Nonfarm Payrolls: Reading Jobs, Wages, Revisions, and Market Reactions

For educational purposes only; not investment advice. Investing may result in loss.

Direct answer

“Nonfarm payrolls” usually means the monthly change in seasonally adjusted payroll employment reported in the U.S. Bureau of Labor Statistics Employment Situation. It comes from the establishment survey and covers jobs at nonfarm businesses and government agencies. It excludes categories such as farm workers, private household workers, proprietors, and the self-employed from this payroll measure.

The headline is a change in payroll jobs, not a count of unique people who found work that month. One person can hold more than one job. The report should be read with prior-month revisions, unemployment, labor-force participation, average hourly earnings, average weekly hours, and industry detail.

CES counts employees who worked or received pay for any part of the reference pay period that includes the 12th of the month. Someone on unpaid leave or strike for the entire pay period is not counted in CES employment, while paid leave is counted. These timing rules matter when interpreting temporary disruptions.

Two surveys and the market channel

Payroll employment, earnings, hours, and industry detail primarily come from the establishment survey. Unemployment, labor-force participation, employment status, and demographic detail come from the household survey. Their populations, concepts, samples, and estimation methods differ, so monthly payroll growth and household employment can diverge without either series being automatically wrong.

In the household survey, a person is counted once even if holding several jobs. The headline U-3 unemployment rate classifies as unemployed people who had no employment, were available for work, and actively searched during the prior four weeks, with an exception for people on temporary layoff expecting recall. People without a job who do not meet those conditions are outside the labor force, not unemployed.

The first CES estimate is revised in each of the next two monthly releases as more employer responses arrive. The annual benchmark then re-anchors sample estimates primarily to near-complete unemployment-insurance records from the Quarterly Census of Employment and Wages. A preliminary benchmark estimate does not revise the official monthly history; the final benchmark is incorporated later. Seasonally adjusted history can also change when seasonal factors are re-estimated.

CES cannot immediately sample every newly opened or closed establishment, so its estimation process includes a net birth-death model. The published payroll estimate already includes that process. Subtracting a posted model amount from the headline does not produce a cleaner or official alternative estimate.

Markets compare the whole release with expectations and prices already embedded in Treasury yields, currencies, and equities. Strong jobs and wages can support consumption and earnings, but may also raise expected policy rates and discount rates. Weak data can lower rate expectations, yet hurt stocks if it signals a sharper earnings slowdown.

Surprise and revision example

Assume consensus expects:

Jobs, revisions, and wages all appear stronger than expected. Treasury yields might rise if investors infer a tighter policy path. Long-duration growth stocks may fall as discount rates rise, even though stronger employment can support revenue.

Now suppose payrolls add only 90,000, unemployment rises from 4.0% to 4.2%, and participation rises from 62.5% to 62.8%. Part of the unemployment increase may reflect more people entering the labor force and looking for work, not only layoffs. The components change the interpretation.

Do not simply add the current-month surprise and prior-month revisions as though they were one month’s job creation: they refer to different reference months. Also keep month-over-month and year-over-year wage rates distinct, state whether a rate has been annualized, and check average weekly hours. Average hourly earnings cover private nonfarm payroll employees and can move because the industry and worker mix changed, not only because individuals received raises.

Reading checklist

• Compare the current payroll change with consensus and prior-month revisions.
• Review three- and six-month averages rather than declaring a trend from one release.
• Separate establishment-survey payrolls from household-survey unemployment and participation.
• Check wage growth, weekly hours, and aggregate weekly payrolls, not wages alone.
• Identify which industries added jobs and whether hiring is broad or concentrated.
• Distinguish nominal wage growth from real purchasing-power growth after inflation.
• Note strikes, weather, seasonal adjustment, population-control changes, and benchmark revisions.
• Use the official BLS release calendar; publication dates can change around holidays.
• Record the data vintage: first preliminary, second preliminary, third estimate, preliminary benchmark indication, or benchmarked history.
• Do not remove the net birth-death adjustment by hand; evaluate the published estimate, revision history, benchmark evidence, and uncertainty together.
• Check whether a quoted employment level or change is seasonally adjusted; do not compare a seasonally adjusted monthly change with an unadjusted level as if they shared one basis.

Average wages also have composition effects. If many lower-paid jobs disappear, the average can rise even without broad individual pay increases. Payroll growth needed to keep unemployment stable also changes with population and labor-force growth, so fixed historical thresholds age poorly.

Common misconceptions

• “Positive payroll growth is always bullish for stocks.” Rates and valuation can offset the earnings effect.
• “Payroll jobs equal employed people.” The establishment survey counts jobs; multiple jobholders matter.
• “Unemployment and payrolls come from one survey.” They come from separate surveys.
• “The first estimate is final.” Monthly and annual revisions are integral to the data.
• “A weak month proves recession.” Sampling error, temporary disruptions, and revisions require a broader trend.
• “The same payroll number has the same market meaning.” Consensus, wages, revisions, positioning, inflation, and policy regime change the reaction.

Related topics

• Monetary Policy Transmission
• Interest Rate Impact
• Market Index

Authoritative sources

• Employment Situation Technical Note - U.S. Bureau of Labor Statistics
• Employment Situation Quick Guide - U.S. Bureau of Labor Statistics
• CES Frequently Asked Questions - U.S. Bureau of Labor Statistics
• Nonfarm Payroll Employment Revisions - U.S. Bureau of Labor Statistics
• Monetary Policy Goals and Strategy - Board of Governors of the Federal Reserve System