Nobody Is Killing the Liberal Arts

The obituary being written over these numbers is for liberal education. It is the wrong patient.

Nobody is killing the liberal arts. There is no committee, no memo, no vote. What is happening is narrower, more legible, and considerably harder to argue with: over fifty years, American higher education stopped being something a state funds and became something a student buys. Everything else in this essay is a consequence of that sentence.

The Vote Nobody Took

Since 1980, net tuition and fee revenue per full-time-equivalent student has risen in every state but Nevada, and by more than 100 percent in forty-one of them. That is the sector-wide migration of cost from the public ledger to the household one, executed over four decades without a single vote taken on the question as such. In fiscal 2025, per-student state and local funding fell for the first time since 2012 — from $12,205 to $12,082, inflation-adjusted — even as total appropriations hit a record $130.7 billion. Record spending, falling per-student support: that is what a system looks like when the money grows more slowly than the obligation, and the difference is billed to families and to institutions that have no one else to bill.

The consequence is not mainly that college got expensive. It is that the buyer changed. When a legislature funds an institution, the institution is a public asset and its worth is argued in public terms — civic capacity, regional development, the general good of an educated population, none of which have a price. When aid becomes portable and follows a student, the institution is a vendor and the student is a customer spending borrowed money. Customers spending borrowed money need to know the return. Once that is the arrangement, an outcomes-and-earnings standard is not an ideological turn against the humanities. It is the only remaining way to regulate a voucher system. Somebody was always going to write it down.

The Closure Map

The closures are not distributed randomly, and the pattern is the tell. They land on small private nonprofits, tuition-dependent, disproportionately in the rural Northeast and Midwest. More than a hundred institutions of that profile currently sit at elevated risk.

Tuition-dependent is the operative phrase, and it is a balance-sheet term, not a moral one. It means an institution with no endowment buffer, for which enrollment is not one revenue line but the entire revenue structure. A shortfall of a few percent in a freshman class is not a bad year at such a college; it is an existential event, because there is nothing else on the asset side to draw against. The demographic cliff did not create that fragility. Four decades of cost migration created it. The cliff is merely the load that finds it.

Then the map does the sorting.

Roughly 13 million Americans already live in places — largely the Midwest and Great Plains — where the nearest university is beyond any reasonable commute. About 3 million adults live more than twenty-five miles from a college or university and lack internet fast enough to substitute an online program. And the National Bureau of Economic Research finding on this is not soft: Black, Hispanic, and economically disadvantaged students respond to living far from a public two-year college primarily by forgoing college enrollment altogether. Not by traveling further. Not by going online. By not going.

Distance, for a specific population, is not an inconvenience. It is a decision, already made, by geography.

Which means the closures are not fungible. A college that closes in a thick market — greater Boston, the Twin Cities, the Philadelphia corridor — subtracts an option. A college that closes in a thin one subtracts access, and subtracts it from exactly the students for whom the institution was the only mechanism that ever converted a rural zip code into something else. The closures are concentrated in thin markets. That is the whole finding, and it does not require anyone to have intended it.

The Rule That Wrote It Down

Workforce Pell was enacted in the reconciliation package signed July 4, 2025. Its provisions are generally effective July 20, 2026, with optional early implementation from July 1. It extends federal need-based aid, for the first time, to programs of 150 to 599 clock hours — eight to fifteen weeks — subject to a dual-approval gate and a set of outcomes standards keyed to completion, job placement, and earnings. Programs must articulate into a related credit-bearing certificate or degree. Programs that do not place graduates into measurably better-paying work do not qualify.

State the case for it at full strength, because it is strong. Millions of working adults need a credential in twelve weeks, not four years, and excluding them from the country's principal need-based aid program while subsidizing a four-year degree they have no four years to spend was a real and defensible grievance. The outcomes gate is a genuine consumer protection, aimed squarely at the predatory short-program industry that has fed on federal aid every previous time eligibility widened. Anyone claiming Workforce Pell is a plot against the humanities is claiming something the statute does not support and the drafting history does not suggest.

