September jobs report: 29,000 new jobs, and a weaker summer than first reported

US employers added 29,000 nonfarm payroll jobs in September, the Bureau of Labor Statistics reported on October 2, well below economists’ expectations. Unemployment reached 4.2%, while downward revisions to July and August showed that summer hiring was weaker than initially reported.

The biggest change was to July: an estimated gain of 21,000 jobs became a loss of 10,000. August’s gain fell from 162,000 to 133,000. Together, those revisions removed 60,000 jobs from the previously reported summer total, according to BLS.

What September’s jobs report actually says

Economists had expected 84,000 additional jobs, ABC News’s Max Zahn reported. September’s 29,000 gain was also below August’s revised 133,000. That revised figure is the relevant comparison; ABC’s initial account used the earlier August estimate of 162,000.

BLS described both payroll employment and the unemployment rate as having “changed little.” The unemployment rate rose from 4.1% to 4.2%, according to ABC, but remained within the 4.1%–4.3% range recorded since March. BLS put the number of unemployed people at 7.1 million.

Average hourly earnings on private payrolls reached $37.81, up 0.1% over the month and 3.0% over the year. Those wage figures come from the employer survey, alongside the payroll employment count, rather than the household survey used to calculate unemployment.

Why July and August’s job estimates changed

The BLS release attributes the revisions to additional employer reports and recalculated seasonal factors. Initial payroll estimates are revised in each of the following two months as more responses and updated adjustments become available. September’s 29,000 figure is therefore a first estimate, subject to the same routine process.

The 60,000 reduction changes the estimated employment picture for July and August. It is not an additional count of jobs lost during September. BLS’s explanation does not specify how much of that adjustment came from additional employer responses and how much came from seasonal recalculation.

Monthly revisions are separate from annual benchmark revisions. BLS explains that the annual process aligns payroll estimates with nearly complete employment counts drawn from unemployment-insurance tax records. That broader check helps control sampling and modeling errors; it is distinct from the updates to July and August in this release.

Why payroll jobs and unemployment use different surveys

The jobs report combines two measurements. Employers provide information about nonfarm payroll employment, hours and earnings. Households provide information about people’s labor-force status, including the unemployment rate and demographic detail. BLS’s survey comparison explains why those measures need not move identically.

One distinction is jobs versus people. A person with multiple payroll jobs can be counted more than once in the employer survey. The household survey counts an employed person once and includes agricultural workers and unincorporated self-employed workers who fall outside the payroll measure.

Neither survey continuously counts employment throughout the month. The household reference period is generally the calendar week containing the 12th; employers report for the pay period containing the 12th. Different coverage, timing and estimation methods help explain why the resulting measures can diverge.

Unemployment is not calculated by counting benefit recipients. BLS generally counts people without jobs who are available and actively seeking work, along with people on temporary layoff even if they are not actively searching. That definition is separate from eligibility for unemployment benefits.

What the data-quality review found

An independent Government Accountability Office review published in June found that expert stakeholders generally considered the jobs report useful. They also said occasional large revisions could make it less useful for timely decisions. GAO reviewed agency documents, studies and performance data and interviewed officials and 14 stakeholders.

GAO identified falling survey response rates as a risk to data quality. Agency studies found limited effects on household-survey accuracy, it said. BLS had not released its employer-survey assessments because of methodological limitations. These were findings about the statistical system, not an assessment of September’s figures.

GAO recommended stronger external feedback and publication of payroll-survey nonresponse assessments. BLS generally agreed. A July update on GAO’s page said a new study was expected by the end of 2026; it did not establish that the study had been completed.

What the figures mean for the Federal Reserve

The release follows September’s Federal Reserve interest-rate increase, which ABC reported was intended to curb inflation. Higher borrowing costs can slow hiring, but ABC noted that such effects typically emerge after a lag of several months. September’s payroll estimate alone does not establish the effect of that decision.

Nor does the report determine the Fed’s next move or establish a recession. The immediate findings are narrower: hiring came in below expectations, summer estimates were reduced, and unemployment remained within its recent range. September’s preliminary payroll figure can still change as the routine revision process continues.

BLS schedules the October employment report for November 6 at 8:30 a.m. Eastern, or 13:30 UTC. That release is the next scheduled monthly update to the employment picture.