Non Farm Payrolls

Total nonfarm payroll employment is the jobs headline of the Employment Situation release from the Bureau of Labor Statistics (BLS), the federal labor agency. It comes from the establishment survey (CES-N, a monthly survey of employer payrolls) drawn mainly from unemployment insurance tax records, running monthly since October 1915. It covers payrolls paid for any part of the period including the 12th day, from about 122,000 employers and 666,000 worksites, about 1/3 of jobs. It counts jobs, not people, excludes farm, self-employed, household, military, and intelligence workers, while the household survey (a survey of homes) gives unemployment.

Employers report the pay period including the 12th day, and the release normally follows on the 1st Friday after at 8:30 a.m. ET under embargo. Each release also shows household unemployment results, private-sector pay and hours, diffusion gauges of breadth with 50 as balance, and tables B-1 to B-9. Each month's initial estimate is preliminary (built on incomplete returns) and turns final only after 2 monthly revisions for late reports. Each February benchmarks March to universe counts (full job records), spreads the gap over the prior 11 months, and revises 5 years of adjusted data.

Total change sums industry cells, and each cell adds its estimated jobs with no fixed weights.

• Mining, logging and construction: covers extractive sites and building projects and their workers.
• Manufacturing, durable and nondurable: covers factories making long-lasting goods and consumable goods.
• Trade, transport, warehousing and utilities: covers wholesalers, retailers, movers, warehouses, and utility providers.
• Information and financial activities: covers media, telecom, data services, banks, insurers, and property firms.
• Professional, education, health, leisure and other services: covers offices, schools, clinics, hotels, restaurants, and repair shops.
• Government, federal, state and local: sums near-universe payroll counts, not a random sample.

The February 3, 2023 report showed January payrolls up 517,000, above the 2022 average gain of 401,000. Unemployment held at 3.4% with 5.7 million jobless, while leisure and hospitality added 128,000 with 99,000 in food services. It still stood 495,000 (2.9%) below its February 2020 level, while professional services added 82,000 and health care 58,000.

Government added 74,000, including 35,000 post-strike university returns, while hourly pay rose 0.3% to $33.03 (+4.4% over 12 months) and hours rose 0.3 to 34.7. November 2022 gained 34,000 in revision and December 2022 gained 37,000, together 71,000. The benchmark added 568,000 (0.3%) as NAICS 2022 moved about 10% of jobs.

1. Cell link relative: Multiply the prior month's cell estimate by the weighted matched-sample ratio (firms reporting both months), then add a birth-death forecast. For example, weighted sample jobs of 820,000 rising to 828,200 give 1.01. An estimate of 1,000,000 times 1.01 makes 1,010,000, plus 4,000 gives 1,014,000, a gain of 14,000.
2. Birth-death adjustment: Dead firms' jobs stay in the estimate to stand in for unborn firms. An ARIMA model (a time-series forecasting tool) then forecasts the leftover net gap from universe data, updated quarterly.
3. Other series: Women, production-line and nonsupervisory (non-managerial) workers, hours, and pay use ratio or smoothed estimates favoring the prior month by about 90%. Average weekly hours divide total hours by jobs; average hourly earnings divide total payroll by total hours.
4. Aggregation: Estimates sum bottom-up from detailed industry cells through 17 sectors and 10 supersectors into total private and then total nonfarm.
5. Benchmark: Each February, March employment is replaced by QCEW universe counts and the gap is spread over the prior 11 months. Later months are re-estimated and 5 years of adjusted data are revised.
6. Seasonal adjustment: Monthly seasonal adjustment uses Census X-13ARIMA-SEATS software with yearly models, fresh factors for 3 months, and calendar corrections. Corrections cover week length, pay-period length, election poll workers, and floating holidays.