The U.S. Jobs Report: Payrolls, Unemployment and Revisions Explained

An employment headline can say that payrolls increased while the unemployment rate also rose. Both statements can be correct. The Employment Situation combines information from two different surveys, with different units and definitions. Understanding that design is more useful than treating every apparent discrepancy as an error. For households, employers and readers following U.S. markets, the report is a set of signals about labor demand and participation, not a single score.

Payroll jobs and employed people are different units

The establishment survey measures jobs on nonfarm payrolls. A person holding two payroll jobs can be counted in both. The household survey asks about people and includes employment arrangements that are not covered in the same way by the payroll survey. The unemployment rate comes from the household survey, not from subtracting payroll losses from a fixed workforce. Before comparing changes across the two surveys, identify what is being counted and whether the populations match. These conceptual differences matter even when both series are estimated accurately.

Read the unemployment rate with participation

The unemployment rate relates people classified as unemployed to the labor force. People outside the labor force are not automatically counted as unemployed. If more people begin looking for work, unemployment can temporarily rise even as employment increases. If people stop looking, the rate can fall without an equally encouraging employment story. Check labor-force participation and the employment-population ratio alongside the rate. None of these measures alone explains why a particular person entered or left the workforce; that requires additional evidence.

Understand normal revisions

Early payroll estimates use information available by a publication deadline. Later responses and other source information can change the estimate. Benchmark revisions serve another purpose by aligning payroll estimates with more comprehensive employment counts. A revision is not automatically evidence of misconduct. Keep a record of the initially published number and the latest vintage when evaluating a forecast or a past headline. Comparing a forecast made before the release with a revised series without saying so mixes information sets and can misrepresent what observers knew.