September added 29,000 jobs, and July turned negative again
Payrolls rose 29,000 in September 2026 and unemployment edged up to 4.2%. Revisions cut July and August by 60,000, putting July back below zero.
- What happened
- The Bureau of Labor Statistics reported on October 2, 2026 that nonfarm payrolls rose 29,000 in September 2026 and the unemployment rate was 4.2%, up from 4.1% in August. The same release revised July from +21,000 to −10,000 and August from +162,000 to +133,000, a combined 60,000 fewer jobs than previously reported.
- Why
- The payroll figure is a survey of employers that is re-estimated twice as late responses arrive. By October 2, 2026, July had been published three times: −23,000, +21,000 and −10,000. The household survey, a separate sample of about 60,000 households, showed employment up 406,000 in the month.
- The numbers that matter
- +29,000Nonfarm payrolls, September 2026 (released Oct 2, 2026)4.2%Unemployment rate, September 2026−60,000Combined revision to July and August 20263.0%Average hourly earnings, 12 months to September 2026
- Why it matters
- Average hourly pay rose 3.0% in the year to September 2026, against consumer prices up 3.4% in the year to August 2026. Longer weeks lifted average weekly pay by 3.6%, which is the figure closer to a household's actual paycheck.
- What happens nextas we saw it on 7 Oct 2026
- The September 2026 Consumer Price Index and the Real Earnings release are both scheduled for October 14, 2026. The October 2026 Employment Situation is scheduled for Friday, November 6, 2026, at 8:30 a.m. ET, and will carry the second revision to September.
July has been published as a loss, then a gain, then a loss. September's first estimate is the smallest gain of the three months.
| July 1st (Aug 7) | -23 |
|---|---|
| July 2nd (Sep 4) | 21 |
| July 3rd (Oct 2) | -10 |
| August 1st (Sep 4) | 162 |
| August 2nd (Oct 2) | 133 |
| September 1st (Oct 2) | 29 |
By October 2, 2026, July 2026 had been published three times. On August 7, 2026 the Bureau of Labor Statistics said the economy lost 23,000 jobs that month. On September 4 it said July had gained 21,000. On October 2 it said July lost 10,000. That third figure is the one the Bureau treats as settled, until the annual benchmark.
The same release, The Employment Situation for September 2026, published at 8:30 a.m. ET on Friday, October 2, 2026 on its scheduled date, reported that employers added 29,000 jobs in September and that the unemployment rate rose to 4.2% from 4.1%. The Bureau's own summary of the month is that both “changed little”.
The headline: 29,000 jobs and 4.2%
“Both nonfarm payroll employment (+29,000) and the unemployment rate (4.2 percent) changed little in September,” the Bureau reports. The gain follows “an average monthly gain of 45,000 over the prior 12 months.” The unemployment rate “has remained between 4.1 and 4.3 percent since March,” with 7.1 million people unemployed.
Private employers added 46,000 jobs. Government shed 17,000, after adding 44,000 in August. The Bureau says employment “in all major industries changed little over the month”, so the industry detail is a description of small moves, not of turning points.
| Sector | July 2026 | August 2026 | September 2026 |
|---|---|---|---|
| Total nonfarm | −10 | 133 | 29 |
| Total private | 28 | 89 | 46 |
| Government | −38 | 44 | −17 |
| Construction | 18 | 16 | 11 |
| Manufacturing | 20 | 15 | 9 |
| Health care and social assistance | 13.3 | 35.1 | 23.0 |
| Financial activities | −10 | −9 | −7 |
Health care added 17,000 jobs in September, against an average of 33,000 a month over the prior 12 months. Manufacturing rose 9,000 and is “up by 72,000 since a recent low in December 2025.” Financial activities lost 7,000 and is “down by 129,000 since a recent peak in May 2025, with most of the job loss in insurance carriers and related activities (−90,000).”
July −10,000, August +133,000 and September +29,000 sum to 152,000. Divided by three, that is 50,667, which the Bureau prints as a three-month average of 51,000. Against the 12-month average of 45,000, the latest quarter is neither a sharp acceleration nor a contraction.
July, measured three times
The payroll number comes from a survey of about 119,000 businesses and government agencies covering about 622,000 worksites. Not all of them report in time for the first estimate. The Bureau publishes what it has and revises each month twice.
