Paying for university: the impact of increasing costs on student employment, debt and satisfaction.

Metcalf, Hilary

The costs of higher education in the UK have shifted increasingly from the state to the student (and students' families). In 1998, a fee contribution of 1,000 [pounds sterling] per annum was introduced for new entrants to full-time degree courses. This paper examines its effect on debt, term-time employment and student satisfaction. The analysis uses data from a survey of two cohorts of students and identifies how the impact varied with student and course characteristics. Fees led to an increase in student debt (particularly for disabled students and for students who did not receive financial support from their families) and a decline in student satisfaction. No general impact on term-time employment was identified, but term-time employment increased for students who did not receive financial support from their families. Whilst for these two groups inequality was increased, fees appeared to lead to greater equality, in terms of term-time employment, between children of graduate and non-graduate parents. The paper discusses the implications for the introduction of top-up fees in 2006.

Key words: Higher education; university; funding; fees; top-up fees; equality; disabled people; ethnicity; gender; finance; social class; term-time working; student; employment; debt; satisfaction; policy; disadvantage.

JEL Classification: 1; 12; 120; 122; 128; 129.

Introduction

Over the past two decades in the UK, the costs of higher education have shifted increasingly from the state to the student (and students' families). Restrictions on access to state benefits were followed by the reduction, then abolition, of grants (which were replaced by subsidised, but size-limited, loans). Redistribution of costs culminated in the reintroduction of a contribution towards fees, reversing a policy of free tuition for fulltime degree course students which had existed for over 40 years. In 1998, new entrants to full-time degree courses were charged up to 1,000 [pounds sterling] per annum as a contribution towards fees. The charge, which has been raised in line with inflation, is means-tested and, for standard age entrants, means-testing is based on parental income and a limited range of parental commitments. The fees contribution represents a large increase in cost to students (or their families), given the median expenditure of full-time students (aged under 25) in 1998/9 was 5,225 [pounds sterling] (Callender and Kemp, 2000). This cost will rise substantially from September 2006, when universities which meet certain conditions will be able to charge fees of up to 3,000 [pounds sterling] per annum.

The mare political interest in the impact of fees on students has been on their effect on participation, with concern that higher costs would disproportionately discourage those from poorer families and nontraditional students from going to university (Education and Employment Committee, 2001b). However, the effects of fees, and the shift of higher education costs to the student, are potentially much wider, affecting the composition of the student body in other ways, affecting students' activities whilst at university (and hence their higher education experience) and affecting outcomes. In terms of participation, the rate of return to higher education decreases with cost and, to the extent that students take this into account, the composition of students will alter towards those with lower discount rates and towards those with higher expected earnings. The composition may also alter in terms of the extent to which non-pecuniary aspects of university are important (although the direction of change is not clear a priori). Thus, as the cost of university rises and the rate of return falls, universities may find that the percentage of students for whom intrinsic interest in the subject is very important changes, as may the percentage for whom higher education is primarily a career tool.

In terms of the university experience, not only may this change because higher costs affect the composition of university students, but also because increased costs alter students' activities whilst at university. In this respect student time use and, particularly, student employment has received most attention. Interest has focused on term-time (as opposed to vacation) working, both due to its historic novelty (Metcalf, 2003) and due to its greater importance for the educational experience, particularly to its possible effect on studying. Term-time employment has grown, with nearly half of students employed term-time in 1998/99 (Callender and Kemp, 2000). Growth in term-time employment is directly related to financial pressure, including family financial support (Metcalf, 2003), and is reversely related to social class (Barke et al., 2000; Callender and Kemp, 2000; Connor et al., 2001). The extent to which term-time working impacts on studying is unclear. Although students tend to report term-time employment to be detrimental (affecting study time, attendance and completion of assignments (Lindsay and Paton-Saltzberg, 1996; Cubie, 1999; Barke et al., 2000; Education and Employment Committee 2001a and 2001 b; Ford et al., 1993; Curtis and Shani, 2002), some studies have found no impact on time devoted to study ($$ref). Instead, leisure time and sleep may suffer instead. Whether this is detrimental to the educational experience has not been explored, although term-time working and financial pressures have been found to lead to stress and to poorer mental and physical health (Roberts et al, 2000). Students, and their families, face choices in how they respond to higher costs, including through their use of debt, financial restraint and living at home. These choices will be conditioned by individual preferences and cultural factors. Students who place strong emphasis on their career goals are less likely to be employed term-time, whereas those for whom leisure and socialising is particularly important are more likely to work term-time (Metcalf, 2003).

There is, as yet, no definitive evidence on the impact of the effect of fees (or of university costs more generally) on educational outcomes, nor on post-university economic activity. What is clear is that, with increasing costs, debt amongst graduates has grown. Full-time students in their final year in 1995/96 expected to owe 2,404 [pounds sterling] when they left university (Callender and Kempson, 1996) and those in their final year in 1998/99 3,462 [pounds sterling] (Callender and Kemp, 2000). This had reached 9,970 [pounds sterling] for final year students in 2003 (UNITE/MORI 2003). Most studies have found debt to be inversely related to social class (Barke et al., 2000; Callender and Kemp, 2000; Callender and Kempson, 1996), although the recent UNITE/MORI (2003) survey did not find this. Unsurprisingly, debt was lower for those who lived at home (Barke et al., 2000; Callender and Kemp, 2000).

This paper seeks to extend our current understanding of the effect of shifting the cost of higher education towards students (and their families) by exploring the effect of an increase in costs (namely, the introduction of the fee contribution) and by taking into account student preferences in identifying effects. The paper examines three outcomes: behaviour whilst at university (in terms of participation in term-time employment); debt on graduation; and student satisfaction with university. Term-time, rather than all, employment was analysed because the former was considered to have a greater effect on the university experience, being not only more likely to affect studying but also to affect the leisure activities which form part of university life.

