How Westminster got away with slashing “devolved” spend on higher education

Scotland had around nine per cent of the British population.

By 2024–25, the student loan subsidy spending that the official accounts record for Scotland came to 49 per cent of England’s rate per head.

Add up the thirteen years since fees in England hit £9,000, and it comes to 29 per cent.

Those numbers measure different things – one was grant paid straight to universities, the other is what government now counts as public spending (ie the subsidy) on student loans.

But put them side by side and they tell a story. Over thirty years, the public money behind UK higher education has drained out of a system in which Scotland, Wales and Northern Ireland all started with at least their population’s share – usually more – and into one where the Treasury decides how much each nation may draw, measured against a benchmark of the Treasury’s own construction, using calculations it won’t publish.

And at no point did anyone announce, or fight much over, a cut.

Politicians in the devolved nations noticed bits of it as it happened – the Scottish Parliament was warning about exactly this six years before England’s £9k fee reforms landed.

But the squeeze was split across multiple sets of accounts, measured by a statistic that didn’t exist until 2019, and tightened through documents that get no vote and no debate. Each bit looked defensible on its own. The upshot less so.

Grant designs

Before 1992 there was no “Scottish” or “Welsh” university funding total to argue about. A single body – the University Grants Committee, and later the Universities Funding Council – handed out the annual grant across Great Britain as a whole, and the money followed institutions and what they did: how many students each was funded to teach, which subjects, how strong the research. Wherever those institutions happened to stand.

A Scottish university educating a student from Surrey was Scottish activity as far as the funding was concerned.

Scotland did well out of that machinery. A November 1991 parliamentary answer records Scottish universities taking 15 per cent of the grant in each of the three years to 1988–89, and 14 per cent in the two years after that – for a nation with around nine per cent of the British population.

The same answer records that 19.2 per cent of the students at those universities in 1990–91 came from England, and another 13.8 per cent from overseas. Scotland had – and still has – far more university than its population alone would predict.

When the law changed in 1992 – polytechnics became universities, and the single funder was split into national funding councils, HEFCE for England, SHEFC for Scotland and HEFCW for Wales – the GB system had to be carved up. The carve-up worked by transferring the cost of the institutions that already existed in each territory.

For 1993–94, the published figures put the three councils’ funding at around £2.8bn for England, £416m for Scotland and £179m for Wales. Dividing the money by population instead would have given Scotland something like £310m. Nobody proposed doing that – the foundation of the settlement was who already had what.

Scotland’s new council also absorbed a second system – the “central institutions” that the Scottish Office, the UK government’s Scotland department, had long funded directly, from Napier, Paisley and Robert Gordon to the art schools and the teacher education colleges. Wales’s version of the same tidy-up included lifting six higher education institutions out of local council finance, with £34.6m deducted from the Welsh local government settlement to cover them.

Northern Ireland needed no carve-up at all, because its universities had been funded territorially ever since Ireland was partitioned. By 1981 the Department of Education for Northern Ireland was paying Queen’s more than £21m a year in grant, at levels recommended by the UGC but paid through Northern Ireland’s own machinery.

So each nation entered devolution in 1999 with a funding baseline assembled from decades of accumulated institutional history – in Scotland’s case, one sitting well above its population share. And the formula that would govern everything afterwards had no interest in relitigating any of it.

That formula – Barnett – works like this. Take each devolved government’s block grant, the lump of money it gets from the Treasury to run everything, as given. Then, whenever England’s spending on a comparable service rises or falls, adjust each block by a population-based share of the change.

The Scottish Affairs Committee said Barnett “determines changes in spending, not the total allocation” – while the Lords Constitution Committee described the starting point as a historic level that had never been reassessed against need.

Basically, Barnett never asks what anyone deserves. It just asks what England’s departments are getting this year compared with last.

Home a loan

Student loans ran very differently, and as regular readers will know, they still do. The cash that went out of the door as loans came from the Treasury, through a category that came to be called “annually managed expenditure” (AME) – a demand-led pot, driven by forecasts, that expanded or shrank to match whatever each nation’s student finance system actually lent.

What hit devolved accounts instead was the expected loss on that lending – the slice everyone assumed would never be repaid. The Lords committee recorded in 2002 that AME allocations were “not calculated or varied according to changes in the Barnett formula” at all. Loans lived outside the formula.

The Scottish Parliament record shows the distinction earning its keep. In May 2000, officials costed the abolition of tuition fees – around £42m of extra cash spending, partly offset because students would need about £30m less in loans. Except that only half of every pound lent counted as a real cost, so £30m less lending saved just £15m on paper. Net cost, £27m.

