Socio-economic inequalities in English schooling under the Coalition Government 2010–15

The reduction of socio-economic inequalities in school outcomes was a major priority of the Coalition Government in England from 2010–15. In this paper we examine the Coalition’s policies and spending, including an analysis of the distributional effect of its pupil premium policy. We also look at trends in outcomes up to 2014. We find that although the pupil premium had a modest overall effect of distributing more money to schools with poorer intakes, this was nested within a wider set of policies which have disadvantaged low income families and children, and that there is evidence of socio-economic gaps widening on some indicators.


Introduction
The Coalition's term of government between 2010 and 2015 was remarkable for the scale and pace of its reform in education: the transformation of the school system, the extensive reform of curriculum and assessment at all levels, and the overhaul of teacher training. This was an exceptional period of policy and one characterized largely by conflict with 'the educational establishment' -namely, teacher unions and education academics.
In this paper we focus only on one part of this policy: the attempt to reduce socio-economic inequalities in educational outcomes. The reduction of such inequalities had been a prominent feature of Labour education policies between 1997 and 2010, especially in the latter years, and it was a surprise to some that the Conservative Party -not typically associated with this agenda -also made a big priority of it leading up to the election, with claims that inequalities had worsened under Labour (Conservative Party, 2008). The first pledge for schools in the Conservative Manifesto was to 'improve standards for all pupils and close the attainment gap between the richest and poorest' (Conservative Party, 2010: 51); while the Liberal Democrats also had inequalities at the heart of their manifesto, albeit in a slightly different guise -emphasizing the need to reduce the effect of family background so that every child could 'receive an excellent education, to unlock children's potential and to ensure that they can succeed in life' (Liberal Democrat Party, 2010: 33).
This paper examines what happened next. What policies did the parties implement in coalition to address educational inequality, and with what effect? The paper draws on our earlier review of the Coalition's record on schools (Lupton and Thomson, 2015), published as one of a suite of working papers examining different aspects of the Coalition's social policy record and based primarily on analysis of policy documents and administrative data, including Department for Education (DfE) attainment and school funding statistics and the National Pupil Database (NPD). All of these papers took a national overview and drew primarily on quantitative evidence. A particular contribution of this paper is an original analysis of the distributional effect of the Coalition's pupil premium policy. More local and in-depth qualitative studies could no doubt deepen the understanding gained.
In this article we focus on schools up to the age of 16 -although we have elsewhere argued that early years and further education must also feature in any serious strategy to reduce inequalities and increase social mobility (Stewart and Lupton, 2015). We also focus on socioeconomic inequalities, principally because the work derives from a project looking at the impact of policy changes on poverty and inequality. This is not to deny the existence or importance of other educational inequalities -such as between students of different genders, ethnic origins, and learning needs -although there has been noticeably less policy focus on these in recent years. A broader analysis would be an obvious next step but is beyond the scope of the current paper.

The Coalition's inheritance
Before analysing what the Coalition did, we briefly review the situation it inherited. Socioeconomic inequality in educational outcomes is not new -it has long been a feature of the English education system (Douglas, 1968;Central Advisory Council for Education, 1967). However, the persistence of the phenomenon has been brought into clear view since the early 2000s, when the NPD made it possible to monitor pupil achievements by free school meal (FSM) status -although not social class -eligibility. In addition, the OECD PISA 1 database made possible international comparisons, both of overall standards and social inequalities.
When the Coalition took office in 2010 there remained very large gaps between the attainments of children eligible for FSM and those who were not: for example, 20.2 percentage points at GCSE at the level of 5 A*-C grades, and 27.6 percentage points at the higher level of 5 A*-C grades including English and maths (hereafter 5 A*-CEM). Socio-economic inequalities were also relatively high by international standards (Jerrim, 2012a).
Conservative claims that this situation had been getting worse were not substantiated. In fact, since 2003/4, and particularly since 2008, there had been a marked reduction in the FSM/ non-FSM gap, both at the end of primary school and at GCSE 5 A*-C, although not at 5 A*-CEM (see Figure 2 later). Blanden and Macmillan (2013) have also demonstrated that inequalities at higher levels of attainment -for example, being in the highest attaining fifth of pupils at GCSE level -had not reduced, while Jerrim (2012b), analysing PISA data for the period 2001 to 2009, also reported a pattern of narrowing socio-economic inequalities among lower but not higher attainers.
