A monthly jobs headline is an estimate made while the reporting process is still underway. The September 2026 employment report illustrates why the first number is not always the last: it revised July’s payroll change from a gain of 21,000 to a loss of 10,000, and cut August’s gain from 162,000 to 133,000. Those changes did not mean the earlier reports were fabricated. They reflect additional information and updated calculations arriving after the initial release. Bureau of Labor Statistics
The first estimate has a deadline
The establishment survey asks businesses and government agencies about payroll employment, hours and earnings. Its results are based on a sample, and the first estimate must be published before every respondent has necessarily reported. Waiting indefinitely would make a monthly indicator less useful; publishing early means the estimate is provisional. The September release says the employment estimate is revised in each of the next two months as more sample reports arrive. Bureau of Labor Statistics
This is a different measure from the household survey, which asks about people’s labor-force status and supplies the unemployment rate. Payroll changes are not a count of every person who has a job: the establishment measure covers nonfarm payroll employment, with a different scope from the household measure. A headline about payroll revisions therefore should not be read as a direct recalculation of the unemployment rate. Bureau of Labor Statistics
What changed for July and August
The September report’s revised history makes the distinction between months essential. July’s initially reported increase of 21,000 became a decrease of 10,000, a downward revision of 31,000. August’s initial increase of 162,000 became 133,000, down 29,000. Taken together, the two months had 60,000 fewer payroll jobs than previously reported. September itself was reported as a gain of 29,000. Bureau of Labor Statistics
Those are not three competing estimates for one period. Each figure describes a distinct month, and the July and August revisions change what readers now know about those past months. A useful way to read a release is to note both the new month’s estimate and the revised values for the two preceding months. That keeps a changed historical baseline from being mistaken for a new-month change.
Reports and seasonal patterns matter
More employer and agency reports are one reason estimates move. A second is recalculation of seasonal factors, which adjust recurring calendar patterns in the data. The Bureau of Labor Statistics identifies both additional reports and seasonal-factor recalculation as causes of the July and August changes. Seasonal adjustment helps make month-to-month comparisons more informative, but its calculations can be updated as new observations become available. Bureau of Labor Statistics
The same general revision pattern appears in another labor-market series, though it measures different things. In the August 2026 Job Openings and Labor Turnover release, July openings were revised upward by 64,000 to 7.3 million, while July hires were revised upward by 92,000 to 5.1 million. The release also attributes monthly revisions to new business and agency reports and recalculated seasonal factors. Openings, hires and payroll employment are not interchangeable measures. Bureau of Labor Statistics
Use the number as a vintage
An estimate is best understood as a dated snapshot of the available evidence. The September payroll figure was the estimate published in the September report; later releases may update it. That does not make the initial number meaningless. It gives a timely reading, while the follow-up figures improve the account of the period as information accumulates. The report also notes that annual benchmark revisions use near-complete employment counts from unemployment-insurance tax records to re-anchor estimates and help control sampling and modeling errors. Bureau of Labor Statistics
For everyday readers, the practical distinction is between a trend and one release vintage. A revised month may alter the apparent pace of hiring without changing anyone’s individual job history. The August JOLTS report similarly records 7.1 million openings and 5.2 million hires for August, while showing a revision to July’s separate estimates. Tracking periods and measures carefully makes revisions legible: they refine the statistical record rather than announce that the economy changed retroactively. Bureau of Labor Statistics
Sources & further reading
- September 2026 employment report ↗Bureau of Labor Statistics
- August 2026 job openings and labor turnover ↗Bureau of Labor Statistics
