The Future of Higher Education and Work: The ROI Divide

If you are trying to decide whether college is “worth it,” you are really asking two questions: whether finishing a credential pays off, and whether this program at this price beats the alternatives. Those are not the same question, and the sources below measure them differently.

The figures come from the U.S. Department of Education’s College Scorecard, the Bureau of Labor Statistics, Census surveys, and related public releases. EDsmartData.com gathered and structured the program-level tables; this article explains what they mean for students and families. When we say typical pay, we mean the national median (half of programs or people fall above, half below), not a guarantee about your offer.

Return on investment (ROI) here means whether typical earnings eventually beat what you paid and time spent in school. A strong campus brand does not guarantee a strong program outcome.

For related reading on EDsmart: degrees with weak ROI and high AI job exposure, whether college is worth it by field and debt, and how we rank programs.

If you only read three things

• Compare programs, not slogans. Two bachelor’s pathways at the same degree level can mean very different first-year paychecks and loan stress.
• Finishing usually beats enrolling. Adults with some college and no credential often earn closer to high school graduates than to degree holders in worker surveys.
• Use the numbers as a compass, not a contract. National medians hide differences by city, school, and career path. Verify the same field on every campus you are considering.

Start here: Degrees tied to strong labor demand and lighter debt are gaining ground on ROI. Degrees with weak early pay, heavy loans, or low completion are losing ground, even when the campus brand sounds prestigious. The split is about program economics, not whether college “works” on average.

How to use this article

What it’s for: Answering why some degrees are becoming more valuable while others are losing ground, with sourced numbers you can check against your shortlist of schools.

• Five reasons → Why the divide opened up
• What’s changing now → Trends you can see in the data
• Why advice changed → Why “go to college” stopped being one answer
• Pay by field → Where the pay ladder breaks apart
• Debt stress → When debt outruns the first paycheck
• Finishing vs. enrolling → The credential you never finished
• AI context → How AI fits in (without the hype)
• Checklist → What to check before you commit
• Charts & tables → debt-to-earnings by major, AI exposure by major, college payoff (2026)

Why some degrees are gaining value while others lose ground

The headline is not one mystery. Five forces show up repeatedly in College Scorecard, BLS, Census, and completion data:

1. Labor demand. Computing, engineering, and nursing sit at the top of typical early pay in Scorecard ($173,344 for Computer Science vs. $43,834 for Computational Science among published bachelor’s fields). BLS puts typical STEM occupation pay near $103,580 vs. $48,000 for non-STEM jobs.
2. Debt vs. first-year pay. ROI weakens when loans eat the first paycheck. Fine Arts program medians run about 1.6× debt to one year’s typical pay; Computer Science families near 0.4× in the same Scorecard build.
3. Completion. About 40.4 million adults have some college and no credential. Among bachelor’s campuses, typical pay 10 years after entry is $44,130 when six-year completion is below 40% vs. $78,207 when completion is 80% or higher.
4. Graduate school and licenses. Many “modest” bachelor’s paths feed licensed careers that need more school. Scorecard shows typical pay about $142,745 four years out for Law vs. $88,472 for Insurance at the master’s level in the same pull.
5. Task change and AI. Automation pressure is uneven. Licensed health and hands-on fields still need credentials; routine office and some entry-level tech tasks face more competition. See How AI fits in for the exposure index by field family.

None of these alone explains every major. Together they explain why averages look fine while individual programs diverge.

What’s changing now

We do not yet publish a multi-year Scorecard-by-major trend file in this build. The signals below mix levels (who is ahead today) with trajectories (where pay moves after graduation).

• Pay spreads are wide and visible. Among 48 published bachelor’s fields with early-career figures, 2 show typical pay below the Education Pays high-school worker benchmark. The ladder is not flat.
• Graduate earnings climb faster in some fields than others. Census LEHD PSEO shows typical pay from year 1 to year 5 rising about 46% for Computer Science vs. about 57% for visual and performing arts (national medians, different cohorts).
• Computer science is high-paid but more competitive at entry. Scorecard lists Computer Science at $173,344 one year out, while Georgetown CEW and others report tougher entry-level hiring as enrollment grew. High typical pay and a harder first job market can both be true.
• AI since 2022. Tool use on campus rose quickly; labor-market pay still rewards scarce technical skills and regulated jobs. Fields with heavy routine digital work may face more wage pressure over time.
• Wealth and pay still rise with education on average. College Board Education Pays 2026 reports a 61.7% typical pay premium for bachelor’s workers vs. high school graduates. That average hides the program-level split described above.