The extent to which these improvements were attributable to policy is disputed (Whitty and Anders, 2014;Lupton and Obolenskaya, 2013;Heath et al., 2013). Certainly, Labour had substantially increased spending on schools, resulting in lower pupil-to-teacher ratios and new buildings, particularly in the poorest areas (Lupton and Obolenskaya, 2013). Centralized school improvement initiatives (including the National Strategies) and major investments in teacher training and development (including leadership development and school-to-school collaboration) had increased the standard of teaching across the board. The House of Commons Select Committee concluded in 2010 that England had some of the best qualified and best trained teachers ever (Whitty, 2014). Funding in general became more targeted towards disadvantaged schools (Chowdry and Sibieta, 2011a), and Labour's academies programme and its major capital programme, Building Schools for the Future (BSF), were both targeted towards the poorest areas. The 'Every Child Matters' (ECM) framework required multi-agency working and a broader focus on multiple outcomes, such as health, safety and enjoyment, as well as academic achievements. In addition, there was a range of further targeted programmes such as Excellence in Cities, the City Challenges, Schools in Challenging Circumstances, the Extra Mile initiative, and the Narrowing the Gap programme, involving multi-agency working around disadvantaged and vulnerable children, and a range of area-based grants. Following evidence that it appeared to be non-FSM children in high FSM schools who were the prime beneficiaries, the period from 2008 saw an increase in schemes that targeted individual children: Every Child a Reader, Every Child a Writer, and Every Child Counts.
However, sceptics pointed out that the rapid increase in GCSE attainments for lower attainers was largely due to the uptake of vocational qualifications equivalent to GCSEs, and perhaps that schools were increasingly pushing students towards these qualifications in order to boost low overall scores (e.g. Wrigley and Kalambouka, 2012) -despite the fact that they typically generate a lower return in the labour market than traditional GCSEs (Dearden et al., 2004). In other words, they suggested both that these were not real improvements and that they might potentially have a negative effect on the life chances of less advantaged students. That the improvements seen in domestic examinations were not reflected in the international assessments over the same period (Jerrim, 2012b) lent some weight to the 'grade inflation' argument. There were also critics who argued that socio-economic inequalities would not narrow substantially without greater reductions in wider economic inequalities and changes to the school system. In particular, many argued that increasing emphasis on qualifications as the sole valued outcome of education, combined with increased pressure on school performance, was leading to increased disengagement of lower attaining pupils, the prioritization of pupils at grade boundaries, loss of curriculum breadth, and teaching to the test (see West, 2010, for a useful review).
In short, some progress had begun to be made under Labour, but gaps remained very wide. While there was emerging evidence that targeted support to individuals could be beneficial, there were also those who argued that much wider systemic changes would be needed to make a serious impact on the problem.

Closing attainment gaps: Pupil premium
The Coalition's flagship policy to reduce the socio-economic attainment gap was the pupil premium -a per capita grant paid to schools and academies for each pupil eligible for FSM. At the same time a range of existing central government programmes -including the 'narrowing the gap' elements of the National Strategies, education/health partnerships, start-up costs for extended schools, and other area-based programmes -were discontinued.
Unlike these programmes, the pupil premium was not ring-fenced to certain activities. The government's intention was that schools should decide how to use resources to best close attainment gaps. It established a new charity, the Education Endowment Foundation, to build up and disseminate knowledge of successful interventions that schools could use. Schools were required to publish information about how they were using the pupil premium on their websites and were held accountable both through inspection and performance measures. A new inspection framework required inspectors to consider how well schools provide for different groups of pupils, including boys and girls, minority ethnic groups, children with disabilities, and those eligible for FSM; and Ofsted announced that from 2013, it would be re-inspecting 'outstanding' schools where the attainment of children on FSM was deemed too low. From 2011, school performance tables have included indicators of attainment and progress of disadvantaged pupils, and the gap between their attainment and that of others.
