The burden of student loan debt: differences in socioeconomic background and attitudes towards higher education

Abstract Increased reliance on student loans to finance higher education (HE) raises concerns about unequal access and repayment burdens. However, who borrows and what motivates students and parents to borrow remain unclear. A longitudinal survey tracked parents and students from high school into their early career to explore the effects of socioeconomic background and attitudes towards HE on taking out student loans in Japan. The findings reveal that, first, families’ economic circumstances are negatively associated with taking out student loans. Second, students, especially those from lower socioeconomic backgrounds, may choose a two-year rather than four-year institution to reduce debt. Third, students and their mothers differ in their motivations for borrowing: students’ perceived risk of failing to graduate discourages them from borrowing money, whereas mothers attach more importance to the cost–benefit of student loans. These results have policy implications for alleviating debt burden and promoting equal access to HE.


Introduction
The cost of higher education (HE) has been rising sharply worldwide (Johnstone and Marcucci 2010).Researchers argue that HE increases earning potential substantially enough that graduates can pay off the cost (Hout 2012); however, the steep rise in costs has raised social and political concerns about socioeconomic and educational inequality.In particular, the privatisation of HE costs, which encourages dependence on unequally distributed available family resources, arguably reinforces educational inequality, leading to student experiences that differ according to socioeconomic background (Antonucci 2016).Given the rising number of borrowers, student loans are increasingly important for financing HE as a policy instrument towards privatisation (Johnstone and Marcucci 2010).
Who, then, is borrowing, and how much do young people owe in student debt?What encourages students and their parents to acquire student loans, or discourages them from doing so?These questions are vital because whether increasing student loans promotes or deters access to HE, especially for students from lower socioeconomic backgrounds, relates to educational equality.Moreover, even if students complete HE, the debt load of student loans may negatively influence individuals' lives.In fact, an extensive review focused on England and the USA shows that student loan indebtedness may restrict individuals' choices related to postgraduate education, their career, home ownership, and family formation and negatively affect their health (De Gayardon et al. 2018).Thus, who carries more student loan debt and what motivates students and their parents to borrow are critical to inequality and risk among young adults today.
Japan provides an important case for examining how individuals respond to the marketisation of HE.Massive expansion of HE has been achieved through development of the private sector, which relies heavily on tuition fees for revenue (Amano 1997;Kariya 2013).Correspondingly, parents have traditionally played an important role in covering the financial cost of HE (Yano 2015).As of 2019, compared to the Organisation for Economic Co-operation and Development (OECD) average (22%), the household share of expenditure put towards tertiary educational institutions was high in Japan (52%; OECD 2022).Furthermore, the scale of available student loans has increased dramatically since the late 1990s, suggesting that individuals are expected to shoulder an increasing portion of the burden of HE costs.
Accordingly, the questions of who carries more student debt burden and what factors lead to motivation and aversion to receiving student loans also relate to how the government, parents, and students should share the financial cost of HE; in the era of mass HE, these questions are commonly critical to HE policy in many countries.For example, HE policy in Japan has followed that of the UK, including a reduction of government spending and an increase in funds with specified purposes under neoliberal reforms (Brazzill 2021).In terms of individual financial burdens, British universities have come to resemble their Japanese counterparts over the past two decades in that they have been transferring the financial cost from the government to students (Goodman 2013, 38-39).
Therefore, this study seeks to develop an understanding of HE funding policy and its relationship with socioeconomic inequality and risk by analysing a unique longitudinal survey administered in Japan that followed a cohort of students, along with their mothers, through high school and into their chosen careers.Specifically, this study addresses the following questions: 1. What effect do students' socioeconomic backgrounds and type of HE institution have on taking out student loans?2. What are students' and parents' attitudes and perceptions that affect the acquisition of student loans?In particular, does borrowing behaviour reflect cost, benefit, and risk calculations that include non-monetary aspects? 3. Do students' and parents' perceptions and attitudes have distinct roles in terms of taking out student loans?If so, what is the difference in the effects of students' versus parents' perceptions?

