X Close

Centre for Education Policy and Equalising Opportunities (CEPEO)


We create research to improve the education system and equalise opportunities for all.


Archive for the 'CEPEO' Category

Using targeted pay uplifts to reduce teacher shortages

By Blog editor, on 11 May 2023

By Dr Sam Sims

CEPEO recently launched New Opportunities, our evidence-based manifesto for equalising opportunities. In this blog series, we are highlighting one of our policy proposals each week. This post will take a fresh look at the reasons why we have a shortage of teachers in England and outline the evidence for using targeted bonus payments to mitigate the problem.

Each year, around 800,000 students graduate from universities in England. And each year the government tries to lure around 30,000 graduates into teaching. Putting aside the pandemic years, the government has failed to do this every year since 2015.

An important for this is the declining attractiveness of teaching as a profession. Over the very long run, as private sector real wages have increased, the competitiveness of teachers’ pay has declined. This process has accelerated over the last decade, as austerity has seen teacher real pay fall by 5% or more, while wages elsewhere have retained their value. More recently, many professional occupations have seen a big increase in working from home, something which is largely incompatible with teaching. Teaching is not what it used to be.

An effective solution would be to dramatically increase teacher pay. Indeed, in 2019 the government announced that it would increase starting salaries by 24% over just three years. However, a series of delays to the policy, combined with sustained double-digit inflation, has rendered this once radical policy somewhat modest.

The sheer cost of a blanket increase appears to have put off the government. Keir Starmer has also refused to rule out below-inflation pay rises for public sector workers. Is there a more cost-effective, more politically palatable way to address the shortages?

There are two broad ways to cut the costs of a blanket increase. The first is to focus pay increases on the phases (secondary) and subjects (maths, physics) where shortages are most severe. The second is to focus pay increases at the career-stage (early career) in which it makes the largest difference.

In 2018, the government announced a pilot of one such policy, known as Retention Payments. These provided a £2,000 (8%) bonus to maths and physics teachers in the first five years of their careers. The policy was only available for teachers working in 42 (of the 343) local authorities in England, providing a convenient comparison group against which to gauge the impact of the policy. Asma Benhenda and I did just this, and found that eligible teachers were 23% less likely to leave the profession in a given year.

The longer-term effects of the policy are less clear. But the hope is that, by retaining more teachers during the early-career period when they are most likely to leave, the effects will be sustained. Early-career payments for maths teachers have now been in place nationwide for  several years and, notably, maths has gone from being a subject that used to have among the most severe shortages to a subject with relatively minor shortages.

Early-career payments have since also been rolled out to other shortage subjects, namely physics, chemistry and languages. However, there remains more to do. Despite the early career payments in physics, it still has the worst shortages of any subject. The retention payments for physicists should therefore be increased in value and/or duration. Computing still does not qualify for early-career payments at all, despite having the second worst shortages of any subject. There is therefore a strong case for expanding coverage of early-career payments to computing and other severe shortage subjects.

If the government is committed to providing enough specialist teachers for all pupils, and they are not willing to increase teacher wages generally, then increasing the value and coverage of targeted payments must be their policy priority.



We must reform school admissions to ensure all pupils can access high-quality education

By Blog Editor, on 4 May 2023

By Jake Anders

In April, CEPEO launched New Opportunities: our evidence-based policy priorities for equalising opportunities. As part of an ongoing series, each week we are highlighting one of our priorities and the reasoning and evidence behind them. This week, we are focusing on school admissions, an area where we see huge disparities in access to high-performing schools by socio-economic status. In this blog post, we discuss two key reasons for this — the importance of distance to school in admission criteria, and the continuing existence of grammar school systems in parts of the UK.

Pupils from more advantaged backgrounds are more likely to attend schools that get better results in national tests. In London, pupils eligible for free school meals (FSM) attend, on average, schools where 59% of pupils achieve 5 passes or higher at GCSE, compared to 65% for non-FSM students. The gap is wider outside London at 8 percentage points between FSM and non-FSM eligible young people. This means that non-FSM students have access to schools where there is a higher chance of achieving academic success, which can have significant implications for their future prospects.

People sometimes suggest that this is because less advantaged families differ in their approach to choosing schools. But analysis of families’ preferences for secondary schools suggests this is not the main cause of the difference. Families of FSM pupils are only slightly more likely than more advantaged families to express a preference for only a single secondary school, or to make their closest school their first preference. This suggests little systematic difference in the degree of active engagement with school choice.

