Payroll vs. Household Survey: A Methods Guide
A guide to what the BLS methods page covers: the payroll survey's scope, its sample rotation, and the household survey it is compared against.
The BLS methods page is a documentation hub, not an analysis piece. Its job is to help you find the methodology behind the monthly Employment Situation numbers and the measurement issues attached to them, including the birth–death model, population controls, and comparisons between the two surveys (c01, c02). If you are reading the monthly report and want to know what the headline numbers actually count, this page points you to the underlying documentation.
What the payroll survey counts
The Current Employment Statistics (CES) program—also called the payroll or establishment survey—is a monthly survey of individual worksites from businesses and government agencies, drawn from a sampling frame of unemployment insurance tax accounts across the United States (c05). From that sample it publishes employment, hours, and earnings estimates at national, state, and metropolitan levels, down to detailed industries (c06).
The single most important definitional point sits here: CES employment series estimate nonfarm wage and salary jobs, not employed persons. An individual holding two jobs is counted twice (c09). The series also excludes employees in agriculture, private households, and the self-employed (c10). So when a payroll number is described as "jobs added," it is a count of positions at surveyed worksites, not a count of people working.
On hours and earnings, CES historically covered only production and nonsupervisory employees. Series covering all employees' hours and earnings were officially added on February 5, 2010, with estimates beginning in March 2006 (c07, c08).
The comparison the page links to
The page organizes its material into three tracks: payroll survey methods, household survey methods, and a direct comparison of payroll and household survey employment (c04). It also links to a separate set of frequently asked questions about employment and unemployment estimates (c03).
A quick orientation, based only on what the payroll side documents:
| Feature | Payroll survey (as documented) |
|---|---|
| Unit surveyed | Individual worksites from businesses and government agencies (c05) |
| Sampling frame | Unemployment insurance tax accounts nationwide (c05) |
| What it estimates | Nonfarm wage and salary jobs, not employed persons (c09) |
| Double counting | An individual with two jobs is counted twice (c09) |
| Excluded workers | Agriculture, private households, self-employed (c10) |
Sample design and rotation
The current CES sample design has been in place since 2003 and was developed in consultation with survey-design experts from universities and other leading statistical agencies (c16). The entire sample is redrawn annually, and a supplemental sample of new business births is selected midway through the year (c17). About one-third of the sample is rotated out each year and replaced with newly selected businesses (c18).
That rotation schedule is worth knowing before you treat month-to-month swings as pure signal. The sample itself is partly replaced every year, which is separate—per the documentation—from the birth–death model the page flags as a measurement topic (c02). The page does not, in these claims, explain how the birth–death adjustment is calculated.
Why the data matter to users
The BLS documents why CES gets attention: a wide array of public and private policy makers use it because it is one of the earliest indicators of economic conditions each month (c13). Major users include many government agencies and entities, financial markets in the United States and around the world, and business and academic analysts, researchers, and forecasters (c14). For anyone watching a monthly release, this is the stated basis for the timeliness premium—not a claim about what direction markets move.
- Before comparing a payroll figure with any other labor measure, check whether it counts jobs or persons (c09).
- Remember the exclusions: agriculture, private households, self-employed (c10).
- Keep the units straight: the series is nonfarm wage and salary jobs at surveyed worksites (c05, c09).
- Note the rotation window when reading revisions or turning points (c17, c18).
A useful mental check: if payroll employment rises while a person-based measure like the household survey diverges, the documented exclusions and double counting are the first place to look—a person with two payroll jobs adds two to CES but one worker to a person-based count. The page itself does not resolve divergences; it points to the comparison material (c04).
What the source does not say
Several things a market reader might expect are simply not in these claims. The page does not detail the household survey's methodology at the same level—it only signals that household survey methods and a payroll-versus-household comparison are available (c04). It does not quantify the birth–death model, explain how population controls are set, or give any adjustment factors. It does not state release dates, revision magnitudes, or seasonal adjustment procedure, beyond noting the series dates for all-employee hours and earnings (c07).
Most importantly for readers who watch labor data as a market input: the source does not discuss the Federal Reserve, inflation, bitcoin, or crypto markets at all. Any link between this monthly release and asset prices is outside what these claims establish. The evidence supports saying CES is one of the earliest monthly indicators that policy makers and financial markets use (c13, c14); it does not support saying how markets react.
How to use the page
Treat it as an index. Start with the payroll methods track if you want definitions of the establishment survey (c04, c11, c12). Use the household track and the comparison page when a jobs figure and a persons figure diverge. Use the FAQ link for definitional questions on employment and unemployment estimates (c03). The page's stated purpose is navigation to documentation, so the most reliable use is to confirm definitions before interpreting a number—not to explain why a number moved.
Sources
Informational only. Not financial, legal or investment advice. Check the primary sources listed above before you act on anything here.