Socioeconomic backgrounds and borrowing through student loans
Although human capital theory is a common framework for understanding the impact of financial aid (Goldrick-Rab, Harris, and Trostel 2009), two sociological theories relating to educational inequality, among others, provide important insight into borrowing behaviour.
First, rational choice theory inherits the idea from human capital theory that students and/ or parents choose educational alternatives, including leaving education, based on the costbenefit analysis of their respective options.Notably, in sociology, costs and benefits are not limited to economics (Boudon 1974;Breen and Goldthorpe 1997;Erikson and Jonsson 1996).Second, research informed by Bourdieu's theory (Bourdieu 1984;Bourdieu and Passeron 1977) argues that students' process in choosing a university will differ depending on their habitus, highlighting the importance of the social context within which complex educational decisions occur (Archer, Hutchings, and Ross 2003;Bathmaker et al. 2016;Reay, David, and Ball 2005).These theories help explain who borrows money for HE and how much they borrow.Because student loans borrow from future income to address current financial constraints, students with disadvantaged socioeconomic backgrounds and/or limited resources are assumed to be more likely to take out student loans for HE.Previous studies have investigated whether such demographic and socioeconomic characteristics as parental income, education, and ethno-racial group are related to the decision to take out student loans as well as to the amount borrowed (e.g.Addo, Houle, and Simon 2016;Cunningham and Santiago 2008;De Gayardon, Callender, and Green 2019;Furquim et al. 2017;Houle 2014).Research in these areas has frequently been conducted in the USA and the UK, where taking out student loans has become prevalent (Boatman, Callender, and Evans 2022;OECD 2022).Hillman (2015) reviewed who borrows and how much with respect to parental income, showing that low-income students tend to rely on student loans in the USA.However, Houle (2014) demonstrated that students from middle-income families are more likely to shoulder a larger student-loan burden than those from either low-income or high-income families.Addo, Houle, and Simon (2016) also found that parents' wealth negatively affects the amount of student loan debt among young white adults in the USA, suggesting that higher parental wealth is used as protection from debt.
Students' socioeconomic backgrounds relate to more than finances, one example being the important factor of parental education.Houle (2014) demonstrated that if their parents are high school graduates or with lower education level, students are more likely to accumulate loan debt than those with at least one college-educated parent.Similarly, regarding parental education, Furquim et al. (2017) showed differences in the probability of taking out student loans, as well as in the amount borrowed, between first-generation and non-first-generation students in the USA.The former showed a higher probability of borrowing and tended to accumulate more debt than their non-first-generation peers.The effect of various indicators of students' socioeconomic background and ethnicity on the take-up of student loans was found in the UK context as well (De Gayardon, Callender, and Green 2019).
Given that student loans are generally taken out in order to deal with financial constraint, it is unsurprising that students from disadvantaged socioeconomic backgrounds are more inclined to take out student loans than peers from advantaged socioeconomic backgrounds.Nevertheless, different indicators of socioeconomic backgrounds might have a distinct relationship with student loan debt.Indeed, while parental education has a linear effect, parental income has a non-linear relationship with student loan debt in the USA (e.g.Houle 2014).Additionally, the effects of parental education may reflect not only socioeconomic background but also parental knowledge and information drawn from parents' HE experiences (e.g.Furquim et al. 2017).Thus, following previous findings on the effect of parental education and economic resources, this study focuses on whether parental education, household income, and household savings affect the amount of student loan debt.

