Federal Student Loans Helped Inflate College Prices. The New Earnings Test Is a Useful Correction.

The headline version of the new federal higher-education rule is that Washington is about to ban loans for low-paying degrees. That is catchy, politically combustible, and wrong in three important ways. The rule does not ban degrees. It does not contain a statutory blacklist of majors. And its ordinary sanction is not the immediate withdrawal of every form of federal aid.

What the rule actually does is more defensible and more limited: it conditions a program’s continued access to federal Direct Loans on whether its former students clear an earnings benchmark. That is an attempt to discipline a market in which federal credit can increase what students are able to pay and, under some conditions, what schools are able to charge. The economic premise has real evidence behind it. The leap from that premise to “federal loans caused college inflation” does not.

The mechanism: credit can become tuition

Student lending changes more than a student’s bank balance. It changes effective demand. When the government raises the amount a student can borrow, the student’s immediate ability to pay rises even if family income and savings do not. In a market with constrained seats, opaque quality, weak price competition, or schools that know the aid formula, some of that extra purchasing power can be captured by the institution through a higher price.

The strongest causal evidence is narrower than the slogan but still substantial. David Lucca, Taylor Nadauld, and Karen Shen studied changes in federal student-loan maximums and estimated that tuition rose by about 60 cents for every additional dollar of subsidized-loan capacity. They found a smaller positive response to increases in unsubsidized limits, with the largest effects among relatively expensive private institutions and two-year or vocational programs.

That is a policy-change estimate, not a law of nature. It does not mean every college raises tuition 60 cents whenever any student receives a dollar. The study identifies an average response around particular changes in federal credit limits, in particular parts of the market. But it establishes the central point: schools can capture a meaningful share of expanded federal credit.

Stephanie Cellini and Claudia Goldin approached the question from another angle. Comparing for-profit programs around Title IV eligibility boundaries, they found eligible programs charged about 78 percent more than comparable ineligible programs. That is powerful evidence that access to aid can be capitalized into price in that sector. It is not proof that the same markup exists at a public flagship, a nonprofit liberal-arts college, and a cosmetology school.

Graduate lending supplies another warning. Sandra Black, Lesley Turner, and Jeffrey Denning found that the Graduate PLUS expansion increased borrowing and prices without a significant overall improvement in access, persistence, or degree receipt. Again, the estimate concerns a defined policy expansion. Together, these studies support a bounded conclusion: when federal credit becomes more generous and institutions face limited competitive restraint, part of the subsidy may flow to schools rather than remaining entirely with students.

Why loans did not cause all college inflation

The long-run sticker-price numbers are ugly. In 2025 dollars, College Board reports that published tuition and fees from 1995–96 to 2025–26 rose from $2,810 to $4,150 at public two-year colleges, an increase of 48 percent; from $5,940 to $11,950 for in-state students at public four-year institutions, up 101 percent; and from $25,820 to $45,000 at private nonprofit four-year institutions, up 74 percent.

But the recent record does not behave like a simple story in which ever-rising federal lending mechanically produces ever-rising tuition. From 2015–16 to 2025–26, inflation-adjusted published tuition fell 10 percent at public two-year colleges and 7 percent at public four-year institutions, while increasing just 2 percent at private nonprofit four-year institutions. Total inflation-adjusted student and parent borrowing peaked at $163.9 billion in 2010–11 and stood at $102.6 billion in 2024–25. Federal undergraduate borrowing fell by more than half from 2009–10 to 2024–25.

Sticker price is also not net price. For first-time, full-time in-state students at public four-year institutions, average net tuition and fees after grant aid peaked at $4,450 in 2012–13 and was estimated at $2,300 in 2025–26, in 2025 dollars. That encouraging number excludes most of the cost of being a student: housing, food, books, transportation, and other living expenses. College Board estimated the corresponding net cost of attendance at $21,340.

Nor can recent public-college pricing be explained simply by continued state retreat. State and local funding per student increased by more than 40 percent between 2011–12 and 2023–24 after falling during and after the Great Recession. That does not erase earlier disinvestment or differences among states. It does show why a three-decade price increase cannot be assigned to one federal loan program without attending to time period, sector, grants, appropriations, enrollment, amenities, labor costs, and local market power.

Credit can also buy something valuable. In a study later published in the American Economic Review, Black, Denning, Lisa Dettling, Sarena Goodman, and Turner used a borrowing-limit discontinuity to show that greater loan availability increased degree completion and later-life earnings and improved loan repayment among credit-constrained students. Restricting credit can restrain prices; it can also keep capable students from finishing. Both effects belong in the analysis.

What the legal rule actually does

Congress created the framework in §84001 of Public Law 119-21, the Working Families Tax Cuts Act—also called the One Big Beautiful Bill Act—signed July 4, 2025. The Department of Education’s July 1, 2026 final rule implements it across sectors and credential levels.