Although previous research indicates an increase in term-time employment as costs to the student have risen, it has not demonstrated a causal link. Whilst fees (along with any other increase in costs) might be expected to increase paid work amongst students (as students seek to minimise debt or raise disposable income), other factors may lead to a decrease in term-time employment. First, there is likely to be a compositional effect: fees may result in a change in the composition of students towards those who value higher education more highly (either in itself or in respect of its future employment effects) and, therefore, towards those who are less likely to work term-time. (In other words, because the rate of return has been reduced, those with the lowest rate of return would no longer go to university.) Second, fees may change behaviour amongst those who do go to university, leading students to place greater weight on post-education employment returns (in order to repay debt). If students expect educational achievement to raise post-education employment returns, this, also, will lead to a reduction in term-time employment. Thus, the impact of fees on student employment is not clear cut, a priori, and the effect will be conditioned by students' preferences.

Similarly, fees may affect satisfaction and debt both through compositional effects on who goes to university and through their response to fees once there. In respect of student satisfaction with higher education, increased costs may not only alter the university experience (due to, for example, increased employment amongst students), but fees may change the relationship between students and universities, from a pupil/teacher relationship to that of consumer/supplier. Being charged for education (rather than just having to pay for maintenance) may alter attitudes towards higher education, with students assuming the role of a consumer, weighing up value for money. Whether such a change takes place or not, increased costs and the prospect of debt may lead students to consider more critically the quality of the services provided by their university and to take costs into consideration in their judgement of how worthwhile was their university experience.

This paper explores these issues for full-time, standard age students and considers whether cost increases may have affected the quality of the higher education experience and outcomes for university students. The analysis is based on a survey of university students and so can examine compositional aspects of change only indirectly. Three issues are examined: term-time employment, expected debt on leaving university and satisfaction with university. The analysis had two main foci: to identify whether there had been any change in behaviour since the introduction of fees and to identify whether any changes have equity implications. The main equity issue considered is equity between university students. Equity issues of access are also touched upon but, owing to data limitations, the issue of equity between graduates and non-graduates could not be considered.

The data

The data is from an ad hoc survey designed and administered by NIESR. The survey comprised a random sample of two cohorts of third year students in four universities. The universities were selected to typify universities with different funding constraints. These were proxied by 'A' level scores of undergraduates and by the research assessment exercise (RAE) average for the university. Universities were selected from each quartile for these scores: University, A, an old university in the highest quartile, University B, an old university in the second quartile, University C, a new university in the 3rd quartile and University D in the lowest quartile. Cohort 1 was surveyed in Spring 2000 and Cohort 2 in Spring the following year. Cohort 1 students had entered university, pre-fees, the majority entering in 1997, and Cohort 2 students had entered after the introduction of fees, mainly in 1998. The survey achieved a response rate of 50 per cent. The data used in this paper relate to standard-age students (aged under 21 at the start of their course).

As fees affect only the behaviour of students who are subject to fees, the analysis was restricted to those in Cohort 2 subject to fees and those in Cohort 1 who would have been subject to fees had fees been introduced the previous year. Thus, in Cohort 1, students from families whose income fell below the means-tested limit for fees needed to be excluded. The survey did not collect data on parental income, but collected data on parental economic activity, including occupation. Parental income was proxied by the mean wage for the two-digit occupational group by gender. Students from families with a total parental income of more than 500 [pounds sterling] per week were assumed to pay fees. Those whose parents were not employed were assumed to fall below the fees threshold. For consistency, the same approach was applied to Cohort 2 students. The estimation resulted in 71 per cent and 67 per cent of students in Cohort 1 and Cohort 2 respectively being deemed as paying fees. The figures are of a similar order to that found in a survey of student finances, in which 63 per cent of students aged under 25 paid fees (Callender and Kemp, 2000).

All data reported are for those who were deemed as having paid fees (Cohort 2) or were deemed as would have been paying fees had fees been in place (Cohort 1). The characteristics of respondents are given in the Appendix.

Analytical approach

The analysis aimed to identify whether there were differences in outcomes between two cohorts of students: the last to enter before and the first to enter after the introduction of fees. The approach taken was, first, to identify significant differences between cohorts using crosstabular analysis. Multivariate analysis (logistic regression or OLS, as appropriate) was then used to identify whether differences (or lack of differences) were related to changes in the composition of students or to changes in behaviour. Data were pooled across cohorts, with each model including a cohort dummy and cohort interaction terms for every independent variable. The model was reduced to include variables where there was a significant cohort effect and other significant variables.

Three outcomes were examined:

* term-time employment;

* expected debt at the end of the course; and

* satisfaction, proxied by whether the student had any regrets about having gone to university.

The models contained a number of common independent variables. To save repetition these are described here with variations described in the appropriate sections below. Each model contained the following independent variables:

* Indicators of financial pressure, proxied by resources (whether the family provided financial support, receipt of various grants), including indicators of parental income (fathers" wage, father' employment status, occupational class, whether the student attended a state school), whether the student lived with parents (e.g. parents provide financial support, live with parents, receive other funds), the length of the course and the degree to which financial considerations had affected choice of university. Term-time working and debt were expected to increase with financial pressure and hence with fees and satisfaction was expected to decrease with financial pressure and hence with fees.

* Indicators of future earning power, proxied by subject of study (grouped into four: science and engineering degrees; social sciences and vocational degrees; humanities degrees; and art degrees). Term-time working was expected to decrease and debt and satisfaction to increase as expectations of future earnings rose.

* Indicators of educational demands, proxied by the subject of study (grouped according to the average time demands placed on attendance. In descending order: art and architecture; science and engineering degrees; other) and by the university (qualitative research in the four universities found that University A made fewest concessions to assist the combination of term-time employment and studying and University D made most; the relative rank of the other two was unclear). Term-time working and satisfaction was expected to decrease and debt to increase with greater educational demands.