The pupil premium started in 2011/12, at £488 per pupil eligible for FSM, with additional premiums for children looked after by the local authority (looked after children or LAC) and those with a parent in the armed services. In the following year, eligibility was widened to those who had been eligible for FSM at any time in the last six years (known as 'Ever 6'). Sums have increased each year and since 2013/14, the funding has been loaded in favour of primary pupils. In 2013/14 (the last year for which we show analysis in this paper), for each 'Ever 6' pupil, primary schools received £953 and secondary schools £900, and by the current year (2014/15) these figures had risen to £1,300 and £935 respectively.
The pupil premium policy as implemented appears to vary in intent and design from the policy trialled at the time of the election, which aimed to ensure that poorer children were educated in the best schools (HM Government, 2010), or placed in smaller classes (Liberal Democrat Party, 2010). An independent evaluation conducted during 2012/13 suggested that schools were using considerable discretion with the extra money: compensating for losses of other funding to continue to provide support that had previously been in place, and targeting it on the basis of educational need rather than strictly pupil premium eligibility (Carpenter et al., 2013). This was also reflected in Ofsted's first report on the issue (based on a survey of head teachers) in 2012, which found that only one in ten had significantly changed the way they supported pupils from disadvantaged backgrounds (Ofsted, 2012). A subsequent report by Ofsted heavily criticized school-wide approaches and called for the money to be specifically targeted on the eligible pupils (Ofsted, 2013). By September 2014, Ofsted was reporting that more schools were using the funding 'well' -that is, targeting it on the eligible pupils. The most frequent use of the funding was to pay for additional staff (teachers and teaching assistants) to deliver one-to-one or small group tuition. Secondary schools were more likely to employ additional teachers and primary schools to employ teaching assistants. Additional staffing was allowing schools to offer a range of interventions, including booster classes, reading support, raising aspiration programmes, or to reduce class sizes. Secondary schools were frequently engaging 'learning mentors', while in primary schools, funding was sometimes used for specialist support for pupils' language and communication skills. Support for after-school, weekend, and holiday sessions, and to enable educational visits, was another common use of the money (Ofsted, 2014).
Some early evidence suggested, then, that schools were using the pupil premium to replace funding lost from other grants. A key Coalition agreement pledge, however, was to fund the pupil premium from 'outside the schools budget' for children who were disadvantaged, so that it would constitute 'extra money' to help these children. Table 1 shows that, in its first year (2011/12), the value of the pupil premium, at around £0.5 billion, was considerably lower than the value of the area-based grants discontinued by the Coalition, at around £0.9 billion. These included parts of the School Development Grant, extended school start-up costs, music grant, Assessment for Learning, the co-ordination of the National Strategies, education health partnerships, and a range of others (see Chowdry and Sibieta, 2011b, for a longer list). In its second year, it was substantially higher than it had been the previous year, mainly due to the extension of the definition of 'disadvantaged children' to include those who had ever been on FSM in the past six years. However, it was still slightly less than the abolished grants. This changed in 2013/14 when the overall spend for pupil premium rose to £1.25 billion, and it has increased further to £2.15 billion in 2014/15 (DfE annual report 2013/14, pupil premium final allocations, 2014/15). 2 DfE annual reports (2010/11, 2011/12, 2012/13, 2013/14) (www.gov.uk/government/ publications/department-for-education-annual-report-and-accounts-financial-year-2010to-2011 ; www.gov.uk/government/publications/department-for-education-consolidatedannual-report-and-accounts-2011-12; www.gov.uk/government/publications/dfe-consolidatedannual-report-and-accounts-2012-to-2013, all accessed on 5 May 2015; and www.gov. uk/government/publications/dfe-consolidated-annual-report-and-accounts-2013-to-2014, accessed 12 May 2015). (2014)  At the school level, the pupil premium appears to have had some redistributive effects. In Figure  1, we show percentage changes in school-level grant income per capita, splitting primary and secondary schools into groups based on their proportions of FSM pupils in 2013/14. These data are for maintained schools only -they are not available in the same format for academies. 3 This has implications for the size and type of the sample by 2014 -particularly for the secondary level, with only around 40 per cent of the total number of secondary schools still in the data set at this point. It is difficult to assess the extent to which this changes the characteristics of the sample by 2014, because many school-level characteristics will have changed over time. However, since the academies sample (not observed) will include schools in deprived areas converted to academies under Labour, as well as more advantaged ones converted under the Coalition, it is reasonable to assume that this group represents a wide range of schools. In addition, the geographical distribution of the schools remaining in the analysis by 2014 is similar to that in 2010. Estimates from the National Audit Office for the period 2010/11 to 2014/15 and using a different classification of schools show smaller increases and larger decreases but broadly the same redistributive pattern (National Audit Office, 2015). From Figure 1, we see that the least deprived group of secondary schools experienced real terms losses in income of around 0.7 per cent, while more deprived schools had real terms increases -of around 3.8 per cent for the most deprived schools. Note that these are not even-sized groups of schools. The least deprived group, with median losses, makes up about 30 per cent of the total. For primary schools, the least deprived schools (about 40 per cent of all primary schools) experienced a small increase in grant funding (of around 2.6 per cent), while the most deprived schools experienced a larger increase (of around 10.8 per cent). These increases are larger than reported in our previous analysis up to 2012/13 (Lupton and Thomson, 2015), because of the increase in the per-pupil value of the pupil premium since then. Notes: Maintained schools here means those schools maintained by the local authority and so does not include academies. Data for academies is not directly comparable to that for maintained schools.

