The master’s expansion
The doctorate is the small, slow stream. The master’s is where American graduate education actually grew — and it grew in a particular direction. This page follows that expansion through three federal collections in turn: how many places were added and who filled them, the visas that carried them, and what the whole thing is worth in tuition.
Every figure is from IPEDS, the State Department or NSF. Where a number is an estimate rather than a measurement, it is labelled and its assumptions are printed alongside it.
1 · The expansion in absolute numbers
Core-STEM master’s degrees conferred each year, split by residency. Both lines rise. One rises far faster, and from a much smaller base.
Loading…
IPEDS Completions, award level 7, CIP families 11, 14, 15, 26, 27, 30, 40 and 42. The series starts in 2008 because the residency columns it depends on do not exist in earlier files. The dip in 2022 and the jump after it are a real cohort effect rather than a reporting change: those students enrolled in autumn 2020 under the pandemic travel restrictions, the count of reporting institutions moves smoothly across it, and the swing sits almost entirely in the non-resident line.
3 · The visas that carried it
F-1 student visas issued by US consulates, by fiscal year. This is issuance, not enrolment — it counts the document, at every level of study from language schools upward. What it shows well is the shape of the pipeline and where it turns.
Loading…
US Department of State, Nonimmigrant Visa Issuances by Visa Class and Nationality, fiscal years 1997–2024. The 2020 collapse is consular closure during the pandemic, not a change in demand. India overtakes China in 2022 and has stayed ahead since, and its 2023 issuance is almost exactly three times its 2019 level. That is the clearest single feature in the series, and it lines up with the master’s jump in chart 1.
What this chart cannot do
Visa issuance cannot be matched to a degree programme. A person may receive a visa and not enrol, enrol at a level other than the one intended, or renew across several years and appear more than once. The correspondence with chart 1 is a correspondence in shape and timing, not a link between named records.
4 · What tuition actually did, over twenty years
All four published rates, in constant 2024 dollars. This chart exists because the shorter window in the next one is misleading on its own: read 2014 onward and it looks as though tuition simply fell. It did not. It climbed steeply, peaked, and has been falling against inflation since — and the master’s expansion begins almost exactly at the turn.
Loading…
IPEDS Institutional Characteristics, academic-year files 2004–05 through 2023–24, median across reporting institutions. In-state undergraduate tuition and fees rose 36% in real terms between 2004 and its 2012 peak, then gave back nearly all of it. Graduate out-of-state peaked later, in 2017, and is now back to within a percent of its 2004 level.
Why this is the honest way to read it
Each line is a median across the institutions reporting that particular rate, so the undergraduate panel includes institutions with no graduate programme. It is not weighted by enrolment: it describes the typical sticker price at a typical institution, not what the typical student pays. Published rates also ignore aid, and undergraduate discounting is heavy — which is exactly why the tuition estimate later on this page is confined to master’s study, where almost nobody holds a waiver.
The turn matters for the argument. Once the published price stops rising, an institution that wants more tuition income has to enrol more people rather than charge more — and the growth in chart 1 is almost entirely non-resident.
5 · What a place costs
Median published out-of-state graduate tuition and required fees across all reporting institutions — the rate a non-resident student is charged. Shown both as billed at the time and in constant 2024 dollars, because the two tell opposite stories.
Loading…
IPEDS Institutional Characteristics, academic-year files 2014–15 through 2023–24 (TUITION7 + FEE7). The sticker price rose in nominal terms and fell against inflation. Whatever drove the revenue growth in chart 6, it was not price.
6 · What the expansion is worth
One year of published out-of-state graduate tuition and fees, multiplied by the number of non-resident core-STEM master’s degrees each institution conferred that year. Since the price per place fell in real terms, all of this growth is volume.
Loading…
A master’s normally runs eighteen months to two years, so the figure for a whole graduating cohort is roughly one and a half to two times the annual figure shown — on the order of five to seven billion dollars at the 2023 level.
This is an estimate, and here is exactly what it is
It is a sticker price, not receipts. It multiplies a published rate by a degree count. It is not net of discounts, waivers or institutional aid, and no institution collects the full published rate from every student.
Why sticker is defensible here and would not be for undergraduates: NSF’s 2024 graduate-student survey puts master’s assistantship rates at 7.0% for research assistantships and 6.9% for teaching assistantships, so roughly 86% of master’s students hold neither — and it is assistantships that carry the tuition waivers. Master’s study is overwhelmingly paid for by the student.
It is not federal money. Only 5.0% of full-time master’s students have federal support as their primary source, against 24.7% of doctoral students. This is a tuition story, not a research-funding story, and the two should not be run together.
7 · Against everything else the sector charges
The estimate from chart 6 set against total net tuition revenue at every US institution, for every student at every level, from the audited finance files. Both indexed to their first year, because a figure in the billions and a figure in the hundreds of billions cannot honestly share an axis.
Loading…
IPEDS Finance, F1B01 (public, GASB) plus F2D01 (private not-for-profit, FASB), net of discounts and allowances, in constant 2024 dollars; for-profit institutions are excluded. In real terms the sector’s total net tuition revenue fell about 9% across this window, from roughly $192B to $174B, while the estimated amount billed to non-resident core-STEM master’s students roughly doubled. The reporting-institution count declines about 1% a year through consolidation, smoothly, so the fall is not an artefact of coverage.
Read the comparison carefully
The two series are not the same kind of number and are not subtractable: one is an estimate of gross billings for one segment, the other an audited net total for everybody. The comparison is about direction — one segment growing sharply against a sector total shrinking in real terms — and that is all it supports.