Get the Ricochet app
Ricochet will work fine in a browser on your mobile devices, but for the best experience please download our app.
Last month I wrote that the headline jobs number — the Establishment Survey's PAYEMS figure — had been systematically more optimistic than reality warrants, and that the Household Survey's full-time employment measure was tracking closer to the truth than the number every financial journalist in America was leading with.
July's report is Exhibit A.
Let's start with what the headlines are saying. The Establishment Survey reported a decline of 23,000 jobs in July. That's the number most outlets led with, and it's bad enough. But buried in the same report was May's number, revised down by 66,000, and June's by 37,000. That's an additional 103,000 jobs quietly subtracted from the prior two months — with no fanfare, no correction notices, no prominent placement. In a single report, the Establishment Survey just walked back 126,000 jobs. That's the revision story playing out in real time, exactly as advertised.
Now look at what the Household Survey is saying.
Full-time employment: My go-to economic statistic (BLS data series LNS12500000) found the count of people who usually work 35 hours or more fell by 106,000 in July, from 133,659,000 to 133,553,000. That's the third consecutive monthly decline. Since March, full-time employment has dropped by more than one million. Year over year (June 2025 to June 2026), full-time employment is down 1,529,000.
One and a half million full-time jobs, gone in a year. The headline number missed it.
The unemployment rate fell to 4.1% in July. At least that's good news, no?
Not really. The labor force participation rate dropped again, down another tenth of a point to 61.4%, following last month's drop to 61.5%. The labor force itself is shrinking. People are not finding jobs and celebrating — they have stopped looking. When workers exit the labor force, they exit the unemployment calculation too. The rate falls not because the labor market is improving but because the denominator is shrinking.
This is the second consecutive month that the unemployment rate has declined for the wrong reasons. It's a pattern now, not a blip. This was the story during the Obama administration too. The unemployment rate fell because of a declining labor force.
Part-time employment for economic reasons, for workers who want full-time work but can only find part-time, changed little. So those million-plus vanished full-time jobs aren't showing up as involuntary part-timers. They're showing up as labor force exits. These are people who have given up.
Note: A word on methodology, because it matters here.
The Establishment Survey (PAYEMS) and the Household Survey (LNS12500000) are genuinely independent measures. They use different samples, different methodologies, and they measure different things. The Establishment Survey counts job slots: a person holding two part-time positions counts twice. The Household Survey counts people who are actually employed full time. Over time, in a healthy labor market, they tend to tell roughly the same story. But they are not telling the same story right now.
Over the thirty months from January 2023 through May 2025, the Establishment Survey reported, in its first release, a cumulative gain of 6,714,000 jobs. After revisions, that figure was cut to 4,136,000. The BLS quietly subtracted 2,578,000 jobs — 38% of the original headline — through the revision process. Over the same period, full-time employment as measured by the Household Survey grew by 2,870,000. The noisier, less prestigious, smaller-sample survey was closer to the eventual revised truth than the number the financial press was leading with every single month.
That trend is not reversing. It is accelerating.
To be clear about what I am and am not saying: I am not claiming intentional manipulation. The two surveys have always diverged, and they will continue to. Measurement is hard, and revisions are a normal part of the process.
What I am saying instead is that for the better part of three years, the headline number has been painting a materially rosier picture of the American labor market than the underlying data supports. Full-time employment, the kind of job a person can actually build a life around, has been deteriorating while the headline number reported gains. Those gains have been systematically revised away over time. And now, with the headline number itself turning negative and the revisions running at 103,000 in a single report, the two surveys are finally, belatedly, pointing in the same direction. Down.
The full-time employment number has been trying to tell us this for months. But few are listening.
Discussion (26)
To read this bio, please join Ricochet.
Geez, ekosj, that is really discouraging. Any thoughts about the underlying reasons?
To read this bio, please join Ricochet.
One fact. The rest is speculation ...The fact: Part of it is the reduction in Federal Government jobs ... down by about 400,000 since Trump2.0. So, it's not an unmitigated bad thing. But those former government employees haven't found other work and that's not good.
Specualtion ... Could also be because of out-migration of illegals ... but if that's true it means that those jobs haven't been filled by native workers. That's not good either. And companies have begun to shed human jobs because of AI. That will accelerate. Lastly, the offshoring of US jobs to India and the Philippines etc. has not stopped. Why hire one person in NYC for 150,000 plus benefits when you can hire 3 offshore for a total of 75,000?
To read this bio, please join Ricochet.
I have wondered whether anyone has come up with a Zoom/remote impact measure. I work in downtown Washington DC which is exhibiting a major downturn in the office and retail space market (but a small boom in residential conversions of that space).
Since COVID, many people did not come back to 5-day/9-5 and so they don't shop or buy lunches or use local dentists, doctors, accountants (spaces that require staff) etc as often or at all. I am so old, I can remember very crowded sidewalks near the big department downtown stores, countless small shops and lunch counters that were always crowded. Other than tourists near the monuments of Smithsonian, there are not many people out and about compared to long ago.
I know the VA had people who left DC altogether during COVID but still somehow had the same desk job. The policy changed to requiring a mandatory once-a-month in-person appearance to go with remote work. (Would love to see a productivity study of that outfit..) In any event, lots of government office space appears empty.
How do you redraw an economic model where no one commutes with all of the attendant economic activity that goes with local travel and in being present somewhere other than home?
Economics is not my area but I suspect that these specific changes have more than just a small ripple effect.
To read this bio, please join Ricochet.
Just thinking out loud here ... wouldn't that just move all that economic activity to spaces nearer to the remote workers rather than in the urban center? I mean remote workers still eat lunch, go to the Dr and dentist and accountant....shop whatever ... just not in an urban center. Intuitively am drawn to the idea that remote work should have an negative impact overall... but now that I try to detail what they are, I'm not so sure.
To read this bio, please join Ricochet.
As for the impact of working from home, I was thinking commercial construction such as for office space would be down due to office space gluts. People would also make their own lunches and socialize less after work hurting some retail business and they spend less on transportation, work attire etc. On a lighter note every Quality Learing Center that shuts down due to loss of subsidy only makes the commercial building glut worse. When I asked grok though it looks like the data center building
bubbleboom compensates for other commercial construction downturns.It indicates the major losses in the last 2 years are reported in
IT/Technology, ( Anecdotally My middle aged daughter just lost her web designer job this week so I believe it. ) government (yeah), manufacturing(oh oh) and Financial services.
Home building is only meh. This is where I would have expected the deportations to have the biggest impact.
Do you have contrary or better data?
You're seeing five of 26 comments. Sign in, or join for just $5 per month, to read them all and add your own.