Government Payrolls

Government Payrolls is the government part of United States payroll employment, covering civilian jobs on employer payrolls. It comes from Current Employment Statistics (CES), an employer payroll survey, in Table B-1 of the Employment Situation, the monthly jobs report. It covers civilian federal, state, and local employees paid for the pay period including the 12th, with postal workers counted under federal.

The survey draws its sample from unemployment insurance tax records, covering about 121,000 agencies and 631,000 worksites. That is about one-third of payroll jobs, while government data come from full payroll counts with no probability sample.

It excludes military staff, four excluded intelligence agencies, farm workers, the self-employed, household workers, and unpaid leave. Paid leave counts, and a worker paid for any part of the pay period including the 12th is counted. Hours and earnings are published for the private sector only. Arabic financial coverage calls this total the government sector.

Responses cover the pay period including the 12th, and publication is embargoed until 8:30 a.m. ET. The report normally lands on the first Friday after the reference month, with an advance schedule posted. Dates can shift around federal holidays or after extraordinary events.

The same release carries household results such as unemployment and participation, plus establishment results for jobs, hours, and earnings. State data follow 2 to 3 weeks later. Each estimate starts preliminary and is revised twice as more payrolls arrive, with seasonal factors for the prior two months recalculated. March estimates are re-anchored yearly to near-universe tax records, with a 5-year revision of adjusted data.

Government Payrolls sums 3 parts with no fixed weights, so each part moves the total by its size and hiring.

• Federal: civilian federal payrolls, split into postal and non-postal parts, and the smallest of the 3 parts.
• State: state payrolls, split into education and non-education parts, with strong school-year seasonality.
• Local: local payrolls, split into education and non-education parts, and the largest of the 3 parts, led by local education.

The September 2024 release, published October 4, 2024, put payrolls up 254,000 against a 12-month 203,000 average, with government adding 31,000 against a 45,000 average. Local added 16,000 and state added 13,000, both still trending up, while federal stayed nearly flat with 2,000, so government trailed rather than drove growth. Local education leapt from 6,997,400 in July to 8,101,700 in September without adjustment, against 8,108,200 to 8,139,700 after adjustment. That gap shows the school-calendar trap, as education falls about 20% at each spring term end, while Hurricane Francine left no clear payroll effect.

1. Matched sample: Keep only units reporting payrolls for both the prior and current month, dropping closed units and nonrespondents.
2. Atypical correction: Remove units with abnormal jumps from the base, compute the estimate, then add the unit back so it represents only itself.
3. Weighted link-relative: Multiply the prior estimate by the payroll ratio of current to prior reports, adding nothing for government births and deaths. For example, matched payrolls of 8,450,000 rising to 8,461,200 give a ratio of 1.0013254, lifting 23,390 thousand to 23,421 thousand, up 31,000. Certainty units carry weight 1, meaning each represents only itself.
4. Birth-death: Add nothing for government because the model for firm openings and closings covers private industries only.
5. Aggregation: Sum detailed industry cells upward, so federal plus state plus local equals the government total. For example, 2,999 plus 5,483 plus 14,939 equals 23,421 thousand.
6. Seasonal adjustment and benchmark: Apply concurrent seasonal adjustment with correction for the 4-week versus 5-week effect using X-13ARIMA-SEATS, then re-anchor March to tax counts yearly.