Why Unemployment Rates Differ Across Groups

What this means

The national unemployment rate is an average, and an average is a summary of a distribution, not a description of anyone in it. When you break the national figure apart by age, by race and ethnicity, and by gender, you find rates that differ from each other and from the headline number. Economists call this disaggregatedBreaking a single overall statistic apart into rates for specific subgroups of the population. data, and the Bureau of Labor Statistics publishes it every month.

Several distinct forces contribute to these differences, and they are not mutually exclusive. Sorting them out is the analytical work of this lesson.

Work experience matters mechanically. Employers use a track record to reduce their uncertainty about a candidate, and workers new to the labor market do not have one. This is a large part of why teenage and young-adult unemployment rates sit above rates for prime-age workers in essentially every period. Younger workers are also more likely to be searching for a first job or moving between short-term jobs, which puts them in the counted-as-unemployed category more often.

Education, training, and human capitalEducation, training, experience, and skills that make a worker more productive. matter next. Unemployment rates fall as educational attainment rises. But access to education and training is itself unequal across groups, so differences in attainment translate into differences in unemployment even when nothing else is at work.

Industry and occupation composition matters and is easy to overlook. Groups are not evenly distributed across the economy. If one group is concentrated in construction or hospitality, which contract sharply in downturns, that group's unemployment rate will swing harder in a recession than a group concentrated in health care or public administration, without any difference in individual behavior or ability.

Caregiving responsibilities matter. Time spent out of paid work to care for children or older relatives interrupts continuous experience and shapes which jobs a worker can accept, and these responsibilities have historically fallen unevenly by gender. This affects hours, occupation choice, and reentry into job search.

Finally, discriminationTreating otherwise similar workers differently in hiring, pay, or promotion because of a characteristic such as race, ethnicity, gender, or age. is a documented factor. Researchers test for it directly, most rigorously through audit and correspondence studies that send employers otherwise identical applications differing only in a signal such as the name on the resume, then measure callback rates. Studies of this design have found differences in callback rates, which is evidence of unequal treatment of equally qualified applicants. Discrimination is one factor among the several listed here, and careful analysis attempts to measure how much of a gap each factor explains rather than assuming any single one accounts for all of it.

Two cautions about reasoning here. First, group averages tell you nothing about any individual person. Second, a gap that remains after controlling for education, experience, and industry is often called unexplained rather than proven to come from a specific cause, and honest analysis says so.

Why it matters

If you only ever see the national rate, you will systematically misjudge the labor market that you or someone you know is actually entering. A young person searching for a first job faces a different market than the headline suggests, and knowing that in advance changes how long you expect a search to take and how you prepare for it.

It matters for policy too. Job training programs, apprenticeships, childcare support, anti-discrimination enforcement, and youth employment initiatives are all aimed at different factors on the list above. Deciding which is worth funding requires knowing which factors are driving a given gap, which is an empirical question that can only be settled with data.

Try it

1. Before touching any data, write down your prediction. Rank these groups from highest to lowest expected unemployment rate: workers aged 16 to 19, workers aged 25 to 54, workers with less than a high school diploma, and workers with a bachelor's degree or higher. Seal your predictions; you will check them at the end.
2. Go to bls.gov and open the most recent Employment Situation news release. Record the headline national unemployment rate along with the month and year of the data. Every figure you collect from here on must carry that same month and year.
3. Collect rates by age. Find the current unemployment rate for teenagers aged 16 to 19 and for workers aged 25 to 54. Record both with the source table.
4. Collect rates by race and ethnicity. Using the appropriate BLS table, record the current unemployment rate for each group the table reports separately. Copy the categories exactly as the BLS labels them rather than inventing your own groupings.
5. Collect rates by gender. Record the current rates for adult men and adult women as the BLS reports them.
6. Collect rates by educational attainment for workers 25 and over, from less than a high school diploma through bachelor's degree and higher.
7. Build one table containing everything, with a column for the figure, a column for the exact BLS category label, and a column for the source table and date. Then compute the gap in percentage points between the highest and lowest rate in each of your three groupings.
8. Now explain rather than describe. For each gap you measured, write a paragraph proposing which factors from the lesson could contribute: work experience, education and training access, industry and occupation composition, caregiving responsibilities, or discrimination. For each factor you name, state what additional evidence would be needed to confirm it is operating. Where your data cannot distinguish among several possible explanations, say so explicitly.
9. Test one explanation against evidence. Use the educational attainment table to ask how much of one gap could plausibly be accounted for by differences in attainment. Explain in writing why comparing overall rates cannot settle this, and what a study would have to hold constant to isolate a single factor.
10. Return to your step 1 predictions. Note which were right, which were wrong, and what specifically in the data corrected you.

Check yourself