How we built this view

We combined College Scorecard program and campus fields (May 2026 build), BLS occupational wages (May 2024), Education Pays worker medians, Census ACS and LEHD graduate outcomes, O*NET technology skills, and Clearinghouse completion counts. Derived rankings, debt-to-earnings cuts, and automation exposure scores live on EDsmartData.com, with cohort rules on the College Scorecard ROI methodology page.

Across 3,886 campuses with both debt and earnings fields, median loan debt is about 0.34× one year of typical pay 10 years after entry; about 16.9% of those campuses show debt above half of one year’s earnings.

Why “go to college” stopped being one answer

For decades, “go to college” was treated as universal advice. That fit an era when the typical pay gap between degree holders and everyone else was large and fairly stable.

The numbers now show a sharper split inside higher education. The credential you finish, the major you choose, the campus you attend, and the debt you carry often explain outcomes better than enrollment alone.

College Board Education Pays 2026 (Census CPS) puts typical full-time pay for bachelor’s workers without a graduate degree at $81,800 vs. $50,600 for high school graduates, a 61.7% premium. That average spans every major. Scorecard program data show much wider spreads when you compare fields directly.

Where the pay ladder breaks apart

Among published bachelor’s programs nationally, College Scorecard lists about $173,344 in typical pay one year after completion for Computer Science vs. $43,834 for Computational Science when both fields report figures, roughly a 4.0× gap.

Degrees gaining ground (typical early pay and demand)

• Computing and engineering fields lead the Scorecard early-career ladder (see chart).
• Registered nursing (CIP 51.38) lands in the top 50 with typical early pay above $100,000 where published.
• STEM occupations still pay more than non-STEM jobs in BLS data, even when individual majors vary.

BLS Occupational Employment and Wage Statistics (May 2024) put typical pay near $103,580 in STEM occupations vs. $48,000 in non-STEM occupations. Individual majors still vary inside those buckets.

Computer science: valuable on the ladder, tougher at the door

Computer Science leads published early-career program medians in Scorecard ($173,344). That tracks one program cohort, not every graduate’s first offer.

At the same time, Georgetown CEW has reported rising unemployment among recent computer science graduates as enrollment grew. Hiring cycles, layoffs in finance and tech, and automation of entry-level tasks can make the first job harder even when typical program medians stay high. PSEO national data still show typical pay rising from year 1 to year 5 for CS graduates (about 46% in our pull). Think of CS as high upside with more competition at entry, not a single “easy” path.

When debt outruns the first paycheck

Low early pay is not the same as a “bad major.” Some fields face wage compression, uneven program quality, graduate-school dependency, or weak local demand. ROI often loses ground when debt is large relative to the first paycheck, not only when pay is low.

Degrees under pressure (weak early pay or heavy debt)

• Education and liberal arts families show typical debt at 63% and 76% of one year’s typical pay in Scorecard medians.
• Fine Arts shows the heaviest stress in our family table: debt near 1.6× one year’s typical pay.
• Education Pays worker medians span about $44,000 (performing arts) to $87,000 (computer science) before loans.

For an interactive cut by major, see debt-to-earnings by major on EDsmartData.com.

Graduate programs are a separate ladder. In the May 2026 Scorecard pull, typical pay for dentistry and taxation master’s programs differs by nearly 3×. Many licensed careers still require graduate school even when bachelor’s medians look modest.

External researchers (including FREOPP-style program ROI work) have estimated that about 23% of bachelor’s programs could show weak returns once completion and forgone wages are counted, while well under 1% of listed institutions show negative 10-year ROI in the campus-level Scorecard build on EDsmartData.com. Headlines that mix program and campus definitions can talk past each other.

The credential you never finished

“Some college, no credential” is its own category in government statistics. It is not halfway between high school and a finished degree.

National Student Clearinghouse figures cited on our college dropout rates page put the group at about 40.4 million adults. BLS Employment Projections (2021) show typical weekly pay near $935 for some college with no degree vs. $1,432 for bachelor’s or higher.

