In a surprise to analysts, the United States economy lost jobs last month, per the Bureau of Labor Statistics (BLS). Economists polled by Bloomberg expected the economy to gain 80,000 jobs, and July’s Current Employment Statistics (CES) survey indicates that we lost 23,000 instead. Any jobs figure from the previous month should always be presented with the caveat that it will be revised later, due to the nature of how this survey is conducted. It’s more complex than this, but put simply: larger firms have more resources and time to respond to this government survey, while smaller businesses do not and typically report late. Every new jobs report contains revisions to past ones, and for July, the BLS also noted that “With these revisions, employment in May and June combined is 103,000 lower than previously reported.”
Another important caveat to note is that the genesis of this data comes from businesses filling out a survey and sending it back to the BLS, and response rates have dipped from around 70% before 2020 to around 60% today. There is simply less data in the jobs data than there was before, and any good statistician can tell you what happens when you have less data: more variance. It’s not that it’s filled with lies or voodoo accounting, but that the methodology of the CES can lead to big swings in the data before it settles on a final figure because many of your local mom and pop shops probably haven’t filled out their July survey yet.
This nature of inherently incomplete monthly jobs reports and their repeated downward revisions was what had Trump fuming last year after he fired the head of the BLS and tried and failed to put in an unqualified hack into one of the country’s most important jobs, and he no doubt is upset again today at being defeated by one of the few books he’s had access to that he has not cooked. The market is taking today’s news as a splash of cold water on the concept of near-term Fed rate hikes, but not a jobs crisis. Despite its warts I will get into at the end of this article, the jobs market has actually looked pretty solid before this week’s bad report, with many experts like Natixis chief economist Chris Hodge looking at many months of data and concluding that “This is a labor market that is exhibiting stability but not heat.”
The two-year U.S. Treasury yield, typically the one most sensitive to Fed policy, is down a little over 1% as I write this. A significant move but nothing out of the ordinary, especially given the volatility Trump has unleashed in debt markets since he spiked the price of oil by losing a war to Iran’s third stringers. Yields on Treasuries across the board are down on the news of this report, down more at the short end of the yield curve around the one-year U.S. Treasury yield, a reflection bond traders getting spooked in the opposite direction of where Trump’s new Fed Chair toady scared them last week.
Trump stooge and alleged economics knower Kevin Hassett went on state TV today to do his best Kevin Bacon in Animal House routine and tell markets that all is well. This is not how this works, Kevin. Teachers don’t get fired at the end of every school year then rehired two months later.
Hassett on why the economy lost 50,000 government jobs: “It was one of those weird seasonal things. Because there was a lot of snow in the winter, a lot of schools had to stay open longer for the snow days, so the teachers got laid off in July instead of the end of June. Heheheh.”
— Aaron Rupar (@atrupar.com) August 7, 2026 at 8:18 AM
It’s a funny thing to say given how close he is to making a real point rooted in common knowledge I was taught in my introductory finance classes, and this clip is a great example of how supposed economic knowers like Hassett are entirely devoted to telling familiar lies Trump’s rotting brain can understand over existing in anything resembling objective reality. There are seasonal job losses based around school closures, as bus drivers, maintenance crew, kitchen staff, and lots of other supplemental jobs lose their work for the summer after the kids go home. But that annual routine is already baked into the model because it is often written into their contracts, while teachers are signed year-round.
The good news for Trump that does make this more complex than just a negative headline jobs number is that the unemployment rate went down, from 4.2% to 4.1%. I know a lot of people in Will Stancil’s mentions will see “lost jobs” in this headline and use it to buttress their case that we’re in month 37 of a deep recession the media won’t tell you about, but the jobs data you may be citing to make the case we are in Great Depression 2 is telling you that the headline figure does not remotely tell the entire story here.
“In July, employment in health care continued its upward trend (+22,000) but at a slower pace than the average monthly gain over the prior 12 months (+36,000),” wrote the BLS. The doomers do have a point in that ever since Trump launched his trade war last April, job growth is concentrated in just a couple industries, especially healthcare as the largest generation of the 20th century ages into retirement, while many other industries have seen little to no job growth. But again, it’s not easy to write a neat summary of what’s going on here, as the BLS noted that “Among the unemployed, the number of people on temporary layoff increased by 153,000 to 921,000 in July. The number of permanent job losers changed little at 1.7 million.”
I’m going to tag in someone sitting in front of a Bloomberg Terminal who knows far more about finance than I do to explain why there are likely serious issues with the CES numbers under the hood, in large part because they contradict the mild uptick in actual unemployment claims, macro economist George Pearkes.