Schools, Pupils and their Characteristics
Authors' calculations to convert 2013/14 to 2009/10 prices using HM Treasury deflator series (December 2013). Schools with unrepresentative funding (e.g. those in the process of closing at the time of financial reporting) have been excluded from calculations.
Schools were categorized into six bands by the percentage of children eligible for free school meals in 2013/4. The national averages for % FSM were 17% (primary) and 14.6% (secondary) in 2013/14, compared with 16% (primary) and 17% (secondary) in our sample (source data from Schools, Pupils and their Characteristics, 2014: www. gov.uk/government/statistics/schools-pupils-and-their-characteristics-january-2014, accessed 1 May 2015).
A rather more complex picture underlies these headline results. Within each band, the range is very wide and some schools have seen significant losses. Table 2 shows the number of schools that gained and lost money between 2009/10 and 2013/14, broken down by FSM band and the average gain/loss. It is clear that the redistribution of funding has not taken place directly in line with FSM percentages. This is partly because the pupil premium is paid per capita for those who have ever been on FSM in the last six years, so schools can have high percentages of these 'pupil premium eligible' children without necessarily having a high FSM percentage in any one year. Schools like this perhaps did not qualify for extra assistance under the old model of deprivation funding because they were not deprived enough but, under the new pupil-level funding, attract additional funding. Variation in school rolls year-on-year will make a difference. There will also be schools which, owing to their particular circumstances, have lost more in abolished grants than they have gained in pupil premium, and those which have been affected by other changes to the school funding formula (in which the criteria used to determine the within-local authority distribution of funding were simplified). Thus, although the pupil premium has had a positive effect for disadvantaged schools as a whole, its effect has not been uniform, and it can by no means be regarded as a panacea for the problem of wide socio-economic attainment gaps.

The other part of the story: Wider education and social policies
The introduction of the pupil premium moved the issue of educational disadvantage to centre stage in the Coalition's schools policy programme. It has been a prominent policy which clearly signalled that the government was taking action. However, it has also been an isolated policy -a rare example of investment in the life chances of disadvantaged children among a broader range of policies which have reduced family incomes and depleted services. In its emergency budget of June 2010, the Coalition announced some critical decisions about the shape of its public spending. One was that over three-quarters (77 per cent) of the contribution to its deficit reduction target would come from public spending cuts, not from tax increases. Another was that spending on the NHS, schools, and pensions would be protected from these cuts. Protection of these very large spending areas meant very severe cuts to budgets in non-protected areas of service spending (particularly at local government level), and to the non-pensions element of the social security budget -often described as welfare benefits.
The range of cuts to welfare benefits has included: an overall cap on the total amount families could receive; tighter limits on Housing Benefit for private tenants; cuts for social housing tenants of working age deemed to have spare bedrooms; reforms to Council Tax benefit, making more low income households eligible to pay the tax; changes to tax credits, making them less generous; and abolition of the Social Fund, which gave emergency grants and loans to people with low incomes. In addition, tighter conditions were imposed for disability benefits, and the administration of many out-of-work benefits has been made much tougher, including much greater use of 'sanctions' imposed on unemployed and other claimants for not meeting particular job search requirements (see Hills, 2015, for a more extended account).