* Value placed on higher education, proxied by the factors which had affected their university choices (future employment prospects, interest in their subject and the importance of social and leisure activities) and by whether the student's father had a degree. The former as an indicator of the value placed on education was suggested by the strong link between higher education participation and parental education (Burnhill et al., 1990; Jesson and Gray, 1990; Payne, 1995; Robertson & Hillman, 1997). Indeed Metcalf (2003) found that father having a degree was more strongly inversely associated with term-time employment than was family income or occupation, suggesting more than just a financial effect on term-time employment. Tern>time working was expected to decrease and debt and satisfaction to increase the greater the value placed on higher education. In terms of the specific variables, term-time working was expected to decrease and debt and satisfaction to increase with the importance of future employment prospects and subject interest. The expected relationship with the importance of social and leisure time was not clear. For term-time working and debt this would depend on whether time or money requirements for leisure predominated. Term-time working was expected to be inversely related to whether the student's father had a degree,

* Personal characteristics: gender, whether the student was disabled or not and ethnicity (owing to the small numbers of ethnic minority respondents, ethnicity was classified as ethnic minority and ethnic majority).

In addition, for appropriate models, participation in paid employment, earnings from paid employment and the size of expected debt at the end of the course were included. The variables used in the models are tabulated in the Appendix.

Findings

Term-time working

Cross-tabular analysis showed that student employment had increased across the two years, from 74 per cent employed at some time during the year for Cohort 1 and 84 per cent for Cohort 2 (table 1). There was no significant increase in the percentage of students working term-time, nor in the hours of term-time employment, which averaged thirteen hours per week (for those employed) in both years, although the pattern of working changed. The increase in employment was almost wholly confined to vacation work. This suggests that, whilst students sought to minimise debt or increase spending power, fees had led to a compositional change towards students who were less likely to work term-time or had changed student behaviour, with an increase in the value placed on educational achievement.

Logistic regression suggested that the fees had affected the propensity to work term-time. An increase in term-time working was marked for students whose fathers were graduates, a group who had had very low rates of term-time working in Cohort 1 (table 2). There was also a significant increase in term-time working amongst those whose university choice had been affected by financial considerations and those who lived at home. Both groups had already had a higher propensity to work term-time prior to the introduction of fees. Thus, strategies to reduce costs (through university choice and through living at home) did not appear adequate to obviate the necessity of working term-time. It seems likely that both groups were already subject to greater financial pressures and that higher university costs had further increased pressure, leading to a rise in term-time working.

A number of other factors appeared to affect term-time working, although these were unchanged by the introduction of fees.

Students who did not receive financial support from their parents were more likely to work term-time. At the same time, the other indicators of financial pressure did not appear to affect term-time working. In particular, there was no difference in term-time working by receipt of hardship grant. This alone might suggest that hardship grants were well-targeted and of appropriate size. However, in combination with the finding on parental support, the lack of difference in term-time working by receipt of hardship grant raises the question of whether hardship grants are appropriately targeted.

Attitudes also appeared to play a part in term-time working. In the main, these were as hypothesised. Those for whom subject interest had been very important in their university choice were less likely to work term-time. Similarly, those for whom future employment considerations had been very important in their choice of university were less likely to work term-time. This suggests that such individuals were more willing to swap immediate income to achieve their employment plans.

Future employment considerations in terms of likely earnings also appeared to affect the term-time employment decision, in that those with lesser employment expectations (i.e. those studying humanities and, a fortiori, art) were more likely to work term-time. Contrary to expectations, students for whom quality considerations were very important (i.e. who for whom this had been very important in choice of university) were more likely to work term-time. We were unable to explain this pattern.

The findings on the effect of educational demands on term-time working were more mixed. Those studying social sciences and humanities, the courses with the lowest attendance requirements, were more likely to work term-time, as hypothesised. However, the pattern of term-time working by university was less easily explainable. The percentage of students working term-time differed by university, with term-time working at about half the rate in University A compared with the other universities (table 3). However, the multivariate analysis identified university differences between University C vis-a-vis Universities A and D and University B vis-a-vis Universities A and D only (i.e. the large difference in term-time working between Universities A and D was accounted for by the differences in composition of the students, rather than any other 'university effect'). From qualitative information about the four universities, it seems unlikely that this pattern reflects differences in the demands placed on students by the universities (as hypothesised), as University D certainly appeared responsive to students who were employed term-time, whereas University A appeared to be the least responsive. (An alternative explanation of this pattern might be expected to lie in differences in the local labour markets: it may have been easier for students at Universities B and C to gain employment than at Universities A and D. Although the study did not collect detailed information on the local labour markets (relating to student jobs), this explanation seems unlikely, as only 1 per cent of students said they were not working because they could not find a job (or a suitable job)).

What do these results say about equality? As the above results suggest that term-time employment tends to be undertaken due to financial constraints, such employment must be seen as a disbenefit to the student. Thus the rise in term-time working amongst groups who are already financially constrained (those who are concerned about costs when they apply to university and those who live at home) is likely to be indicative of an increase in inequality. However, the effect is not all one way: the increase amongst students with graduate fathers can be seen as equalising the higher education experience. Having a graduate parent is one of the greatest influences on going to university at all and, for our four universities, this group had a substantially lower rate of term-time employment prior to the introduction of fees. Thus the increase in term-time employment amongst this formerly privileged group may be seen as bringing their higher education experience more into line with that of other students.

Debt

Debt had grown across the two cohorts. More students in Cohort 2 than Cohort 1 expected to be in debt by the end of their course, 89 per cent and 84 per cent respectively. The average size of the expected debt (for those expecting to be in debt) had grown substantially from 5,371 [pounds sterling] (1) to 8,739 [pounds sterling] (table 4), an increase of 3,368 [pounds sterling], similar to the cost of fees for most fee-paying students.

