Have the changes introduced by the 2004 Higher Education Act made higher education admissions in England wider and fairer?

‘Widening participation’ and ‘fair access’ have been contested policy areas in English higher education since at least the early 1990s. They were key facets of the 2003 White Paper – The Future of Higher Education – and the subsequent 2004 Higher Education Act, with stated objectives that the reach of higher education should be wider and fairer. In particular, there has been considerable concern about admissions to ‘top universities’, which have remained socially as well as academically exclusive. The principal policy tools used by the Act were the introduction of variable tuition fees, expanded student grants, discretionary bursaries and the new Office for Fair Access (OFFA). This paper draws on publicly available statistics to assess whether the changes implemented by the 2004 Act have indeed made access to English higher education wider and fairer in relation to young people progressing from state schools and colleges and from lower socio‐economic groups. It concludes that, while there is some evidence for modest improvements, these have been concentrated outside the ‘top universities’, which have seen slippage relative to the rest of the sector. The paper concludes with a discussion of the reasons why financial inducements appear to be a flawed and naive approach to influencing student demand.

class gap', where 'people from lower socio-economic backgrounds make up around one half of the population of England, but represent just 29 per cent of young, fulltime, first entrants to higher education' (NAO 2008, 6). The 'widening participation' agenda has been extremely prominent in the 2000s, with an estimated £392 million invested annually from the public purse into recruiting and supporting students from groups that are under-represented in higher education (NAO 2008).
While one focus has been on widening participation, a parallel policy concern about 'fair access' to 'top universities' has also received significant attention. This issue has centred on the imbalance of admissions between institutions, in the wake of the integration of polytechnics and colleges into a single higher education sector in 1992. Broadly speaking, those institutions with the most rigorous entry requirements drew their student body from private schools and middle-class households, while those that catered for low-ability levels tended to recruit more heavily from statefunded schools, further education colleges, minority ethnic communities and workingclass households (Ainley 1994;Modood and Shiner 1994;NCIHE 1997).
The political agenda of the late 1990s and early 2000s was thus twofold: to increase the proportion of students drawn from lower socio-economic groups and to work towards a more equitable demographic distribution within the sector. Components have included the government-funded Aimhigher initiative to raise aspirations among schoolchildren (DfES 2003a), a major review of admission practices (Schwartz 2004), targeted additional financial support (DfEE 2000), institutional outreach work and other approaches designed to break down real and perceived 'barriers' (NAO 2002;Universities UK 2005;Gorard et al. 2007). This paper will analyse the impact of the 2004 Higher Education Act on the fair access and widening participation agenda, investigating the policy levers used by the government and their effect on entry to higher education, especially to the highest status universities. 3

The 2004 Higher Education Act and after
Despite this policy interest in widening participation and fair access, higher education in England in the 2000s continued to be highly stratified, with institutions pursuing strongly contrasting missions and attracting very different student populations, both in terms of prior attainment and demographic background. The 'Russell Group' of universities occupies what is perceived as the top position within this hierarchy, typified by large civic institutions with a strong research basis, very taxing entry requirements and a high demand for each available place. There are currently 16 Russell Group universities in England, 4 comprising around 13% of the higher education institutions and accounting for 25% of the full-time young first degree population. The Russell Group has no official status (it is a closed membership organisation), but it exerts considerable influence on government policy by dint of its market share, reputation for excellence, lobbying power and prestigious alumni.
The Russell Group institutions' reputation for excellence, however, does not extend to their performance in social inclusivity. Concern has long been voiced that 'top universities', as typified by the Russell Group, are socially as well as academically exclusive. The Dearing Report (NCIHE 1997) noted the demographic differences between the former polytechnics and older universities. These were drawn into even sharper focus by a series of reports by the Sutton Trust, which found that 'the chance of getting into a top 13 university is approximately 25 times greater if you come from an independent school than from a lower social class or live in a poor area and is about double what it should be' (Sutton Trust 2000, 1). They later referred to the 'missing 3,000' who are 'a potential waste of talent' because they attend lowerstatus universities than they might (Sutton Trust 2004, 1). While these reports did serve to make clear the stark differences in social mix between the various tiers of the higher education hierarchy, they focused on the admission procedures and the approach of universities without seeking to understand the deeper social and cultural processes and individual choices which underpin student demand for higher education. They were also founded on an elitist agenda which presupposed that the 'top universities' offer the courses which are best suited to the most talented young people, underplaying the very high-quality provision that exists in other universities, especially in specialised and vocational areas, as well as the role of individual preferences.