“The change in total nonfarm payroll employment for July was revised down by 31,000, from +21,000 to −10,000, and the change for August was revised down by 29,000, from +162,000 to +133,000. With these revisions, employment in July and August combined is 60,000 lower than previously reported.”
| Month | First estimate | Second estimate | Third estimate |
|---|---|---|---|
| July 2026 | −23 (Aug 7) | +21 (Sep 4) | −10 (Oct 2) |
| August 2026 | +162 (Sep 4) | +133 (Oct 2) | Due Nov 6, 2026 |
| September 2026 | +29 (Oct 2) | Due Nov 6, 2026 | Due Dec 4, 2026 |
The Technical Note states: “It is only after two successive revisions to a monthly estimate, when nearly all sample reports have been received, that the estimate is considered final.” Once a year the series is also benchmarked. In the release of February 11, 2026, “the seasonally adjusted total nonfarm employment level for March 2025 was revised downward by 898,000.”
The Technical Note gives a 90-percent confidence interval for the monthly change in total nonfarm employment “on the order of plus or minus 122,000.” Applied to the first estimate: 29,000 − 122,000 = −93,000 and 29,000 + 122,000 = +151,000. On the Bureau's own terms, September cannot be distinguished from a small loss or from a gain of 100,000.
None of this means the number is wrong. It means the first print is the start of a measurement. See how data revisions work and our earlier piece on July's first revision.
What the household survey says
The unemployment rate comes from a different survey, a sample of about 60,000 households. It counts people, not jobs, and includes the self-employed, farm workers and private household workers, whom the employer survey excludes. Someone with two jobs appears twice on payrolls and once in the household count.
The labour force rose to 170,262,000 and employment to 163,152,000. Participation was 61.8%, up from 61.6% in August, and the employment-population ratio was 59.2%, up from 59.1%. The unemployment rate rose because the labour force grew a little faster than employment: 7,109,000 ÷ 170,262,000 = 4.18%, published as 4.2%.
The household sample is far smaller than the employer survey's. For employment itself, the passage of the Technical Note quoted here gives no interval. It puts the 90-percent interval for the monthly change in the number unemployed at about ±425,000, and for the unemployment rate at about ±0.3 percentage point, at an unemployment rate of around 6.0%. The 0.1-point rise to 4.2% is well inside that band, which is why the Bureau calls it little changed.
The year-on-year comparison needs a correction
Set September 2026 against September 2025 in Summary Table A and the household survey appears to have lost 999,000 people from the labour force and 504,000 from employment, while the unemployment rate fell from 4.4% to 4.2%. Part of that is not economics. It is a change in the population estimates the survey is weighted to.
Updated population controls entered the household survey with revised data for January 2026, published with the February 2026 release on March 6, 2026. Applied to December 2025, they lowered the civilian labour force by 1,417,000, employment by 1,432,000, participation by 0.4 point and the employment-population ratio by 0.5 point, and left the unemployment rate unchanged. The Bureau notes that comparisons across the adjustment “may be affected.”
Participation fell 0.7 point between September 2025 (62.5%) and September 2026 (61.8%); 0.4 point of that is the size of the population-control effect measured on December 2025. The employment-population ratio fell 0.5 point (59.7% to 59.2%), exactly the control effect. The effect was measured on December 2025 levels, not on September, so this is an approximate reconciliation, not a Bureau figure. The 0.2-point fall in the unemployment rate is not explained by the adjustment, which left that rate unchanged.
The widest of the Bureau's six underutilisation measures, U-6, was 7.6% in September 2026, against 7.7% in August and 8.1% in September 2025. Our piece on the six unemployment rates explains what each one counts; the glossary covers unemployment measures.
Pay: 3.0% an hour, 3.6% a week
Average hourly earnings for private-sector employees “edged up by 5 cents, or 0.1 percent, to $37.81” in September 2026, and are up 3.0% over 12 months. Production and nonsupervisory employees averaged $32.60 an hour, up 7 cents. The average workweek held at 34.4 hours. August's average was revised from $37.75 to $37.76.
Hourly: 1.030 ÷ 1.034 − 1 = −0.4%, using 3.0% earnings growth to September 2026 and 3.4% CPI inflation to August 2026 (the September 2026 CPI is scheduled for October 14, 2026, so the months do not match). Weekly: 34.4 hours × $37.81 = $1,300.66 in September 2026, against 34.2 × $36.70 = $1,255.14 in September 2025, a rise of 3.6%. Then 1.0363 ÷ 1.034 − 1 = +0.2%. The weekly figures are products of the two published averages, not the Bureau's own weekly-earnings series.
