About Those Manufacturing Employment Numbers…

What the new jobs data can tell us, and what it can’t.

Last month, the Wall Street Journal published a story straightforwardly titled “U.S. Manufacturing Is in Retreat and Trump’s Tariffs Aren’t Helping.” Exhibit A, meant to indicate as self-evident that tariffs are killing U.S. manufacturing and thus failing in their stated purpose: the fact that American “[m]anufacturers shed workers in each of the eight months after Trump unveiled ‘Liberation Day’ tariffs.”

The piece is a classic of the genre. News outlets, elected officials, think tanks, and other assorted members of the D.C. “policy community” keep asking the same question: What about these manufacturing employment numbers? Don’t they mean tariffs aren’t working?

The numbers are true enough. U.S. manufacturing lost 93,000 jobs between March 2025 and February 2026. It is also true that a decline in manufacturing jobs certainly can indicate something seriously amiss. For example, after the United States normalized trade relations with China in 2000, the American manufacturing sector shed three million jobs in three years.

This did indeed portend a sector in crisis. Manufacturing output, measured appropriately, stagnated, a fact that people living in the affected communities did not miss even if economists did. Eventually, manufacturing productivity flatlined and began a long decline, meaning America got worse at making things efficiently. Fair enough, then, to ask whether manufacturing job loss suggests something wrong with American economic policy now as well.

The problem with the argument the WSJ and others are making, however, is that it misunderstands the logic of reindustrialization. The rebuilding of an atrophied American economy is a slow process that must proceed in stages; tariffs are a vital part of a toolkit intended to prompt that slow, staged process. The real question is whether this logic holds in the face of reality. A reasonable evaluator, rather than a partisan polemicist, would therefore ask several questions.

For example: How was the sector performing before these tariffs? How quickly should we expect manufacturing jobs to increase after one year, if this were working? Are there any signs that indicate something positive happening in manufacturing? The answers (which are, in order: even worse; not at all; yes) suggest that the state of manufacturing employment does not undermine the logic of reindustrialization at all. Working through each question in turn helps clarify why.

First up: past jobs numbers. The argument that declining employment after tariffs demonstrates harm to American industry rests on the implied premise that manufacturing jobs were in a better state before. This was manifestly not the case.

As explained by analyst Alan Tonelson, the Bureau of Labor Statistic (BLS)’s annual benchmark revision, released with its January 2026 jobs report, contains a telling story. American manufacturing lost 81,000 jobs in the first 11 months of Trump’s second term—but lost 179,000 jobs in the preceding 11 months. Breitbart economics editor John Carney, in a direct rebuttal to the WSJ’s February story (helpfully titled “Why the Wall Street Journal Missed the Trump Manufacturing Boom”), explains the point: “Manufacturing employment losses during the Trump tariff period were less than half those recorded during the comparable pre-tariff Biden period. The trajectory improved, not deteriorated.”

This improvement happened despite countervailing pressures from the Trump administration’s severe immigration enforcement efforts, which would reasonably be expected to worsen the manufacturing employment situation. Estimates of how much of the unauthorized labor force works in manufacturing vary; a 2025 analysis from the Center for Migration Studies suggests around 11%. The Dallas Fed reports that unauthorized migration has turned net negative since February 2025, and estimates a “net loss of about 49,000 unauthorized immigrant workers in July [2025] alone,” while the Pew Research Center estimates that 1.2 million immigrants (including both those in the U.S. legally and illegally) had left the workforce by July.

As Carney points out, net negative migration also results in new pressures within the workforce still in the United States, as manufacturing employers compete for workers from a smaller labor pool. Manufacturing job loss has slowed even under these conditions.

American manufacturers also want to hire. Monthly job openings in the manufacturing sector spiked by over 100,000 from a low of 389,000 in April 2025 to a high of 495,000 in January 2026. This does not suggest a sector eager to shed its workers. Quite the opposite: the sector is desperate to grow its workforce. Its difficulty in doing so holds many implications for what policymakers could do better when it comes to workforce development and education, among other things. But skyrocketing demand for manufacturing labor, combined with slowing job loss despite significant enforcement pressure on a meaningful part of the manufacturing workforce, does not suggest a sector in freefall.