What, then, actually explains the difference? The disparity is driven by the fact that more affluent families are more likely to live in the proximity of good schools (including due to deliberately moving house to be near a good school), combined with admissions rules that prioritise distance in their decisions.

Schools typically apply admissions rules that consider the distance from prospective pupils’ homes. While this makes sense if all schools are equally good for all pupils, given the reality of disparities in school quality, it ends up limiting the ability of some pupils (disproportionately from less advantaged backgrounds) to access the best school to which they could reasonably travel. This means that disadvantaged families are limited in their ability to access schools with the characteristics they desire.

This is particularly important as parents and pupils seem to do a good job of picking schools for their children, if they are able to exercise that choice. Recent work has found that pupils who get into their first choice school do better than if they attend one of their lower-ranked schools, and this boost is not explained by any differences in overall effectiveness between the two schools.

Another feature of our school admissions system that is a major disruptor to fair access to high-quality schools for all pupils is grammar schools.

Grammar schools (which are allowed to select their students based on tests purporting to measure academic ability) are highly socially selective. In the areas where a grammar school operates there are stark differences in attendance at that grammar school by socioeconomic status. Just 6% of pupils from the most deprived backgrounds attend a grammar school. It is not until the 90th percentile of the socioeconomic status distribution that we see more than half of students attending a grammar school. The top percentile group, however, has a grammar school attendance rate of 80%.

And this is not just because of correlations between academic attainment and socioeconomic status. Pupils with the same level of attainment in their end of Key Stage 2 tests (taken in the same school year as grammar school entry tests are sat) are much more likely to go on to attend a grammar school if they are from advantaged backgrounds. This suggests that high-attaining young people from less advantaged backgrounds are less likely to be taking the grammar school entry tests, or are doing less well in those tests than we would expect from other measures of their attainment. This latter factor could well be explained by the big differences in private tutoring by family income.

And if you live in a grammar school area then missing out on a place matters for long-term life chances. High-attaining pupils living in such areas who miss out are less likely to go on to higher education. If they do, their chances of attending a high-status university and achieving a good degree classification are lower compared to equivalent pupils who went to grammar schools.

Across both of these issues, reforms to school admissions could make a significant difference in equalising opportunities. Reducing the importance of distance to school and, hence, the link between family income and school attended could make a significant difference to life chances. Even better, requiring schools to prioritise applicants who are eligible for the pupil premium, or, more radically, introducing a degree of random assignment of pupils to schools within certain areas would help to level this aspect of the education playing field.

Jake Anders is Associate Professor and Deputy Director at UCL Centre for Education Policy & Equalising Opportunities, and Principal Investigator of the COVID Social Mobility & Opportunities study (COSMO).

Why we should introduce a post-qualification applications (PQA) system for post-18 education

By Blog Editor, on 27 April 2023

Gill Wyness

The UK is the only country in the world in which young people make their university applications before they have received their exam results. Instead, pupils apply to university with grades that have been predicted by their teachers.

The problem is predicted grades aren’t always accurate. Analysis of UCAS data has shown that only 16% of applicants achieve the A-level grades they were predicted to achieve, while 75% of students are over-predicted. The same research showed that among equally high attaining students, disadvantaged students receive less generous predictions compared to more advantaged students, highlighting that there are systematic differences in predicted grades across student groups.

Teacher predicted grades have always been controversial, but were thrown into the spotlight during the pandemic, when exams were cancelled, and teacher predictions were pupils’ only means of proving their achievement levels. Of course, there will always be some degree of error in predicting student grades. Research has shown that even when relying on machine learning and advanced statistical techniques, it is only possible to accurately predict the grades of 1 in 4 students from their attainment in previous years and characteristics. This demonstrates that teachers are certainly not to blame for inaccuracies in predicted grades, and also highlights that this difficult and time-consuming task may not be an ideal use of their limited resources. But it also demonstrates that we cannot simply shift the responsibility of the assignment of predicted grades away from them easily.

The fact that there are systematic errors in predicted grades is important. Predicted grades are an integral process of students’ decision-making, being the main piece of evidence they submit to university courses. The fact that high attaining disadvantaged students and state school students receive less generous grades than their more advantaged and independent school counterparts is likely to have consequences for their decisions on which courses to apply to. High attaining disadvantaged students are more likely to ‘undermatch’ and enter courses which are less selective than expected, given their grades, which leads to higher chances of dropping out, receiving a lower class degree, and earning less in the future.