It is also worth being exact about what this rule does not do, because the sloppy version of this argument is available and it is wrong twice over.

It does not close colleges. The institutions shutting down are small tuition-dependent privates; the programs Workforce Pell funds sit at community colleges and short-program providers. These are substantially different sets of schools. The rule opens no competing revenue line inside a closing college. It opens one next door, at the sector that college was already losing to.

And it did not start the reallocation. That is most of a decade old and plainly visible before the statute was drafted. Community colleges grew 4.0 percent in fall 2025, outpacing every other sector — public four-years managed 1.9 percent, private nonprofit four-years 0.9 percent. Within the two-year sector the split is sharper: vocational-focused institutions are up nearly 20 percent since spring 2020, to 871,000 students, and undergraduate certificate enrollment stands at 752,000, a 28.3 percent increase since fall 2021. The high- and mixed-transfer two-years — the traditional academic track, built to hand a student to a four-year institution — grew 1.6 and 5.6 percent respectively and remain below where they stood in spring 2020. A trend running since 2020 cannot have been caused by a rule effective in 2026, and reading it that way would be the same error the obituary-writers make: inferring a cause from an outcome that happens to sit near it.

The smaller claim is the durable one. Workforce Pell does not originate the reallocation; it ratifies it. It is the increment at which a fifty-year shift in who pays finally acquires a written federal definition of what the payment is for, and a revenue stream attached to that definition.

The point is not intent. The point is what becomes measurable.

An eligibility regime that conditions aid on placement and earnings inside a fifteen-week window has, as a byproduct of ordinary administrative necessity, written down a definition of what education is for. Everything scoreable on that instrument now has a federal revenue stream attached and a federal metric certifying its worth. Everything not scoreable on it is not banned, not disparaged, not even mentioned. It is simply unscored — and that is the general law here, the one that reaches well past higher education: what an instrument does not measure is not prohibited, it is vacated. No enemy, no censorship, no argument anyone has to lose. Just an absence from the ledger, and what is not on the ledger cannot be funded, cannot be defended, and in time cannot be seen.

Which returns us to the provost. Unscored programs, at institutions where every line must justify itself against a five-year enrollment projection, are what gets cut in the March budget meeting. Not by an enemy of the liberal arts. By an administrator with a spreadsheet and a fiduciary duty.

What the Humanities Numbers Actually Say

The decline in humanities degrees is real and it is not new. Depending on which fields go in the bucket, humanities graduates are down between 16 and 29 percent since 2012. English majors fell by roughly a third from 2009 to 2020; history by about 35 percent. Fewer than one in ten graduates took a humanities degree in 2020, and the last time American colleges produced this few was 2002.

Here the honest reader raises a hand, and the objection is fatal to a lazier version of this essay. The humanities share of bachelor's degrees peaked at 17.2 percent in 1967 and fell hardest through the 1970s and early 1980s — during the maximum expansion of public higher education funding, the broadest access the American system has ever provided, and decades before the demographic cliff or any outcomes gate. The steepest decline on record happened under conditions that are the opposite of the ones described above. A chart of that period, held up alone, ends the argument.

It ends the wrong argument. The 1970s decline was a different mechanism, and a good one. In the late 1960s more than one in five degrees earned by women were in the historical humanities, second only to education — not because those fields were freely chosen from an open field, but because most of the others were closed. Women were a semi-captive constituency for the humanities in the same way they were for teaching. When law, medicine, and business opened, the share collapsed, and it should have. That is a decline in the aggregate produced by an expansion of freedom, and it looks identical on a chart to a decline produced by a contraction of options. Which is exactly why the chart cannot settle anything, and why the mechanism has to be named each time. The post-2012 decline is not women leaving a captive field. It is institutions without endowments cutting what they cannot score.