Using OLS to investigate the effect of characteristics suggested that the predicted expected debt by the end of the course for the average (2) student rose from 4,353 [pounds sterling] prior to the introduction of fees to nearly 7,262 [pounds sterling] (3) once fees had been introduced (table 5). With a few exceptions, the rise tended to be similar across students irrespective of characteristics. Growth tended to be greater for those who did not receive financial support from their families, a group who, prior to fees, already had had higher than average expected debt. For such students, the predicted expected debt had risen from 5,406 [pounds sterling] to nearly 9,488 [pounds sterling]. Fees also appeared to hit two other disadvantaged groups: disabled students, whose expected debt rose by an additional 2,706 [pounds sterling], and students whose fathers were plant and machine operatives (SOC 8), an additional rise of 3,343, [pounds sterling] (4) resulting in predicted expected debt of over 9,968 [pounds sterling] and nearly 10,605 [pounds sterling] respectively, by the end of the course.

The only relationship found between paid work and debt was for those who worked every week. Prior to fees, those who worked every week term-time tended to have higher debt. However, for the post-fees cohort, this relationship was reversed (although the total effect was far smaller). It is possible that prior to fees, this relationship was due to those with higher consumption demands being more likely to work, whereas post-fees, the dominant causal factor was to minimise debt. However, further research would be necessary to draw confident conclusions about the causes of this pattern.

A number of other factors appeared to affect debt, although these were unchanged by the introduction of fees. Not surprisingly, debt rose with length of course: by an average of 1,567 [pounds sterling] per year. Living with one's parents appeared to be an effective way to keep debt low: students living with their parents tended to have 3,505 [pounds sterling] less debt, resulting in average expected debt of only 847 [pounds sterling] for Cohort 1 and 3,756 [pounds sterling] for Cohort 2. Predisposition also played a part in the expected acquisition of debt: students for whom social factors were a very important influence on university choice (who might, therefore, be expected to have higher expenditure) tended to have more debt, but only by an average of 549 [pounds sterling]. There was evidence of a class effect, with students who had been to private schools expecting lower debt, by 1,079 [pounds sterling]. We would suggest this was a result of greater support rather than frugality.

Hardship grants did not appear to provide adequate compensation to equalise finances across students, as those in receipt of hardship grants expected to incur higher debts by the end of the course (an additional 2,198 [pounds sterling]). However, this suggests that hardship grants were reaching those under greater financial pressure. The size of the debt had not risen with the introduction of fees, despite the percentage of students receiving hardship grants not increasing (5) (see Appendix).

In terms of equity, these results show a particularly large effect of the increase in costs of higher education on the indebtedness of two groups of already disadvantaged students: disabled students and those without financial support from their families. Disabled students will be further disadvantaged through greater difficulty in repaying their debt due to, on average, lower earning power. In addition, the higher indebtedness of state school students suggests a further inequity regarding higher education costs more generally. At the same time, hardship grants do not fully compensate for the financial difficulties of those in their receipt. Thus the current system appears to result in greater inequality in debt for those leaving higher education than previously.

Satisfaction

As discussed above, increased costs and, through introducing a buyer/seller relationship, fees, may have led students to consider more critically the services provided by their university. The hypotheses were that increased costs either through raising critical assessment of higher education provision or through changing the opportunity cost would have reduced satisfaction.

'Satisfaction' was proxied by the extent to which students regretted having gone to university. The survey provided information on a range of satisfaction measures, including satisfaction with individual university facilities and study provision. However, whether the student had any regrets was considered a particularly good measure of satisfaction. Firstly, the measure did not depend on a researcher-defined choice of factors of importance (for example, satisfaction with accommodation, satisfaction with library facilities), but allowed the measure to reflect what was important to the student. Secondly, it incorporated consideration of the alternative (of not having gone to university).

Very few students regretted going to university, per se. However, a large minority had some regrets. This had grown between cohorts, from 29 per cent for Cohort 1 to 34 per cent of Cohort 2 (table 6).

The reasons for regret were similar across cohorts, with a small percentage in each giving financial reasons and dissatisfaction with the nature of the experience or the learning (table 7). Debt was the most frequently cited reason for regret, cited by 14 per cent and 18 per cent of Cohorts 1 and 2 respectively. Being financially worse off having been to university was also a cause of regret. Dissatisfaction with the higher education experience was less common. However, none of these differed significantly across the two years and the only reason for regret which did differ significantly (poor value for money) caused regret amongst a very small percentage of students, 4 per cent and 8 per cent for Cohorts 1 and 2 respectively.

Analysis using logistic regression identified few differences between cohorts in the relationship between each characteristic and regret, suggesting that much of the growth in regret across cohorts was due to changes in the composition of students. The only significantly different effects were by subject and by whether university choice was affected by the ability to live at home. Those doing arts and architecture, prior to fees, were less likely to have any regrets about going to university, but this group was more likely to have regrets after fees had been introduced. This group had the highest formal demands on their time at university and it is possible that the increasing pressure, as higher education costs rose, to combine study, paid employment and leisure was most keenly felt by this group. The converse was true of those whose university choice had been influenced by the ability to live at home: prior to fees this group was more likely to have regrets about higher education, but, after fees less likely. One might expect the opposite, as there was likely to be an increase in those living at home out of financial necessity rather than preference. However, it may also be that those now considering living at home are more dedicated to going to university and so less likely, to feel regret. Further research into this choice and the effect on the higher education experience would be useful.

Although there was little evidence of fees affecting satisfaction, there was evidence of financial concerns affecting satisfaction. Those whose university choice had been affected by financial considerations were more likely to regret having gone to university. Interestingly, paid employment during term-time reduced the likelihood of regret, unless paid employment interfered with study (when it increased regret).

