The System That Cannot Pay For Itself

Chapter 3

The Cross-Subsidy Sleight of Hand

Why did universities become so dependent on international students?

In a great many universities, international fee income is described as a cushion — a commercial success, a useful extra, something at the edge of the institution that softens a difficult year.

It is not at the edge. It is load-bearing. Government controls the price of most home undergraduate teaching while the cost of delivering it rises; research funders routinely do not meet the full cost of the work they fund. The surplus from uncapped international fees is what closes both gaps, and once it has closed them it has already become posts, repairs, library provision, student support and research time.

This chapter takes apart the reassuring stories that grew around that arrangement — the cushion, the margin, the volume, the diversification, the offshore campus — and asks the question a sane governing body would ask before betting the budget on any of them.

In many universities, international fee income is not a cushion. It is part of the structure. Remove it and the issue is not that the room becomes less comfortable. The wall moves.

Regulated to Fail, Chapter 3

Move the dial

One column of the sector's accounts covers the losses on the others. Drag international recruitment and watch what happens to the rest.

Surplus on international and other uncapped teaching
£3,232m
The deficits it is asked to cover
−£5.23bn
Sector position
−£2.00bn

As reported for 2023–24, the sector was already £2.0bn short on a full economic cost basis — before any change to international recruitment.

For context: 119 of 279 providers (42.7 per cent) forecast a deficit for 2025–26 on their own assumptions, and 56.6 per cent of forecast growth in course fee income to 2028–29 is expected to come from overseas fees.

Source: Annual TRAC 2023-24: sector summary and analysis by TRAC peer group (Office for Students, retrieved 2026-08-14); Financial sustainability of higher education providers in England 2026 (Office for Students, retrieved 2026-08-14)

Measure: £ million, full economic cost basis. Coverage: 128 higher education institutions in England and Northern Ireland (TRAC); 279 providers in England (OfS forecasts). 2023–24 outturn, with OfS forecasts to 2028–29.

What this chart does not show (4)
  • This is arithmetic on a sector aggregate, not a forecast, and not a model of any institution.
  • It assumes the other activity lines are unchanged. In practice a recruitment fall also reduces some costs — but far more slowly than it reduces income, which is the point the chapter makes.
  • TRAC uses full economic cost. A TRAC deficit is not a cash shortfall in the same year.
  • The surplus is not evenly spread. Institutions most exposed to international recruitment would move much further than the sector average.

How the figures were prepared: The interaction is deliberately simple arithmetic on published figures: the reader's chosen percentage is applied to the published non-publicly funded teaching surplus, and the sector position is recalculated. Nothing else is modelled.

View as a table
Sector position on a full economic cost basis as the surplus on non-publicly funded teaching changes. TRAC 2023–24 baseline.
Change in international fee incomeSurplus on that activitySector position
-50%£1,616m−£3.62bn
-40%£1,939m−£3.29bn
-30%£2,262m−£2.97bn
-20%£2,586m−£2.65bn
-10%£2,909m−£2.33bn
0%£3,232m−£2.00bn
+10%£3,555m−£1.68bn
+20%£3,878m−£1.36bn

The two deficits, and what covers them

What each activity costs against what it earns. Teaching British students loses money. Research loses more. One column pays for both.

Surplus or deficit by activity, TRAC 2023–24Publicly funded teaching runs a deficit of £1,693 million at 89.2 per cent cost recovery, and research a deficit of £5,367 million at 66.0 per cent. Non-publicly funded teaching — largely international students — returns a surplus of £3,232 million at 143.1 per cent. The sector position is a deficit of £2,003 million.← Loses moneyMakes money →Teaching British students89.2% of full economic cost recovered−£1,693mResearch66.0% of full economic cost recovered−£5,367mOther trading activity98.1% of full economic cost recovered−£136mInternational and other uncapped teaching143.1% of full economic cost recovered+£3,232mOther non-commercial906.4% of full economic cost recovered+£1,962mSector position: −£2,003m95.7% of full economic cost recovered, against 93.6% the year before.
  • Deficit
  • Surplus

Source: Annual TRAC 2023-24: sector summary and analysis by TRAC peer group (Office for Students, retrieved 2026-08-14); Research financial sustainability: issues paper (UK Research and Innovation, retrieved 2026-01)

Measure: £ million surplus or deficit; cost recovery as a percentage of full economic cost. Coverage: 128 higher education institutions in England and Northern Ireland. 2023–24.

What this chart does not show (5)
  • TRAC uses full economic cost, not cash accounting. A TRAC deficit is not the same as a cash loss in the year.
  • Covers 128 institutions in England and Northern Ireland only. Scotland and Wales are not included.
  • 'Other (non-commercial)' shows a very high recovery rate because it carries little attributed cost; it is not a meaningful trading surplus.
  • Sector aggregates conceal a wide spread. Individual institutions differ sharply from these figures.
  • The five activity lines sum to -£2,002m against a published sector total of -£2,003m. The £1m difference is rounding in the source table, not an error here.

How the figures were prepared: Figures reproduced as published. No rebasing or adjustment. Percentages are cost recovery against full economic cost.

View as a table
TRAC 2023–24, 128 institutions in England and Northern Ireland.
ActivitySurplus / (deficit)Cost recovery
Publicly funded teaching−£1,693m89.2%
Research−£5,367m66%
Other (income-generating)−£136m98.1%
Non-publicly funded teaching+£3,232m143.1%
Other (non-commercial)+£1,962m906.4%
Sector position−£2,003m95.7%

Where this goes next

This is the question the chapter opens with and the mechanism it identifies. What it concludes — and what follows from it — is in the book.

Get the book