That distinction is the whole essay, and it is worth holding onto: the number is the same, the cause is not, and only one of the two is a loss.

A liberal education has not become less valuable. It has become less defensible — at the institutions that are required to defend it. At Princeton, no one has to argue annually that a classics department pays for itself; the endowment settles the question before it is asked. At a tuition-dependent regional college running a thin operating margin against a 2029 enrollment projection, someone has to make that argument every single year, in a room where the counter-argument is a nursing cohort with a waiting list. They lose eventually. Not because the argument is bad. Because they have to keep making it.

This is what it looks like when something turns back into a positional good. It is not disappearing from the country. It is re-concentrating in the places that never had to justify it by wage return — which is, to a first approximation, where it lived in 1940, before the GI Bill and the Higher Education Act built the middle of the system that is now thinning out.

The Counterweight

Here is the fact that cuts against all of this, and it should be stated flat rather than caveated away.

The same 2025 reconciliation act that created Workforce Pell also taxed the top of the system. The endowment excise moved from a flat 1.4 percent to a three-tier structure keyed to endowment per domestic tuition-paying student: 1.4 percent from $500,000 to $750,000, 4 percent from $750,000 to $2 million, and 8 percent above $2 million, applying only to institutions enrolling at least 3,000 tuition-paying students, at least half of them domestic. Harvard, Yale, Princeton, Stanford, and MIT sit in the top band. Harvard's exposure has been reported above $200 million a year.

That is a policy taking real money from precisely the institutions a sorting thesis says are winning. The gradient is not one-directional, and an argument that pretends otherwise is not worth reading.

But notice where the cost lands. The reported institutional responses have been hiring freezes and pressure on financial aid budgets. An excise on investment income does not touch the positional value of the credential — the thing that actually does the sorting, in employer screens and graduate admissions and professional networks, none of which reprice by a nickel because Harvard paid a tax. It touches the discretionary spending, and the largest piece of discretionary spending at a wealthy university is the aid office. Taxing the endowment while leaving the credential untouched is a tax on the mechanism that partially un-sorts, levied to fund nothing that sorts less.

The Top of the Funnel

None of the machinery above touches admissions, where the sorting was already finished before any college closed.

Opportunity Insights examined twelve highly selective private colleges across 1999 to 2015 and found that students from families in the top 1 percent were about twice as likely to attend an Ivy-Plus institution as middle-class students with comparable test scores. Among applicants with identical scores, those from the top 1 percent were 34 percent more likely to be admitted. Legacy preference is the obvious channel and the one that draws the legislation, but it is not the load-bearing one: strip the legacy preference out entirely and legacy applicants remain roughly 33 percent more likely to be admitted, because the advantage travels through everything else in the file — the private-school pipeline, recruited athletics in sports that require a childhood of paid coaching, and non-academic ratings that read as character and are substantially a readout of what a family could purchase. Abolishing legacy admissions removes the label. It does not remove the applicant's file.

So the system sorts at intake, and it sorts at exit through the labor market's reading of the credential. What is being rebuilt right now is the middle — the segment assembled between 1944 and 1975 that let a student enter from a town of eight thousand and come out somewhere structurally different. That segment is thinning by closure at the bottom, by eligibility rule in the aid regime, and by the slow arithmetic of tuition dependency, all at once, and all of it downstream of the same shift: from funding institutions to funding students.

The debate ahead will be conducted in the vocabulary of rescue: which colleges can be saved, which mergers pencil out, whether the humanities can be marketed better to eighteen-year-olds. That framing assumes a system that broke.

Read the closure map and the eligibility criteria together and a different reading becomes available — a system doing exactly what its funding structure asks of it, and doing it competently. Nobody has to want this outcome. Sorting does not require an author. It requires only that the cheapest thing to cut and the hardest thing to score be the same thing, in enough budget meetings, for enough years running.