The majority of these changes came into effect from April 2013 (two years later than the introduction of the pupil premium). There have been some offsetting policy changes, notably the introduction of universal free school meals for infants (from 2014), which evaluation suggests is more likely to have educational benefits for children from less advantaged homes (Kitchen et al., 2013). Some lower income families have also benefited from increases in the Income Tax personal allowance. But overall, families on low incomes have experienced sizeable net losses of income. The Institute for Fiscal Studies (Browne and Elming, 2015) estimates that tax-benefit reforms have meant an average loss of £1,000 to £2,000 per family, with larger families, lone parent households, and couples with one or no earner feeling the largest effects. Comparing these figures with the pupil premium data, we can see that a low income family with one child at secondary school and one at primary would have lost on average only a little less from their household budget than their children's schools will have gained in pupil premium funding. While it is not yet clear exactly to what extent these changes will affect children's educational attainments, some negative effects seem likely. Cooper and Stewart's (2013) systematic review of the literature on the effect of family income on educational outcomes demonstrated a clear relationship between additional family income and improved outcomes. Whether the same effects will operate in reverse remains to be seen. Early studies of the effects of welfare reform, while not focusing specifically on children and schools, have revealed that families have been cutting back on food, heat and electricity, and selling belongings, as well as relying on food banks (Power et al., 2014). Moreover, while schools spending was protected overall, other services affecting children have been cut. Local authorities have seen cuts to their budgets of around one-third (Hastings et al., 2015), with the most affected services being those that are not statutorily determined -such as libraries and community and youth services -leaving schools, with relatively protected or increased budgets, to take on more of the burden of support to children from low income families.
Our earlier working paper (Lupton and Thomson, 2015) documents in full the Coalition's broader policies towards schools, including the rapid expansion of the academies programme (and the introduction of Free Schools) and the reform of teacher training, so we do not repeat all of these here. Two areas, however, are particularly pertinent to issues of socio-economic disadvantage.
One is the reform of, and cuts to, the schools capital funding system. Capital spending makes up a relatively small part of the schools budget, but can have significant symbolic as well as material effects. Under the previous Labour Government, a large new capital programme (BSF) had been initiated, which proposed a total replacement, over a period of 15 to 20 years, of the entire secondary school stock, starting first with schools in the most disadvantaged areas. Its ambitions went beyond physical improvements to include area regeneration and wider participation. Proposals for new BSF schools had to demonstrate not just newer, better designed, and more sustainable buildings, but the ways in which these buildings would enable innovative high quality teaching and learning, raise standards, be accessible to local communities, and be founded on extensive local collaboration and parental involvement (Mahony and Hextall, 2013). Under the Coalition, BSF was abruptly cancelled, following concerns about under-delivery and lack of value for money (National Audit Office, 2009), and replaced with a new programme targeting capital allocations based on the need for pupil places (a Targeted Basic Need Programme) and on the condition of the local estate (a Priority Schools Building Programme); in other words, concentrating on the state of the buildings rather than seeing school capital spending as a route to achieving redistributive goals. Overall capital spending on schools had fallen by 57 per cent by 2013/14, compared with its 2009/10 value.
The second, and much more important area of policy reform, concerns curriculum and assessment. Despite the fanfare around the pupil premium, it represents only a very small proportion of overall school spending (initially 1.3 per cent in 2011/12, rising to 2.9 per cent by 2013/14), suggesting that the mainstream work of schools is likely to be much more important for children's outcomes. A key feature of the Coalition's approach to this work was to make curriculum more academic and assessment tougher, motivated by a belief that standards were too low, both by comparison with other nations (DfE, 2010) and as a preparation for life after school (DfE, 2014). The DfE-commissioned Review of Vocational Education (Wolf, 2011) was particularly critical of vocational 'equivalents' at GCSE, arguing that they were not equipping young people for Level 3 courses ('A' Level or equivalent), nor were they regarded as valuable by employers.