The model identified a difference by institution (those students at the old, prestigious university were less likely to have any regrets than those at any of the other three universities). Attitudes towards university appeared important. However, this seemed more related to having strong positive reasons towards university than having particular attitudes. Thus, those for whom employment considerations, social/leisure considerations or subject interest were very important were less likely to regret going to university, with employment (followed by social/leisure) most influential. In addition, those who took into account university quality in their choice of university were less likely to have any regrets, suggesting the benefit of well-informed choices or of discerning students.

There was no evidence of differences across social groups, although disabled students were more likely to regret going to university.

Thus it appeared that, if fees (or other cost increases), had led to a reduction in satisfaction, this was due to changes in the composition of students, rather than a reduction in the likelihood of satisfaction for any given type of student. There was no evidence of any effect on equality. However, the evidence did suggest that financial concerns could reduce satisfaction. Interestingly, in terms of satisfaction, term-time working was found to have positive effects, so long as it did not interfere with studying.

Discussion

At the start of this study, we hypothesised that fees (and other increases in costs) would have increased student debt, decreased student satisfaction with higher education and affected term-time working. The impact was expected to be greater on those who were already disadvantaged in terms of having fewer financial resources and to be greater on groups which were less attached to higher education (i.e. had lower participation rates in higher education).

Fees (or the general increase in costs) certainly appeared to have raised expected post-university debt and, for the average student, quite straightforwardly, by about the size of the fees. This is not as mundane a finding as it might at first appear. The most plausible explanation of this is that the composition of students has not changed towards those who can and do (or whose parents can and do) absorb these fees, i.e. it suggests that the feared impact on the composition of students, towards students from higher income families has been small. (6) It also suggests that the increase in cost has been passed directly to the student, which might be assumed to have been the intended outcome. However, this cost transfer is interesting in equity terms, as the charge is based on means-testing one group (parents) but the debt (and repayment) falls on another group (ex-students) and is not means tested by their means. (7) Fees (or the general increase in costs) also appear to have affected satisfaction with higher education, leading to a slight increase in those who had regrets about having gone to university. This appeared to be due mainly to a change in the composition of students (towards types who were more likely to have regrets about higher education) rather than due to fees resulting in a growth in dissatisfaction for any given type of student.

The differential effects of fees (or other increases in costs) on certain groups had a number of equity implications. Fees (or other increases in costs) appeared to exacerbate debt and term-time working where financial disadvantage was already present (for example, where concern about finance had affected higher education choices or students did not receive financial support from their parents). Whilst it is not clear that term-time working reduces the benefits of higher education, it was clearly seen as a disbenefit, per se, by students, and debt too would be a disbenefit. However, whilst fees seem to have raised debt and term-time working for those with financial difficulties, there was 710 evidence of a differential effect on satisfaction with higher education. At the same time, the impact of fees (or other rises in costs) on the term-time working of students with a graduate father appears to indicate a reduction in inequality, as the incidence of term-time working amongst this group rises from low to average levels. The impact on debt was greater for disabled students. This may be due to a smaller percentage of this group being able to reduce debt through paid employment, resulting in any increase in costs flowing directly into debt.

Fees (or other increases in costs) appeared not only to have increased the percentage of students living with their parents (see Appendix), but seemed likely to have changed the composition of this group also. This is suggested by the increase in satisfaction with higher education amongst those living at home and the rise in term-time working amongst this group, combined with the increase in the percentage living at home. However, it is also feasible that the rise in satisfaction is due to fees leading to a greater appreciation amongst students of the financial benefits of living at home, whilst the increase in term-time working may be because this group (which includes many who live at home for financial reasons) have fewer resources with which to cover increased costs or have a greater aversion to debt.

Whilst the main focus of the study was on the impact of fees (and other concurrent rises in costs), the analysis identified other factors which influenced debt, term-time working and satisfaction which have public policy implications. There was some evidence that finance affected whether the student worked term-time (affected the quality of the experience of higher education), the size of the debt at the end of the course and overall satisfaction with higher education. Hardship grants did not fully compensate to prevent higher levels of debt amongst those receiving hardship grants. If equality of experience at university is considered important, then the finding approach needs to compensate more fully those with financial difficulties.

The finding that satisfaction was higher amongst students who went to the old, successful university deserves more investigation. Whilst it could be simply that University A was simply better at satisfying its students than the other three universities, it raises the question whether the funding system (across research, as well as teaching) results in students at the old, successful universities being better provided for, which, given the greater concentration of students from higher social classes in the old universities, has equity implications.

The finding that satisfaction was greater amongst students with more positive reasons for going to university (whether they were education, employment or social) may merely indicate that certain types of people enjoy university more. However, it may also argue for better preparation for students (and potential students) and increased information, so that more students make more positive higher education choices.

Whilst it appears that fees (and other higher costs) have affected behaviour and outcomes and that there are a number of other financial aspects about higher education of concern, individual choice and control over some of these factors should not be ignored. Choice was apparent in the decision to work term-time, as those who placed greater importance on study and future employment appeared to be able to survive with less term-time employment. At the same time, some (but not all) students appeared able to choose to reduce debt through living at home, whilst others it appeared chose slightly higher debt to finance their social life.

The effects identified above are for fees of 1000 [pounds sterling] per annum. In September 2006, universities which satisfy the Office for Fair Access (OFFA) on how they will 'safeguard and promote fair access--in particular for students from low income groups--through bursary and other financial support and outreach work' will be able to charge fees of up to 3,000 [pounds sterling] per annum ('top-up' fees) (OFFA, 2004). The emphasis for satisfying OFFA is on bursaries for the 'poorest students' (i.e. those eligible for the full Higher Education Maintenance grant) to bridge the gap, if any, between the grant and fees (OFFA, 2004). Currently, details of proposed fees and the related support provisions are not available. However, it seems likely that fees of 3,000 [pounds sterling] per annum will be introduced for nearly all first degree courses at universities (although the approach to be taken by Further Education Colleges for degree courses is unclear) (The Guardian, 6 January 2005 and discussion with OFFA).

