Funded by the Nuffield Foundation
Under the current higher education funding system in England, the government pays around £22 billion to fund the education of each cohort of around 490,000 England-domiciled full-time undergraduate students studying anywhere in the UK. This covers spending on students at higher education institutions or at alternative providers that are eligible to charge full tuition fees. It includes payments to universities (largely tuition fee payments funded by government student loans, but also teaching grants) and to students towards their living costs while at university. In the long run, the government gets back part of this initial outlay as graduates make repayments on their student loans. Under current rules and economic projections, the long-run cost will be mostly borne by graduates, but how much exactly is hugely uncertain. (Students from other parts of the UK are eligible for greater government support, so for them a larger share of the long-run cost is borne by the taxpayer.)
The higher education system in England is funded primarily through tuition fees. Due to binding caps on the level of tuition fees that institutions can charge, nearly all courses currently cost between £9,250 and £9,535 per year (with the undergraduate fee cap set to rise to £9,790 for the academic year starting in 2026 and to £10,050 in 2027). Direct grants for universities only amount to around £1,050 per student per year on average, with only some ‘high-cost’ subjects attracting a substantially higher level of direct government funding. This means that, for most courses, total resources per student amount to just over £10,500 per year.
For most England-domiciled students, all of this funding is initially provided by the taxpayer. This is because England-domiciled students studying for their first undergraduate degree can take out government-subsidised loans to cover the full cost of tuition fees. In addition, they are eligible for so-called maintenance loans to cover part of their living costs. These loans accrue interest, and are repaid on an income-contingent basis: graduates repay 9% of their income over a repayment threshold and any outstanding loan is written off at the end of the repayment period with no adverse consequences for graduates.
Consequences of the funding model
The current system ensures that students attending university for the first time do not need to pay any up-front fees to attend university and have access to living cost support in the form of maintenance loans. However, the amount of maintenance loan a student is eligible for depends on their parents’ income, and even the maximum rates are low compared with living costs in many parts of the UK. As a consequence, most students spend more than the maximum maintenance loan, and many students who do not receive parental transfers rely on part-time work to fund their studies.
For graduates, this system means that those with low earnings later in life contribute very little towards the cost of their degrees, and will not clear their loans by the end of the repayment period. For these graduates, the system therefore functions like a graduate tax: moderate changes to the repayment rate, repayment threshold or repayment period essentially constitute tax rises or cuts. In contrast, moderate changes in the loan interest rate have no effect on their lifetime repayments.
For high-earning graduates, who can expect to clear their loans before the end of the repayment period, the system functions more like a standard loan contract. For these graduates, the key parameter of the system is the interest rate: the higher the interest rate, the more they repay, provided they do not make voluntary repayments. In contrast, moderate changes to the repayment rate, repayment threshold or repayment period will mostly affect how quickly these graduates pay off their loans rather than how much they repay in total.
On the university side, a key consequence of this system is that the bulk of teaching funding is not centrally controlled and therefore not subject to short-term government budget pressures: spending on student loans automatically rises in proportion to the number of loan-eligible students admitted. This is advantageous for universities but makes it more difficult for the government to control total spending.
A more subtle effect, highlighted in our previous work and emphasised in the 2019 Augar Review, is that funding varies less across subjects than the cost of provision: in many cases, lab-based subjects are likely to be cross-subsidised by social science and humanities subjects that are cheaper to teach. This gives universities an incentive to shift provision towards subjects that are less expensive to teach, which may not align with government priorities.
Past changes to the system
Since the current system of university funding in England was introduced in 2012, it has been subject to constant tinkering by successive governments. According to stated government policy, the fee cap is meant to increase with (RPIX, the Retail Prices Index excluding mortgage interest payments) inflation every year, but until 2025 it was in fact only increased once (in 2017). As a result, per-student spending has been declining in real terms and universities have been particularly exposed to rising costs, complicating their financial planning. The fee cap did increase with forecast inflation in 2025 and it is now government policy that the fee cap will continue to increase each year in line with inflation.
On the repayment side, the terms applying to a particular loan – including the repayment threshold, the interest rate and the repayment period – vary depending on when that loan was taken out. Those who started university between 2012 and 2022 have ‘Plan 2’ student loans. The repayment threshold for these loans was supposedly linked to average earnings, but this has been subject to repeated changes, with a large unexpected increase to £25,000, repeated cash-terms freezes, and a change in default indexation from average earnings to RPI inflation. At the Autumn Budget 2025, the government announced that the Plan 2 repayment threshold will be frozen in cash terms at its April 2026 level of £29,385 for three years, and will then increase in line with RPI inflation from April 2030.
