from 152.5 million jobs in 2022 to 158.4 million in 2025, an increase of 5.9 million, and from 142.2 million in 2020, an increase of 16.2 million. Those workers were found.

Demand for them points one way across the same years. Job openings averaged 11.2 million a month in 2022 and 7.1 million a month in 2025, 4.1 million fewer. The quits rate fell from 2.8 percent of employment to 2.0 percent. Unemployment and underemployment together, on the U-6 measure, rose from 6.9 percent of the labour force in 2023 to 8.0 percent in 2025. A shortage of workers does not usually present as fewer vacancies, fewer resignations and more underemployment at once.

That is a real finding and it is available to anyone. It says nothing whatever about who took the jobs behind that increase, and I am not going to write as though it did. What it does say is that across these years these series do not describe an economy short of people willing to work, which is the proposition the migration argument is usually deployed to settle.

What to do with the numbers instead

Three things follow. When a figure attributing capacity to migration cites a labour-force series, ask which series, because the attribution is in none of them. When the attribution genuinely bears on a decision, it needs a source carrying a nativity field, and no such source is held here; naming the absence costs less than approximating around it. And the labour-force series keep their value for the question they do answer, which is whether workers were available, not where those workers were born.

This Commentary carries no chart, and the omission is deliberate. The line that would be plotted is the participation rate, and putting it on the page is the error the piece is about. A chart here would be read as the migration signal within a paragraph of the sentence denying that it is one.

The honest sentence is short. The labour-force data on this site records how many people are working, looking and quitting, and records it well. It does not record where any of them came from, and no arithmetic performed on it will recover that. I would rather publish the limit than an estimate built on a denominator.