The fair access theme was picked up and partly challenged in contemporary governmental documents, which found that 'the very significant differences in patterns of application are the main cause of differential access to universities, while admission procedures appear to be far less significant' (DfES 2003a, 12). In other words, it suggested that it was not that higher-status universities purposefully discriminated against applicants from lower socio-economic groups or state schools, but that these people were far less likely to apply, even when they had attained the highest levels of pre-entry qualifications. Top universities were challenged to 'understand why it is that students may be put off from applying to their institution' and to actively 'encourage a broader range of applications' (DfES 2003a, 13). There is a key tension here with the Sutton Trust position, as outlined above.
Steering a careful path through these contrasting perspectives, the 2003 White Paper, The Future of Higher Education, effectively skirted around the issue. On the one hand, it asserted that: 'Education must be a force for opportunity and social justice, not for the entrenchment of privilege' (DfES 2003b, 67), promoting efforts made by a number of Russell Group members to ensure 'fair access' for 'nontraditional students' by using transparent and holistic admission procedures. On the other hand, it failed to challenge the status quo which led to the social segmentation within higher education, pushing the widening participation agenda towards lowerstatus institutions with the creation of two-year foundation degrees (Jones and Thomas 2005).
The resulting independent report into university admissions (Schwartz 2004) sought to establish five principles that underpin fair access: transparency; ability to select by achievements and potential; reliability and validity; minimisation of barriers; and a professional approach. It concluded that the system was 'generally fair' with 'room for improvement' (Schwartz 2004, 4). Among the suggestions were approaches that acknowledged the different educational origins of students, concluding that seeking 'latent talent and potential, which may not fully be demonstrated by examination results, is a legitimate aim for universities and colleges which seek to recruit the best possible students' (Schwartz 2004, 5). This egalitarian position was somewhat undermined by the addition of the three words: 'regardless of background'. In other words, the applicant's background was to be disregarded. Schwartz therefore stopped short of a more radical approach to address fairness of access, encouraging the greater use of contextual information, but leaving individual institutions to formulate their own admission policies and steering them away from using 'bias' to socially engineer their student mix. It is difficult to see, given the comments quoted above from The Future of Higher Education (DfES 2003b), how this would significantly impact on fairness; patterns in application would continue to cause differential access.
Given these various policy contexts, there was significant unease surrounding the 2003 White Paper and the eventual passage of the 2004 Higher Education Act. Academics, commentators and politicians (e.g. Sanders and Goddard 2003;Callender and Jackson 2005;NatWest Bank 2005) saw a strong danger that the partial deregulation of tuition fees would discourage students from under-represented groups from seeking entry to higher education, particularly to Russell Group universities which were widely expected to charge the highest possible tuition fees (£3000 per year 5 ). The initial signs were that this was indeed the case (Lipsett 2005). The government sought to offset these dangers by offering large grants (up to £2700 per year) to students from low-income households, allowing the payment of fees to be deferred through a loan and increasing the post-graduation income threshold for loan repayments (DfES 2003b). This was coupled with a new regulatory body, the Office for Fair Access (OFFA), which was conceived to 'commit universities to increase the take up of university places from the most disadvantaged groups' (DfES 2003b, 85) through the requirement to produce formal Access Agreements with the threat of financial penalties for those deemed not to be doing enough.
The Access Agreements were conceived in broadly two sections. The first would lay out the institution's outreach activities, while the second would describe their tuition fee policy and the concomitant response to the requirement on universities to reroute a proportion of their additional income into financial bursaries for students from lower-income households (DfES 2003b). The initial purpose of these new bursaries was 'to ensure that no additional financial barriers to entry are created' (DfES 2003a, 20) with the prevailing concerns about participation and fairness being explicitly acknowledged in the Secretary of State's instruction letter to the first Director of Fair Access (Clarke 2004) and OFFA's subsequent guidance to institutions (OFFA 2004).
The aims of OFFA were variously laid out to be to 'safeguard and promote access' and 'to protect the poorest students' with the expectation of 'the most [effort], in terms of outreach and financial support, from institutions whose records suggest that they have furthest to go in securing a diverse student body' (Clarke 2004). At its inception, the OFFA mission was therefore to protect the current admission profile in terms of social mix, and also to see progression on participation and access, especially from the 'top universities'. However, as McCaig and Adnett (2009) note, this stopped short of placing a requirement for specific improvements onto individual institutions, wherever they may have been in the hierarchy. For this reason, among others, OFFA has been repeatedly challenged for being light-touch and toothless (Thomson 2003;Baty 2004;Jones and Thomas 2005;Gill 2009) and its sanctions over universities have never been used (NAO 2008).