Since 2010, changes to assessment have been proposed or implemented at all levels from age 5 to 18. For the youngest age, a baseline measure in reception year will be introduced in 2016 to replace the Early Years Foundation Stage Profile. A new test (a phonics screening check) was introduced in 2012 for children at the end of Year 1, and since 2013, failure to reach the required standard in this test has triggered extra support and a re-test at the end of Year 2. New externally-set but internally-marked tests will be introduced at the end of Key Stage 1 in 2016 in mathematics and reading, and grammar, punctuation, and spelling (GPS). Key Stage 2 tests have been retained, but changed. From 2013 there has been no external assessment of the composition element of writing, but a new test in GPS has been introduced. From 2016, test results will be expressed as scaled scores (rather than threshold levels) with a score of 100 marking the 'expected level' (which 85 per cent of pupils are expected to meet).
GCSEs have been comprehensively reformed, both in content and form, to make them 'more challenging' (DfE, 2014). The mode of assessment has been changed from a modular system to one of assessment at the end of the course, and with examinations as the default mode. Most subjects will be untiered, and marked on a new scale from 1 to 9. Subject content has also been changed, with the intention of making the exams more demanding and requiring students to demonstrate competence in reading and writing at length, and in mathematical skills. Following the Wolf Report, the number of qualifications that count towards school performance tables was significantly reduced, and there were changes to the way that they were counted: each qualification would only count for one GCSE, and a cap was introduced on the contribution that non-GCSE qualifications could make to a student's overall points score.
Some of these changes have already been made. In particular, students sitting GCSE science in summer 2012 were the first to encounter more demanding syllabuses, and those taking exams in summer 2013 also faced revised qualifications in single science subjects. In the same year, speaking and listening assessments were no longer counted in GCSE English grades, and a stronger weight was given to written exams over controlled assessments. In English literature, history, geography, and religious studies exams, marks were awarded for spelling, punctuation, and grammar. Students starting GCSEs in September 2012 and completing them in 2014 were the first to take all-linear exams, and during the course of their GCSE year, the Secretary of State also announced that only one attempt at the exams would be counted in league tables (for English, maths, modern languages, history, geography, and the sciences, with other subjects to follow). This had an immediate deterrent effect on the practice of 'early entry' in November 2013. The additional 'Wolf' changes to performance tables came into effect in 2014. However, the major overhaul of programmes will not take effect until after the next election, with the new programmes being taught from 2015 (English and maths), 2016 (other larger subjects), and 2017 (all other subjects).
While one ostensible aim of these reforms was to put an end to 'grade inflation' and to give all young people access to meaningful qualifications, there are good reasons for concern that they might have a detrimental effect for some disadvantaged students, at least in the short term. In recent years, the least advantaged students have relied more on vocational subjects to reach GCSE expected levels (House of Commons Education Committee, 2014), possibly because they have been pushed into them by schools keen to raise their results, but also possibly because they have found them more engaging and motivating. A group of one hundred academics publicly argued that moving to a 'knowledge heavy' curriculum could lead to early demoralization, and to difficulty for children in relating abstract ideas to their own experiences and lives, as well as failing to develop the skills that will be needed in the labour market (The Independent, 2013).
In summary, the Coalition's term in office has been marked by, on the one hand, the high profile introduction of a redistributive funding mechanism and increased targeting of effort on individuals from poorer families, and on the other, a set of wider social policies which have had the effect of reducing the incomes of such families and the wider services available to them, along with changes to curriculum and assessment to increase academic content and make examinations harder. It is an approach which relies heavily on an academic-focused school system to rescue low income students and provide them with access to improved life chances, rather than one which invests in the foundations of secure childhoods, putting students in a better position to learn and to make choices. It shifts responsibility, in some respects, from the wider welfare state to schools.

Trends in socio-economic inequalities
Since many of the Coalition's curriculum and assessment reforms have not yet been implemented, and the key welfare reforms only came into effect in April 2013, it is really too early to tell the effect of this policy regime on attainment gaps and differences in other child outcomes. The period up to 2013 can be broadly considered as 'welfare, curriculum, and assessment stable', while the examinations taken in 2014 give an indication of trends under the first year of assessment changes and large-scale welfare reforms. Table 3 shows the basic trends for the main Key Stage 2 and GCSE performance measures since the Coalition came to office.  Notes: Data are for all pupils in state-funded schools.
