The July employment report landed on 7 August with a number almost nobody modelled: American employers cut 23,000 jobs in a month economists expected to add 80,000. Then the S&P 500 closed higher.
Both halves of that sentence are worth sitting with, because the second one is not irrational.
The quarter's worth of damage in one print
| Metric | July 2026 | Expected |
|---|---|---|
| Nonfarm payrolls | −23,000 | +80,000 |
| June payrolls (revised) | +20,000 | +57,000 as first reported |
| Unemployment rate | 4.1% | 4.2% prior |
| Average hourly earnings, 12-month | 3.2% | 3.5% |
The revision matters as much as the headline. June was originally reported as a 57,000 gain and is now 20,000. Two months that were supposed to show a labor market gently decelerating instead show one that has stopped adding workers.
The unemployment rate fell, and that is not good news
Unemployment dropped to 4.1% from 4.2%. In a normal expansion that is a strong number. Here it is a mechanical artifact: the labor force participation rate declined again. The unemployment rate counts people who are actively looking for work. When someone stops looking, they leave the calculation entirely, and the rate improves without a single job being created.
A falling participation rate alongside falling payrolls is the combination that our team treats as a genuine deterioration signal rather than a soft patch. One number is describing people who cannot find work; the other is describing people who have stopped trying.
Why the market rallied
Average hourly earnings rose by two cents. The twelve-month rate came down to 3.2%, the slowest since May 2021. That is the number that changed the trading day.
For most of 2026 the market's live question has not been whether the Fed cuts — it has been whether the Fed is forced to hike again. Wage growth at 3.2% takes a great deal of pressure off that scenario. Odds on the CME FedWatch gauge for a September move fell to 44%, and October to 58.3%.
The bond market moved first and moved cleanly: the 2-year Treasury yield slipped more than four basis points to 4.204%, its lowest level since 17 July. Equities followed — S&P 500 up 0.3%, Nasdaq Composite up 0.9%, Dow up 67 points.
What our team is actually watching now
The reflex reading is "bad data, rate relief, buy equities." That reflex has worked for two years and it has a specific expiry date: it works while weak labor data reduces the cost of capital faster than it reduces earnings.
The honest position is that we do not yet know which side of that line July sits on. One month of negative payrolls with a downward revision is not a recession. It is also not the picture of an economy that absorbs a hike.
Three things would change our reading:
- Participation stabilising while payrolls stay negative. That would mean the weakness is demand for labor, not supply of it, and the unemployment rate would start rising fast.
- Wage growth below 3%. Below that level the inflation argument for holding rates high mostly dissolves, and the conversation moves entirely to growth.
- A second consecutive negative print. One month is noise plus revision. Two is a trend, and the equity market does not price the second one the way it priced this one.
For now the market has decided that a labor market losing altitude is worth more in discount-rate relief than it costs in forward earnings. That trade has a floor under it — until the earnings side starts confirming what the payroll side is saying.