U-3 vs. U-6 Unemployment: What the Rates Count

The U.S. Bureau of Labor Statistics (BLS) publishes six alternative measures of labor underutilization, labeled U-1 through U-6. U-3 is the familiar headline unemployment rate. U-6 is the broadest measure in that series. They are not competing estimates of one identically defined group: each rate draws a different boundary around people who are unemployed or whose available work is limited.

The measures come from the Current Population Survey (CPS), a monthly survey of households. The CPS classifies people by their activity during a reference week and their recent job-search behavior. It does not define unemployment by whether someone receives unemployment-insurance benefits. A person can meet the survey definition without receiving benefits, and benefit eligibility rules do not decide whether the person is counted in U-3. See the BLS [CPS concepts and definitions]({source:blsCpsConcepts}) and its guide to [how unemployment is measured]({source:blsGovernmentMeasuresUnemployment}).

This article focuses on the U.S. definitions. Other countries may use different survey methods, age limits, labels, and rules. A U-6 rate should therefore be read with its country and statistical definition attached, not treated as a universal unemployment formula.

The U-3 formula is:

U-3 = unemployed people ÷ civilian labor force × 100

The civilian labor force is the sum of employed people and unemployed people in the covered civilian population. For the CPS, the underlying civilian noninstitutional population is generally age 16 and older; it excludes people on active duty in the armed forces and people living in institutions. That larger population is not the U-3 denominator. Someone enters the labor force by being classified as employed or unemployed.

In broad terms, a person is unemployed in the CPS if they had no job during the reference week, were available for work, and actively looked for work during the preceding four weeks. A temporary layoff with an expectation of recall is a defined exception to the active-search requirement. A person who wants a job but did not search recently may be outside both the employed and unemployed counts, depending on the detailed classification. The [BLS CPS guide]({source:blsGovernmentMeasuresUnemployment}) explains these categories.

This denominator matters. Dividing by the whole adult population would answer a different question. The labor-force participation rate uses the civilian labor force divided by the civilian noninstitutional population. The employment-population ratio uses employed people divided by that same broader population. Those measures help provide context when U-3 changes.

The BLS sequence adds groups in stages:

Discouraged workers are a subset of marginally attached people. A marginally attached person wants a job, was available to work during the reference week (unless temporarily ill), and looked for work at some point in the previous 12 months—or since the end of their last job if it ended within those 12 months—but did not search during the most recent four weeks. A discouraged worker meets those conditions and gives a labor-market reason for not searching, such as believing no suitable work is available. U-5 includes more than the discouraged subgroup; calling every marginally attached person “discouraged” loses that distinction.

U-6 adds one more group: people who are employed part time for economic reasons. These people are already employed and already in the labor force. They are added to the numerator as a form of underutilized labor, but not added again to the denominator. The category generally covers people who worked 1–34 actual hours in the reference week for an economic reason, such as slack work or inability to find full-time work. The BLS definition usually also requires that they want and are available for full-time work. It does not include everyone who voluntarily prefers a part-time schedule.

The full [BLS definition of the alternative measures]({source:blsCpsConcepts}) lists the exact groups and formulas. U-4, U-5, and U-6 are successive statistical definitions, not separate surveys of unrelated populations.

Suppose an invented CPS-like example has a civilian noninstitutional population of 1,000 people, of whom 600 are in the labor force: 576 employed and 24 unemployed. Outside the labor force, 18 people are marginally attached, including 7 discouraged workers. Of the 576 employed people, 12 are part time for economic reasons.

The population of 1,000 is included to show the distinction between denominators. It is not used directly in the U-3 through U-6 formulas.

Every count and rate in this table is hypothetical, chosen only to demonstrate the definitions. They are not current BLS readings, estimates, or a forecast. The arithmetic also shows why U-6 is not calculated by simply dividing its expanded numerator by the original U-3 denominator: marginally attached people enter its denominator too.

In this example, the 1,000-person civilian noninstitutional population makes the labor-force participation rate 600 ÷ 1,000 = 60%, and the employment-population ratio 576 ÷ 1,000 = 57.6%. These are different statistics with a different denominator from U-3. If the same 24 unemployed people were counted against 1,000 adults, the result would be 2.4%, but that would not be the official U-3 definition.

The U-6 rate is about 4.7 percentage points above U-3 here. That gap cannot be read as “4.7% discouraged” or as the number of people who need jobs. It reflects marginal attachment, economic part-time work, and the larger denominator together. The gap can change because any of those pieces changes.