Student loans: cost-benefit calculations and the complex decisions they involve
As human capital theory holds, enrolment in HE is regarded as an investment decision, and substantial variation is to be expected among individual earnings (Avery and Turner 2012).Thus, expectations for future employment and earnings may affect students' choices in whether to borrow and how much.Bachan (2014) found that UK undergraduate students who expect higher earnings after graduation are more likely to accumulate debt than their peers with lower earnings expectations.Conversely, Oosterbeek and Van den Broek (2009) demonstrated that a subjective indicator on expected earnings does not affect whether students borrow money for HE in the Netherlands.Although the effects of subjective indicators are inconclusive, these studies are valuable for exploring the effects not only of students' socioeconomic backgrounds but also of such wide-ranging subjective variables as earnings, risk, and uncertainty on taking out student loans.
Nevertheless, some have argued that viewing student loans in terms of cost-benefit calculation is overly simplistic (Archer, Hutchings, and Ross 2003;Reay, David, and Ball 2005), instead emphasising the complexity of students' indebtedness experiences and the risk involved in HE decisions (Archer and Hutchings 2000;Jones 2016;Reay, David, and Ball 2005).Students from working-class backgrounds in the UK perceived the economic benefit of HE, but they saw dropping out of HE as a major risk, partly because of their disadvantaged position in relation to the overall process of HE including applying to, participating in, and completing HE and then getting a job (Archer and Hutchings 2000).Similarly, Christie and Munro (2003) argued that attending university and incurring student loan debt is an inherently risky option for students with limited economic and cultural resources in the UK.
However, showing the complexity and ambivalence in student attitudes, other studies reveal that students from disadvantaged backgrounds view student loans as a worthwhile investment and regard student loan debt as a normative and inevitable aspect of current student life (Esson and Ertl 2016;Evans and Donnelly 2018;Harrison et al. 2015).Esson and Ertl (2016) demonstrated that the majority of potential undergraduate students in England who faced tuition fee increases in 2012 did not worry about student loan debt along with the increase in tuition fees.Notably, however, Esson and Ertl found that this result was owing to students' perceived lack of alternatives; they regarded HE as the only ticket to secure future employment.
Given their perception that not attending college is risky, along with current uncertainty in the labour market (Clark, Hordósy, and Vickers 2019;Esson and Ertl 2016), students from lower socioeconomic backgrounds might feel that they have no choice but to apply for student loans.Middle-class students take for granted that they will attend university, following their habitus (Bathmaker et al. 2016;Reay, David, and Ball 2005).Thus, most students, even if they have concerns about dropping out, are unlikely to consider not attending college.Students also regard university debt as socially acceptable, differing from other types of debt, such as commercial bank loans in the UK (Clark, Hordósy, and Vickers 2019;Evans and Donnelly 2018).As such, students are unlikely to evaluate educational alternatives purely from a monetary cost-benefit perspective but may regard attending HE as the inevitable pathway for obtaining employment.Therefore, a better understanding of student borrowing may be achieved by exploring wide-ranging perceptions of and attitudes towards HE, including the risk, the non-monetary benefits of HE, and the jobs that are most valued.
Examples are considerations about future jobs upon graduation for balancing student loan costs and the risk of dropping out of HE.
However, previous studies mostly focused on students who already have HE experience, thus lacking a longitudinal perspective.Therefore, two problems arise.First, it is difficult to exclude the possibility that the college experience causes changes in attitudes towards student loan debt.Davies and Lea (1995) examined differences in attitudes towards debt and the amount thereof between first-year, second-year, and third-year UK college students.They found that the amount of debt rose sharply between students' first and second years, while tolerant attitudes increased between the second and third years, thereby implying that students' indebtedness experience yields more tolerant attitudes towards debt.
Second, parents' perceptions and attitudes towards student loans play a critical role in borrowing decisions (Burdman 2005;Harrison et al. 2015;West et al. 2015).Burdman (2005) showed that in the USA debt aversion often tends to originate not from students themselves but from their parents.If the parents' perceptions and attitudes differ from those of their children, the complexity of decision-making explored in previous studies might be partially attributable to the students' and parents' different motivations for borrowing.Much as Stopforth and Gayle (2022) measured both parents' and children's cultural capital to refine the effect of cultural capital on the General Certificates of Secondary Education (GCSE) outcomes in the UK, analysing these different units -students and their parents -simultaneously might provide a new understanding of the mechanisms of taking out student loans.Therefore, given that parental contribution to HE costs is more expected than ever (Johnstone and Marcucci 2010), this study explores both parents' and students' attitudes and perceptions that may prompt and/or discourage them from taking out student loans.