The alternative to teacher predicted grades, used by every other major education system worldwide, is a post-qualification application (PQA) system. This would allow students to make university applications after they have taken their A-level exams and received their results. This system would be more accurate, fairer, and bring the UK in line with the rest of the world in allowing students to make these life changing application decisions based on full information.

It is possible to move towards PQA with minimum disruption to the current education system. Two proposals for achieving this are as follows: the first option would be to condense the final exam period to 4 weeks, and accelerate exam marking to 7-8 weeks. Examinations would take place in early May. Students would return to school afterwards, receiving their results in mid-July, in time for an in-school ‘applications week’. Universities would then have a month to process and make offers at the end of August, and students would have a short time to accept their favoured choice.

A second option would be to shorten the school summer holidays, allowing pupils to sit their exams and receive their grades during term time, and then make their university applications before the school holidays begin. Alternatively, first year university students could start later, giving them more time to process applications. We outline these options in more detail in a previous blog post.

We cannot ignore the flaws of a system that grants and denies young people the opportunity to have full information on their achievement levels before making such important and life-changing decisions. Put simply, if we were starting from scratch, no-one would design a system like this one. Young people deserve the chance to have their applications assessed in light of their actual achievements.

Should we abolish personal statements from the university application process?

By Blog Editor, on 19 April 2023

By Dominic Kelly and Gill Wyness


The personal statement – a key element of the university admissions process – has come under increasing scrutiny in recent years. The main criticism levied at the personal statement concerns fairness:  students from disadvantaged backgrounds may have fewer extra-curricular experiences to highlight in their personal statements and are less likely to have access to high quality advice and guidance when writing their statement.

So, it is welcome news that UCAS recently announced the personal statement is to be reformed, by replacing the free text personal statement into a structured format consisting of six short questions. While these reforms may address some of the fairness issues that many are concerned about, these reforms do not go far enough, and a better solution would be to abolish the personal statement altogether.

What’s the problem with personal statements?

Under the current rules, students applying to university through the UCAS system must submit a personal statement alongside their educational attainment (based on predicted A-level or equivalent grades). The personal statement is a free-text essay, with no specific question, just some loose advice, and a word limit of 4,000 characters. The essay is automatically sent to all the courses an applicant applies to.

There are three key reasons personal statements have been criticised – that they may favour students from more advantaged backgrounds, that they may not be great predictors of ability, and that they may put candidates under unnecessary stress.

Looking first at whether personal statements favour more advantaged students, research examining large samples of personal statements (Jones, 2012, 2013, 2015) revealed clear differences between the statements of independent/selective school applicants and those from non-selective state schools. Independent/selective school pupils had access to many more work experience opportunities to discuss, which were also broader and more diverse. The extreme differences in the extra-curricular activities that those from higher SES backgrounds have has been recently documented by Park et al (2023) whose study of US college applicants found that White, Asian American, higher SES, and private school students listed more extra-curricular activities, reported more top-level leadership roles, and reported more activities reflecting accomplishments and distinctions. Importantly, among those who listed undertaking an activity, Black and low SES students were just as likely to list having played a leadership role in the activity, suggesting that disadvantaged students are just as likely to have leadership qualities, but simply have less resources available to try different activities.

As well as having a wider range of experiences to draw from, the statements of independent school pupils were also longer, and contained fewer spelling errors and punctuation errors. One reason this may be the case is that independent school pupils may receive more guidance and assistance in writing their statements. There is some evidence for this. A study on 16- to 19-year-olds’ efficacy at ‘selling themselves’ in personal statements suggested that this was directly related to differences in resources and training provided by their educational institutions. It is also likely that those from richer backgrounds would be more able to take advantage of private consultancies to help them craft their statements, as they do with personal tutoring, for example.   A 2009 Guardian article reported evidence of independent school pupils receiving more help with personal statements, with a teacher admitting “of course we help our students with their personal statements, their parents are paying £7,000 a term!”

Turning to the usefulness of personal statements as a way to assess candidates, there are a small number of papers examining this issue. For example a small meta-analysis showed that ratings of personal statements were poor predictors of academic achievement, over and above standardised test scores and prior attainment. Arguments have also been made that rather than functioning as a holistic assessment of university suitability or preparedness, they function solely as assessments of writing skills, again casting doubt on their usefulness in identifying good candidates. There is also evidence that a more structured approach to rating personal statements yielded negligible predictive validity for first year grades and dropout rates.

