How the jobs report works: payrolls and unemployment
How the monthly jobs report is built: the employer payroll survey, the household survey, who counts as unemployed, why figures are revised and how to read both.
The jobs report is the common name for the Employment Situation, a monthly release from the Bureau of Labor Statistics (BLS) that reports how many jobs employers added or cut and what share of the labor force is unemployed. It is built from two separate surveys that measure different things, which is why its headline numbers can point in different directions in the same month.
In brief
- The establishment survey, formally Current Employment Statistics (CES), counts jobs on employer payrolls.
- The household survey, formally the Current Population Survey (CPS), counts people and produces the unemployment rate.
- Both are estimates drawn from samples, and both are revised or benchmarked.
- The BLS publishes the release date in advance on its release calendar.
What the jobs report is
The Employment Situation appears on the BLS release calendar alongside the Consumer Price Index, the Producer Price Index and other national releases. Readers usually quote two figures from it: the change in nonfarm payroll employment and the unemployment rate. The first comes from the employer survey and the second from the household survey. The report also contains industry detail, hours and earnings, and a range of labor-force measures, so a careful reading goes well beyond the two headline numbers.
The employer survey (CES)
The BLS says the Current Employment Statistics program produces detailed industry estimates of nonfarm employment, hours and earnings for workers on payrolls. Each month it surveys about 119,000 businesses and government agencies, representing about 622,000 individual worksites. The sample is drawn from unemployment insurance tax accounts, and the BLS says it covers about 30 percent of total universe employment.
Three features shape how to read it.
- It counts jobs, not people. A person with two payroll jobs is counted twice.
- It excludes some workers. Agriculture, private households and the self-employed are outside the nonfarm payroll count.
- It is an estimate. The BLS surveys a large sample and projects it to the full universe of employers.
The establishment survey is the source of the payroll change that headlines the report.
The household survey (CPS)
The Current Population Survey is a monthly survey of about 60,000 eligible households covering the civilian noninstitutional population. It measures whether people age 16 and over are employed, unemployed or not in the labor force. Because it asks about people rather than payroll records, it includes groups the employer survey misses, such as the self-employed, agricultural workers, private household workers and unpaid family workers.
Status is measured for a reference week, which is usually the calendar week, Sunday to Saturday, that includes the 12th of the month. The employer survey uses a payroll period that includes the 12th as well, so the two surveys describe a similar moment.
Who counts as unemployed
The CPS definitions are specific, and misunderstanding them is the source of most confusion about the unemployment rate.
- Employed. A person who worked at least one hour as a paid employee or in their own business or farm during the reference week. People temporarily absent from a job count as employed, and so do unpaid workers who put in at least 15 hours in a family business or farm. A person with several jobs is counted once.
- Unemployed. A person who was not employed during the reference week, was available for work and either made at least one specific active job-search effort in the prior four weeks or was on temporary layoff and expected to be recalled.
- Labor force. The employed plus the unemployed, age 16 and older.
- Not in the labor force. Anyone who is neither employed nor unemployed under those rules, for example a full-time student or a person who has stopped looking.
The unemployment rate is the number of unemployed people as a percentage of the labor force. The labor force participation rate is the labor force as a percentage of the civilian noninstitutional population age 16 and older. A person who stops searching leaves the labor force, so the unemployment rate can fall even when no one has found a job. That is one reason participation is reported alongside it.
A simple hypothetical shows the arithmetic. If a community has 100 people in the labor force and five are unemployed, the rate is 5 percent. If one of the five stops looking and leaves the labor force, the rate becomes 4 out of 99, a little over 4 percent, even though nobody gained a job. The numbers are invented for illustration only.
Why the two surveys can disagree
The surveys differ in unit, scope and method, so a gap is normal.
| Feature | Employer survey (CES) | Household survey (CPS) |
|---|---|---|
| Unit measured | Jobs on payrolls | People |
| Source | Business and government payroll records | Interviews with households |
| Self-employed | Excluded | Included |
| Multiple jobholders | Counted once per job | Counted once |
| Produces | Payroll change, hours, earnings | Unemployment rate, participation |
The BLS publishes a quick guide that sets out these differences and cautions readers that month-to-month changes in either survey are subject to sampling error. A monthly payroll gain can coexist with a flat unemployment rate, and the reverse is also possible.
Revisions and benchmarks
The first number is not the last. For the employer survey, the BLS says the first preliminary estimate comes about three weeks after the reference period, a second preliminary follows the next month and the final sample-based estimate comes two months after the initial release. Estimates are then held constant until annual benchmarking, which recalculates nearly two years of not seasonally adjusted data using March unemployment insurance records.
The household survey works differently. Seasonal factors apply to the current month immediately, and earlier months are not revised until year’s end, when the most recent five years of seasonally adjusted data are revised. Annual population control adjustments enter with January estimates, and the BLS typically does not revise historical data for them.
The practical lesson is to read a payroll change together with the revisions to the previous two months. A modest headline can hide a large revision, and the reverse.
How to read the report
- Start with the BLS release. The text and tables are the source. Secondary coverage tends to compress them.
- Check both surveys. Note the payroll change and the unemployment rate, then look at participation and the employment-population ratio.
- Look at the revisions. Compare the prior two months with what was reported before.
- Read the industry detail. The report breaks payroll employment out by industry, which shows where hiring or cutting is concentrated.
- Consider hours and earnings. Average hours and average hourly earnings help show whether employers are adding hours or workers, and whether pay is rising.
- Remember the margin of error. Sample surveys carry sampling error, which the BLS discloses in its technical notes.
Related labor-market releases
The jobs report is the most visible labor release, but not the only one. The BLS also publishes the Job Openings and Labor Turnover Survey, which measures openings, hires and separations, and state and metropolitan-area employment releases. The Employment Situation is the national monthly summary. Our economic calendar guide shows where each of these sits among the other agencies’ releases.
Common misconceptions
A rising unemployment rate always means job losses. It can also rise when more people enter the labor force and begin searching.
The payroll number counts new people employed. It counts jobs. One person holding two jobs adds two to the count.
The unemployment rate counts everyone without a job. It counts only people who meet the definition, including active job search. People who are not looking are outside the labor force.
Revisions mean the first estimate was wrong. The first estimate uses the data available at that time. Later estimates incorporate more complete responses and administrative records.
One month shows a trend. Month-to-month changes are noisy. Three-month and longer comparisons give a steadier picture, though the BLS does not prescribe any particular smoothing.
FAQ
When is the jobs report released?
The BLS posts its release dates on its release calendar, in month and list views, with the time in Eastern Time. Check that calendar for the date of the next Employment Situation.
What is the difference between the payroll number and the unemployment rate?
The payroll number comes from the employer survey and counts jobs. The unemployment rate comes from the household survey and measures the share of the labor force without work that is looking for it.
How big is each survey?
According to the BLS, the employer survey covers about 119,000 businesses and government agencies and about 622,000 worksites. The household survey covers about 60,000 eligible households.
Why are payroll numbers revised?
More survey responses arrive after the first estimate, and annual benchmarking aligns the sample estimates with unemployment insurance tax records. Revision is built into the design.
What does “seasonally adjusted” mean here?
Seasonal adjustment removes regular patterns, such as holiday hiring, so that month-to-month changes are comparable. The BLS publishes both adjusted and unadjusted data for many series.
Does the jobs report include the self-employed?
The household survey does. The employer payroll count does not.
Prepared with AI assistance from public sources and reviewed under our editorial policy. Not investment advice.