Education Pays worker medians run about $50,600 (high school), $61,300 (associate), and $81,800 (bachelor’s). ACS 2024 table B20004 (adults 25+) shows typical earnings in the past 12 months near $40,153 (high school), $47,260 (some college or associate), and $70,044 (bachelor’s).

Among bachelor’s-granting campuses in Scorecard, six-year completion rates and typical pay 10 years after entry move together:

Typical pay differs by about $34,077 between the lowest and highest completion bands. That gap can exceed the typical associate–bachelor’s spread in Education Pays.

How AI fits in (without the hype)

AI is not a reason to skip education by itself. It can change tasks, hiring, and which skills stay scarce. We separate tool use on campus from typical labor-market pay after graduation.

Fields where AI supports scarce technical work may keep strong demand. Fields where AI replaces routine office tasks may see pay pressure. Regulated health and other licensed jobs still require credentials regardless of chatbots.

We mapped O*NET technology skills to major families. Healthcare and arts-related groups score lower than computing and business on the exposure index below. The index describes task content, not a forecast of job loss.

See the AI exposure by major chart for the full ranking.

Paths that are not a four-year degree

Associate degrees, skilled trades, certificates, and community college programs can beat selected bachelor’s fields on typical pay with less debt. Trade ROI pages for electricians, HVAC, and nursing are on EDsmartData.com. College Scorecard also lists non-degree credentials, though a systematic national certificate ranking is still in progress. Employer-paid training is real, but it is not modeled in the public tables used here.

What to check before you commit

Replace “Is this a good school?” with questions about your program:

1. What is the completion rate for students with a background like yours?
2. What is typical pay one and four years after finishing this program in Scorecard?
3. What is typical loan debt among completers compared with one year of typical pay?
4. Where do graduates work? Census LEHD PSEO offers a national view for selected fields (table below).
5. How exposed are those occupations to automation in O*NET? Use the index above as context, not a verdict.

At the campus level, the Scorecard ROI build on EDsmartData.com shows a typical 10-year surplus near $143,847 overall, with about 0.8% of listed institutions negative after average net price. Typical surplus near $169,141 at public colleges vs. $107,804 at for-profits reminds you that sector still matters alongside major choice.

What to remember

1. Degrees gain ground when pay, debt, completion, and job demand line up. They lose ground when any of those break.
2. Major choice can move typical early pay by about 4.0× among published Scorecard bachelor’s fields.
3. Finishing a credential usually beats “some college, no credential” on typical pay in worker surveys.
4. Loan stress rises when debt is a large share of one year’s typical graduate pay (Fine Arts near 1.6× in our family table).
5. High-completion campuses cluster at much higher typical pay 10 years after entry than low-completion peers.
6. Computer science shows high typical pay and tougher entry-level competition at the same time.
7. AI shifts tasks and hiring unevenly. Licensed and scarce-skill paths still matter.

What we still cannot measure cleanly

Public data are strong on typical pay, debt, and completion, but several headline questions still lack a single national series. Until those land, treat gaps as unknowns, not zeros.

• We do not yet have a national underemployment rate by major (e.g., NY Fed SCE). This report uses Scorecard “not working 1 year” as a program-level proxy only.
• ACS table B20004 groups some college and associate credentials; SCNC-specific earnings still come best from BLS/CPS tabulations.
• We have not yet pulled every certificate and apprenticeship program from Scorecard in a systematic way.
• We do not yet publish a multi-year Scorecard-by-CIP trend file in this build; the “what’s changing” section uses cross-section levels plus graduate-earnings trajectories (PSEO) and external labor-market reports.

Sources

Derived Scorecard tables, interactive charts, and cohort notes are on EDsmartData.com. Definitions: College Scorecard ROI methodology.

1. EDsmartData.com: institution- and program-level ROI builds, debt-to-earnings by major, AI exposure by major, and related pages (May 2026 build).
2. U.S. Department of Education College Scorecard (program and campus fields).
3. Bureau of Labor Statistics Occupational Employment and Wage Statistics (May 2024 national file).
4. College Board Education Pays 2026 and U.S. Census Bureau Current Population Survey.
5. Federal Reserve Survey of Consumer Finances (2022 wave).
6. NCES and National Student Clearinghouse (completion and some-college-no-credential counts).
7. U.S. Census Bureau LEHD Post-Secondary Employment Outcomes.
8. O*NET 28.0 Technology Skills (automation exposure index).