How the unemployment rate can fall without new jobs

A falling unemployment rate can accompany job growth. It can also reflect a change in who is looking for work. The reason is in the denominator. The U.S. unemployment rate divides the number of unemployed people by the labor force, which includes people who are working and people counted as unemployed.

See who goes into each group

In the U.S. Current Population Survey, a person without work is generally counted as unemployed if available for work and actively looked for a job in the four weeks ending with the survey’s reference week. People on temporary layoff who expect to be recalled can qualify without a search. The Bureau of Labor Statistics separates people outside the labor force, a group that can include retirees, students, carers and people who want work but did not actively search in the relevant period.

Imagine 100 people in the survey population. Sixty have jobs, five are unemployed and 35 are outside the labor force. The unemployment rate is 5 divided by 65, about 7.69%. The employment-to-population ratio is 60 divided by 100, or 60%.

Move one person without adding a job

Suppose one unemployed person stops actively looking long enough to move outside the labor force and has no job to return to. The example now has 60 employed, four unemployed and 36 outside the labor force. The unemployment rate drops to 4 divided by 64, or 6.25%. The employment-to-population ratio stays at 60%.

This fictional comparison isolates one change. A real monthly release reflects many people moving into work, losing jobs, entering the labor force and leaving it. A change in the rate is a reason to look at the underlying counts.

Read three numbers together

• The unemployment rate asks what share of the labor force is unemployed.
• The participation rate asks what share of the population is working or counted as unemployed.
• The employment-to-population ratio asks what share of the population is employed.

Keep the geography and reporting period consistent. Unemployment insurance claims measure a different set of events, so they are a separate clue. Start with the survey counts behind the headline and then ask whether more people were employed, fewer were looking, or both happened at once.