Any doubts about the validity of personal statements are likely to translate into inconsistencies in their use. There is little research examining how personal statements are used by university admissions teams – so we have no real understanding of the extent to which they are trusted by admissions departments and used seriously for decision-making. A greater concern for UCAS than what applicants are writing should be how university admissions staff are assessing their writing, as well as the biases that these staff implicitly or explicitly have regarding low socioeconomic and ethnic minority groups.

Finally, examining the issue of stress, a recent report analysed a large number of personal statements from students from underrepresented backgrounds, as well as conducting surveys and interviews of these students. Applicants from under-represented backgrounds were found to experience a number of challenges when writing their essays, with some spending 30-to-40 hours on their essays, with obvious knock-on effects to their studies and free time. Particular challenges arose from the free-form nature of the essay – such as writing the opening paragraph, developing an effective flow / structure arise, and uncertainty from the lack of an explicit question.

What is UCAS’ Proposed Solution?

UCAS current proposals are to replace this free text personal statement with a structured format consisting of 6 open questions:

  1. Motivation for Course – Why do you want to study these courses?
  2. Preparedness for Course – How has your learning so far helped you to be ready to succeed on these courses?
  3. Preparation through other experiences – What else have you done to help you prepare, and why are these experiences useful?
  4. Extenuating circumstances – Is there anything that the universities and colleges need to know about, to help them put your achievements and experiences so far into context?
  5. Preparedness for study – What have you done to prepare yourself for student life?
  6. Preferred Learning Styles – Which learning and assessment styles best suit you – how do your courses choices match that?

This structured format does have some improvements over the existing personal statement. For example, Jones, Fryer and Westlake (2023) argue that shorter, more guided questions are likely to reduce the stress burden on applicants, and are more transparent, making the essay easier to write. However, there is still likely to be a significant effort involved in writing the statement, given the number of questions.

Simply breaking the personal statement down in this way is unlikely to overcome the many other issues discussed above.

First, the issue of unfairness will remain. Pupils from better-off backgrounds will still have more experiences and activities to draw on in answering the questions. In fact, reducing the remit of short-answer questions to more specific topics could further highlight the lack of extra-curricular opportunities some students have to draw upon, especially since students will then be forced to provide an answer for each topic regardless of whether they have anything they feel important to say. Given a lack of ways of verifying applicants’ information, survey research suggests that there is already an established culture of lying or embellishing on personal statements, which could be exacerbated if applicants feel forced to list opportunities. Issues with spelling and grammar and greater sophistication of language will remain, even in this environment. And independent school pupils can still avail themselves of extra help in answering the questions – in fact a structured format could even make it easier for independent schools to assist their pupils in completing the form. A further issue concerns the rise of ChatGPT software, which seems particularly suited to this new style of short question, now potentially posing an existential threat to the personal statement.

Second, the issue of whether shorter questions are likely to alleviate stress is also questionable. There is not an existing literature on the differences in stress between writing long-form personal statements compared to shorter questions; i.e., there is not a hypothetical study where participants experienced two conditions and levels of stress were compared. Assumptions that shorter questions are less stressful are based on inferences from qualitative data about the current UCAS application. Until such a study is conducted – or the results of the natural experiment that UCAS propose by changing the questions are analysed – this remains an assumption. There is a possibility that issues of opening sentences, structure and word limits will apply to each of the 6 proposed short responses, ultimately multiplying the stress compared to one response.

Finally, as has already been criticised elsewhere, any reference to the concept of ‘learning styles’ (the idea that students have inherent differences that require them to be taught the same concepts in different ways for the instruction to be effective) should be omitted. The concept of learning styles has been debunked but persists as a ‘neuromyth’ which is at best pointless and at worst harmful.

Removing the personal statement altogether

In order to move towards a fairer, more equal applications process, we believe the personal statement should be removed from the university application process altogether. This would not be an unusual situation. Many countries, such as the Republic of Ireland, operate a completely blind process where grades are the only admission criteria.

A potential criticism (as pointed out by Jones, Fryer and Westlake) is that admissions would then be purely based on academic grades (plus an academic reference), meaning inequality could still arise if students’ grades (and teachers’ perception of them) do not accurately reflect their true ability. This is especially likely among more disadvantaged students, who have typically received far less investment in their education than their more advantaged peers. However, retaining the personal statement is unlikely to help with this problem, and may even compound it, if both grades and personal statements favour more advantaged students. A further potential problem is that the personal statement can be used as a means for students to highlight their extenuating circumstances – but this option could be retained without the personal statement.