A lower U-3 rate can result when unemployed people find jobs. It can also occur when people classified as unemployed stop meeting the survey’s recent-search criteria and leave the labor force. In the second case, the unemployed count and labor-force denominator can both fall, even though the people did not move directly into employment. That is one reason to read U-3 beside labor-force participation and the employment-population ratio rather than in isolation.

The same arithmetic applies to increases. A person entering the labor force and actively searching may be counted as unemployed before finding a job, temporarily lifting U-3 even as participation rises. A single rate does not show those transitions. The CPS provides several related measures, but they still summarize populations and do not follow every worker’s complete history.

Changes in U-6 can add another dimension. If some employed people take fewer hours than they want because of economic conditions, U-6 may rise even when they remain employed and do not change U-3. If marginally attached people begin actively searching, they may move into the labor force and be classified as unemployed before securing work. These examples illustrate how classifications can shift a rate; they do not establish the cause of a particular monthly move.

U-6 is useful when the question is broader than “What share of the labor force is unemployed under the official definition?” It includes two groups that U-3 omits: people marginally attached to the labor force, and people who are employed but working part time for economic reasons. A larger U-6 reading relative to U-3 signals that those additional groups matter in the measured labor-market picture.

It is not a complete measure of hardship, job quality, or unmet demand for work. It does not tell you each person’s income, preferred hours, family circumstances, occupation, location, or ability to find a better match. It also excludes people who want work but do not meet the marginal-attachment criteria, and it does not treat all part-time work as involuntary. Someone can be economically secure and work part time by choice; someone else may face serious constraints while still meeting the U-3 definition.

U-6 is not a replacement for U-3, and a higher rate is not automatically the “true” unemployment rate. The two answer different questions. For a national headline, U-3 is the standard rate. For a broader look at labor underutilization, U-6 can add context. Neither number alone gives a complete account of the labor market.

The Employment Situation release presents the alternative measures in [Table A-15]({source:blsAlternativeMeasuresTable}), including seasonally adjusted and not seasonally adjusted series. Match the adjustment status before comparing values: seasonally adjusted with seasonally adjusted, or not seasonally adjusted with not seasonally adjusted. Seasonal adjustment attempts to account for recurring calendar patterns; it does not remove every unusual event or uncertainty.

Also check the reference month, the exact series, and whether a number is a level, a month-to-month change, or a year-over-year comparison. A percentage-point difference between rates is not the same as a percent change in one rate. For example, a move from 4% to 5% is an increase of 1 percentage point and 25% relative to the starting rate.

CPS rates are survey estimates, not a census of every person. BLS applies survey weights and population controls, and estimates can be affected by sampling and nonsampling error. Small month-to-month movements may not represent a clear underlying change. BLS re-estimates seasonal factors annually, which can revise recent seasonally adjusted history. Separately, annual population-control updates enter the estimates in January and can shift population and labor-force levels. The [BLS Handbook of Methods for the CPS]({source:blsCpsHandbook}) describes the survey design and estimation process. When comparing a historical chart with a release, note the data vintage as well as the rate.

Do not substitute unemployment-insurance claims for U-3 or U-6. Claims count benefit filings under program and state rules; CPS unemployment is based on household survey classification. They can move differently because their populations, definitions, and data collection differ.

When a release headline says unemployment rose or fell, first identify whether it is U-3 and which month and adjustment status it uses. Then compare U-6 to see whether marginal attachment or economic part-time work changes the broader measure. Check U-4 and U-5 if the release table helps locate whether discouraged or other marginally attached workers are contributing to the gap.

Next, look at participation and the employment-population ratio. These do not explain every move, but they help distinguish a change in unemployment from a change in how many people are in or out of the labor force. For interpretation, state the definitions and the period instead of describing a higher U-6 as proof of a single cause.

Finally, separate measurement from forecasting. These indicators describe survey classifications for a reference period. They do not guarantee what the next report will show or, by themselves, establish an investment outcome. For adjacent U.S. macro measures, see the guide to nominal and real GDP and the comparison of CPI, PCE, and the GDP deflator. The jobs-report and macro-event guide addresses release context separately; it is not a substitute for the BLS definitions.

The most accurate short description is simple: U-3 measures unemployment within the civilian labor force, while U-6 adds marginal attachment and economic part-time work under a broader formula. Reading both alongside labor-force participation and employment helps preserve what each statistic can—and cannot—say.