The monthly jobs report actually uses two surveys
When the U.S. employment report is released, markets often see nonfarm payroll growth, the unemployment rate and the labor force participation rate presented together. These figures appear in the same report, but they are not all calculated from one survey.
The Bureau of Labor Statistics uses two major monthly surveys. The Current Employment Statistics survey, or CES, is commonly called the establishment or payroll survey. The Current Population Survey, or CPS, is commonly called the household survey. Both measure the labor market, but their units, coverage and estimation methods differ, so their monthly movements do not have to match.
The establishment survey mainly measures payroll jobs
The headline change in nonfarm payroll employment comes primarily from the CES establishment survey. Businesses and government agencies report payroll information that BLS uses to estimate nonfarm payroll jobs, hours and earnings.
The key concept is closer to payroll jobs than to a deduplicated count of individuals. If one person works at two different establishments and appears on both payrolls, that worker is counted at each establishment. A headline such as “200,000 nonfarm jobs added” therefore means that payroll jobs increased by roughly 200,000 on net; it does not necessarily mean 200,000 previously jobless people found work.
The establishment survey is also a sample-based estimate and is revised as additional information becomes available. For more background, see How a Single Annual Revision Can Rewrite a Year's Employment Story.
The household survey mainly measures people
The CPS household survey asks a sample of households about the labor-force status of people age 16 and older during the reference period. Eligible people are classified as employed, unemployed or not in the labor force. Here, the core unit is the person.
Someone holding two or more jobs is generally counted only once in the household survey's total employment measure. The household survey also has broader coverage in several areas, including unincorporated self-employed workers, agricultural workers, unpaid family workers and private household workers who are outside the payroll survey's main scope.
The national unemployment rate, labor force participation rate and household measure of employment are produced from this survey. That is why the unemployment rate should not be treated as if it were calculated from the same business sample that produces headline payroll growth.
Why can the two employment measures move in different directions?
Suppose businesses add many part-time payroll jobs and many of those jobs are taken by people who already have another job. The establishment survey can record additional payroll positions, while the household survey's number of employed people may change much less because those workers were already classified as employed.
The reverse can also happen. If a multiple jobholder loses one job but keeps another, a payroll job can disappear from the CES measure while the person remains employed in the CPS measure. Different sample sizes, coverage, reference periods and estimation procedures can create additional short-term divergence.
How can payroll growth be strong while unemployment rises?
Once the two-survey structure is clear, apparently conflicting combinations become easier to understand. Payroll jobs can rise in the establishment survey while employment, unemployment or labor-force participation in the household survey moves differently, allowing the unemployment rate to rise at the same time.
That does not automatically mean one survey is wrong. They answer different questions. CES is especially useful for payroll-job and industry trends, while CPS shows how many people are employed, unemployed or participating in the labor force. Industry breadth is another useful layer; see Are Non-Farm Payrolls Still Strong When Job Gains Are Concentrated in Few Sectors?
A nonfarm payroll report is more than one number
The headline payroll change describes the net movement in payroll jobs. The unemployment rate, participation rate and household employment measure provide a different view based on people and labor-force status.
A clearer way to read the jobs report is to identify whether each statistic comes from CES or CPS first, then compare jobs, employed people, unemployed people and the size of the labor force. Using both surveys together gives a much fuller explanation of why a single employment report can contain signals that look strong in one place and weak in another.