A year later, officials explained that better repayment data had let England cut its assumed loss on lending to 40p in the pound. Scotland moved to a more cautious 45p – and even that drop from 50p banked a £10m budget saving. The less you expected to lose, the less lending “cost”.

In 2008, Scotland cut planned lending by around £30m in a shift from loans towards grants, and an MSP – a member of the Scottish Parliament – asked whether £30m had been freed to spend elsewhere. It hadn’t. The lending was Treasury money, met in full, so lending less just handed it back. What Scotland kept was the £10m or so by which its own budget’s share of the expected losses fell. A pound lent and a pound spent were never the same thing.

Was there a ceiling? The machinery was certainly shared – repayments were collected by the Inland Revenue because, as Scottish official Gillian Thompson told MSPs in 2000, running a separate Scottish collection system would have been cumbersome and expensive, and Scottish ministers couldn’t give the UK tax authority instructions anyway.

She doubted Scotland would necessarily be free from Treasury oversight of repayment terms, and that December, deputy minister Nicol Stephen told a committee that decisions on how the UK-wide repayment scheme operated ultimately rested with UK ministers.

But nothing published in the early years says devolved lending may cost no more than English policy would cost in the same nation. Treasury approval, forecast scrutiny, shared machinery – all present from the start. A numerical ceiling pegged to England’s policy – nowhere to be found.

Out of sight, out of Barnett

Then, when the coalition came to power, England rebuilt its own system on the other side of the excel sheet. The 2010 Spending Review – the exercise where the Treasury sets departments’ budgets for years ahead – cut English higher education spending outside research from £7.1bn to £4.2bn by 2014–15. That was a £2.9bn, 40 per cent reduction in precisely the kind of visible, budgeted spending Barnett tracks – and when England cuts spending Barnett can see, the devolved blocks shrink with it.

Government simultaneously projected that universities’ combined income from grant and loan-financed fees would rise from £9bn to £10bn. Total support for English universities went up. The share of it Barnett could see collapsed.

Scottish politicians had sort of seen it coming in the previous decade. Fiona Hyslop told the Scottish Parliament in January 2004 that England’s new top-up fees would generate no knock-on money for Scotland, because fee income paid by students wouldn’t count as public expenditure. By June 2010 the chamber was hearing the same warning about the coming English reforms, and in December Alex Salmond predicted that the consequences for Scotland would be substantial.

Those warnings went nowhere, and at the time that was fairly rational. Devolved governments were fighting austerity-era battles over their whole budgets, and the loss on loans wouldn’t officially count as spending until the statistics rules changed in 2019. The standing deal still read as “we’ll cover your lending, and you can always run something kinder with your own money”. And anyway, pre £9k fees, there wasn’t much of a subsidy, at least in the long term. Most of the money would be repaid eventually.

So what arrived in place of a population share of England’s new loan billions was ringfenced cover for loan losses – money that could only be used to recognise, in the accounts, that some of each nation’s own lending would never come back. It was money in a currency that couldn’t buy anything – no staff, no buildings, no bursaries. Just the bookkeeping of debt.

Northern Ireland’s 2011–15 budget carried £0m, £5m, £20m and £36m of it across four years, with anything unused returned to the Treasury. Scotland’s ringfenced student loan line rose by £192m in a single 2012–13 budget document, from £88.6m to £280.6m.

Wales took £21.8m in the same document, then claimed £326.45m from the reserve the following year, taking its line from £93.6m to £420.1m. All substantial sums, all spendable on nothing.

Northern Ireland shows what declining to follow England now cost. Its Executive’s budget had assumed £68m of higher education savings by 2014–15 – around £28m from grant efficiencies and £40m from raising fees to English levels.

The Executive kept fees low instead. The £40m never materialised, and pressures of £15m, then £30m, then £40m a year had to be found from ordinary day-to-day spending.

The cap on student numbers that Northern Ireland had operated since 1994 – described to an Assembly committee this spring as “essentially, a cost-control measure” – went on holding participation down, because places paid for by grant come out of a fixed budget.

England, whose additional students were now financed through demand-led lending that barely counted as spending at all, abolished its number controls from 2015.

Diverge and rule

That left lending as the only funding route the devolved nations had that grew automatically with demand – and the published rules governing it then hardened, one Statement of Funding Policy at a time. The Statement is the Treasury’s rulebook for funding the devolved governments. And no parliament votes on it.

2010’s Statement drew the first line. Ordinary ups and downs in the lending forecasts would be covered, while increases caused by devolved policy decisions had to be met from devolved budgets.

By 2015 the test had turned comparative – lending would be covered where devolved policies were “broadly equivalent” to UK government policy and generated “broadly similar costs”, with anything more generous falling on devolved budgets.