The Japanese context
In Japan, HE institutions consist of specialised training colleges, junior colleges, colleges of technology, and universities.The length of study in specialised training colleges and junior colleges is predominantly two years, and that in universities is four years, while in colleges of technology it is five years after graduating from lower secondary education, and the latter two years are regarded as HE.Students must complete upper secondary education and pass an entrance examination to enter these HE institutions.Most students who enter HE enrol immediately after completing upper secondary education: some students who fail to pass the entrance examination postpone their entrance and retake the examination the following year.
Currently, most high school graduates attend one HE institution or another, although the majority of students participate in university.For example, the percentages of high school graduates in March 2014, the same cohort of students in our data, who progressed to university, junior college, and specialised training college in April 2014 were 48.0%, 5.3%, and 17.0%, respectively (MEXT 2014).These figures show that alternatives other than HE are limited in Japan, especially given the economic stagnation since the 1990s, which severely hit the youth labour market: young adults, especially high school graduates, have more difficulty than do preceding generations in finding secure employment (Brinton 2011).
With respect to earnings immediately after leaving education (ages 20-24 years), high school graduates, junior college graduates (including from colleges of technology), and university graduates (including postgraduate education), on average, earn annually 3.171 million yen, 3.092 million yen, and 3.353 million yen, respectively (MHLW 2018). 1 The differences by educational level increase with age: at ages 50-54 years, when the differences are largest, the mean annual earnings of high school graduates, junior college graduates, and university graduates were 5.02 million yen, 5.297 million yen, and 8.553 million yen, respectively (MHLW 2018).
As HE expansion in Japan, which first occurred between 1960 and 1975 and then again from 1985 onwards, has been led by the private sector (Amano 1997;Kariya 2013), this sector accounts for a larger part of institutions and student bodies.In 2014, of 781 universities and 2.55 million undergraduate students, 77.2% and 77.4%, respectively, belonged to private institutions.Concerning junior and specialised training colleges, private institutions accounted for 94.6% of the 131,000 junior college students and 95.6% of the 589,000 specialised training college students (MEXT 2014).
Tuition fees at private universities are approximately twice as high as those at national and public universities.On average, the annual tuition fee was 1,051,200 yen at private universities, 497,300 yen at national universities, and 520,200 yen at public universities; the average annual living expenses were 643,000 yen at private universities, 869,200 yen at national universities, and 770,100 yen at public universities.The differences in the living expenses between private and national university students can be accounted for by the fact that while 64.7% of private university students lived at home, only 31.9% of national university students did so (JASSO 2018).So far, parents have been expected to pay for HE, including living expenses (Yano 2015).In addition, the majority of students have part-time jobs: more than 80% of university students engaged in some part-time work in 2016 (JASSO 2018).
Nevertheless, the number of students taking out student loans has climbed drastically in the past 20 years, since the Japan Scholarship Foundation (in 2004, reformed and renamed the Japan Student Service Organization [JASSO]) introduced a new scheme of low-interest student loans in 1999.Figure 1 shows the percentage of students who receive JASSO student loans by type of HE institution.The new programme expanded the scale of recipients and relaxed the eligibility criteria -based on household income and academic achievement -for Category 2 student loans, which bear interest, and the amount of borrowing available to students also increased.The growth of the number of student loan recipients is considered to have contributed to increasing university enrolment over the past 20 years of economic recession (e.g.Breaden and Goodman 2020).Consequently, more students shoulder a debt burden than ever before: combining Categories 1 and 2, about 40% of students across types of HE institutions have taken out a loan in the past decade.With more students incurring loans, difficulty in repaying these debts after graduation is a serious concern (Ouchi 2015), partly because the youth labour market has deteriorated, as shown by the rising rates of unemployment and non-standard employment (e.g.Ohta 2010).
Notably, financial aid in Japan seems to be less complex than in other countries, such as the USA.First, the JASSO manages student loans for most students (JASSO 2018).Second, few students receive grants that do not require repayment; those who receive a tuition fee waiver or reduction are limited. 2Third, students can apply for JASSO student loans before they enter HE through their high school with assistance from their teachers, and the process is uncomplicated.Fourth, although JASSO student loans have eligibility criteria, most students who apply for Category 2 student loans receive them.Additionally, except for a small minority of Category 1 loan recipients, repayment is not income-contingent; the repayment schedule is fixed.Therefore, the risk of high debt is substantial, especially relative to low incomes.