Second: How quickly should manufacturing jobs be expected to increase? The answer requires thinking in stages. The first step on the reindustrialization road is to create demand and induce investment; tariffs contribute to this effect by raising the cost of reliance on imports relative to investment in domestic industry. Reindustrialization prompted by tariff pressure therefore necessarily entails short-term pain.

Meeting this increased demand requires increased capacity utilization. In the immediate term, we should expect it to be hard for an atrophied manufacturing labor market to swiftly respond to that need. Manufacturers could be expected instead to increase hours (which they are doing) and boost productivity (which they are also doing).

In the longer term, we should expect growth in employment as new investments pay off and new capacity comes online. A revitalized manufacturing sector should be expected to create jobs—but not in the first year. The example of recent industrial policy (another vital tool of reindustrialization) is instructive. The CHIPS and Science Act was enacted in 2022; asking in early 2023 where all the chip-making jobs were would have been nonsensical. Almost four years later, however, one academic study estimates the policy had “national direct employment effects of approximately 15,000-16,000 jobs in the core semiconductor sector and indirect effects of 15,000-30,000 jobs in related sectors.” No one can build (and then man) a factory overnight.

Third and most important: Are there any signs of a manufacturing sector gradually reorienting itself towards increased production? The answer here is encouraging: yes.

The ISM Manufacturing PMI measures growth or contraction in the sector based on surveys of purchase managers; a PMI above 50 indicates growth. After 12 months of sub-50 readings, the manufacturing index hit 52.6 in January 2026 and 52.4 in February. These are the highest readings since August 2022. They did remain negative through 2025, a point much noted by tariff critics, who now seem strangely silent on the potential import of such a positive turn. S&P Global’s U.S. Manufacturing PMI, for its part, was consistently positive in both 2025 and early 2026 after trending negative in 2024, and seems to have therefore been ignored by the pundits.

Furthermore, the industrial production index is at its highest level since 2019, and has been on an upward trend since November 2024. Manufacturing output has been on an upward trend since 2025 as well, versus a downward trend in the preceding years. The counter-trend spike in demand for manufacturing labor indicates a sector eager to grow. The sector is responding to that pressure with a counter-trend spike in productivity. Manufacturing productivity increased 2 points in 2025—the largest annual increase since 2010, as the BLS helpfully highlights in its March 2026 report. This is healthy; increasing pressure to get better at making things is a key element of a sensible reindustrialization agenda.

None of this compares favorably with the shallow argument that tariffs are destroying American manufacturing. They clearly are not. What seems more likely is that, upon discovering that American manufacturing is not in the unambiguous freefall they predicted, critics have cottoned to a bad argument instead.

There are many quite reasonable criticisms to make about the current trade regime’s relationship to manufacturing. Some we at American Compass have made ourselves. One is that tariffs are insufficient on their own to achieve the desired scale of revitalization, and that when imposed without a comprehensive suite of other policy measures (public-private financing for critical industry and real workforce development solutions, to name just two), they create more pain than needed for less upside than intended. Another is that tariffs implemented in a haphazard, opaque, and unpredictable manner make it harder for manufacturers to invest than necessary, thus offsetting some of the potential gains tariff pressures produce. But these criticisms come from within the frame that tariffs can work, as the data is indicating they can.

Those opposed to tariffs in principle, on the other hand, are stuck grasping at any data that might uphold their narrative that tariffs cause nothing but harm. The real irony, of course, is that when a massive and precipitous decline in American manufacturing jobs really did indicate a crisis in the sector, the free trade absolutists were the first to remind the public that economic transformation takes time, and that painful tradeoffs would be worth it in the long run.

They were wrong; their version of the long run was even worse. Here, however, we seem headed in the right direction. And at the very least, the economists who lectured the American people about the wisdom of achieving general equilibrium over multiple generations through free trade, despite the pain along the way, should refrain from judging an alternative policy a failure because a negative job trend didn’t turn positive overnight.