In 2010, industrial action meant that the state school participation rate for KS2 tests was 74%.
We show the trends for reading (and not English) at KS2 because the changes to KS2 tests over this period mean that there is not a stable time-series for English.
For most measures, the attainment of pupils on FSM rose every year from 2010 to 2013. Moreover, until 2013, the overall picture was one of narrowing gaps between FSM and non-FSM pupils. In general, however (as shown in Figure 2, which shows a longer time trend), these increases represent a continuing trend rather than a step-change. In other words, there is no indication of a pupil premium effect -perhaps not surprising, since our earlier analysis shows that it is not until 2013/14 that the sums involved exceeded the grants abolished, and one would in any case expect some 'bedding' down while schools learned how to use the new funding to optimal effect. To look at these trends in a little more detail, at KS2, both maths and English gaps fell until 2012. The FSM gap in reading scores (the only consistent element of the English test over this period) then fell to around 12 percentage points. The gap in GPS cannot be measured over the same period, but was considerably higher (at 17 percentage points) in 2014 than the gap in reading. For maths, the gap between FSM and non-FSM continued narrowing in 2014. At GCSE, the FSM gap at 5 A*-C also narrowed year-on-year until 2013. At the higher level of 5 A*-CEM, it also narrowed in 2011 and 2012, before opening up very slightly again in 2013 due to improved performance of the non-FSM group. As Figure 2 shows there was no real break in trend here from the Labour period. The EBacc (not shown) was entered and achieved by higher proportions of non-FSM than FSM students, and the gap actually widened over time.
The 2014 GCSE results show a very different story. In Figure 2, we show the official results, using the 'new counting rules' in which each qualification only counts for one GCSE; there is a cap on the contribution of vocational qualifications to the overall score and only a student's first attempt at qualification is counted. On the 5 A*-C measure, the FSM/non-FSM gap returned in 2014 to its 2006 level, suggesting that all the gains made since then were due to a combination of students taking more vocational qualifications, or ones with higher equivalent value, or having several attempts at an examination. Certain groups of students seem to have benefited particularly from these practices. In 2014, at the 5 A*-C level, the performance of white boys on FSM was down 29.2 percentage points from 2013 (from 64.8 per cent to 35.6 per cent), and performance of children with special educational needs (SEN) on FSM went down 32.8 percentage points (from 49.4 per cent to 16.6 per cent). Interestingly, the performance of FSM students in London fell less than in other parts of England.
Thus, much of the increase in the FSM/non-FSM gap in 2014 can be accounted for by the counting rule changes. However, as Table 3 shows, not all of it can. The DfE has also published a set of results using the old rules. On these terms, at 5 A*-C, there was also a fall in attainment, which was more pronounced for FSM students -8.1 percentage points -than overall (4.8 points). The gap between FSM and non-FSM pupils widened from 16 percentage points in 2013 to 19.8 in 2014. When English and maths are included, there has been less change, with a fall of just 0.9 percentage points for FSM students and 0.4 for non-FSM students, resulting in a widening of the gap of just 0.5 percentage points from 26.7 to 27.2. This suggests that the change to the linear format of assessment, the removal of speaking and listening from the English assessment, and the disincentive to sit examinations early and to take vocational qualifications have had, in combination, a negative effect on the attainment of lower attainers from poor families, although a negligible effect for higher attainers. This demands further investigation.