One intended consequence of the 2004 Higher Education Act was to create a market based on price competition for tuition fees. However, this policy objective was not realised with nearly every institution charging the maximum amount of £3000 from the outset. In contrast, it could be argued that a competitive market in bursaries has emerged with a myriad of schemes being put in place by higher education providers to further their individual missions in the context of their Access Agreements. Widely different sums of money have been made available to students using widely different eligibility criteria. A conservative estimate is that 350 separate schemes came into existence, focusing on income, academic merit, geographical location, school type, disability, ethnicity and so on (Harrison, Baxter, and Hatt 2007). Just under £100 million was dispersed in this way in 2006/2007, accounting for around a quarter of the total additional tuition fee income.
Research by McCaig and Adnett (2009) has revealed a clear stratification within the market for income-contingent bursaries with higher-status universities (including the Russell Group) tending to offer roughly twice as much per student, on average, as lower-status institutions (NAO 2008). However, because of their entry profile, the numbers of bursaries offered by top universities are relatively low, with lower-status universities offering bursaries to a larger pool of students. McCaig and Adnett (2009) found that, contrary to the Secretary of State's expectations (Clarke 2004), there was no evidence that 'top universities' were investing more than those with an already diverse entry profile, although there was significant variation between institutions. Furthermore, they note that institutions have tended to reduce their expenditure on income-based bursaries since 2008/2009 in favour of scholarships based on merit, 'academic potential' or subjects which struggle to recruit. In a contrary view, Callender, Wilkinson, and Hopkin (2009) have recently argued that bursaries have been particularly efficacious in widening access to Russell Group institutions, where students report that the bursary offer is influential in their decision-making processes.
At the end of the decade, widening participation and fair access continued to be key policy issues, and there were signs that positions had begun to harden, with more critical comment emerging about progress made by Russell Group universities. A National Audit Office report into widening participation found that with reference to sector-wide widening participation benchmarks (these will be explained in more detail below), 'on average, Russell Group institutions performed significantly below their benchmarks, and [the former polytechnics] performed significantly above theirs' (NAO 2008, 20), leading to very different social mixes. The report of a governmentsponsored panel on access to the professions (PFAP 2009) also drew strong attention to the disparities between Russell Group and other institutions in terms of their entry profile and the implications for students' career prospects.

Methodology
This paper seeks to answer two research questions about the effects of the 2004 Higher Education Act, including the student financial support changes and the inception of OFFA and Access Agreements. These issues lie at the heart of the widening participation and fair access agendas which have been recurrent policy themes for at least 10 years. The particular questions are: (1) To what extent has the Act helped to widen participation? This will be investigated through analysing two of the national widening participation performance indicators, which measure admissions of students from state-funded schools and colleges and from lower socio-economic groups.
(2) To what extent has the Act helped to promote fair access? This will be investigated through analysing the performance of the Russell Group universities on the same indicators relative to other institutions with specific reference to nationally calculated benchmarks.
These questions will be investigated through analysing the data published annually by the Higher Education Statistics Agency (HESA 2010). The dataset provides sector-wide performance indicators on widening participation for each of the institutions in England.
The particular dataset used here relates to young (aged under 21 on entry) full-time undergraduate students on their first degree course (i.e. it excludes sub-degree level courses). This subset comprises around two-thirds of the total full-time higher education entrants. This was selected in order to limit the possible confounding factors, such as differing financial support arrangements for older students and the creation and growth of foundation degrees, and to thereby offer a broadly homogenous population across the period of analysis. The timeframe used is the six academic years from 2003/ 2004 to 2008/2009 6 and the dataset refers to new entrants. 7 Three widening participation performance indicators are published annually for each institution: (1) The proportion of entrants coming from a state-maintained school or college (in contrast to fee-paying private schools). This indicator has been developed in response to concern that certain universities, such as the Russell Group institutions, attract a disproportionate number of their students from private schools. Across the sector, the proportion from state schools and colleges has risen from 86. For the purposes of this study, the institutions in the HESA dataset have been divided into three groups representing: (1) the Russell Group institutions, 8 (2) a 'Pre-1992' group of institutions comprising those that had university status prior to 1992, and (3) an 'Other' group comprised of higher education institutions not falling into the other two groups, including those gaining university status since 1992, higher education colleges and small specialist colleges. The data has been analysed for the six-year-period from 2003/2004 to 2008/2009, during which the new financial and regulatory regime announced in the 2004 Higher Education Act came into operation. The trends in the proportion of entrants who come from state schools and lower socio-economic groups have been analysed for each of the three groups so that the results can be compared.