The 2020 edition tightened it again – administrations offering more generous terms would generally need to fund costs above a population share of the equivalent UK government programme themselves.

There was still room to move, and governments used it. Wales, following its Diamond review of student finance, scrapped its tuition fee grant, pushed tuition costs into loans and redirected the cash into grants for living costs – a redesign whose sums explicitly traded real cash spending against paper loan losses.

Wales also spent years cancelling up to £1,500 of maintenance loan for borrowers who made their first repayment, at a paper cost later put at £25m–£27m a year.

Scotland’s 2018 move to “Plan 4” repayment terms – graduates would only start repaying above a salary threshold heading for £25,000, with debts written off after 30 years instead of 35 – had been costed by its independent review at around £27m a year for the threshold change alone, and once implemented knocked £338m off the assessed value of the existing loan book.

The Treasury went on providing the ringfenced cover regardless, and Audit Scotland records only that proposed changes had to be discussed with the Treasury and included in forecasts. What test the Treasury actually applied before waving Plan 4 through sits in correspondence nobody has published.

Then, in November 2023, the Treasury wrote the ceiling down:

the UK Government will require the devolved administration to demonstrate that their scheme costs the same or less than it would cost if they were to apply UK Government policy in their respective nation.”

The test assesses the loan system on its own rather than student support as a whole, so a devolved government can’t offset a more expensive loan scheme with cheaper grants elsewhere – despite Wales having been handed student support powers in 2006 on the promise of “full discretion over the financial levers for higher education on a whole system basis”.

A scheme can also fail the test without its government deciding anything, because the comparison is with England’s policy – and England keeps moving. Budget cover continues through a “lead-in period” only where the Treasury is satisfied that policy is being adjusted as quickly as practicable. Adjusted, that is, towards England.

Wales is the proof that it hurts. England’s switch to cheaper “Plan 5” repayment terms cut what English lending costs, while Wales kept the “Plan 2” terms it regards as fairer to lower- and middle-earning graduates – so the benchmark fell while the Welsh scheme stood still.

In March the Welsh Government told a committee in the Senedd – the Welsh Parliament – that a new UK-wide repayments model had pushed its 2025–26 bill for expected write-offs to around £367m, beyond the share of England’s forecast costs that the Treasury will cover, and warned that “even modest further growth in loan outlay or divergence from English repayment policy is likely to prove unsustainable”.

Welsh modelling now shows projected lending sitting, within modelling error, at the Treasury’s ceiling. A power written down in 2023 is setting live policy in Cardiff in 2026.

Sealed with a loan

Three systems entered territorial funding at or above their population weight – Scotland spectacularly so. Devolution froze those positions into baselines, and Barnett then eroded them whenever England cut comparable spending – which after 2010 England did on a heroic scale, removing £2.9bn of teaching grant Barnett could see and replacing it with lending Barnett can’t.

The one thing left growing was Treasury-managed lending, and the rule on that went from oversight, to broad similarity, to England-or-less. On the spending the official national accounts now attribute to student finance, the years since 2012–13 work out at £1,437 per head in England, £857 in Wales, £519 in Northern Ireland and £419 in Scotland.

The change has been slow – Barnett erosion that read as austerity, lending rules that read as plumbing, ringfenced cover that read as money arriving – and the statistic that pulls them into one number only became possible when the Office for National Statistics started counting expected write-offs as spending in 2019, and applied the treatment backwards through the years.

The rule changes were buried in Statements of Funding Policy, which receive no vote and no debate, and the workings stayed in private letters between governments – so a minister who did object would have objected in the files the Treasury now declines to release. When Scotland’s £192m or Wales’s £326m appeared in a budget document, it looked like a win. The first public admission that the ceiling binds came in March this year, from the one government that had hit it.

Some questions remain – what test was applied to Plan 4 in 2018, what conditions attached to the 2006 Welsh transfer, and whether November 2023 wrote down what officials were already doing or invented a harder rule as the money tightened.

Some of that paper trail certainly exists. Responding to a freedom of information request this year, the Treasury confirmed that it holds the devolved governments’ lending and write-off forecasts, modelling updates, ministerial submissions and documents comparing devolved lending with England’s – and withheld every page, citing exemptions for “relations within the UK” and ongoing policy development, refusing even summaries. The refusal was upheld at internal review in July.

Three devolved governments are now reviewing how to fund higher education. Each is working inside machinery whose history is a series of tightenings nobody announced, and whose current workings are officially none of their voters’ business. Whether any of the reviews has the nerve to demand the file is a better test of them than anything they conclude about fees.