The removal of the personal statement should be paired with a continuing push towards more contextual admissions. For example, in cases where there are several applicants with similar grades, places should be filled based on a contextual admission strategy (e.g., applicants on Free School Meals, from schools that traditionally send few applicants to university, etc., should be favoured). And beyond simply dealing with ties, students from low SES backgrounds should be given grade discounts. As we outline in a recent blog post, there is a clear economic rationale for the use of contextual admissions, to “level the playing field” at this crucial life stage. Any remaining ties in grades could be filled based on random assignment, which has also been shown to be a fair system of assignment when allocating individuals with the same levels of achievement.

The UK’s university applications system has remained unchanged for many years, and this reform is a unique opportunity to improve the fairness of the system. However, UCAS proposals do not appear to go far enough to achieve this goal.


A substantial childcare package from the Chancellor – but children’s needs take a back seat

By Blog editor, on 15 March 2023

Claire Crawford and Laura Outhwaite

We had to wait almost an hour to hear details of the heavily trailed package of childcare reforms announced by the Chancellor in the Budget today, but when the moment finally arrived, it lived up to its pre-programme billing. This was the most significant investment in childcare for years.

But while the Chancellor presented the reforms as part of the “Education” pillar of his 4 Es of economic growth and prosperity, they should really have been part of the “Employment” pillar. These reforms were all about childcare as a route to higher labour market participation and working hours, for mothers in particular. There was nothing about the implications for children, their development and wellbeing, or the inequalities in access and child outcomes that these reforms would cement. And little on the importance of quality, which is vital to the success of these reforms.

The most substantial element of the reforms was the one leaked last night – the introduction of 30 hours per week of free care during term-time for children 9 months and older in working families. For a Chancellor seemingly interested only in the likely impact on mothers’ labour supply, our earlier work comparing the impact of offering 0 vs. 15 vs. 30 hours of free care per week during term-time offers some good news. We found that offering 30 hours per week of free care during term-time was effective at encouraging more mothers into work, while the offer of 15 hours per week had no effect. Although, as we have pointed out previously, it does seem odd for a policy focused on supporting working families to apply during term-time only. When stretched across a whole year – which is what many working families using private care have to pay for – it is really only 22 hours per week, and then only during certain hours of the day and excluding ‘extras’ like lunch.

More concerningly, these reforms will undo what little progressivity there was left in the system, funnelling more money to support children from potentially very high-income working families, while those in non-working families receive less support. The existing 15 hours per week of free care for the 40% most disadvantaged 2-year-olds will be dwarfed by the new reform. Only children from non-working households amongst the 60% better-off families will not be entitled to any free care at age 2. And, just as with the existing 3-4-year-old offers, those in families with household income of nearly £200k per year will receive more government support than non-working families on Universal Credit, who receive an 85% subsidy on their childcare spending (now up to a higher cap and paid upfront, rather than in arrears).

There are already large differences in skill development and school readiness between children from richer and poorer families by the time they start school, driven in part by differences in use of high quality early education. As we outlined in an earlier briefing note, high quality early education, including from as young as age 1, can benefit children’s development, especially for those from more disadvantaged backgrounds. But these reforms will bolster the amount of time spent in early education for those from better-off backgrounds, while doing little to improve the participation rates of those from more disadvantaged backgrounds. This is likely to exacerbate, rather than address  inequalities in child learning and development.

The benefits of early education found in the literature are heavily reliant on the provision available being of high quality. And in a move that will be hugely welcome to providers, the Chancellor also announced a rise in the funding rates paid for existing hours of free care. We’ll leave our friends at the Institute for Fiscal Studies to crunch the numbers on what the figures mean for the hourly rates paid to providers over the next few years. But it was notable that there was no mention of the funding rates to be paid for the new provision. This will be absolutely vital in determining how many providers choose to offer the free hours, and what quality of care they will be able to provide.

And quality is crucial to the effects of these policies. Not only for children’s development, but also for parents’ labour force participation. Parents will be more likely to use the provision available – especially for younger children – if they are happy with the quality of care their child is receiving, making it vital for labour supply decisions. Moreover, evidence suggests that a child attending a low-quality setting can have negative effects on the wellbeing of both parents and children. For all of these reasons, it is essential that providers are paid sufficiently for the care they deliver.

