College wage premiums vary by group—and the gap has narrowed since the pandemic

What researchers mean by the college wage premium

The college wage premium is the percentage by which workers with a bachelor’s degree (or more) out-earn workers with only a high school diploma. It is one of the most cited statistics in debates over whether college still pays off—but the headline number hides wide variation by race, ethnicity, gender, and geography.

In a widely cited San Francisco Fed Economic Letter (FRBSF Economic Letter 2023-22, August 2023), Leila Bengali, Marcus Sander, Robert G. Valletta, Cindy Zhao report that the national premium peaked near 79% in the mid-2010s before edging down to about 75% by 2022—a decline of 4 percentage points. The national college wage premium edged down after the pandemic; researchers attribute much of the flattening to faster wage gains for high school graduates rather than slower pay growth for college graduates, especially among Black and Hispanic workers.

Methodology note: SF Fed estimates use CPS microdata with regression adjustments. EDsmart figures below use published Education Pays and ACS medians — comparable in direction but not identical in level.

National snapshot: Education Pays and Census ACS

Education Pays 2026 reports median pretax earnings for full-time, year-round workers age 25 and older. In 2024, bachelor’s degree holders earned $81,800 versus $50,600 for high school graduates—a 62% premium and a $31,200 annual gap.

Census American Community Survey table B20004 (2024) shows a similar pattern for workers with earnings in the past 12 months: bachelor’s median $70,044 vs high school $40,153 (74% premium). ACS combines 'some college' and associate's degree in one row; it does not isolate 'some college, no degree' like CPS ASEC. Use NSC/BLS for SCNC-specific population counts.

See also our article on how the dollar gap widened while the percent premium held near 62% across Education Pays editions since 2004.

Premiums differ sharply by race and ethnicity

Education Pays 2026 Figure 2.5 breaks out median earnings and bachelor’s-to-high-school ratios for Asian, White, Black, and Hispanic workers. The premium is largest for Asian men (95%) and narrowest for Black and Hispanic men (56% each). Among women, premiums range from 58% (Hispanic) to 71% (White).

These cross-sectional gaps reflect differences in majors, graduate degrees, occupations, and labor-market experience—not just the label “bachelor’s degree.” The SF Fed authors note that Asian workers’ higher premium partly reflects postgraduate attainment and field of study. Black and Hispanic workers see smaller premiums in part because tight labor markets since 2020 lifted high school graduate wages relatively faster.

For broader equity context, see college earnings gaps by gender and race and the parent report’s equity section.

Has the premium fallen? SF Fed CPS trends vs Education Pays

Published Education Pays editions show the national percent premium in a relatively narrow band—62% to 67%—from 2004 through 2024, even as the dollar gap widened with inflation and nominal wage growth.

The SF Fed authors plot a regression-adjusted CPS premium that runs higher than Education Pays medians and shows a clearer rise-and-fall: about 68% in 2000, a peak near 79% in the mid-2010s, and about 75% in 2022. The chart below overlays our digitization of their Figure 1 with Education Pays income-year markers (open circles).

Chart note: Annual values transcribed from Figures 1–3 in el2023-22.pdf (rendered PNGs). Anchor years match values stated in the letter text (e.g., 68% in 2000, ~79% peak mid-2010s, 75% in 2022). Intermediate years are visual estimates from the published charts, rounded to the nearest whole percent.

Premiums also diverge by race and ethnicity in the SF Fed series. Asian workers see the largest gap (above 120% at the 2019 peak in Figure 2); Black, Hispanic, and White workers cluster nearer 70–80%, with Black and Hispanic premiums edging down since the mid-2010s.

Why did the gap narrow? Figure 3 in the letter compares compound annual real wage growth before and after 2011. For Black and Hispanic workers, high school graduate wages accelerated more than college graduate wages after 2011—a reversal of the pre-2011 pattern—consistent with tight labor markets lifting pay at the bottom of the education distribution.

Short-term premiums can mislead enrollment decisions. Research cited in the SF Fed letter notes that the college premium roughly doubles over a typical career as experience and advancement compound—so a temporarily narrower gap does not necessarily mean college stopped paying off.

Geographic variation in college outcomes

Worker-level premiums from CPS differ from institution-level earnings reported in the College Scorecard. EDsmart Data’s state profiles aggregate Scorecard medians for bachelor’s-predominant campuses. National median earnings 10 years after entry are $53,734, but state medians span a wide range.

Note: College Scorecard median institution-level earnings 10 years after entry among bachelor's-predominant campuses in each state — not the same as CPS worker medians.

Explore state-level tuition, debt, completion, and earnings on our state higher education profiles.

What this means for students and policymakers

• College still pays on average, but the payoff is not uniform. Group-level premiums of 56% to 95% mean enrollment advice should be tailored—not one-size-fits-all.
• Tight labor markets compress premiums by lifting wages at the bottom of the education distribution. That is good news for high school graduates in the short run, but it does not eliminate the lifetime earnings advantage documented across decades of CPS research.
• Program and completion matter. Institution-level Scorecard data show that where you study and whether you finish strongly shape outcomes—especially in states at the bottom of the earnings distribution.
• Compare sources carefully. CPS worker medians, ACS household surveys, and Scorecard cohort earnings answer different questions. Our guide to conflicting college earnings numbers explains the distinctions.