Findings
The headline data for widening participation presented above therefore suggests that access to full-time degree-level courses in England for young students has become slightly wider overall since 2003/2004. By 2007/2008, more young students from state schools and colleges and from lower socio-economic groups were attending such courses than five years previously; a trend that continued into 2008/2009 for the former. To place these statistics in context, around 42% of the population are drawn from NS-SEC groups 4-7, while an estimated 7-14% of young people attend private schools prior to their entry to university; no exact figures exist and the proportion with some experience of private education is doubtlessly higher than the proportion whose whole schooling has been via this method (Lockhart 2009).
We now turn to contrast the average entry profile of the Russell Group universities in England with the remainder of the English higher education sector with reference to the proportion of students recruited (1) from state schools and colleges, and (2) from NS-SEC groups 4-7. For this analysis, two comparator groups have been constructed.
The first is a 'Pre-1992' group of those institutions that had university status prior to 1992, which broadly form the stratum immediately below the Russell Group in terms of reputation. These are typically medium-sized institutions with a strong research base. Twenty-four institutions are included in this group, comprising around 25% of the total student population (and therefore of a similar overall size as the Russell Group).
The second is an 'Other' group of higher education institutions composed of those which are not members of the Russell Group or the Pre-1992 group. This is a very heterogeneous group, dominated by the former polytechnics and colleges which became universities in 1992 or in the 18 years since. It also includes a number of higher education colleges and small specialist colleges; this 'Other' group comprises around half of the contemporary higher education sector by student numbers.
In order to construct the average entry profile statistics for the Russell Group, Pre-1992 group and Other group presented below, the individual performance indicators for each institution were extracted from the HESA datasets and arithmetically weighted by the size of their entry cohort before being aggregated into either the Russell Group, Pre-1992 group or Other group average. This was undertaken for each of the six academic years, where data was available.

State schools and colleges
The average percentages for entrance from state schools and colleges for the two groups are presented in Table 2 and Figure 1. The data shows that, after a brief dip in 2004/2005, there has been a steady rate of improvement in all three groups, with the difference between the Russell Group and the English average growing slightly, demonstrating that the pace of improvement in this group is lower than in others.
There is no significant change in the relative entry profiles between the groups in 2006/2007 compared to 2005/2006, which would be concurrent with the changes brought about by the 2004 Higher Education Act. This suggests that the implementation of the Act has not affected the proportion of students entering each group from state schools and colleges.

NS-SEC groups 4-7
An identical approach was taken with regard to the proportion of students recruited from lower socio-economic groups and this is presented in Table 3 and Figure 2. Once again, we find with this indicator that the gap between the Russell Group and English average has grown across the period in question, from 8. 9% in 2003/2004 to 10.4% in 2007/2008. The proportion of students from these backgrounds recruited by Russell Group universities remained roughly constant at around 19%, while there were significant increases in 2005/2006 and 2006/2007 for both the Pre-1992 group and the Other group. It is notable that the rises for these groups occur either side of the key year of policy change rising from the 2004 Higher Education Act. In other words, these two groups experienced an increase in their proportions of students recruited from lower socio-economic groups across the point where tuition fees were increased and bursaries and Access Agreements were introduced, while there was no change for Russell Group institutions.

Using the location adjusted benchmarks
The other tools for examining questions concerning widening participation and fair access are the performance indicator benchmarks published as part of the HESA data. For each of the three widening participation indicators, a complex calculation is undertaken by HESA to determine what the indicator might be expected to be, based on those features of each individual institution which are not components of the performance of the institution, specifically including the subject, qualification and age profiles of the student body. 9 'Location-adjusted' benchmarks that adjust for the geographical locations from which the student body is drawn are also published alongside the basic benchmarks. It is this latter benchmark which is referred to the analysis below. There was a 1% of its student body from NS-SEC groups 4-7, under-performing against a basic benchmark of 29.1% based on its subject mix and the age and prior qualifications of its students. However, the locationadjusted benchmark was set at 26.3%, recognising that the University recruits disproportionately from the affluent south-eastern region of England, where there are fewer households from NS-SEC groups 4-7. Once location has been taken into account, it therefore is assessed as recruiting very close to the student mix by socio-economic classification that might be expected, relative to the higher education sector as a whole.
The location-adjusted benchmarks therefore provide a yardstick for assessing the performance of individual institutions in relation to the three widening participation indicators. In general, the Russell Group universities have tended to perform significantly below their benchmarks, in line with ongoing concern about equality and fairness (NAO 2008). In the 2007/2008 dataset, 10 of the 15 institutions were significantly underperforming relative to their benchmark for entry from state schools and colleges, compared to two significantly exceeding the benchmark and three achieving close to it. For the NS-SEC indicator, five universities were underperforming, none were exceeding and 10 were close to the benchmark.