Data
The 'Survey among High School Students and their Mothers, 2012' , conducted in Japan, yielded the data (Nakazawa and Fujihara 2015).The sample comprised pairs of second-year high school students and their mothers who were randomly drawn from an access panel owned by a survey company.The sample reflected the distribution of population by residential area, city size, and gender (Fujihara 2015).Since the original survey was conducted in 2012, the students have been surveyed several times, but just one follow-up survey was administered to mothers in January 2017.
The current study uses the original survey and a second follow-up survey conducted between December 2019 and March 2020, when the respondents were aged 23 or 24 years. 3 The original survey included 1070 pairs of students and mothers (response rate 68.6%), and then 559 individuals responded to the second follow-up survey. 4As this analysis focuses on the take-up of student loans, individuals who had not attended any HE institution were excluded.After eliminating missing values, the final sample size was 377.The available sample size is small, but a couple of aspects are unique: the dataset was longitudinally derived from a cohort of high school students followed into HE and their early careers, as opposed to only current HE students.In addition, not only socioeconomic background but also perceptions and attitudes that might affect the acquisition of student loans were collected from both students and their mothers during upper secondary education.This procedure afforded insight into some of the more subjective aspects of borrowing by examining prospective HE students' and parents' attitudes.

Variables
The dependent variable is the total amount of student loan debt, which includes the amount already repaid (in million yen).This variable was obtained from the second follow-up survey, when about six years had passed after respondents graduated from high school in March 2014.
Independent variables include socioeconomic background -parental education, household income, and household savings -perceptions and attitudes that could affect the acquisition of student loans, and type of HE institution attended.These variables were collected in the original 2012 survey except for type of HE institution, which was reported in the second follow-up survey.
Parental education indicates the educational level of parents.Fathers' and mothers' education were combined into a variable with three categories: high, father and mother both have HE experiences (four-year universities or junior colleges); intermediate, only one parent had HE experience; and low, neither mother nor father attended an HE institution.
Household income and savings directly measure the family's economic resources.Household income is total annual income presented in three categories: low, 0-6 million yen; medium, 6-8.5 million yen; and high, 8.5 million yen or more.Household savings is the amount of savings at the time of the original survey, also divided into three categories: low, 0-1.5 million yen; medium, 1.5-6 million yen; and high, 6 million yen or more.Each category of household income and household savings includes about one-third of the respondents.
High school students and their mothers were asked about perceived financial cost as well as benefit and risk with regard to three types of HE institutions -university, junior college, and specialised training college -specifically: ' Attending the HE institution (1) carries a high economic burden, (2) gives a high social reputation, (3) gives a chance of getting a preferred occupation, and ( 4) is difficult to graduate from' .Respondents were asked to answer with a five-point Likert scale ranging from disagree to agree.Answers to each item were rated from 0 to 4, with higher scores indicating agreement.This study used the responses regarding the HE institution type that students attended.
Although direct monetary benefits are not included above, the data have rich subjective variables.To complement the perceptions of cost-benefit and risk, this study employs job attributes that individuals find important in their future careers, considering that they might think about future jobs in terms of balancing the cost of student loans.It is also reasonable to assume that college graduates plan for jobs based on their educational credentials.High school students and their mothers were asked to rate how much they value the following three items using a four-point Likert scale: an occupation and/or workplace that is well paying; that offers opportunity to gain marketable skills; and that provides a chance to use specific skills and specialised knowledge.The responses were rated from 0 to 3, with higher scores indicating respondents' value on the job attributes.For example, if respondents who value well-paying occupations are more likely to take out student loans, this suggests that expectation of monetary benefits is related to decisions about student loans.
The first institutions students entered after graduating from high school were used to define the categories of HE establishments because transfer between institutions is relatively rare in Japan.Most specialised training and junior colleges are private (as shown in the 'The Japanese context' earlier), whereas universities are either national/public or private and are thus divided accordingly to examine differences between sectors.Thus, HE institutions consisted of specialised training colleges, junior colleges, private universities, and national or public universities.
Control variables are students' gender (0 = female, 1 = male), their number of siblings, and their residential region in 2012.Descriptive statistics for each variable are presented in Table 1.

