We’re increasingly relying on Wales to get intel on where the Westminster government’s thinking is on higher education finance.
As DK notes elsewhere on the site, Vikki Howells, Minister for Further and Higher Education in the Welsh Government, has announced an additional £18.5m in capital funding for estate maintenance and digital projects to reduce costs, improve sustainability, and enhance the student experience.
But the other thing that happened was disclosed in the papers outlining the 2024–25 supplementary budget process. That contains a change in the student loan resource budget from £277m in October to £428m now. That feels like a big thing, and it’s worth unpacking why.
Imagine for a minute that you own a pub, and you let students run up a tab. Then imagine that you expect them to gradually pay you back over decades – “pay us back when you can afford it mate”, or an “income contingent” loan.
If you were trying to work out how well off you are now by thinking of how much that group of students will pay you back, you’d have to guess how much money they’ll have and when, and how much of that they’ll give you at a given point.
But you’d need to guess how much that future money will be worth.
The thing about student loans – and how they’re treated in the national accounts (including the “devolved” national accounts of the nations) – is that that’s broadly how it’s done.
Before 2020, the government accounted for student loans in a way that understated their real cost. Every loan was recorded in the national accounts as if it would be fully repaid, plus interest.
But a significant portion of loans would always be written off, but this wasn’t properly accounted for. This made the government’s books look better – despite knowing that many students would never repay in full. That “fiscal illusion” allowed George Osborne to spare higher education from austerity while your local council was cutting bin collections.
Then ONS forced the government to acknowledge that much of the money lent to students wouldn’t be repaid. The removal of the “fiscal illusion” means that now, every loan is split into two parts – the part expected to be repaid, which is counted as an asset, and the part likely to be written off, which is counted as spending immediately.
That sounds sensible – it makes the true cost of student loans more transparent in government accounts. But then, things got complicated – because that calculation has to be constantly revised.
A scenario
Imagine a student owes you £10,000 in 30 years. How much is that worth to you today? If you expect high investment returns elsewhere, future money is worth less today because you could invest it and earn more. Conversely, if expected returns are low, future money retains more value in today’s terms.
So if you expect high investment returns elsewhere, you might say: “That £10,000 in 30 years is worth only £3,000 to me today.” If you expect low investment returns, you might say: “That’s still worth £7,000 to me today.”
And so for the government, the “discount rate” is how they decide how much future student loan repayments are worth in today’s money.
The higher the discount rate, the less future repayments are worth today and the more expensive student loans appear in the accounts. The lower the discount rate, the more valuable future repayments seem, and the cheaper student loans look on paper.
Every so often, His Majesty’s Treasury (HMT) revises the discount rate – and a wander down the rabbit hole reveals that it’s gone up again – which means the government is now valuing future student loan repayments less in today’s terms. Hence the change in Wales.
Why? Because the government’s own borrowing costs have risen.
Lend us a fiver mate
The government raises money by selling bonds (gilts), and the interest rates on those bonds (gilt yields) have gone up. Money is loaned to the government but those loaning it expect higher returns.
Because the government is borrowing more money, lending to it is riskier. Inflation is higher, so investors want more interest to protect the value of their money. And there’s more economic uncertainty, so investors demand higher returns to compensate for potential risks.
Then, as other investment options become more attractive, the government needs to offer better rates to compete for investors’ money. All of that combined makes investors ask for higher interest rates when lending to the government.
So the government now pays more to borrow money. Then when it lends it out, like those lending to the government, it expects higher returns to match its increased costs. That makes the government value future student loan repayments less in today’s money. It’s like saying, “If I’m paying more to borrow, I need to earn more when I lend.”
As a result, the government sees student loans as less valuable now, even though students will still repay the same amount in the future – and so when that discount rate changes, it suddenly makes loans look more expensive in government accounts.
So what?
In theory, if the interest rate on student loans goes up, expected total repayments grow faster. It should make the loan book appear more valuable today – because the government would record a lower upfront cost for issuing the loans.
Then if it wanted to reduce the reported cost of student loans, you’d think they would increase interest rates, which lowers the “cost” of issuing loans in the accounts. Graduates would pay more, but in the government accounts, loans would look cheaper to taxpayers.
But in 2023, Rishi Sunak’s reforms actually lowered interest rates for new students – from RPI+3% to RPI. This means future repayments won’t grow as quickly, so the value of the loan book shrank. Why did he do that with no obvious political benefit?
The weird thing is that lowering interest rates on student loans can actually make them look less expensive for the government in the short term, even though the actual long-run costs have increased.
That’s because the government only records the part of loans that won’t be repaid as a cost upfront, while treating the rest as an asset. Because lower interest rates mean students are expected to repay less overall, the unpaid portion of loans (write-off) shrinks – so the immediate cost to the government looks smaller.