Using the HESA datasets for the five-year-period prior to 2008/2009, the simple arithmetic difference between actual performance and location-adjusted benchmark was calculated for each institution, giving a measure of the extent to which they are under-performing (a negative figure) or over-performing (a positive figure). Averages weighted by size of intake were calculated for the Russell Group, Pre-1992 group and Other group of institutions and these are presented in Table 4 and Figures 3 and 4. We find that, on average, the Other group and the Pre-1992 group tended to perform slightly above their benchmarks for both indicators, while the Russell Group performed well below its collective benchmark, particularly for admissions from state schools and colleges.  Against the schools benchmark, the average performance of Russell Group universities declined from being 5.20% below the benchmark in 2003/2004 to 6.21% below in 2007/2008. In contrast, the performance of the Pre-1992 and Other groups remained broadly static. The major change in Russell Group performance coincides directly with the change in student financial support arrangements in 2006/2007. This analysis provides specific evidence for a decline in the proportion of admissions from widening participation groups into Russell Group universities, relative to the benchmark expectations of institutions with their particular mix of students. At the same time, the performance of both remaining groups has improved to some extent. In terms of performance against the NS-SEC benchmark, there is evidence for deterioration since a high point in 2004/2005 in the Russell Group and Pre-1992 group. In contrast, the average performance of the Other group has been broadly static across the period in question. There is no evidence here for a change in either direction associated with 2006/2007. This evidence notwithstanding, it could be argued that the aggregation and averaging process used in this paper masks a diversity of performances within the Russell Group. For the final piece of analysis, we turn to look at how each of the 15 Russell Group universities progressed relative to their benchmarks taking 2004/2005 and   2007/2008 as census years, as these avoid the years immediately before and after the financial support changes; the anticipation and then implementation of policy changes could, arguably, provide some anomalous results in the short term. In relation to state schools, six of the 15 Russell Group universities improved their performance relative to their benchmark, with nine deteriorating. With reference to recruitment from lower socio-economic groups, only two institutions improved, with 13 declining. While this does speak to diversity within the Russell Group, it remains clear that, with respect to these two benchmarks, the majority of members performed worse in 2007/2008 than in 2004/2005, before the changes to the student financial support system laid down in the 2004 Higher Education Act were implemented.

Discussion
We begin the discussion with a summary of the findings presented above: (1) Across the English higher education sector as a whole, between 2003/2004 and 2008/2009, there have been modest improvements in the recruitment of young full-time students to first degree courses from state schools and colleges and lower socio-economic groups.
(2) Considering the basic proportions, the national improvement in attracting young people from state schools and colleges has been located universally across Russell Group, Pre-1992 and Other institutions, albeit at different rates, while the increase in students from lower socio-economic groups has been driven solely by institutions outside the Russell Group. (3) Using the location-adjusted benchmarks, the relative performance of the Russell Group universities has declined across the period in question for both indicators, but especially with respect to recruitment from state schools and colleges, where it has fallen rapidly since 2006/2007. (4) Looking behind the aggregated figures, we also find that the majority of Russell Group universities declined in terms of their performance relative to The original purpose of this paper was to consider the impact of the policy changes introduced through the 2004 Higher Education Act on the recruitment of students from non-traditional backgrounds, bearing in mind the findings of the Dearing Report (NCIHE 1997). The subsidiary purpose was to consider this question with specific reference to 'top universities', as represented by the Russell Group, where specific concern about fairness of access has been voiced over many years (e.g. NCIHE 1997;Sutton Trust 2000;NAO 2008;PFAP 2009). The changes instituted in 2006/2007 might be hypothesised to have a number of possible outcomes. A substantial increase in tuition fees, from £1200 to £3000, might be expected to have a dampening effect on demand for higher education, especially for the most price-sensitive students in less affluent households (Callender and Jackson 2005;Pennell and West 2005). However, the introduction of grants and bursaries for this very group, sometimes of a greater amount that the tuition fee, might be expected to act to ameliorate this or act as an incentive (Pennell and West 2005), while the other provisions contained within the Access Agreements (e.g. outreach work) were designed to boost demand specifically from under-represented groups.
The findings of this study suggest that, in terms of young entrants to degree courses at least, the radical changes in financial support have not adversely affected participation from the core groups of concern. The overall proportion of students from state schools and colleges and from the lower socio-economic groups has continued to rise, pointing to the success of the Aimhigher programme and other governmental, regional and institutional initiatives. Indeed, 2006Indeed, /2007 was the year of greatest growth in the proportion of students from NS-SEC groups 4-7. Arguably the proportions may have grown faster without the 2006/2007 changes, but without a control group, this is impossible to test.