Distribution of student loans and type of HE institution
Table 2 presents statistics on the total amount of student loans by type of HE institution.In total, more than 40% of respondents who attended HE institutions took out student loans.Although the percentage of borrowers appears somewhat higher among university students than among junior college and specialised training college students, the differences were non-significant (χ 2 = 2.925, df = 3, p = 0.403).No differences were observed between sectors among those who attended a four-year university.These results are generally consistent with a survey conducted by the JASSO in 2014 and 2016 (JASSO 2018), the respondents of which were university and junior college students, although our data show a somewhat lower rate of borrowers in junior college.
The amounts of student loan debt differ according to type of institution attended.Among respondents who included non-borrowers, university students, on average, had higher student loan debt than did their junior and specialised training college counterparts (F(3, 373) = 4.484, p < 0.01); the same situation prevailed among borrowers, with the mean debt for university students being higher (F(3, 163) = 5.731, p < 0.001).This is a reasonable result, given that students study longer at a four-year university than at a junior or specialised training college, where they usually study for two years.Indeed, according to JASSO (2018), annual income from 'scholarships' , which are mostly loans, 5 for university and junior college students was almost the same: 385,300 yen and 387,500 yen, respectively.Thus, although the sample size of these data is not large, the JASSO statistics suggest that the current data accurately captured HE students' borrowing behaviour.

The effect of socioeconomic background and the type of HE institution
Addressing the first research question, the focus now turns to students' socioeconomic backgrounds and the type of HE institution.About 55% of respondents did not borrow, indicating that a relatively large proportion of respondents have zero for their amount of loans.That is, the outcome is censored at zero; thus, the ordinary least squares estimate could be biased.Hence, the Tobit model was adopted for censored outcomes (Breen 1996;Long 1997).A more elaborate statistical model that separates the decision to borrow and the amount borrowed could have been used (e.g.Furquim et al. 2017), but the Tobit model simply describes the effects of many independent variables, including students' and mothers' wide-ranging attitudes towards HE, on the propensity to take out student loans, as shown by Bachan (2014).Table 3 presents the results of Tobit models for amount of student loan debt.Model 1 estimated the effects of socioeconomic background, showing that higher household income and savings protect children from shouldering larger debt.Put differently, students from families with limited economic resources were prone to building up debt.However, parental education did not statistically significantly affect borrowing after controlling for household income and savings.In this respect, although it is non-significant, coefficients of parental education show a positive sign; thus, students with high parental education may accumulate more student loan debt.Perhaps students with highly educated parents but limited economic resources were more likely to progress to HE by taking advantage of student loans (Furuta 2022).Thus, holding household economic resources constant, those with high parental education may tend to take out student loans.Furthermore, the number of siblings had a positive effect, implying that more children mean that student loans play a larger role in obtaining HE.
Model 2 added the type of HE institution to Model 1.Even after controlling for socioeconomic background, the type of HE institution was associated with the amount of debt, replicating the findings presented in Table 2. Students attending junior/specialised training colleges had lower debt than did those attending private university.This is almost certainly owing to the length of study, as no significant difference was found between national or public and private university.Those attending four-year university shoulder a larger debt burden than their counterparts from junior/specialised training college.Therefore, attending a two-year institution rather than a four-year university may be a financial strategy to reduce debt burden.
Figure 2 displays the marginal effect of each independent variable on the expected amount of debt calculated from Model 2: the effect of a change in each independent variable on the change in the expected amount of student loan debt among respondents, including non-borrowers.This is because the coefficients of the Tobit model show the effect of each independent variable on an underlying latent variable, which is a propensity to acquire student loans in this analysis, holding other independent variables constant (Breen 1996;Long 1997).Regarding household economic resources, youth from low-income families accumulated 851,000 yen more debt than those from high-income families.Similarly, those with low household savings had about 1.2 million yen more debt than those with higher household savings.For each additional child, the expected amount of debt increased by 347,000 yen.Furthermore, respondents who attended private or national/public universities owed 804,000 yen and 677,000 yen, respectively, more debt than those who attended specialised training colleges.Thus, the amount of student loan debt depends on household financial resources and the type of HE institution.

