We also make use of two additional variables-living at home and paid work during term-time-that capture two possible ways in which students can reduce or avoid the need for loans. Living at home while at university is divided into three categories: never, partially and always. The middle category includes all those who entered higher education at wave 6 and changed accommodation in wave 7. Similarly, the variable coding for work during term-time includes a category for irregular work.
The models presented below use probit for consistency. Part of this research aims at analysing the importance of the nature of student loans. To do so, we estimate two models simultaneously in Table 3, a design that is only possible for categorical outcomes when using the probit transformation. Consequently, the other model has also been estimated using a probit.
Each column in Table 2 presents model estimates that address the first part of our three research questions regarding the key determinants of loan take-up
Table 1 describes students who do and do not take out loans. It strongly supports the hypothesis that these two student groups are different, especially regarding parental wealth and debt aversion. While there is little to no difference in parental social class between student borrowers and non-borrowers, the other indicators of wealth show the expected relationship. Student non-borrowers are more likely to come from wealthier backgrounds, i.e. from a family who owns their house, had attended a private secondary school and are living in an area with a lower IDACI score. Debt aversion is more pronounced among non-borrowers, with an average score of compared to for borrowers-a significant difference. Finally, a smaller proportion of students living at home borrow for higher education. This result is also true of those who work during term-time, though the difference is smaller.
Borrowers and non-borrowers differ by gender: 89% of the female students are borrowers compared to 92% of male students. There is also a small but non-significant difference by religion, with a higher representation of religious students among non-borrowers, but little difference by ethnicity is observed.
Table 1 also provides descriptive statistics for the subsample of tuition fee loan borrowers and for the subsample of maintenance loan borrowers. It can be noted, for example, that students who lived at home were less prominent (66%) in the sample with maintenance loans than they were in the sample with tuition loans (78%) payday loans in Bloomington MN. However, because most students take out both types of loans (84% of borrowers in our sample), the observed differences are relatively small.
Results from the probit analysis of student loan take-up
Two analyses are conducted. The first, using a probit model, describes who does versus does not take out student loans in general. When reported in the text, marginal effects are calculated at the means. Estimates and marginal effects are reported for model 2, except when discussing debt avoidance mechanisms, which are accounted for in model 3.
The models show that, independently of other controls, family wealth does matter in student loan take-up. Students who attended private schools, or whose parents own their house, earn more, or live in less-deprived areas, have lower probabilities of taking out student loans. Housing tenure has a particularly substantial effect. The probability of borrowing is 7.8 percentage points higher for a student whose family’s house is being bought on a loan compared to a student whose family own their house outright-with all other variables in the model being set at the means.
Higher parental education also relates to higher student loan take-up, though in the opposite way to that observed for other proxies for family resources. This result, however, is no longer significant once debt avoidance mechanisms are accounted for (model 3). This finding suggests that parental education significantly raises loan take-up but only for those not living at home. Footnote 6 Interestingly, parental social class makes no difference in loan take-up in any of the models, corroborating the descriptive results in a way consistent with earlier studies. Overall, the results tend to support the hypothesis that family wealth is related to lower student loan take-up and thus that students from richer backgrounds can escape the burden of student loans.