These Plan 2 (2012–22) loans normally accrue interest at a rate between the rate of RPI inflation and RPI inflation plus 3%, depending on a borrower’s income (the interest rate is capped at the ‘prevailing market rate’, which is based on market interest rates for unsecured consumer credit). The lower and higher interest rate thresholds – which determine how much, if any, of the additional 3% interest is applied to a given loan – are also set to be frozen for three years. Any outstanding loan is written off 30 years after graduation.
Amongst the latest entry cohort under the 2022 loan system, we estimate that on average, borrowers will repay around £55,800 towards their student loans (in today’s prices). We estimate those in the lowest 10% of lifetime earnings will pay back around £9,500, because they will rarely earn above the repayment threshold. At the other end of the scale, average student loan repayments of the highest-earning half of graduates are likely to be around £74,200. They can expect to pay back much more than they borrowed in real terms because of interest accrued on their loans.
The strong link between repayments and lifetime earnings under the 2012–22 loan system is illustrated in Figure 1. This picture looks somewhat different when repayments are considered as a percentage of average lifetime earnings. As a share of their lifetime earnings, repayments from the lowest-earning 10% of graduates (1.4%) are set to be similar to those of the highest-earning 10% (1.2%). Those in the lower-middle of the graduate earnings distribution repay the most, at around 2.6% of their lifetime earnings. Overall, we expect just under half of the latest entry cohort repaying under this system not to clear their loans by the end of the repayment period.
Major reforms announced in February 2022 mean the system looks quite different for those who started university from 2023 onwards, who instead have ‘Plan 5’ student loans. For these cohorts, the repayment threshold will be lower at £25,000 and is set to be frozen in nominal terms until 2027. The repayment period will be longer at 40 years, up from 30 years. And the maximum student loan interest rate will be lower, too: interest will be charged at the rate of RPI inflation, instead of a maximum of RPI inflation plus 3%. If RPI reform goes ahead, this means that from 2030, the interest rate on student loans for these cohorts will be the rate of consumer price inflation.
Under the new system, student loans will become slightly more expensive for those with lower earnings, but will be cheaper on average for most middle- and high-earning graduates (see Figure 2). Low-earning graduates can expect to repay more as a result of the lower repayment threshold and the extension of the repayment period. The cut in interest rates will counterbalance this effect for higher-earning graduates, who can expect to pay off their loans in full and to accrue and repay substantially less interest under the new system.
We expect the 2025 university entry cohort to repay less on average, at £44,300 in today’s prices (see Figure 3). That this is lower than for the 2022 starting cohort, even had they been on the same loan plan, reflects that later cohorts can expect to borrow less in real terms as a result of past freezes in the tuition fee cap and declines in the real-terms generosity of maintenance support.
Amongst the 2025 university entry cohort, we expect four-fifths to clear their loans by the end of the repayment period, including almost all of the highest-earning three-fifths of graduates. Those clearing their loans will repay around £50,000 on average – roughly the same amount as they borrowed in real terms.
The Plan 5 loan system is less progressive amongst borrowers, with the highest lifetime repayments as a share of income from those in the second and third deciles of lifetime earnings (2.7%). The highest-earning 10% of graduates can expect to repay around 0.8% of their lifetime earnings. For the same amount of initial borrowing, expected average lifetime repayments are also lower under the Plan 5 system, making these loans costlier for the taxpayer.
Nonetheless, there was a strong case for reforming the system broadly along these lines. If overall progressivity is government’s aim, that can be more directly achieved through the tax and benefit system. The lower interest rate on the new loans will reduce distortions and increase fairness between high earners who have taken out loans and those whose parents paid for their education. The change could also increase trust in the student loans system, as the new system is a ‘good deal’ for all students, including those who go on to become high earners. This will not have been true for many students with Plan 2 loans.
Accounting for the cost of student loans
Accounting for the cost of student loans is a challenging task. First, it requires a prediction of student loan repayments for decades into the future, which is unavoidably highly uncertain. Second, one needs to take a stance on how future receipts should be valued relative to current expenditure.