The effect of perception and attitudes towards HE
The second and third research questions are addressed next: what motivates prospective HE students and their mothers to take out student loans?Do students' and mothers' attitudes and perceptions play separate roles in borrowing behaviour?To answer these questions, the effects of students' and mothers' subjective variables collected when children were attending high school were estimated to explore whether borrowing behaviour reflects perceived cost, benefit, and risk of attending HE.Specifically, each subjective variable was added separately to Model 2, and respective marginal effects were calculated.
Figure 3 shows the marginal effect of the perceived cost, benefit, and risk of participating in HE.For students and their mothers, the social and occupational benefit did not affect the amount of borrowing, but students' perceived risk did.Remarkably, students who anticipated a higher risk of failing to graduate from HE were less likely to borrow an amount of note: reference categories in some variables are changed from table 3. about 167,000 yen, implying that potential students regard dropping out of HE with a large amount of debt as risky behaviour.Conversely, mothers' perceived risk was not relevant to acquiring student loans.
In a similar way, the effects of future job attributes that respondents value were also estimated (Figure 4).None of students' attitudes towards future job attributes had significant effects.Although prospective HE students generally seem to attach importance to well-paid jobs, future job attributes that prospective HE students value, including skills and knowledge, were not relevant to whether respondents took out student loans.However, mothers' perception of desirable future job attributes for their children significantly differentiated borrowing behaviour.Interestingly, the effects of mothers' attitudes appeared to be in line with cost-benefit calculations: mothers who want their children to acquire marketable skills, make use of specific skills and knowledge, and earn a higher income tended to accumulate student loans.Children with these mothers have about 150,000-220,000 yen more debt than those without these expectations, although 'marketable skills' was statistically significant at 10%.Thus, it seems that mothers expect the economic return on investment in skills to balance the cost of borrowing for HE.
In sum, while Davies and Lea (1995) have demonstrated that indebtedness experiences can change attitudes towards debt among HE students, the current study shows that respondents' perceptions and attitudes, captured during high school, may lead to subsequent borrowing behaviour in HE.Furthermore, the role of subjective variables would differ between prospective HE students and their mothers.Particularly, students' perceived risk and mothers' future job attributes affected the tendency to take out student loans.Therefore, students and parents seem to have different intentions for acquiring student loans.