The test is program-level. For a cohort of completers, the Department uses federal earnings records, primarily IRS data, to calculate median annual earnings in the fourth tax year after completion. The measured group includes completers who are working and not enrolled. That median is compared with an applicable earnings threshold.

There is no single universal high-school comparison. Undergraduate programs generally face benchmarks based on working adults whose highest credential is a high-school diploma. Graduate and professional programs are compared with groups holding bachelor’s degrees. The detailed benchmark can also reflect geography and field under the statutory and regulatory formulas. Compressing all of that to “graduates must out-earn high-school graduates” misstates how graduate and professional programs are treated.

A single miss is not the sanction. A program becomes a low-earning outcome program after failing the earnings-premium measure in two of three consecutive award years. The ordinary result is loss of eligibility for federal Direct Loans. The school may still legally offer the degree. The rule does not automatically erase Pell Grants or every other Title IV benefit the moment a program crosses the line.

Broader aid consequences operate through a separate institution-level administrative-capability standard. At least half of an institution’s Title IV recipients and at least half of its Title IV dollars must not be associated with low-earning programs. Failure in two of three years can bring broader consequences for the institution’s low-earning programs, including Pell or other Title IV implications. That separate route matters, but it should not be conflated with the ordinary program-level Direct Loan sanction.

The rule names no forbidden majors. Reports that social work, art, music, religious studies, teaching-related programs, or cosmetology may be heavily exposed are forecasts based on expected earnings—not statutory categories. The framework covers programs across public, nonprofit, and for-profit sectors and across credential levels. The final rule also delays accountability consequences for certain programs leading predominantly to tipped occupations until their measurement years use earnings from tax years in which the federal no-tax-on-tips policy is operating. The Department will still publish their earnings information during the delay.

Do not replace a federal warning with private debt

This recommendation is analysis: failure of the federal screen is evidence that the program’s typical measured earnings have repeatedly fallen below its benchmark, so substituting less-protected credit compounds the risk the test has identified. It is not proof that the program is worthless, that every individual graduate will earn little, or that a program retaining federal eligibility is safe.

The differences in loan protection are not inference. Federal Student Aid’s own comparison says private loans may require a cosigner, may carry variable rates or rates higher than federal loans depending on the borrower, and may require payments while the student is still enrolled. Private lenders may offer fewer—or no—income-driven repayment, deferment, forbearance, forgiveness, discharge, and hardship options comparable to federal protections. Terms vary, and some private products are less costly or more flexible than others. The point is not that every private loan is ruinous.

The point is that private borrowing may be ruinous over the long term when it replaces federal financing for a high-debt program that failed a repeated federal earnings screen. Removing the federal loan does not improve the program’s earnings. It merely shifts more of the downside to a student who may now have fewer escape valves if the forecast proves right.

The strongest case against the rule

Earnings are not the same as social value. A social worker, teacher, artist, clergy member, or public-interest professional may generate enormous value that never appears in a W-2. A rule tied to wages can punish programs that train people for work society claims to need but chooses to pay poorly. The cleaner policy may be to raise compensation or subsidize the socially valuable service directly, but until that happens, institutions and students bear the collision between public value and market wage.

The fourth-year window is also imperfect. It may miss long-run gains in fields with slow earnings trajectories, while looking generous to fields with an early peak. Regional labor markets, occupational licensing, part-time work, family care, discrimination, and demographic differences can affect observed earnings without measuring instructional quality. IRS records are comprehensive, but precision in measurement is not the same as completeness in meaning.

The behavioral response may disappoint reformers. A college can close a program rather than cut its price. That could eliminate a weak offering—or remove the only local pathway into a public-service field. Institutions may become more reluctant to serve students with lower expected earnings for reasons beyond the school’s control. Students may migrate into private debt, especially when schools market private financing as a seamless replacement. And because the test is about earnings rather than price directly, an expensive program can pass while remaining a bad individual bargain.

Those are not reasons to preserve unconditional federal credit. They are reasons to avoid pretending the metric contains more information than it does. A repeated failure is a serious warning about the relationship between a program and its labor-market results; a pass is not a federal warranty.

What would make this policy better

An improved accountability system would combine earnings with net price, debt at completion, repayment performance, completion rates, and outcomes over a longer horizon. It would publish enough program-level information for borrowers to compare realistic total cost—not just tuition—with the distribution of outcomes, not merely a median. It would also monitor private-loan substitution and program closures so that risk is not simply pushed outside the federal dataset.

For occupations with high public value and persistently low wages, targeted subsidies are better than unconditional credit. Service scholarships, employer support, forgivable public-service financing, or direct appropriations can make training affordable without giving every institution a blank check to raise price. That separates a legitimate social choice—supporting teachers or social workers—from an indiscriminate commitment to finance whatever a school charges.