Does our research shed light on the implications of these changes? Whilst we cannot merely extrapolate the behaviour in response to a 1,000 [pounds sterling] per annum fee (in part, because the larger increase in fees is likely to alter participation in higher education), our research does provide some indicators. It seems safe to say that student debt and term-time working increase and more students will live at home. These are highly predictable. However, the research also suggests that the increase in debt could be substantial. Our finding that, on average, the full fee cost was passed into debt means that the same cannot be ruled out for top-up fees. In this case, the average student in our sample would have debts of around 15,000 [pounds sterling] at the end of their course, nearly triple that of students prior to fees. The research also suggests that student satisfaction with higher education is likely to decline.

In addition, the research suggests groups (in addition to those from very low income families) which may be particularly disadvantaged by top-up fees. The research found that the introduction of fees impacted more on disabled students (in terms of debt) and students (of standard age) whose families' did not provide financial support (in terms of debt and term-time working). This suggests that top-up fees may have a greater impact on these groups, both in terms of participation and the experience of higher education, unless additional financial support is provided. Whilst disabled people may benefit from OFFA's aim that universities should increase access measures for disabled people, the extent to which this will include additional financial provision is unclear. Students unsupported by their families are not mentioned in OFFA's advice on access agreements (OFFA, 2004). Unless additional financial provision is related to students' own financial resources and to students' special needs, not only will the quality of the higher education experience and outcomes become more skewed, but participation will fall amongst certain disadvantaged groups.

Appendix: Independent variables

 % of students sig. diff.
 or mean

 Year 1 Year 2

Indicators of financial pressure
Father's weekly wage (mean),
 [pounds sterling] 828 805
Father not employed 2 1
SOC
 1 35 34
 2 31 30
 3 10 10
 4 3 3
 5 9 12
 6 4 3
 7 2 3
 8 4 4
 9 1 3
Attended a state school 81 79
Family provides financial support 75 70 **
Receives hardship/access grant 8 7
Receives non-hardship donations
 (bursaries, grants) 12 4 ***
Lives with parents 16 21 **
Length of course (years) (mode) 3 3
Financial considerations affected
 university choice 49 49

Indicators of future earnings: course studied
Science 44 40
Social sciences and vocational
 courses 32 31
Humanities 15 21 **
Art 7 11 **

Indicators of educational demands
Indicators of time demands of
 the course
 Art and architecture 8 13 ***
 Science 44 40
 Humanities/social sciences 46 49
University facilitating term-time
 employment: work term-time **
 University A 37 42
 University B 28 18
 University C 19 21
 University D 16 19

Indicators of attitudes towards higher education
Father has a degree 31 31
Factors affecting university choice
 Employment prospects 46 47
 Subject interest 64 64
 Social life/leisure activities 34 37

Personal characteristics
Ethnic majority male 40 35
Ethnic majority female 50 53
Ethnic minority male 4 3
Ethnic minority female 6 8
Disabled 5 3

n 588 567

Table 1. Paid employment, % of students

 Cohort 1 Cohort 2

Employed during the year *** 74 84
 Employed term-time 45 47
 Employed vacations only *** 29 37

Working pattern **
 Works all weeks term-time 23 26
 Works most weeks term-time 14 11
 Sometimes works term-time 8 12
 Never works term-time 55 53
All students 100 100

Notes: Difference between cohorts: *** significant at 1 per cent; **
significant at 5 per cent ; * significant at 10 per cent.

Table 2. Factors associated with term-time working

Dependent variable: paid employment term-time; 0 did not have paid
employed term-time; 1 had paid employment term-time

 Coef. Std. Err. z

Constant -1.0524 0.2180 -4.828
Constant x Cohort 2 -0.6278 0.2244 -2.797
Father a graduate -0.6346 0.2112 -3.005
Father a graduate x Cohort 2 0.6374 0.3049 2.090
Financial considerations affected
 university choice 0.4045 0.1911 2.116
Financial considerations affected
 university choice x Cohort 2 0.6336 0.2788 2.273
Lives at home 0.5351 0.2688 1.990
Lives at home x Cohort 2 1.1127 0.3865 2.879
White female 0.5226 0.1417 3.687
Family does not financially support 0.4233 0.1601 2.643
University B 0.9062 0.1735 5.224
University C 0.7189 0.1753 4.101
Course demands: social sciences/
 humanities 0.3086 0.1661 1.858
Earnings potential: humanities 0.4933 0.2082 2.369
Earnings potential: art 0.8417 0.2419 3.479
Employment considerations affected
 university choice -0.3994 0.1662 -2.404
University quality considerations
 affected university choice 0.4080 0.1755 2.325
Subject interest affected university
 choice -0.4531 0.1502 -3.017
Number of observations = 1112

 P>
 [abso-
 lute
 value
 of (t)] [95% Conf. Interval]

Constant 0.000 -1.4796 -0.6252
Constant x Cohort 2 0.005 -1.0676 -0.1879
Father a graduate 0.003 -1.0484 -0.2207
Father a graduate x Cohort 2 0.037 0.0397 1.2350
Financial considerations affected
 university choice 0.034 0.0299 0.7790
Financial considerations affected
 university choice x Cohort 2 0.023 0.0872 1.1800
Lives at home 0.047 0.0081 1.0620
Lives at home x Cohort 2 0.004 0.3553 1.8702
White female 0.000 0.2447 0.8004
Family does not financially support 0.008 0.1094 0.7371
University B 0.000 0.5662 1.2462
University C 0.000 0.3753 1.0624
Course demands: social sciences/
 humanities 0.063 -0.0169 0.6342
Earnings potential: humanities 0.018 0.0851 0.9014
Earnings potential: art 0.001 0.3675 1.3158
Employment considerations affected
 university choice 0.016 -0.7251 -0.0738
University quality considerations
 affected university choice 0.020 0.0640 0.7520
Subject interest affected university
 choice 0.003 -0.7475 -0.1588
Number of observations = 1112

Notes: The constant shows the effect for a student who has none of the
characteristics listed (i.e. whose father was not a graduate, for whom
financial considerations did not affect university choice, who did not
live at home). The coefficients on the variables show the difference
in effect if the characteristic was as listed, i.e. the student's
father was a graduate, financial considerations did affect university
choice.

