College earnings gaps by gender and race, in federal data

Higher education raises medians for all groups—and gaps remain

Education Pays 2026 (U.S. Census Bureau CPS) reports higher earnings at each education level for major race, ethnicity, and gender groups. It also documents that women and many minority groups earn less than White men at the same credential level on average. Equity analysis requires both facts.

Within-college spread by gender (2024)

Figure 2.6 shows quartiles for college graduates: among women, 25% earned below $51,200 and 25% above $102,200; among men, below $62,300 and above $144,500. Twelve percent of female and 16% of male high school graduates earned more than the median for college graduates of the same sex—overlap matters for counseling, not only medians.

Race, ethnicity, and lifetime patterns

Georgetown CEW The College Payoff (2021) lifetime tables show wide dispersion within credentials by race and ethnicity. Education Pays (U.S. Census Bureau CPS) adds annual CPS detail by group in recent editions. Neither source supports using a single institution’s College Scorecard median as a race-specific graduate earnings estimate—those files are not disaggregated for that use.

Also important: Use Census-based Education Pays or CEW ACS tables for equity reporting—not institution medians alone.

Black and Hispanic adults remain underrepresented among bachelor’s degree holders relative to their share of the population, even as attainment rose in every state from 2000 to 2022. Closing those gaps requires both K–12 preparation and affordable college pathways—not only marketing four-year degrees.

Employers who require degrees for jobs that do not use college skills may widen measured gaps without changing productivity. Pay transparency and unionization also affect observed earnings at the same education level.

EDsmart: gender pay gap by education level · equity section.

Within-group inequality

Education Pays 2026 (U.S. Census Bureau CPS) shows wide quartiles within credentials. Among college graduates, the middle half of women earn between $51,200 and $102,200; for men, $62,300 to $144,500. Race and ethnicity add further dispersion in CEW lifetime tables and in CPS detail in Education Pays.

Implications for policy and counseling

Equity programs should track completion and major choice—not only enrollment. A student can enroll and still face weak financial outcomes if they leave without a credential or enter a low-paying field with high debt.

Use CEW and Education Pays (U.S. Census Bureau CPS) tables for group patterns; use program-level College Scorecard tools for school-specific debt and earnings pairing—not institution medians—for race-specific claims.

Figure 2.5 through 2.7 in Education Pays

Education Pays 2026 (U.S. Census Bureau CPS) breaks out earnings by race, ethnicity, and gender within education levels. The report documents persistent gaps while also showing higher medians at each step of the credential ladder for every group reported. Readers should cite both facts when discussing equity—progress within credentials and gaps across groups.

Occupation and major choice explain part of within-group variation; discrimination and occupational sorting explain part as well. The data cannot fully separate those mechanisms in a single table.

Intersectionality and small sample sizes

Gender and race earnings gaps are often reported separately; in reality, Black women, Latino men, and other groups experience combined disadvantages in some fields. Education Pays 2026 (U.S. Census Bureau CPS) tables break out sex and race where sample sizes allow, but cell sizes shrink for smaller groups. Do not over-interpret a single year’s gap; look for decade-long trends and field-specific splits.

Major choice and occupational segregation

Part of the earnings gap reflects concentration of women and people of color in lower-paying caring professions and part reflects pay discrimination within the same occupation. Policy responses range from pay transparency laws to targeted scholarships in high-demand STEM fields. College earnings gap articles should mention both pipeline and within-job equity.

Closing gaps without ignoring medians

Equity policy often targets completion and debt burdens for Black, Hispanic, and first-generation students. Education Pays shows higher education still raises typical earnings within each group compared with less education—so access and completion support remain economic as well as moral priorities.

Researchers distinguish level gaps (differences at the same degree) from composition gaps (different fields chosen by group). Both appear in national data; interventions differ.

State higher-education agencies increasingly publish disaggregated completion and earnings where sample sizes allow. Use those files for state policy stories; use Education Pays for national worker medians by group.

Equity reporting standards

Report both higher medians with more education and remaining gaps at the same level.

Name whether data are annual CPS or lifetime synthetic.

Avoid ranking colleges by average pay alone when discussing low-income access.

How this article fits The College Payoff, 2026

This piece is one of twenty companion articles to The College Payoff, 2026. The parent report synthesizes College Board Education Pays, Georgetown CEW, Census, BLS, College Scorecard, Federal Reserve SCF wealth tables, and EDsmart institution-level findings. Each breakout answers one question journalists and families ask when those sources are quoted side by side.

When federal sources update on different calendars, start from the parent report’s definitions section, then return here for the topic-specific FAQ. Interactive tables and state pages on EDsmart Data—tuition trends, debt-to-earnings by major, attainment by state, STEM outcomes—let you filter national benchmarks beyond what a single government table provides.

If a number here differs from a chart in the parent report, use the parent report and the underlying federal publication as the authority. Breakouts explain how to read those sources; they are not a separate dataset.