Discussion and conclusion
This study explored socioeconomic differences in student loan debt and the motivations of students and their parents to acquire or refrain from acquiring student loans in Japan, where the private sector, which charges high tuition fees, has been leading the growth of HE (Amano 1997;Kariya 2013).Specifically, this study analysed the data from a longitudinal survey that followed children and their mothers from upper secondary education into their early careers.
Data from this study have shown that socioeconomic background is closely related to taking out student loans.This result supports 'the reproduction of advantage perspective' (Houle 2014), which refers to a negative relationship between parental resources and student loan debt.It is not surprising that a family's material circumstances should affect borrowing for HE, yet even though student loans may compensate for a family's lack of economic resources, risks remain unevenly distributed among individuals from different socioeconomic backgrounds.However, parental education did not have an independent effect after controlling for household income and savings.Thus, different indicators of socioeconomic background may have distinct relationships with student loans.Second, while the rate of students who take out student loans was generally similar across types of HE institutions, the amount of debt differed by institution.Respondents who attended institutions requiring longer study acquired more loan debt: accumulated debt was larger for those attending a four-year university than for those attending a junior or specialised training college.Previous findings have shown that students from lower socioeconomic backgrounds are more likely to attend junior/specialised training colleges than are those from higher socioeconomic backgrounds (Yano 2015).Attending short-cycle institutions enables students from lower socioeconomic backgrounds to be protected from building up large debt.Thus, attending a short-cycle institution can be regarded as a financial coping strategy, but it also may constrain the choice of HE institutions, especially for students from lower socioeconomic backgrounds.This result accords with those from previous studies that demonstrated a variety of 'financial coping mechanisms' , such as choosing an institution near the students' parental homes in the UK (Callender and Jackson 2008;Callender and Melis 2022).This study contributes through its findings on financial coping strategies related to the student loan debt amount.
Third, children's and mothers' perceptions and attitudes had different effects on whether respondents took out student loans.For students, the risk of failing to complete an HE programme played a critical role.The greater the perceived risk of dropping out of HE, the lower the tendency to borrow for HE.In contrast, for mothers, the expectation of jobs for their children that pay well, build marketable skills, and offer knowledge acquisition were important factors in considering student loans.Students with mothers who valued these job attributes in their children's future careers were more likely to take out student loans, implying that mothers attach importance to the monetary benefits of HE.Therefore, family members, to some extent, differ in terms of motivations and intentions for acquiring student loans.
From a theoretical perspective, the finding that students' higher perceived risk of failing to graduate discouraged them from taking out student loans accords with previous findings that emphasise the risk and uncertainty of borrowing for HE in the UK from the Bourdieu theory (e.g.Archer and Hutchings 2000;Christie and Munro 2003).Indeed, as working-class students considered the failure to complete HE and repay student debt to be a key risk (Archer and Hutchings 2000), high school students in this study who worried about the possibility of dropping out seemed to perceive a greater risk associated with accumulating student loan debt.Conversely, mothers' perceptions of desirable job attributes for their children's future careers affected the tendency to take out student loans.This finding indicates that mothers seem to emphasise occupational and monetary returns after graduating from HE institution to balance the cost of accumulating debt, thereby suggesting that there is some consistency with the cost-benefit calculations posited in rational choice theory.Thus, both theoretical perspectives contribute to an understanding of the complexity of decisions regarding student loans.Furthermore, this result may imply that students and their mothers have different time perspectives.On one hand, by paying attention to the HE experiences in the immediate future, high school students might be ignorant and optimistic about the financial burden of HE.On the other hand, their mothers may be concerned about whether their children will earn enough money to repay the debt in the future, thus focusing on the monetary benefit of HE.
Here, parents' perspectives reflect the Japanese context, in which the rising financial cost of HE has been largely accepted (Yano 2015), possibly making them more sensitive than others to the benefit of HE.However, parents' perceptions and attitudes would be critical in countries where their financial contribution to their offspring's HE is increasingly important.Under current funding policy in the UK, for example, it is argued that 'student loan debt is a family affair' (West et al. 2015, 39).Consequently, it is critical to focus on parents' perspectives of student loans as well as their ability to pay for HE.
Thus, for policy implications, parental perspectives on student loans must be considered, even if loan aversion from students can be reduced through information about student loans, as shown in the USA (Evans and Boatman 2019).Effects from information intervention for students likely depend on parents' attitudes in any case, given that loan aversion often stems from parents rather than students (Burdman 2005).Furthermore, in an HE system that depends heavily on individual expenditure, shouldering student loan debt is inevitable for those with limited economic resources.Put differently, students from disadvantaged socioeconomic backgrounds face greater risk even if they participate in HE.Therefore, to reduce inequality and risk, a continuous policy effort to alleviate repayment burden is increasingly required to avoid financial difficulty.Some limitations remain: the sample size was small, respondents' attitudes that may relate to student loans were not comprehensively measured, and qualitative data were unavailable.Moreover, as financial aid policy has been constantly changing, other factors that this study did not deal with might become important.In these respects, although our data may not be ideal, this longitudinal survey directly measured parents' subjective variables along with those of children when they attended high school.This sheds light on different motivations of potential HE students and their parents for borrowing and provides some insight into the complexity of decisions about HE, including take-up of student loans.Future research that addresses limitations of the data, including the lack of qualitative data, is required to better understand current and future financial aid policies.

Notes
1. Annual earnings were calculated by multiplying 'Contractual cash earnings' by 12 and adding ' Annual special cash earnings' .2. A new grant was introduced in 2020 as part of a financial aid reform (JASSO 2021), but this reform was not in place for the cohort of students in this study.3. The survey conducted in 2012 and the second follow-up survey were approved by the Research Ethics Committee of the Graduate School of Human Sciences, Osaka University.4. In this follow-up survey, although a new sample of the same cohort as the original survey was added to address the problem of small sample size, the current analysis used the respondents who were observed in both the original and the second follow-up survey.This is because the original survey provided rich subjective variables that may influence whether respondents took out student loans.5. Indeed, regardless of its name, the 'JASSO Scholarship Loan programme' is not a grant but a loan.

Figure 1 .
Figure 1.Percentage of students receiving Jasso student loans by type of hE institution.sources: Japan scholarship foundation annual report and Jasso annual report.

Figure 2 .
Figure 2. Effects of each independent variable: average marginal effects and 95% confidence intervals.

Figure 3 .
Figure 3. Effects of attitudes towards hE: average marginal effects and 95% confidence intervals.note:Each variable was standardised with a mean of zero and a standard deviation of one.

Figure 4 .
Figure 4. Effects of preferred job attributes: average marginal effects and 95% confidence intervals.note:Each variable was standardised with a mean of zero and a standard deviation of one.

Table 2 .
descriptive statistics on student loans by the type of hE institution.

Table 3 .
tobit models regarding student loan debt.