The UK government’s approach is to model student loan repayments using a prediction model of individual borrowers’ loan sizes, earnings and repayments. This model primarily relies on historical data on borrowers’ earnings from the Student Loans Company and macroeconomic forecasts from the Office for Budget Responsibility. It is used to generate two different statistics:
- the Department for Education (DfE)’s ‘RAB charge’ (resource accounting and budgeting charge), which is a measure of the long-term cost of student loans;
- the Office for National Statistics (ONS)’s ‘write-off share’, which measures the share of student loans that is not expected to be repaid and is used in the UK National Accounts.
These two measures differ in how future receipts are valued relative to current expenditures; they represent two different approaches to splitting student loans into a graduate share and a taxpayer share. Neither of these indicators is ‘better’ than the other; rather, they are answers to different questions.
The RAB charge measures the expected long-run government cost of providing student loans, considering the government’s cost of funding the loans (rather than other government projects). In contrast to the write-off share, the RAB charge crucially depends on an assumed discount rate meant to capture the opportunity cost of government funds. The Treasury has set the relevant ‘financial instrument discount rate’ to RPI inflation minus 0.85%. This negative real discount rate means that, somewhat counterintuitively, the same repayment in real terms now counts for more in the RAB charge calculation the further in the future it is made. (It is somewhat puzzling that the government now favours a negative real discount rate for student loans but still uses a substantially positive discount rate of 3.5% for general policy appraisal.)
The write-off share used in the National Accounts is the share of student loans not expected to be repaid. It is the result of partitioning student loans issued into a grant element (oddly counted as capital spending) and a loan that is expected to be repaid in full, which is treated like other government loans in the National Accounts. This is equivalent to discounting future repayments at the actual interest rates charged on student loans. The write-off share is higher than the RAB charge, because the interest rate on student loans (either RPI, or between RPI inflation and RPI inflation + 3%, depending on the borrower’s earnings) is higher than the discount rate used to construct the RAB charge (RPI inflation – 0.85%).
As the government’s student loan model is not publicly available, we have developed our own model of individual graduates’ loan amounts, earnings and repayments using a different methodology which does not rely on confidential student loans data. We use that model to estimate various measures of the cost of student loans, including the DfE and ONS approaches.
For the 2025 university entry cohort, our estimate of the RAB charge calculated using the government’s preferred discount rate stands at minus 6%, which translates into a long-run taxpayer gain of £1.3 billion, whereas our estimate of the write-off share stands at 10%, which translates into an initial grant element of £2.1 billion. Without any real discounting, we estimate that loans issued to the 2022 entry cohort will cost taxpayers £2.1 billion, with a RAB charge of 10%. It should be noted that there is enormous uncertainty in these estimates: among other things, they depend on policy remaining constant for many decades into the future and on economic forecasts being accurate over the same time horizon.
Partly due to methodological differences between the student loan models, these estimates differ from official government estimates. The difference is especially large for our estimate of the RAB charge: for the RAB charge on Plan 5 student loans issued to full-time students in the 2025–26 fiscal year, the DfE estimate is 30%. The main reason for the discrepancy appears to be that the RAB charge is now very sensitive to assumptions about graduates’ mid-career earnings, and our model is substantially more optimistic about these earnings than the DfE model. One likely reason is that our model relies on earnings differences by age within a year to predict earnings growth, whereas the DfE model relies on observed earnings growth between years for the same individual.
Government borrowing costs
For many years, government had been able to borrow at a lower rate of interest than the interest it charges on student loans. This meant borrowing to lend money to a student who was expected to fully repay the loan was a profitable transaction for government (because the repayments the government made on its extra borrowing were more than offset by the interest it received from the student).
Over the past three years, yields on 15-year gilts (government bonds) have risen substantially, and are now much higher than expected RPI inflation, which is the interest rate charged on new loans. Based on this measure of the government’s borrowing costs, it is now set to lose money on funding even the student loans of those graduates with Plan 5 loans who pay off their loans in full. But whatever precise measure is used, the end of a long period of exceptionally low yields means it is now substantially more expensive for the government to provide student loans.
Neither of the government’s official measures of the long-run cost of student loans reflects this. The ‘financial instrument discount rate’ that underlies the RAB charge is linked to gilt yields, but it is based on a 10-year rolling average and so reflects historical yields rather than current financial market conditions.
Because of the increase in market interest rates, there is now a good case for interest rates on student loans above the rate of RPI inflation, to better reflect actual government borrowing costs. The implicit subsidies for student loan interest rates under the new Plan 5 loan system disproportionately benefit higher earners.