Table 3. Term-time working by university, % of
students

 Cohort 1 Cohort 2

 % of students n % of students n

University A 28 209 24 237
University B 54 158 69 100
University C 53 115 59 117
University D 60 92 65 108

Table 4. Expected debt at the end of the course, (a)
of students/[pounds sterling]

 Cohort 1 Cohort 2

Expected to be in debt ** 84 89
Mean debt, those with an expected debt 5371 8739
Standard deviation, those with an
 expected debt 3061 3881
n 587 567

Notes: (a) for Cohort 2 students who paid fees and their counterparts
in Cohort 1. ** significantly different at the 5% level.

Table 5. Expected debt at the end of the course

Dependent variable: expected debt at the end of course, [pounds
sterling]

 Coef. Std. Err. t

Constant -348.97 737.58 -0.473
Constant x Cohort 2 2908.55 278.41 10.447
Family does not support financially 1052.56 398.17 2.643
Family does not support financially
 x Cohort 2 1173.80 530.12 2.214
Father SOC 8 (plant and machine
 operatives) -1446.71 812.72 -1.780
Father SOC 8 x Cohort 2 3343.25 1177.04 2.840
Disabled -1217.19 714.20 -1.704
Disabled x Cohort 2 2706.27 1147.37 2.359
Works all weeks term-time 941.64 403.00 2.337
Works all weeks term-time x Cohort 2 -973.97 549.38 -1.773
Course length (years) 1567.40 201.89 7.764
Lives at home -3505.87 328.17 -10.683
Went to private school -1078.94 279.28 -3.863
Receives hardship grant 2197.54 426.07 5.158
Social factors very imp influence on
 university choice 548.97 234.55 2.341
Number of observations = 1070
Adjusted [R.sup.2] = 0.3026

 P> [95% Conf. Interval]
 [abso-
 lute
 value
 of (t)]

Constant 0.636 -1796.25 1098.31
Constant x Cohort 2 0.000 2362.25 3454.86
Family does not support financially 0.008 271.26 1833.86
Family does not support financially
 x Cohort 2 0.027 133.59 2214.01
Father SOC 8 (plant and machine
 operatives) 0.075 -3041.44 148.03
Father SOC 8 x Cohort 2 0.005 1033.65 5652.85
Disabled 0.089 -2618.61 184.22
Disabled x Cohort 2 0.019 454.89 4957.65
Works all weeks term-time 0.020 150.87 1732.41
Works all weeks term-time x Cohort 2 0.077 -2051.98 104.04
Course length (years) 0.000 1171.25 1963.55
Lives at home 0.000 -4149.81 -2861.93
Went to private school 0.000 -1626.95 -530.94
Receives hardship grant 0.000 1361.51 3033.57
Social factors very imp influence on
 university choice 0.019 88.74 1009.20
Number of observations = 1070
Adjusted [R.sup.2] = 0.3026

Table 6. Regrets about higher education, % of
students

 Cohort 1 Cohort 2
Had no regrets about going to
 university 71 66
Had some regrets about going
 to university *, of which 29 34
 Regretted going to university
 (unqualified) 2 2
 Regretted going to university
 in some respects 20 25
 Regretted going to this
 university 7 7
n 588 567

Note: * significantly different at the 10% level.

Table 7. Reasons for regrets about higher
education, % of students

Reasons for regret Cohort 1 Cohort 2

Financially worse off 12 13
Debt 14 18
Poor value for money *** 4 8
Haven't learnt what wanted/much 9 8
Not the experience expected 11 13
n 585 567

Notes: Other reasons given by a small percentage of students were not
expecting to get a degree, work would have been more enjoyable and
that university was too much hard work. Difference between cohorts:
*** significant at 1 per cent; ** significant at 5 per cent; *
significant at 10 per cent.

Table 8. Regrets about higher education

Dependent variable: whether had any regrets about going to university;
0 no regrets about going to university; 1 some regrets about going to
university

 Coef. Std. Err. t

Constant -0.2306 0.1975 -1.168
Constant x Cohort 2 0.4470 0.1630 2.742
University choice affected by ability
 to live at home 0.6442 0.2481 2.597
University choice affected by ability
 to live at home x Cohort2 -0.8961 0.3414 -2.625
Course demands: art/architecture -0.8571 0.3965 -2.162
Course demands: art/architecture x
 Cohort 2 1.0041 0.4829 2.079
Disabled 1.2525 0.3281 3.817
University A -0.6910 0.1663 -4.155
Employment prospects affected
 decision to go to university -0.5343 0.1556 -3.434
Subject interest affected decision to
 go to university -0.3185 0.1497 -2.127
Leisure/social considerations
 affected decision to go to
 university -0.4350 0.1438 -3.024
University quality affected
 university choice -0.3396 0.1549 -2.193
Finance affected university choice 0.2905 0.1462 1.987
Worked term-time -0.4766 0.2205 -2.162
Paid work affected studying 0.7728 0.2220 3.481
Number of observations = 1127

 P> [95% Conf. Inter-
 [abso- val]
 lute
 value
 of (t)]