Disaster CES print, really bad. Household numbers look COMPLETELY different.
I think CES is running into major sampling issues, there’s just no way July jobless claims plunged as hard as they did with a negative job creation number. They’re just completely inconsistent.
— George Pearkes (@peark.es) August 7, 2026 at 6:31 AM
Under the hood, warehouse clubs/supercenters/etc cut 20k+ employees MoM…one of the larger industry job losses. I am confident that number is wrong.
Bars and restaurants were another standout, -26.1k MoM. Hard to believe as well but at least possible.
Final notable: state/local education -49k
— George Pearkes (@peark.es) August 7, 2026 at 6:34 AM
Estimated numbers of people permanently laid off and in slack work conditions are also completely inconsistent with CES data.
— George Pearkes (@peark.es) August 7, 2026 at 6:50 AM
The National Women’s Law Center (NWLC) analysis of the report found that “women accounted for 100 percent of the decline in the labor force in July,” which is one of those kinds of figures that is both true and doesn’t tell the whole story, and it opens up a whole can of worms that I think actually does lead us towards an explanation for this mess that goes beyond lower CES response rates and the reasonable doubt they bring with them.
“A total of 165,000 women, aged 20 and over, left the labor force last month—meaning they are neither working nor looking for work—while men’s labor force participation was unchanged,” writes the NWLC. “Overall, the economy lost 23,000 jobs in July. Women accounted for all of these losses, losing 32,000 jobs, while men gained 9,000 jobs. Women make up 50.1% of the workforce.” All these figures are correct, but this is a good example of how this month’s data tells a strange story that frankly doesn’t really make much sense.
Women actually gained 11,000 jobs in the private sector last month, as Table B-5 demonstrates, and you can parse that out and find most of those increases concentrated in the typical gainers like healthcare. Government jobs are where these steep losses come from, as they were down 43,000 women from June. That’s where the accurate NWLC net 32,000 loss figure is derived, but separating it out like this demonstrates how noisy this data is and how throwing all women into one jobs bucket is just not a good way to tell the whole story. BLS data suggests that women comprised 58.4% of all government employees in July, so the jobs lost in local government education are entirely responsible for women’s job losses last month, and this narrative cannot be extrapolated on to the private sector.
Bill Adams, chief U.S. economist at Fifth Third Commercial Bank pointed to the nearly 50,000 total jobs lost in local government education as a “wonky seasonal adjustment fluke” that is rooted in “big swings that come every year in the summer months,” which is a more developed version of the lobotomized story Hassett told. It’s very possible this negative headline jobs figure is just wrong, partially rooted in a borked local government job loss figure that will be revised upwards in the coming months.
But arguing about a flawed one-month snapshot of the labor market I think misses the larger picture this confusing puzzle piece does actually fit into. We’re focused on the numerator in the unemployment rate, the number of unemployed people, but there’s a more developed story unfolding in the denominator, the labor force.
Fed Chair Jerome Powell said last year that Trump’s immigration policies are hurting job growth more than tariffs, and the Dallas Federal Reserve wrote a compelling research paper last October about the break-even employment rate plummeting under Trump’s crackdown (break-even employment is the number of new jobs needed to keep the unemployment rate steady). This chart above is from Anton Cheremukhin, Daniel Wilson and Xiaoqing Zhou’s newest research published at the Dallas Fed at the end of March, and it demonstrates how we really are in a different economy that needs far fewer new jobs to maintain the unemployment rate than the one under Joe Biden. As much as the human rights abuses should be front and center in critiques of Trump’s war on people with more melanin in their skin than Stephen Miller, there have been dramatic economic consequences to his immigration policies, and they are directly tied to the nonstop tepid jobs reports driving him mad.
“The break-even rate peaked at about 250,000 jobs per month in 2023, fell to roughly 10,000 by July 2025, and declined to near zero thereafter, averaging about -3,000 jobs per month from August to December 2025,” wrote the Dallas Fed. “indicating, if anything, a modest net jobs loss over this period.”
This is where you would see the impact of the 165,000 women aged 20 and over leaving the labor force last month, as it lowers the number of jobs needed to keep the unemployment rate steady with fewer people out looking for work (assuming that figure is not a “fluke”). This is how we can live in a world where we lose jobs and the unemployment rate goes down. That actually logically tracks along the trend of break-even employment growth turning negative. This illogic is simply just Trumponomics in action.