Finally, we look at a wider set of outcomes for children and young people. Under the Coalition, wider goals relating to child well-being, as expressed in Labour's ECM framework, were largely dropped. For our earlier working paper (Lupton and Thomson, 2015) we revisited the indicators associated with ECM to try to establish and document trends after 2010. This exercise has its limitations. We did not attempt a critical reading of the framework or seek to problematize the indicators themselves. For some indicators there were no available data, or a very short time series -in some cases only extending to 2012 or 2013. We described trends where it was sensible to do so and without seeking to ascribe statistical significance to a trend. An attempt was made to find official data sources for each indicator, but we did not conduct primary analysis of large administrative data sources, and so some gaps may remain. The full list of indicators and changes is included in Lupton and Thomson (2015). Table 4 summarizes the results of this exercise. For around a quarter of the indicators, it was not possible to establish improvement or decline. Where trends could be established, there was no change in trend or a stable trend for another quarter. Of the indicators where there was evidence of improvement and decline, most improved. Those that show decline include the percentage of children subject to child protection plans for a second (or subsequent) time, child protection cases reviewed within required timescales, SEN statements issued within 26 weeks, and care leavers in employment, education, or training -all vulnerable groups. The education attainment indicators included in the ECM indicator set for LAC have either improved or stayed stable since 2010. However, the government's own impact indicators covering this issue, which are slightly different measures, show outcomes getting worse for this group. This is perhaps not surprising as there has been a large focus on improving educational outcomes for disadvantaged children under the Coalition Government, but large cuts to other local authority services on which these children and their families may rely. One of the most relevant indicators in the ECM series to our discussions here is the child poverty measure of 'Households Below Average Income' (HBAI) for households with children. This can be measured in several ways: to indicate absolute or relative poverty and poverty before or after housing costs. Absolute child poverty, measured before or after housing costs (BHC/ AHC), fell between 1997 and 2005, and then was relatively stable on both measures until 2008. After this, both measures continued to fall slightly until 2011 and rose again until 2012; then the AHC measure continued to rise while the BHC measure stabilized. The AHC measure is thought to be a better indicator of poverty for areas where housing costs are particularly steep (e.g. London) (Belfield et al., 2014) Relative poverty -on both the AHC and BHC measures -fell until 2004, rose again until 2008, and then fell until 2011, before stabilizing. This, however, is indicative of the fall in median income since 2008 and shows that income levels of those at the lowest end of the income distribution fell a little less sharply than for others. The Institute for Fiscal Studies suggests future releases of HBAI data will show an increase in child poverty because of increased social security cuts implemented after April 2013 (Belfield et al., 2014: 57). Whether educational policies can be relied upon to narrow socio-economic inequalities while child poverty is rising must be in doubt.

Conclusion
The Coalition Government, perhaps to some people's surprise, made reducing socio-economic attainment gaps one of its key education priorities. Its key policy in support of this goal was the targeted pupil premium, and the associated investment in research into 'what works', so that schools could choose the right interventions to support students from low income families.
In this paper, we have examined the distributive effect of the pupil premium showing that, overall, it has had a redistributive effect on school funding -although some schools with very disadvantaged intakes have also seen their funding reduced. We have also shown that the pupil premium has had no noticeable effect on educational inequalities to date -but this might be expected at this early stage and should be kept under review.
Our key point, however, is that assessments of a government's record in tackling educational inequalities cannot be confined to its flagship additional policies, but must also include mainstream educational policies and wider social policies affecting the distribution of income and, in particular, the circumstances of the poorest children whose attainment the targeted flagship policies are intended to raise. Results to date show that, at best, these policies in combination have made a very modest impact on socio-economic attainment gaps, with some evidence that they have made things worse for some groups of students -that is, low attainers from low income families and LAC. Moreover, the full effects of the Coalition's welfare reforms are yet to be seen and child poverty is predicted to rise. Post-election debate around socio-economic inequalities in education has largely focused on whether the new Conservative Government will stick to its pledge to retain the pupil premium. A more important question is whether the pupil premium can be expected to have any meaningful impact as part of a suite of education and social policies likely to work in the opposite direction. This situation will need to be closely monitored. Meanwhile, aspiring future governments with intentions to reduce inequalities in school outcomes surely need to see the problem 'in the round' -taking into account family poverty and the mainstream activities of schools, as well as additional interventions sourced from supplementary funding streams.

Notes on the contributors
Ruth Lupton is a professor of education at the University of Manchester. She researches and writes on poverty and inequality with a particular interest in spatial inequalities and the dynamics of disadvantaged neighbourhoods. Her work on education has focused on relationships between poverty and education, including the socio-economic contexts of schooling. Before joining Manchester, she was Principal Research Fellow and Deputy Director of the Centre for Analysis of Social Exclusion at the London School of Economics and Political Science.
Stephanie Thomson is a research associate at the University of Manchester. Her research interests are inequalities in education and innovative research methods in the social sciences. Previous work has explored the relationship between pupils' backgrounds and their attainment in mathematics and how case-based methods of analysis can help to explain this.