It is also apparent that there is a distinct pattern for the Russell Group universities. In common with the general pattern, they have increased the proportion of their intake from state schools, but they continue to lag well behind the Other and Pre-1992 groups in terms of the proportion of their entrants from this educational route. Their rate of growth has also been slower than the national average. The Russell Group universities have not improved their position with respect to students from lower socio-economic groups, as this proportion has remained fixed at around 19%. Similarly, the majority of the 15 institutions are faring worse since the implementation of the 2004 Act relative to the expectations set of them through the location-adjusted benchmark.
It would, however, be wrong to assume that these broad trends can be attributed directly to the changes implemented in 2006/2007. With one exception (i.e. performance of Russell Group universities against the state school and college benchmark), there are no obvious changes in student profiles that occur precisely at this threshold. This is despite the significant changes to student funding and the new fair access expectations placed on institutions through OFFA. It might be argued that such changes would need time to bed down before the policy objectives are met. However, three years have now passed and there is no sign of the types of shift that were sought by the Act. While the outreach activities contained in Access Agreements might be expected to take time to bear fruit, as we have seen, the radical changes to student funding were predicted to have instant and catastrophic results; these have quite simply not come to pass, in terms of initial entry at least.
It could be hypothesised that the academic years either side of significant change may be unusual. Students have a degree of choice about when they choose to enter higher education, as seen through the increasingly common phenomenon of 'gap years' (Heath 2007). They may, therefore, take an active decision to maximise their financial resources by either delaying (for better state support or the chance to accumulate savings) or bringing forward (to avoid adverse state support changes) their entry (Wainwright 2005). No information is available on students who attempt to 'play the system' in this way, but it may go some way to explaining why a single event effect is not found. Another explanation could be that the highly publicised and controversial changes, which had been in the public arena since the publication of the White Paper in 2003, were actually exerting some effect before their implementation. This could occur through heightened public awareness of the costs of higher education or misunderstanding leading to a belief that changes, especially negative ones, had been enacted earlier than reality. In the absence of repeated studies of student attitudes in this period, it is impossible to assess this possibility directly. There was certainly a sharp jump in public opinion concerning access to higher education and student financial support in 2004, captured in the British Social Attitudes reports (National Centre for Social Research 2010). 10 However, the most pragmatic interpretation is simply, of course, that the changes implemented by the Act were not a major factor in defining the student entry cohort in this period.
Russell Group universities have been in a position to offer the most generous bursary packages since 2006/2007. As they attract relatively few students from underrepresented groups, they have positioned themselves in the bursary pseudo-market by focusing on awarding relatively few, but very sizable, bursaries, generally amounting to several thousands of pounds (NAO 2008;McCaig and Adnett 2009). The purpose of this approach has presumably been to answer the Sutton Trust's call (Sutton Trust 2004) to attract a new cadre of very academically able students from under-represented groups who would not otherwise consider applying to the 'top universities'. For example, the University of Manchester Access Agreement opens with a 'commitment to being an open, meritocratic institution that pro-actively seeks out people capable of benefiting from higher education, removes barriers to their participation and in so doing contributes to the expansion of higher education opportunities' (University of Manchester 2009, 1). It is suggested (Callender, Wilkinson, and Hopkin 2009) that students are very cost-sensitive and that the large Russell Group bursaries have been particularly successful in shifting application patterns from students from lower socioeconomic groups. This study provides no evidence to support this proposition across the Russell Group as a whole, though it may hold for individual universities within it. A more convincing interpretation is the prediction made by Jones and Thomas that 'the bursary model … is likely to support exceptional students in elite universities, as opposed the majority of students from under-represented groups in all institutions ' (2005, 626). Note that the inference is about supporting existing students rather than encouraging additional applications. This may explain Callender, Wilkinson, and Hopkin's (2009) finding: students receiving bursaries from Russell Group universities have their decision-making cemented by the financial certainty of the additional money, but they were always the type of student that would make that leap. We will return to this theme later.
An alternative hypothesis is that the Russell Group bursary strategy has helped to ameliorate what could have been a decline in applications from these groups. The overall pattern in widening participation performance over the period in question has been one of relative decline compared to other institutions and, although Russell Group bursaries appear not to have made the distribution of students fairer, they may have helped to avoid it become even less fair than it was.
We need to consider why there has been an apparent policy failure: Why are students from lower-income households seemingly uninterested in the lucrative sums of money being offered by the 'top universities'? There is plenty of evidence from other research which can be drawn upon to address this question.
