Student debt in Scotland by income in 2013-14: a game of two halves (probably four quarters, if we had the data)

This post belatedly does a bit more work to tease out differences in actual borrowing by household income under the arrangements for student support introduced in Scotland in autumn 2013. The published data is less clear on this than it used to be, but it turns out that the statistics still provide a reasonable way to compare students in the lower and upper half of the income distribution, at least.

It explains how:

• at minimum students from incomes below £34,000 accounted for 61.5% of all borrowing, but were only half, probably slightly less, of all students, and
• the average amount borrowed in the lower income group was £5,780 and in the higher one £4,140, a difference of £1,640  a year  (+40%).

Data issues

In contrast to previous years, for 2013-14  it was not possible to tell directly from the data published by SAAS  how debt was distributed across the different income bands.  For 2012-13,  figures on all forms of support were provided in income bands of £10,000.  There were two further categories. One was for those deemed “exempt from contributions” , who are entitled to the highest value loan and grant (mainly mature students, but also those coming out of care, for example).  The other covered those who did not declare any income detail, because they were not seeking any form of mean-tested support.  They came mainly from higher-income backgrounds.

For 2013-14 Table A6 instead has this simpler structure:

While this shows a clear  income-related effect, the figures are incomplete at lower and higher incomes, as the first line combines the borrowing of  those previously classed as “exempt from contributions” (many of whom will have been borrowing sums either side of £6,000) and most of those at incomes over £34,000.  The “£0 – £16,999” category therefore excludes  many of those assessed as having no relevant income at all, while the  “£34,000 and above category” only catches a small number of cases at that level.

However, there is a new set of figures among the 2013-14 statistics which can be used to unpick what was happening at lower incomes a bit further.  This post looks at those.

Comparing claimants for two types of loan

The new information provided in Table A11 splits the student population into two groups:  (a)  those who claimed “income-assessed” loan; and, (b) those who only claimed “non-income assessed” loan.

Those claiming an income-assessed (ie mean-tested) loan by definition must have declared incomes below £34,000.  The rest will mainly be students whose incomes were £34,000 and above.  The second group could in theory contain some students from lower incomes who chose not to claim the higher loan, for whatever reason (perhaps because they were living at home).  In practice, that group appears likely to be small, however: see footnote 1.  So Table A11 appears to provide a pretty good basis for comparing borrowing between those above and below £34,000.  Just under half of students will have fallen into the lower-income group and just over half into the higher one (footnote 2).

The figures show that students claiming a means-tested loan borrowed £264m, out of  total student borrowing of £430m.  So we can say that at minimum students from incomes below £34,000 accounted for 61.5% of all borrowing, but only half, probably slightly less, of all students.  Conversely, the better off half (plus) of the student population took out at most 38.5% of total loans.

Students from homes below £34,000 took a lower share of total debt than in 2012-13: that year they accounted for around three-quarters of borrowing (footnote 3).  That’s not because their debt was falling (far from it: it rose by around 40%, see footnote 4) but because  a substantial rise in the non-means-tested “minimum” loan, from under £1,000 to £4,500, entitled students at higher incomes to anything between a few hundred to several thousands more  (for example, the minimum loan at £47,000 used to be £3,000).   Borrowing rose faster in this group than at lower incomes, as students responded to this.

However,  because their starting point was so much lower, the average debt for the better-off half remained significantly below that for poorer students.  Table A11 also shows that the average amount borrowed in the lower income group was £5,780 and in the higher one £4,140, a difference of £1,640  a year  (+40%).   These average figures are for active borrowers, of which there were more in the lower income half or so of the student population  (45,675 versus 39,980). If non-borrowers were factored in, the differential in borrowing between the two groups would be wider.

Differences within each half of the income distribution

A household income of £35,000 is very different from one of £80,000, as discussed here. If we could split the upper half of the income distribution in half again, we would be likely to find the effect above repeated in miniature, with higher take-up, and probably higher average borrowing,  for those between £34,000 and around £50,000 and the rest: there were some signs of that pattern in last year’s data.  However, from this year all students from homes with incomes over £34,000 appear identically in the SAAS records, so this exercise cannot be done.

Similarly, if we could identify separately those who get maximum support because they have nil income, we would be likely to find a group of students whose borrowing and take-up rates were higher than for the rest of the lower-income group. The data to do that probably does exist, even if not in the official statistics.

Skewed debt: an act of policy, not nature

As ever, it is important to emphasise that the skewing of debt towards poorer students is not some sort of  irresistible natural phenomenon, which is a common automatic reaction among people in Scotland when this effect is drawn to their attention. It is a direct result of policy choice, as comparison with other  parts of the UK shows.

Footnote 1

In 2012-13, a total of 43,100 borrowers were classified as having an income below £30,000.  Separate figures produced for a PQ last year suggested a further 3,000 or so borrowers at incomes  between £30,000 and £34,000, suggesting around 46,000 borrowers from incomes below £34,000 last year in total.

Let’s assume that there has been no significant shift in the absolute number of low income students.  That seems reasonable: total grant claims fell slightly between the two years, but  for reasons to be discussed elsewhere, that is not necessarily a sign that fewer poorer students were in the system.  Let’s also assume the likelihood that low-income students would borrow did not fall: that also seems reasonable.  If anything, with grants being cut the pressure in the system would have been  towards increasing the numbers taking out a loan.   In other words, let’s assume there were still at least 46,000 lower-income borrowers in 2013-14.

In practice, 45,675 students took out a mean-tested loan in 2013-14.  That suggests that this category includes most of the borrowing by low income students – and conversely that most low-income students who borrow, do not restrict themselves to the minimum loan.    That wasn’t necessarily a predictable effect: more self-limiting borrowers might have been predicted at lower incomes, as such students tend to report more debt aversion in qualitative studies.  But what seems to be happening here – and it deserves more testing – is that students at low incomes tend either to engage fully with the loan scheme, or not at all.

There may still be a small amount of borrowing by lower-income students not included in this category, but it would be likely to have a de minimis effect on the figures: 1,000 students borrowing the standard non-means tested amount of £4,500 would account for 1% of total borrowing.

Footnote 2

For 2013-14, again we only know the total number of students supported by SAAS in 2013-14 at incomes below £34,0000 excluding those exempt from contributions (on grounds of low income).  The figure was 43,505, but with so many missing students, it is not much help.

Using a similar approach to footnote 1, we can see that in 2012-13, a total of 54,450 students in receipt of some form of support from SAAS  were classified as having an income below £30,000.  Separate figures produced for a PQ show there were a further 3,730 students between £30, 000 and £34,000, giving a total of 58,150 at incomes below £34,000 in 2012-13.

That was 48% of the total number of SAAS supported students (excluding EU domiciled students, who are ineligible for loans) that year.

Because of the absence of separate data for the exempt group this year, it is impossible to judge whether the proportion of lower-income students was exactly the same in 2013-14.  But from what data we have, not least for grants, there is no obvious reason to assume that there this group has increased as a proportion of the SAAS caseload: around half of SAAS-supported students are likely still to report incomes below £34,000.

Footnote 3

The 2012-13 SAAS statistics show that  students between £0-£30,000, plus those exempt from contributions, accounted for £178m of loan.  In addition, a PQ answer from last year allows us to estimate that 45% of the debt taken out among those between £30,000 and £40,000 fell to students with incomes up to £34,000.  That comes to  £12m. Altogether, that gives total borrowing for all students at incomes up to £34,000 of £190 million, out of a total student loans of £254m.

Footnote 4

£190 million (see note above) divided by 46,000 gives £4,128, as the average for this group in 2012-13, to compare with £5,780 in 2013-14.