The US Bureau of Labor Statistics released the July 2026 employment report this morning.
Every figure here is taken from the BLS release itself, dated 7 August 2026. Even so, errors, inaccuracies or omissions are possible, and the figures may change after publication. Spotted something that looks wrong? Write to me and I will correct it.
The Headline Number
Nonfarm payroll employment fell by 23,000 in July.
In the release's own words, both that figure and the unemployment rate "changed little". That is the phrasing the BLS uses when a move is not statistically meaningful.
| July 2026 | |
|---|---|
| Change in payroll employment | minus 23,000 |
| Average monthly change over the prior 12 months | plus 34,000 |
| Unemployment rate | 4.1% |
| Number of unemployed | about 6.9 million |
A fall of 23,000 jobs in an economy employing more than 159 million people is noise. If that were the whole report, it would not have moved the market.
But That Is Not the Whole Report
The most important thing sits in the paragraph on revisions.
| Month | Previously reported | After revision | Change |
|---|---|---|---|
| May | plus 129,000 | plus 63,000 | minus 66,000 |
| June | plus 57,000 | plus 20,000 | minus 37,000 |
Together, those two months are 103,000 jobs lower than reported at the time.
Why does a revision matter more than the monthly number? Because the monthly number describes one month, and the revision describes an error in the picture. Before the revision, May and June appeared to have added 186,000 jobs between them. After it, they added 83,000. This is not a labour market that weakened in July. It is a labour market that was already weaker than we thought in the two months before.
Where is the weakness coming from? Per the release, July saw declines in local government education, at minus 50,000 jobs, and in retail trade, at minus 19,000. Health care employment continued to trend up.
And Why That Invites a Rate Cut
This is the logic the market was working from.
The Federal Reserve has two mandates: price stability and maximum employment. When inflation is high it raises rates to cool the economy. The problem is that cooling too hard breaks the labour market.
What this report shows is cooling without breaking:
| The figure | What it means | |
|---|---|---|
| Unemployment rate | 4.1% | still historically low |
| Average hourly earnings | $37.62 | up just 2 cents on the month |
| Annual wage growth | 3.2% | not enough to feed inflation |
| Average workweek | 34.3 hours | unchanged |
Wage growth of 3.2% a year is the critical figure. When wages rise fast, workers spend more and prices follow. 3.2% is a pace that does not generate meaningful inflationary pressure, and so it does not block a rate cut.
Put differently: the labour market is cooling, wages are not running away, and unemployment is still low. That is precisely the combination that lets the Fed cut without fearing it reignites inflation.
The Caveat You Should Not Skip
Part of the stability in the unemployment rate comes from a less flattering place.
The labour force participation rate was 61.4%, down 0.7 percentage point since January. The employment-population ratio was 58.9%, down 0.5 percentage point over the same period.
Why does this matter? The unemployment rate counts only people actively looking for work. Someone who gives up and stops looking leaves both the numerator and the denominator, and unemployment looks lower. A 0.7 percentage point fall in participation over seven months means some people simply left the picture. Unemployment of 4.1% looks better than it is when read alongside that figure.
And two more figures pointing the same way: the long-term unemployed, those looking for 27 weeks or more, numbered 1.8 million - 25.5% of all unemployed people. And the number working part time for economic reasons, meaning they wanted full-time work and could not find it, was 4.8 million.
How the Market Reacted
As of 14:13 in New York, mid-session:
| 6 August | 7 August | Change | |
|---|---|---|---|
| Russell 2000 | 298.25 | 301.00 | +0.92% |
| Nasdaq 100 ETF | 714.65 | 720.39 | +0.80% |
| S&P 500 ETF | 768.56 | 771.69 | +0.41% |
| VIX | 15.15 | 14.92 | -1.52% |
| 5-year Treasury yield | 4.39% | 4.36% | -3 basis points |
| 10-year Treasury yield | 4.67% | 4.65% | -2 basis points |
| 30-year Treasury yield | 5.21% | 5.21% | unchanged |
Trading was still open as this was written. These are not closing prices.
Three things in that table tell the story:
First, small caps rose the most. The Russell 2000 is an index of small companies, and they are especially rate-sensitive: they carry more leverage and borrow at higher cost. When the market prices lower rates, they move first.
Second, short yields fell more than long ones. The five-year fell three basis points, the ten-year two, and the thirty-year barely moved. Short yields respond to rate expectations; long yields respond to growth and inflation further out.
Third, the fear gauge fell. A lower VIX means the market is pricing less risk, not more.
That combination is a clear signature of a market pricing a rate cut, not one fearing a recession. If it feared recession, small caps would have fallen first and long yields would have collapsed alongside short ones.
הזווית שלי
דעה אישית של אילן אברמוב - לא ייעוץ ולא המלצה
This report is a good reminder of how misleading a headline can be.
"Minus 23,000 jobs" sounds like bad news, and the market rose. Anyone reading only the headline sees a contradiction. Anyone reading the report sees exactly what the Fed wanted to see: an economy slowing at a controlled pace, without wages running away and without unemployment spiking.
And what I take from it is not a rate forecast. I do not know what the Fed will do, and nobody does. What I do take is a reading rule: in an employment report, the revisions matter more than the monthly number. A monthly figure is a first estimate with a wide error band. A revision is what actually happened.
The second thing I take is the caveat. A participation rate falling 0.7 percentage point in seven months does not make headlines, and it is what turns 4.1% unemployment into a number to read carefully. People who left the labour force are not unemployed by the statistical definition, but they are not working either.
What I will watch in the next report is not the jobs number - it is whether the revisions keep coming in negative. One month of downward revision is noise. Three consecutive months is a trend, and that is a very different story.