The difference is hours. A workweek 0.2 hour longer than a year earlier added more to the average paycheck than hourly rates lost to inflation. See nominal versus real and our earlier comparison of wages and prices for August.
What it means for your money
Three mechanisms, none of them a prediction.
- Pay against prices. An hourly wage that rises 3.0% while prices rise 3.4% buys 0.4% less. On $37.81 an hour, that is the difference between a raise and a small pay cut in real terms. Weekly pay depends on hours as well, and hours can fall as easily as they rose.
- First prints move. July went from −23,000 to +21,000 to −10,000 in eight weeks. Any conclusion drawn from one month's payroll number rests on a figure with a ±122,000 error band and two revisions still to come.
- Search length is a separate number. The unemployment rate has stayed between 4.1% and 4.3% since March 2026, but it counts how many people are looking, not for how long. In September 2026, 1.9 million had been unemployed for 27 weeks or more, 27.1% of all the unemployed. Sizing an emergency fund is, in effect, an estimate of how many months of expenses a search could take.
The Federal Reserve raised its target range to 3.75%–4.00% on September 16, 2026, with August's payroll gain then reported as 162,000. The October 2 release put it at 133,000. Our account of that decision used the figures available then.
Sources
- The Employment Situation — September 2026 (USDL-26-1549)U.S. Bureau of Labor Statistics · Oct 4, 2026 · Primary source
- Employment Situation Summary Table A. Household data, seasonally adjusted, September 2026U.S. Bureau of Labor Statistics · Oct 4, 2026 · Primary source
- Employment Situation Summary Table B. Establishment data, seasonally adjusted, September 2026U.S. Bureau of Labor Statistics · Oct 4, 2026 · Primary source
- Table A-15. Alternative measures of labor underutilization, September 2026U.S. Bureau of Labor Statistics · Oct 4, 2026 · Primary source
- Employment Situation Technical Note, September 2026U.S. Bureau of Labor Statistics · Oct 4, 2026 · Primary source
- The Employment Situation — July 2026 (archived release of August 7, 2026)U.S. Bureau of Labor Statistics · Oct 4, 2026 · Primary source
- The Employment Situation — August 2026 (archived release of September 4, 2026)U.S. Bureau of Labor Statistics · Oct 4, 2026 · Primary source
- The Employment Situation — February 2026 (archived release of March 6, 2026, USDL-26-0367): Adjustments to Population Estimates for the Household SurveyU.S. Bureau of Labor Statistics · Oct 4, 2026 · Primary source
- The Employment Situation — January 2026 (archived release of February 11, 2026): annual benchmark revisionU.S. Bureau of Labor Statistics · Oct 4, 2026 · Primary source
- Schedule of Releases for the Employment Situation, 2026U.S. Bureau of Labor Statistics · Oct 4, 2026 · Primary source
- Schedule of Selected Releases for October 2026U.S. Bureau of Labor Statistics · Oct 4, 2026 · Primary source
- Schedule of Releases for the Consumer Price Index, 2026U.S. Bureau of Labor Statistics · Oct 4, 2026 · Primary source
- Consumer Price Index Summary — August 2026U.S. Bureau of Labor Statistics · Sep 11, 2026 · Primary source
- FOMC statement, September 16, 2026Federal Reserve Board · Sep 16, 2026 · Primary source
Related analysis
July's jobs number went from −23,000 to +21,000. That is not a rounding error.
The Bureau of Labor Statistics revised two months of hiring up by a combined 55,000. One of them changed sign, and it is the figure this publication built an article around.
The unemployment rate has six versions. Here is August 2026 in each
The headline 4.1% is U-3. The same BLS release also put the rate at 1.8%, 1.9%, 4.4%, 5.1% and 7.7%. Who each version counts, and what the gaps have meant.
Pay rose 3.1% in the year to August. Prices rose 3.4%.
Average hourly earnings reached $37.75 in August. Measured against inflation over the same twelve months, the average worker went slightly backwards.
August added 162,000 jobs. The average of the prior twelve months was 31,000.
The strongest hiring month in a year sits on top of a twelve-month average five times smaller. Both numbers are true, and only one of them is a trend.