But because the government borrows money to fund these loans, and its own borrowing costs are rising, a big financial hit comes later when it earns less from student loan interest while still paying more to finance the system.
And when the Treasury gets round to noticing that second part, it adjusts the discount rate – and then the hit appears in the accounts. Cheers, Rishi.
As IFS says, all of the above means that the official statistics are likely constantly understating the true cost of the student loans system to the taxpayer:
The ONS [government accounts] measure focuses only on the share of loans that is not repaid in full and thus misses the spread between government borrowing costs and student loan interest rates, which is now expected to be negative.
And that all then ends up having implications for future reform.
What next?
The Russell Group’s spending review submission doesn’t touch on any of this. Nor does Guild HE’s. Universities UK’s submission is silent on it too.
But Rachel Reeves has set a spending envelope. Her new Public Sector Net Financial Liabilities (PSNFL) measure is a broader measure of government debt that includes both traditional borrowing (like gilts) and other financial commitments, including student loans.
It differs from the old Public Sector Net Debt (PSND) because PSND only counts traditional government borrowing, whereas PSNFL includes all financial assets and liabilities – including the value of student loans.
So increasing the value of student loans – tuition or maintenance – would classify these loans as financial assets under PSNFL. But the Office for Budget Responsibility (OBR) says that practices like that might hide underlying fiscal risks, because loans are recorded as assets regardless of the probability of repayment. If a big(ger) portion of these loans is not repaid that would hit the public finances.
Providing bigger (tuition or maintenance) grants instead of loans would result in immediate government expenditure, increasing the deficit in the short term. That could conflict with the government’s fiscal mandate to achieve a balanced or surplus current budget by 2029–30. Grants don’t add to future liabilities, but they do directly impact day-to-day spending, making it harder to adhere to Reeves’ fiscal rules.
Government investment in university infrastructure – like funding for new buildings and research facilities – is capital expenditure, typically financed through borrowing. Under current fiscal rules, the government has to ensure that Public Sector Net Financial Liabilities (PSNFL) fall as a share of the economy by 2029–30. Investments like that might be crucial for long-term growth, but increased borrowing adds to public debt. So to stay within fiscal limits, the government has to manage the spending to prevent an unsustainable rise in its future financial liabilities.
Politically, it may also want to consider relieving the pressure on young/new graduates – by raising the repayment threshold, or introducing stepped repayments. But then to pay for any or all of the above, it would need to introduce steeper interest rate bands so that higher earners pay more, have a higher repayment rate for higher earners, or look again at early repayment penalties to maintain long-term repayment contributions. And they would all bring political costs.
The iceberg and the tip
And that’s the irony. When the last government gave a (graduate) tax cut to higher earners by cutting interest to RPI, the idea was that they would pay no more, no less than the “real” cost of their HE – and that was paid for by lower earners paying more back, rather than having profit from higher earners subsidising lower earners.
But once the cost of borrowing the money to loan to students starts to exceed inflation, the government loses money on every loan. That’s why from a government point of view, aligning student loan interest rates with the government’s long-term borrowing costs would make more sense.
Having interest rates that are significantly lower than the government’s borrowing costs results in an expensive subsidy spent on the rich – it benefits higher earners who would repay their loans even with higher interest rates. Low-earning graduates, who are less likely to repay their loans in full, get no benefit.
So the rich either pay upfront through the great boomer wealth transfer, or go into the loan scheme and pay less than the loan costs. While everyone, and everything, else suffers.
Just like the student housing problem, financing things through borrowing is fine when the costs of borrowing are low, and crucially lower than inflation. But once that becomes your default way of financing something, it’s like an iceberg – at the tip you have the tuition fee, below that there’s the value of maintenance, below that is the terms you offer for paying you the money back that you lend out, but what ends up driving everything deeper down is the cost of borrowing it to lend out in the first place.
Put another way, and to return to the pub, what I didn’t say at the start was that you as the landlord of the Dickinson Arms used to be able to borrow money at low rates to buy stock, pay staff, and maintain the premises. You offered loans or credit to regulars who weren’t able to pay immediately, figuring that the low cost of borrowing made it a sustainable business model. You offered affordable pints and generous credit terms, confident that the costs of borrowing were lower than the rate of inflation.
But the cost of borrowing is up. Each barrel of beer bought on credit now costs you more to finance than it did before. You continue offering low-interest loans to your punters, but the true cost of financing is being hidden beneath the surface, gradually growing into a larger financial burden. And at some stage, the pub becomes insolvent. That’s the real problem the government faces now over HE reform – one that spending review submissions from sector bodies really did need to address.