Constant 0.243 -0.6176 0.1565
Constant x Cohort 2 0.006 0.1275 0.7664
University choice affected by ability
 to live at home 0.009 0.1580 1.1304
University choice affected by ability
 to live at home x Cohort2 0.009 -1.5653 -0.2269
Course demands: art/architecture 0.031 -1.6342 -0.0800
Course demands: art/architecture x
 Cohort 2 0.038 0.0575 1.9506
Disabled 0.000 0.6094 1.8957
University A 0.000 -1.0170 -0.3651
Employment prospects affected
 decision to go to university 0.001 -0.8393 -0.2293
Subject interest affected decision to
 go to university 0.033 -0.6119 -0.0251
Leisure/social considerations
 affected decision to go to
 university 0.002 -0.7169 -0.1531
University quality affected
 university choice 0.028 -0.6432 -0.0360
Finance affected university choice 0.047 0.0039 0.5771
Worked term-time 0.031 -0.9088 -0.0445
Paid work affected studying 0.001 0.3376 1.2079
Number of observations = 1127

NOTES

(1) The mean debt is of a similar scale to that found by Callender and Kemp (2000) for students completing in 1998/99 (i.e. who would not have paid fees) (4,781 [pounds sterling]).

(2) i.e. on a three year course, receiving financial support from their family, went to a state school, father not in SOC 8, not in receipt of a hardship grant, not disabled, living away from home, not working every week term-time and social factors not 'very important' in their choice of university.

(3) This includes all students, whether they expected to be in debt or not.

(4) Note that the model suggested lower levels of debt for these two groups prior to fees.

(5) Note that we are considering here students who are liable for fees and not those whose fees are waived.

(6) An alternative explanation is that the composition has changed towards higher income groups, but only towards those who are not willing to absorb the costs and so incur higher debt. This does not seem as plausible.

(7) This is not to ignore that the repayment of student loan debt is means-tested.

REFERENCES

Barke, M., Braidford, P., Houston, M., Hunt, A., Lincoln, I., Morphet, C., Stone, I. and Walker, A. (2000), Students in the Labour Market: Nature, Extent and Implications of Term-time Employment among University of Northumbria Undergraduates, Research Report No. 215, London, DfEE.

Burnhill, P., Garner, C. and McPherson, A. (1990), 'Parental education, socio-economic group, and entry to higher education, 1976-1988', Journal of the Royal Statistical Society, 153, 2, pp. 233-48.

Callender, C. and Kemp, M. (2000), Changing Student Finances: Income, Expenditure and Take-up of Student Loans among Full and Part-time Higher Education Students in 1998-99, Research Report no.213. London, DfEE.

Callender, C. and Kempson, E. (1996), Student Finances: Income, Expenditure and Take-up of Student Loans, London, Policy Studies Institute.

Connor, H. and Dewson, S. with Tyers, C., Eccles, J., Regan, J. and Aston, J. (2001). Social Class and Higher Education: Issues Affecting Decisions on Participation by Lower Social Class Groups, Research Report No. 267, London, DfEE.

Cubie, A. (1999), Student Finance: Fairness for the Future. Report of the Independent Committee of Inquiry into Student Finance, Edinburgh, The Scottish Executive, http://www.studentfinance.org.uk/final/

Curtis, S. and Shani, N. (2002), 'The effect of taking paid employment during term-time on students' academic studies', Journal of Further and Higher Education, 26, 2, pp 129-138.

Education and Employment Committee (2001a), Higher Education: student retention, Sixth Report, HC 124, London, House of Commons.

--(2001b) Higher Education Access, Fourth Report, HC 124, London, House of Commons.

Ford, J., Bosworth, D. and Wilson, R. (1995), 'Part-time work and full-time higher education', Studies in Higher Education, 20, 2, pp. 187-202.

Jesson, D. and Gray, J. (1990), Access, Entry and Potential Demand for Higher Education amongst 18-19 year olds in England & Wales, Youth Cohort Report No 12, Sheffield, Employment Department/Training Agency.

Lindsay, R. and Paton-Saltzberg, R. (1996), The Effects of Paid Employment on the Academic Performance of Full-Time Students in a British 'New' University, Consultancy Report, Oxford, Oxford Brookes University. http://www.brookes.ac.uk/schools/social/ psych/staff/lindsay.html

Metcalf, H. (2003), 'Increasing inequality in higher education: the role of term-time working', Oxford Review of Education, 29, 3, September.

OFFA (2004), Producing Access Agreements OFFA guidance to institutions, Bristol, OFFA, http://www.offa.org.uk/pubs/04_01.pdf

Payne, J. (I 995), Qualifications Between 16 and 18: A Comparison of Achievements on Routes Beyond Compulsory Schooling, Youth Cohort Report no 32, Sheffield, Employment Department.

Roberts, R., Golding, J., Towell, T., Reid, S., Woodford, S., Vetere, A. and Weinreb, I. (2000), 'Mental and physical health in students: the role of economic circumstances', British Journal of Health Psychology, 5, 3, September, pp. 289-97.

Robertson, D. and Hillman, J. (1997), Widening Participation in Higher Education by Students from Lower Socio-economic Groups and Students with Disabilities, National Committee of Inquiry into Higher Education, Report 6, London, HMSO.

Stevens, P. and Weale, M. (2004), 'Lazy students? A study of student time use', National Institute of Economic and Social Research, Discussion Paper No. 233.

UNITE/MORI (2003), Student Living Report 2003, Bristol, UNITE. http://www.unite-group.co.uk/UNITEDM.DLL/ Student_Living Report_2003.pdf?/docs/1760.dat

* National Institute of Economic and Social Research. e-mail: h.metcalf@niesr.ac.uk. This paper has been produced as part of a Leverhulme Trustfunded study into the effects of fees on higher education. We are most grateful to the Leverhulme Trust for its support. My thanks also to the other members of the research team: Phil Stevens, Heather Rolfe and Martin Weale.