Quirks in student loan accounting
One important point to note about the ONS accounting methodology for student loans is that reductions in the student loan interest rate have a counterintuitive effect on the write-off share: if the interest rate on student loans is lowered, future loan repayments will fall, but the write-off share will also fall. The reason is that lower interest rates mean that loans become more ‘affordable’ for graduates, so a larger share of each student loan will be paid off with interest and a lower share of each loan will be written off under the ONS’s ‘partitioned loan’ accounting methodology. The paradoxical implication for the public finances is that lowering interest rates and thus making loans more attractive for students also lowers the initial cost of loans for the Treasury.
The same effect arises as a result of the government’s proposed reform of the RPI. Under the plans, the RPI will be calculated in the same way as the Consumer Prices Index including owner-occupiers’ housing costs (CPIH) from 2030. CPIH inflation is typically around 0.9% lower than RPI inflation calculated using the traditional methodology. This will lead to a ‘real’ reduction in the interest rate charged on student loans, as that interest rate is currently linked to RPI. But the effect that will dominate as far as the budget deficit is concerned is the ‘accounting’ change in how much repayments are valued at: the write-off share will fall, because a higher value will be attached to future student loan repayments. (A similar effect could be expected to arise for the RAB charge, as the discount rate used also depends on RPI; however, the real discount rate relative to RPI will be changed to RPI inflation + 0.05% following RPI reform, leaving the nominal discount rate unchanged in expectation.)
RPI reform also has another ‘real’ effect on student loans. As the repayment threshold on student loans will be indexed to RPI, the repayment threshold can be expected to rise more slowly following RPI reform. This reduces the cost of student loans, as graduates will have to repay more from 2030 onwards.
Unresolved issues
Several problems with the system remain unresolved. The most important issue is inadequate inflation adjustment of the parameters of the maintenance system over recent years. After a period in which inflation and nominal earnings growth were unexpectedly high, the system as currently set up produced large ‘stealth cuts’ in maintenance support, which can cause genuine hardship for students on tight budgets.
These cuts arise from two major flaws in the system. The first is that parental earnings thresholds for maintenance loan eligibility have been frozen in nominal terms. The lower income threshold, below which students are eligible for full maintenance loans, has been frozen at £25,000 since 2008, and the higher income threshold, above which students are only eligible for the minimum level of maintenance support, has been approximately frozen since 2016 at roughly £62,000 for students living away from home and studying outside London (different higher income thresholds apply for those living at home or studying in London). These freezes mean that at a given level of real parental earnings, students are entitled to less and less maintenance support over time, with entitlements falling especially fast in times of high nominal earnings growth.
The second flaw is that annual increases in maintenance loan levels are determined by inflation forecasts made years before actual inflation is known, with no provision for errors to be corrected. This was problematic in recent years as there were large differences between actual inflation and such forecasts, such that the value of maintenance loans has fallen substantially in CPI real terms. In 2020–21, students from the poorest families were entitled to borrow £9,203, equivalent to £11,828 in today’s prices. In 2025–26, they are entitled to borrow £10,544, £1,284 less in real terms.
Another unresolved issue highlighted above is that because of a mismatch between funding and the cost of provision, it is still financially attractive for universities to shift provision towards courses that are less expensive to teach. The government recently committed to reviewing direct grants to universities, to ensure this funding aligns with its priorities and reflects future skills needs. This could be an opportunity for government to change the level of direct teaching grants differentially across courses, thus realigning overall funding to better reflect the cost of provision. However, the overall level of teaching grants was cut in 2025–26. With the tuition fee cap set to increase each year, it is possible that teaching grants will become an even smaller proportion of overall teaching funding over time, reducing the scope to differentiate grant funding across courses.
The reform to the loan repayment system for those starting university now arguably also creates a new problem. As shown in Figure 3, the cost of student loans for tomorrow’s graduates with lower-middling earnings (in the second and third deciles of lifetime earnings) is now set to be around 2.7% of their gross lifetime earnings. Partly as a result of the distortions created by the tax system, this high cost may well put off some students from attending university even when it would be good for society as a whole for them to go.
The government could address this issue in various ways. Instead of making changes to the parameters of the student loans system, it could adjust the tax system – for example, by making student loan repayments tax-deductible. Alternatively, the government could bring back more substantial direct grants to students, which are likely to be more salient for prospective students than later-life tax reductions.
