US August payrolls 162,000 — triple the forecast, and July's loss became a gain

The US Bureau of Labor Statistics reported on September 4 that nonfarm payrolls rose 162,000 in August. Forecasts had clustered in the low-to-mid 50,000s. The outcome was roughly triple that, and the largest monthly gain in five months.

To see the size of it you need the trend it sits against. Over the preceding twelve months, US payrolls grew by an average of 31,000 a month. August alone was more than five times that.

But the part of this report that moved markets most was not the headline. It was what happened to last month's number.

1. A loss became a gain

July payrolls were first published as a loss of 23,000. When that figure landed on August 7, markets read it as the beginning of a genuine crack in the US labour market and cut the odds of a September rate rise sharply.

In this report, July became a gain of 21,000 — a 44,000 swing. June moved from 20,000 to 31,000, another 11,000. Together, 55,000 jobs that had been recorded as absent now exist in the data.

Revisions themselves are ordinary. The establishment survey publishes a first estimate from businesses that report by the cutoff, then folds in late responses across two later releases. What makes this one consequential is the direction: it does not merely adjust the number, it removes the evidence that a repricing was built on. July's negative print is gone.

2. The August report in numbers

Two rows deserve a second look.

The labour force grew by 683,000 and the unemployment rate did not move. When that many people enter the market in a single month and the rate holds, it means most of them found work. Unemployment is a ratio whose numerator and denominator move together, so this combination reads as a stronger signal than the headline alone (see "How the unemployment rate is calculated").

Wage growth slowed. Average hourly earnings rose 3.1 percent from a year earlier, down from 3.2 percent in July. More hiring without more wage pressure is the one line in this report that argues against urgency on inflation.

3. Where the jobs came from

The distribution is less impressive than the headline.

• Food services and drinking places +59,000 — five times the 12-month average of 12,000
• Leisure and hospitality overall +62,000
• Local government education +42,000
• Construction +22,000
• Manufacturing +16,000
• Health care +13,000 — below its 12-month average of 32,000
• Financial activities −11,000
• Information −23,000 — three times its average monthly decline of 8,000

Of the 162,000 total, 101,000 came from two lines: restaurants and local government education. August is the month school staffing returns, and local education employment is volatile even after seasonal adjustment.

At the other end, information lost 23,000, and the BLS attributed the decline to computing infrastructure providers, data processing and web hosting. The industry absorbing the largest capital investment in the economy is shedding headcount. Health care undershooting its average belongs in the same reading: health care and government have carried US job growth for the past year, and both softened here.

4. What the market took from it

US stocks and short-dated Treasuries sold off together after the release.

• S&P 500 7,718.60 (−0.38%)
• Nasdaq 26,506.99 (−0.29%)
• Dow 53,414.25 (−271.86 points, −0.51%)
• 2-year Treasury yield at its highest since January 2025

A good jobs number hurting stocks is the defining feature of this cycle. US policy is pointed at a rise, not a cut. After Warsh's August 28 Jackson Hole speech markets began pricing a September move; on September 3 Waller tied the decision to the August inflation print and the odds eased. This report reversed that.

A durable labour market widens the room the Fed has to keep fighting inflation. That is why the good number was read as the bad one.

5. What is still open

• Do not quote a single probability. Reported odds ranged from 53 to 65 percent depending on outlet and timing. The number is inferred from fed funds futures prices, so it moves with the calculation.
• August will be revised twice. July moving 44,000 in this release is the demonstration. Only the direction is settled today.
• The next checkpoint is August CPI, the print Waller named, due before the September 15-16 meeting. Employment has supplied one argument for a hike; inflation supplies the other.
• The 10-year yield is disputed. Some accounts had yields rising after the release; another had the 10-year falling to 4.77 percent from 4.818 percent. Short rates rising faster than long rates is normal in a tightening phase, but the Treasury's daily data settles it.
• Korea gets this a day later. The KOSPI closed 1.64 percent higher at 6,687.21 on September 4, hours before this report landed at 21:30 Korean time (see "KOSPI closes at 6,687.21 on September 4, 2026").