For example, we know that students from under-represented groups travel, on average, less far for higher education than other students (UCAS 2008). The reasons behind this geographical immobility are not yet well-understood, but friendship and family ties to the home area (Patiniotis and Holdsworth 2005;Holdsworth 2006;Clayton, Crozier, and Reay 2009), caring responsibilities (Evans 2009;Clayton, Crozier, and Reay 2009) and financial concerns (Patiniotis and Holdsworth 2005;Davies et al. 2008;Johnson et al. 2009) are likely to be contributory factors. For many such students, this means living in the family home, which is around twice as common among lower socio-economic groups (Patiniotis and Holdsworth 2005;UNITE 2007), while others seek nearby options that permit travel home at weekends. This therefore becomes a limiting factor on choice, reducing the geographical catchment area actively considered by these students. While the Russell Group universities are relatively evenly distributed around England, there are obviously some centres of population that have better access than others. Even where students do move away, financial considerations such as cost of living and access to part-time work are strong considerations (Callender and Jackson 2008).
There are also well-established factors determining institutional preference among non-traditional students. A large corpus of literature in the last 10 years (e.g. Thomas 2002;Archer, Hutchings, and Ross 2003;Yorke and Thomas 2003;Reay, David, and Ball 2005;Lehmann 2007;Gorard et al. 2007;Clayton 2009, 2010) has concluded that the institutional 'habitus' of top universities is alienating and unattractive for many students: 'not for people like us'. A pervading middle-class and elitist ethos acts as a deterrent and leads many academically able students to 'choose' not to maximise the status of the institution that they attend. Clayton (2010, 1115) go as far as to conclude that working-class students need 'almost superhuman levels of motivation, resilience and determination' to attend and thrive at elite universities.
A related feature of the higher education sector in England is that students from minority ethnic communities tend to prefer institutions or locations which already contain a significant proportion of students from their cultural background, or where they have cultural or family links, leading to the concentration of these students in a relatively small number of institutions and effectively acting as another constraint on mobility (Connor et al. 2004;Reay, David, and Ball 2005;Clayton, Crozier, and Reay 2009). Asian students, in particular, are more likely to choose to live at home, regardless of household income levels (Davies et al. 2008).
It would appear from the evidence presented in this study that the financial incentive of bursaries is not sufficient to overcome the pull of the local and familiar or concerns about the ability to fit in and thrive at Russell Group universities. While the historical record demonstrates that increasing costs of higher education have not acted as an overwhelming disincentive for students from lower socio-economic groups (with evidence of widening participation through the 1990s and 2000s), so the offer of lucrative bursaries has also failed to shift their application patterns. Students have seemingly proved largely insensitive to financial pushes or pulls. Harrison, Baxter, and Hatt (2007), Davies et al. (2008) and Callender, Wilkinson, and Hopkin (2009) suggest that this has been due to poor information and misunderstanding, but there is ample evidence discussed above that this, at best, is only a small part of the equation.
This points to a significant failure in public policy around widening participation and fair access. The creation of a financial market for higher education as a tool to achieve these policy aims was a centrepiece objective of the 2004 Higher Education Act. However, no competition in tuition fees has emerged and the market in bursaries appears not to have had anything like the anticipated impact, impacting only a tiny proportion of students; within the group of lower-income students, it is effectively only geographically mobile, socially mobile and academically excellent students who are able to 'choose' to access significantly higher bursaries. The approach of using competitive markets to stimulate and manage student 'choice' has thus been revealed as naive and paternalistic, based on an assumption that students would ignore social and cultural constraints and sound academic judgement for financial inducements. This failure is evidenced by the fact that access to the 'top universities' has remained high on the political agenda since the Act came into effect and shows no sign of diminishing (NAO 2008;PFAP 2009;Attwood 2010).
Arguably, the failure of the bursary idea has simply allowed 'top universities' to continue with business-as-usual, secure in the knowledge that their Access Agreements give the appearance of offering new incentives to students from under-represented groups, without them being taken up in meaningful numbers. If they were successful in stimulating new demand, these universities would have to either invest a higher total of their additional income in their bursaries or reduce the amounts on offer, decreasing their efficacy.

Conclusion
This study provides evidence that, in relative terms, access to the 'top universities' in England, represented by the Russell Group, has narrowed since 2004/2005, and that the features of the 2004 Higher Education Act that were intended to make access to these universities fairer have been largely unsuccessful. This is set against a backdrop of modest improvements in the proportion of higher education students drawn from state schools and colleges and lower socio-economic groups which predate the changes. This articulates well with the analysis undertaken by Jones and Thomas (2005) five years ago, prior to the implementation of the Act. The 'entrenched privilege' sought out by the 2003 White Paper has evaded the policy ostensibly targeted at it.
