Is college worth it? Medians and ROI answer different questions

The “worth it” debate is really two debates

In 2026, public skepticism toward college often cites “low ROI” studies. At the same time, Education Pays 2026 documents a $81,800 median for bachelor’s graduates—far above high school levels. Both can be true because median earnings and return on investment answer different questions.

Median pay describes the middle of the earnings distribution for a defined group in a given year or horizon. ROI subtracts tuition, fees, living costs, and forgone wages while in school, and sometimes weights by the chance of finishing. A school can show solid median cohort earnings while specific programs still fail a cost-benefit test—and vice versa.

What Education Pays shows about pay—not profit

Education Pays is explicit: observed differences in outcomes are not all caused by education alone. Still, the earnings ladder is steep. In 2024, associate’s degree holders median $61,300; bachelor’s $81,800; master’s $100,500; doctoral $125,000; professional $142,300—all for full-time, year-round workers age 25+.

Break-even age models in the same report (Figure 2.3B) ask when cumulative earnings net of college costs exceed a high school path. That is closer to ROI thinking but still not program-level accounting.

Institution-level ROI in federal data (EDsmart build)

EDsmart Data applies a published cumulative-earnings-minus-cost rule to College Scorecard price and earnings fields (see methodology). Among 4,691 institutions with sufficient data at a 10-year horizon:

• 0.8% show negative ROI (39 schools)
• Median surplus: $143,847 (25th percentile $104,028, 75th percentile $175,513)
• At 20 years, no institution in this build shows negative ROI under the same assumptions

Also important: Institution ROI describes the school’s entering cohort—not each major. Do not merge these shares with program-level studies in one column.

Program-level vs institution-level ROI

Federal College Scorecard fields support both institution comparisons and program-level views (see Scorecard data documentation). EDsmart’s institution build finds about 0.8% of schools with negative 10-year ROI (39 of 4,691)—a cohort-level measure, not a judgment on every major.

Researchers who model programs with Scorecard and Census inputs often report a larger share of weak or negative returns than institution-level tables show. That is expected: majors, completion rates, and net price differ within the same campus. The mistake is merging “23% of programs” headlines with “0.8% of institutions” without saying they count different things.

Readers should expect multiple ROI storylines in 2026 coverage. Name the unit of analysis—program, institution, worker, or occupation—before comparing percentages.

Practical guidance for students

Use medians to understand typical pay if you finish and work full time. Use ROI-style tools to stress-test whether this program at this price beats alternatives, including community college, trade pathways, or entering the labor force sooner.

EDsmart publishes ranking colleges by ROI and ties definitions to The College Payoff, 2026. Pair any ROI table with completion rates and net price for your income band—not sticker tuition alone.

Why negative ROI headlines persist

After the pandemic, several researchers refreshed program-level ROI models using College Scorecard and Census data (see federal field definitions). Program-level estimates often show more weak returns than institution-level tables because majors, prices, and completion differ on the same campus—that is a warning about specific programs, not a verdict on every bachelor’s degree.

Education Pays (CPS-based) still shows a large annual earnings ladder by credential. Families should compare ROI tools using the same horizon, price assumptions, and completion rates—not a single viral statistic.

Horizon length changes the story

EDsmart’s institution build shows no negative 20-year ROI among schools with sufficient data under its assumptions, even though 10-year horizons still show a small negative share (0.8%). Longer horizons matter because early-career earnings are noisy and tuition is front-loaded.

Students should compare programs using the same horizon and the same cost assumptions—published price, net price, and local aid offers—not a single viral statistic.

Reading ROI over 10 vs 20 years

EDsmart’s median 10-year surplus is $143,847 across 4,691 institutions; negative share is 0.8%. By 20 years, medians rise above $546,868 with no negatives in this build. Families making decisions about borrowing should consider early-career earnings volatility and non-completion risk, not only long-horizon surplus estimates.

Program-level and institution-level ROI tables answer different policy questions—both belong in informed debate when definitions are labeled.

How to read ROI next to medians

Start with program completion probability and net price after aid.

Compare program-level earnings to program-level debt when Scorecard publishes both.

Treat institution ROI as a screening tool, not a personal forecast.

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.