Did raising tuition fees flatter measurements of young people’s incomes?

The government has announced that the maximum annual tuition fee will be frozen at £9,250; and that the earnings threshold for repayment will jump from £21,000 to £25,000. What’s more, there will be a wide review of student finances to “look again” at this turbulent political issue. While they’re doing that, government statisticians should look again at how changes like these are recorded in our measures of living standards.

The level of tuition fees (for UK students) has no direct impact on our measures of GDP or earnings, but it does have a perverse effect on household income measures that are intended to record living standards, poverty and inequality. The punchline? Cutting tuition fees would – given how these things are currently measured – appear to cut household income growth and increase poverty. Now, that doesn’t tell us anything about whether different tuition fee policies would be good or bad – that is not the point of this technical blog – but it does raise questions about how we should best measure incomes.

The first question to explore is how we should treat student loan repayments when calculating how much money households have at their disposal. If a graduate’s gross earnings are £30,000 and they pay 9 per cent of any income above £21,000 to repay their student debt; should we subtract that £810 loan repayment when calculating their net income?

On the one hand, it could be argued that this is a private loan and that these repayments should be treated just like credit card repayments, for example – i.e. irrelevant to calculating someone’s net income. But I think most people would agree that deducting them from income is appropriate, especially given that these are income-contingent repayments made through the PAYE system and deducted from payslips just like income tax and National Insurance. It’s not disposable income if you have no choice about how to spend it.

One implication is that, as more and more people graduate with student debt – debts that are also larger than in the past – these repayments will be captured in the surveys as a drag on average disposable incomes, working its way up the age distribution. And in the long-term, a cut in tuition fees would boost living standards for graduates – as expected.

What is odd, however, is that when students receive their loans this is counted as income in the government’s (misnamed) Households Below Average Income data: the main source of information on household incomes, poverty and inequality. For example, a student in 2017-18 in England could receive a maintenance loan of £11,002 and this would count as income. This seems reasonable. But their tuition fee loan of £9,250 would also count as income, even if the cash never passes through their hands. So their gross and disposable income in household surveys would probably be £20,252.

The argument in favour of this treatment might be something like “payment of tuition fees is not a tax and it would be wrong for household income stats to capture repayments to the student loans company without also capturing the loans themselves”. But the loans system is hardly a private sector affair: every detail is a matter of (hotly contested) public policy. And we know that a substantial fraction of the value of the loans will never be repaid. Indeed, many people have said that the current loans system is effectively a ‘capped graduate tax’. But the statistical treatment of the loans in household surveys needn’t change completely depending on whether it is or is not explicitly a tax. Moreover, one of the main purposes of the Households Below Average Income data specifically is to determine levels and distributions of poverty, and so counting tuition fee loans as income seems far from ideal.

This approach has particularly odd consequences when the tuition fee regime changes. When tuition was free, students of course had no tuition fee loan ‘income’. The introduction of tuition fees would therefore have immediately boosted student incomes and living standards as recorded by the surveys. And tuition fees (and therefore loans) have increased significantly further since the cap was raised in 2012-13, as Figure 1 shows.

Figure 1: Tuition fee loans are now considerably higher than before 2012-13

Notes: England only
Source: Student Loans Company, Student Support for Higher Education in England 2016

This will have had an effect on student incomes – and indeed incomes overall – as recorded in the Households Below Average Income data. In Figure 2, the large jump in average tuition fees between 2011-12 and 2014-15 can be seen again (though the immediate impact of tuition fee increases can be blunted or delayed by several factors: noise, the fact that students in halls of residence are not surveyed, the time it takes for all students to be affected, and academic vs financial years).

Figure 2: Increases in tuition fees have boosted ‘incomes’ for young people in higher education

Notes: England only. 2015-16 prices.
Source: RF analysis of DWP, HBAI / FRS

Remarkably, average disposable incomes for university students age 18-22 were higher than those of their non-university peers in the last couple of years of data. This would certainly not be true if tuition fee loans were not counted as income.

Moving from individual to household incomes, Figure 3 shows a similar picture with typical student household incomes (equivalised for household size) having jumped up by thousands of pounds over the last few years. An earlier step-change in student incomes between the late 1990s and the 2000s is likely to also relate in part to the introduction of tuition fees.

Figure 3: The household incomes of young people in higher education have jumped up in recent years

Notes: England only. 2015-16 prices.
Source: RF analysis of HBAI and FRS

With such large changes in recorded income for this group (a group that has historically had low incomes), there is also the risk that this treatment could affect important aggregate income statistics, as well as comparisons between age groups, generations and different parts of the country (with different regimes in each nation of the UK). For example, between 2011-12 and 2015-16 in England, the number of people in poverty after housing costs (relative to the UK median) rose by 550,000 overall. But this figure would have been higher were it not for a 140,000 fall in poverty among those in households containing at least one university student. And that fall was driven (at least in part) by the increase in tuition fee loans.

Conversely, the government’s new policy of freezing the fee cap will – according to its own measures – weigh on the incomes of young people (though for graduates the higher repayment threshold will be a significant boost). And if Labour came to power and implemented its policy of abolishing tuition fees, then that would likely be recorded as a hit to average incomes and an increase in working-age poverty.

There is no perfect answer here. Defining ‘income’ can be surprisingly difficult. But the approaches currently used in the government’s invaluable surveys fail a common sense test, given that we want household income data to tell us meaningful things about living standards (note that the ONS’s separate Effects of Taxes and Benefits data currently has a different, but equally flawed, method). The most sensible approach for measuring living standards – and remaining robust in the face of constantly changing policy – would be to deduct repayments when calculating disposable income and not to count tuition fee loans as income.

Obviously these statistical considerations should not have any bearing on the real-world policy choices surrounding student finance. But for researchers like us it is something to bear in mind when looking at income statistics for young people. And for the relevant government statisticians, this choice of methodology is something that should definitely be looked at again.