Payroll employment from the BLS establishment survey, with calendar-aligned cross-year comparison: levels for trend, year-to-date gains for the pace of hiring against prior years, YoY for the cycle. The market never trades the level — it trades the surprise and the direction of revisions.
This view presents U.S. payroll employment from the Bureau of Labor Statistics (Current Employment Statistics). Choose one or more series, a presentation style, and multiple years to compare on the same calendar axis. Values are in thousands of persons.
Styles:
Payroll data are revised regularly; the backend refreshes the full history on updates so revisions are reflected here.
Why it matters: The revisions are the signal: the two monthly revisions (driven by late survey responses) skew systematically negative when the cycle is turning, and the annual benchmark to QCEW unemployment-insurance records — near-census coverage — corrects the birth-death model's guess at net new business formation in one shot, as the famous preliminary -818k benchmark revision of 2024 showed. A run of same-direction revisions therefore carries more information than any single print. Note also that CES counts jobs, not employed persons (multiple job holders count twice), and it regularly diverges from the household survey (see the unemployment page) near turning points — that divergence is itself worth watching. Cross-checking against daily, never-revised withheld taxes is exactly why these three pages sit together on this site.