America lost jobs in July — minus 23,000 against a +83,000 forecast
US payrolls fell 23,000 in July against a +83,000 forecast, with 103,000 more erased from prior months
The three lines
- Nonfarm payrolls fell 23,000 — a decline, not a slowdown, against a +83,000 consensus
- May was cut from 129,000 to 63,000 and June from 57,000 to 20,000 — 103,000 jobs erased
- Unemployment fell to 4.1% — but only because participation dropped: a bad-news improvement
Key questions
- How bad was the July jobs report
- Qualitatively worse than a miss. Payrolls fell 23,000 when the consensus expected an 83,000 gain — a swing of over 100,000. The revisions were heavier still: May was cut from 129,000 to 63,000 and June from 57,000 to 20,000, erasing 103,000 jobs from the recent record. Retail led the declines, down 19,000.
- Why did unemployment fall if jobs disappeared
- Because the denominator shrank. The rate fell from 4.2% to 4.1% not because more people found work, but because discouraged workers stopped searching and dropped out of the labor force. Someone who quits looking is no longer counted as unemployed — a statistically 'better' number for a worse reason.
- What does this mean for the Fed
- It puts a September cut close to fully priced. The employment half of the Fed's dual mandate just turned negative, undercutting the case for holding at 3.50-3.75%. Treasury yields fell immediately and stocks rallied — the market read bad news as a rate-cut signal, at least for now.
Yesterday this page called it "the type of release where interpretation forks." The number left nothing to interpret. US nonfarm payrolls fell by 23,000 in July — against a consensus forecast of an 83,000 gain. Not a slowdown; a contraction. July 2026 enters the record as a month when America's labor market shrank.
1. The footnote worse than the headline — 103,000 jobs erased
The heaviest line in the report is not July's number but the revisions. May's job gain was cut from 129,000 to 63,000; June's from 57,000 to 20,000. Combined, 103,000 jobs vanished from the past two months' record. The labor market described last month as "holding up" had, in fact, already stalled. By industry, retail led the losses at -19,000.
Two days earlier, when ADP's private count printed 44,000 — a six-month low — parts of the market shrugged it off, noting ADP's loose correlation with the official data. This time the direction matched, and the official number was worse.
2. The numbers, illusion included
| Item | Figure | Note |
|---|---|---|
| July nonfarm payrolls | -23,000 | Forecast +83,000 |
| May revision | 129,000 → 63,000 | -66,000 |
| June revision | 57,000 → 20,000 | -37,000 |
| Unemployment rate | 4.2% → 4.1% | Participation-driven — an illusion |
| Retail | -19,000 | Led declines |
| ADP (two days prior) | +44,000 | Same direction, official worse |
The unemployment row needs its own reading. The rate fell to 4.1% in a month of job losses because discouraged workers stopped searching and exited the labor force — and someone who quits looking is not counted as unemployed. It is the textbook case of a number improving for a bad reason; the full mechanics are in today's companion reference, "How the unemployment rate is measured."
3. What remains
The ball is now with the Federal Reserve. With the employment half of its mandate turning negative, the case for holding rates at 3.50-3.75% through September 15-16 has thinned dramatically — the question is how the three-way dissent from July's meeting re-sorts itself around this number. The market's answer — a record-high close on a job-loss report — is dissected in "Bad jobs, good rally." Why the revisions run so large is in "Why jobs numbers get rewritten."