Inevitably, this cannot be the final word on the matter, especially not with further changes to the student financial support system on the horizon. The changes introduced in 2006/2007 will not now have a chance to bed down into a more mature market, with better targeting, better information and more attractive offers to students. However, the system that currently looks set to replace it from 2012/2013 onwards, with tuition fees of up to £9000 per year and a continuing requirement to reinvest a proportion in bursaries and other forms of outreach work, shares many of the same features and the analysis in this study remains relevant in the context. For example, it remains difficult to see how the Russell Group universities could meaningfully improve their financial proposition to academically able students from lower socioeconomic groups without expending significantly more resources. Based on the data presented in this study, institutionally specific financial inducements simply do not seem to impact on recruitment patterns to any identifiable degree -though they may have effects elsewhere on retention or quality of the student experience. Similarly, there is no evidence to suggest that there have been fundamental changes in the culture of these institutions since the introduction of the new regulatory framework under OFFA. In fact, this study suggests that 2006/2007 changes had a minimal impact on the composition of the student body relative to trends that were already underway. Whether this will be the case when tuition fees rise to up to £9000 a year remains moot; it once again appears that most universities look set to charge the maximum in 2012/2013, with little or no market in tuition fees. More data will become available in coming years, and the analysis in this paper will bear repetition and extension in due course.
On this basis, the pseudo-market in bursaries needs to be questioned. It is debatable whether it provides value-for-money for government, institutions or taxpayers, given the huge sums dispensed and the extensive administrative overhead for all parties. One alternative approach would be to re-organise the system through an expanded, but radically simplified, national bursary system that focused on access to, motivation for and retention within higher education rather than recruitment to specific institutions. There is better evidence to support this approach (e.g. West et al. 2006;Harrison, Baxter, and Hatt 2007) than one focused on recruitment. At the time of writing, the details of the proposed National Scholarship Scheme are still unclear, but there is an opportunity for a rethink of what bursaries are really for.
This study has focused on aggregated data as a means of examining the policy implications of the 2004 Higher Education Act. Clearly there is significant scope for additional research to better understand why the 2006/2007 changes did not positively impact on the fairness of the entry profile more than they appear to have done. This study indirectly suggests that components of individual 'choice' beyond finance continue to exert a greater influence on access to top universities, and this needs to be better understood if the next waves of public policy are to succeed in meeting their objectives.

Notes
1. England will be examined in this paper as distinct from the whole United Kingdom because: (a) the financial support systems in Wales, Scotland and Northern Ireland differ, and (b) the location-adjusted benchmarks for widening participation performance (see below) are only calculated and published for England. 2. The definitions of lower socio-economic groups have changed significantly in the early 2000s (from the Registrar General's system to the National Statistics Socio-Economic Classification), but this figure represents the closest corollary of the measure used from the 1960s to the 1990s. 3. To avoid clumsy phraseology, the term 'university' will be used in this paper to mean any institution offering higher education, including colleges of higher education and further education colleges providing higher education courses. 4. The current English Russell Group members are: Birmingham, Bristol, Cambridge, Imperial College, Kings College London, Leeds, Liverpool, London School of Economics, Manchester, Newcastle-upon-Tyne, Nottingham, Oxford, Sheffield, Southampton, University College London and Warwick. 5. The figures provided for tuition fees, grants and bursaries are for the 2006/2007 academic year and have risen in line with inflation since. 6. In reality, data is only available for a subset of these years due to definitional changes which render year-on-year comparison impossible. For 2008/2009, the method used to elicit social class data differs from previous years, while the construction of the benchmarks has also changed significantly. Where such problems exist, these are highlighted in the text. 7. Full details of the definitions used in the dataset are available at: http://www.hesa.ac.uk/ index.php?option=com_content&task=view&id=1447&Itemid=141 8. Excluding Cambridge University, as this is missing data from 2005/2006 which has an identifiable impact on the aggregated time series data. Overall, the pattern for Cambridge University over the period being examined has been a decline in the recruitment of underrepresented groups of students. Its exclusion therefore makes the performance of the remainder of the Russell Group appear better than it would were Cambridge included. 9. Full details of how the benchmark figures are calculated can be found at: http:// www.hesa.ac.uk/index.php?option=com_content&task=view&id=1433&Itemid=141 10. Support for widening access to higher education dropped from 50% in 2003 to 34% in 2004; the lowest level since the question was first asked in 1983. Similarly, the proportion of people believing that higher education should be free for all students dropped from 29% to 23%. These are the sharpest year-to-year changes in both record sets.