Are there providers where graduates tend not to repay their loans?
David Kernohan is Deputy Editor of Wonkhe
Tags
The Times is especially excited about a release of data (following two Freedom of Information Act requests from shadow education minister Neil O’Brien) by the Student Loans Company, purporting to show which universities produce graduates most likely (and least likely) to repay their student loan debt.
As a brief reminder, you usually begin to repay your student loans (of any sort) after what is known as the Statutory Repayment Due Date – usually the start of the tax year after you complete your studies. At this point, provided you have an outstanding loan balance and are earning more than the repayment threshold (that’s £25,000 for the current undergraduate loan system – it is different for others) you repay 9 per cent of the amount you earn above that threshold.
The data released purports to allow you to see the average balance (the amount of loan left to pay) and the average amount that has been repaid. Dividing the latter by the former has allowed The Times (or someone) to calculate the average amount of loans that have been paid back. According to O’Brien “The variations between institutions are absolutely enormous”. And according to HEPI’s Nick Hillman “Questions should be asked by regulators when so few students at some institutions seem to be paying their way”.
Can you safely make such determinations from the data as presented? Well no, of course you can’t. Here’s a couple of caveats that appear to have been missed in the press coverage:
- The data includes all student loans (that’s UG fee loans, UG maintenance loans, PG loans, and everything else) going back to 1998 (the year student loans were invented)
- The data includes information from four discrete funding systems (England, Northern Ireland, Wales, Scotland) that have diverged markedly over time.
- A student is assigned to the last provider they studied at – loans amounts are not disaggregated by provider. So if you did UG at the University of Kent, a Master’s at Birkbeck, and a PhD at Plymouth your debt and repayments are shown related to Plymouth.
I’m sure you don’t need my help in spotting many issues here, but as a starting point:
- There’s no attempt to control for the amount borrowed per student. Students from less well off backgrounds will borrow more maintenance loans, students on longer courses will borrow more to pay fees and maintenance than those on shorter courses.
- There’s no attempt to control for subject of study, level of study, subsequent employment, sex, region of residence or even completion (the data will include students who did not complete their course). We already know from the LEO data releases and accompanying IFS research that all of these things are factors in student earnings.
- Even the by university framing doesn’t hold up – there are duplicates in the data, and Oxbridge colleges are included separately because “This will depend on how an HEP has added their details/courses to the Courses Management Service”
- There’s no attempt to control for the number of cohorts. Clearly students who graduated in the early 00s will have had more time to repay loans than those who graduated last year – if a provider only became eligible for loan-backed fees or student maintenance recently clearly students will have more money still to repay.
- Though there is some attempt to control for sample size (if there are less than five students who have their loan balance assigned to an institution it is redacted) we are still faced with the impact of the choices of small numbers of students being disproportionate at smaller providers – again, this is just how averages work.
When Neil O’Brien tried doing this via a parliamentary Written Question the redoubtable Janet Daby had a fair stab at explaining some of this when she released England only data of a comparable level of utility. It’s a shame the shadow minister didn’t pay attention.
When I saw the story I was looking forward to plotting some new data. Having looked at the data, I have chosen not to plot it – I don’t want to draw further eyes to such useless information.
Wowzers! Talk about selective and unfair quotation.
Here is what The Times actually and accurately reports me saying to them:
Nick Hillman, head of the Higher Education Policy Institute, said that such “granular” information should be in the public domain. “Questions should be asked by regulators when so few students at some institutions seem to be paying their way,” he said. But he added this needed to recognise that there could be “good reasons” for some students not repaying their loans.
“For example, people in low-paid public service jobs may not earn enough to pay much back, people keeping our creative industries going rarely make large sums and people working for charities in poorer countries, not to mention people who with serious health challenges, may all have good reasons for not paying their loans back,” he said.
“It is often the case that people who do not pay back every penny of their student loans are big contributors to society in other ways and they may still be paying more in general taxation than if they had not gone to higher education, so it does not automatically follow that the country has lost out financially.”
I don’t think this answers the point. Seems that what you said to the Times implicitly rests on the idea that the figures actually convey useful information. And, very clearly, they don’t.
The only question that needs to be asked here is why on Earth did The Times think this information was in any way credible.
@Nick Hillman
I think a big problem with the data is that – as it is not a cohort view and is for all borrowers since 1999 – it is comparing apples and pears.
For example, a traditional provider will have had a lot of borrowers in the 90s/00s – pre-2012 and especially pre-2006 students borrowed far less (making it easier to repay in full) and have had up to 25 years to repay. New providers, providers who have seen significant growth post-2012 and part-time providers are clearly disadvantaged by simple comparisons that ignore these caveats.
I think the problem with putting granular data into the public domain is that it invites simple and misleading comparisons to be made. Public bodies therefore need to properly curate any data that is released meaning that there is a cost attached to it if the data is to be useful.
Thanks Peter. I 100% agree. I made some of those points – for example about different institutions – to the media when asked. The ideal is obviously to have officially curated data alongside the raw data for people like DK to play with.
It strikes me that this is very valid data, properly compiled and presented, that should form part of the discussion about the future of HE. It would be very interesting to look at franchise institutions figures compared to the mainstream for example.
Of course, as noted above, there are very good reasons why this data should not be seen in isolation but it does seem to me to be something that used, in conjunction with Graduate Outcomes and properly presented LEO data would give some useful insights into the impact institutions are having