Alongside incentives to encourage more childminders to enter the system and schools to offer more wraparound care, the Chancellor also went ahead with the much-maligned reform to relax staff:child ratios for 2-year-olds from 1:4 to 1:5, bringing England into line with Scotland. As we set out in an earlier briefing note, there is little concrete evidence that this would significantly harm children’s outcomes – but neither is there much evidence that it would significantly reduce parents’ childcare costs, and a clear risk that it could further damage already low workforce morale. Given that today’s announcements may substantially increase demand for childcare for 1-2-year-olds – potentially requiring a massive increase in the staff needed to deliver it – ensuring that the pay and conditions of staff in the early years sector make it an attractive option will be hugely important.

The Chancellor certainly didn’t try to get away with tinkering around the edges of the childcare system. He has gambled on a large expansion of free childcare encouraging more mothers back to work, helping to fill vacancies and plug skill shortages in an effort to increase economic growth. But these short-term benefits for productivity may come at the expense of children from our poorest families missing out on the benefits of early education, potentially damaging productivity in the longer-term.

Why greater investment in early years should be a no-brainer for the Chancellor

By Blog Editor, on 14 March 2023

Claire Crawford & Laura Outhwaite

With the Budget just around the corner, the calls for the Chancellor to tackle soaring childcare costs are growing into a clamour. A recent survey estimated the average cost of a full-time childcare place for a 2-year-old in England at nearly £15k per year – an eye watering 45% of the median annual salary of a full-time employee in the UK. For a parent – unfortunately still usually the mother – considering whether to work following the birth of their child, knowing that such a large chunk of their take-home pay is likely be swallowed by childcare costs will inevitably give them pause. This has implications for the economy in the short-term, potentially drawing valuable labour out of the workforce at a time of high labour demand and skills shortages. It also has implications for inequalities in school readiness and beyond, as those from lower socio-economic backgrounds are less likely to access high quality early years education, which benefits children’s development.

Public opinion seems to be pointing clearly in the direction of greater support for early years. But does the evidence support this position? And how generous would the Chancellor have to be on Wednesday lunchtime to make a material difference to these challenges?

Our new briefing note reviews the evidence on the effects of offering free or highly subsidised childcare for parents and children. It also compares what the evidence suggests might be the optimal policy to what is currently available in England, to highlight what additional investment might be required.

There are two main takeaways:

  • First, there is a strong evidence-based case for significant additional government investment in early childhood education and care (ECEC). The evidence suggests that offering additional free or highly subsidised childcare for parents of 0-2-year-olds could encourage more mothers to work (more) and also benefit children’s development. Part-time provision (e.g., 15 hours per week during term-time) might be sufficient to deliver the benefits for children, but would be unlikely to increase mothers’ labour supply very much; greater investment (offering full-time provision, ideally across the full year rather than term-time only) would be required to achieve this.
  • Second, quality is key. This is vital to maximise the benefits for children’s development, but its importance goes beyond this. Parents will be more likely to use the provision available – especially for younger children – if they are happy with the quality of care their child is receiving, making it crucial for labour supply decisions. Moreover, evidence suggests that a child attending a low-quality setting can have negative effects on the wellbeing of both parents and children. For all of these reasons, it is essential that providers are paid sufficiently for the care they deliver.

What does this mean the Chancellor should do on Wednesday lunchtime? Our reading of the evidence leads us to make two primary recommendations:

  • The funding rate paid by the government to cover the free early education entitlements for 3-4-year-olds and disadvantaged 2-year-olds in England should be increased. This would help to cover cost rises that are largely out of providers’ hands (e.g., energy price rises, increases to the National Living Wage) and reduce the incentive for providers to charge more for privately paid-for hours of care to compensate for the lower rate paid by the government. This would enable providers to reduce fees and/or invest more in raising the quality of care they provide.
  • A more generous childcare subsidy should be introduced for families of 0-2-year-olds, to reduce the financial barriers to work, particularly for mothers, and support children’s development. The subsidy should be progressive – higher (potentially 100%) for families with lower income – to minimise the risk of such reforms widening inequalities. As discussed in more detail in our submission to the Education Select Committee’s Inquiry on Support for Childcare and Early Years, it would be more transparent for such a subsidy to be delivered via something like the tax-free childcare system than by offering more free hours of care during term-time only, whose benefits are less than the headline amount for families who need childcare throughout the year.

It has been widely reported in the media that the Chancellor will announce the cap on childcare costs that can be claimed by Universal Credit recipients will be raised, and that families will be able to claim childcare costs upfront rather than in arrears. While this will undoubtedly be good news for families for whom these restrictions would otherwise prevent them from using childcare or moving into work, if that is the limit of the financial support for childcare announced on Wednesday lunchtime, then the Chancellor will have disappointed many – and gone against the evidence outlined in our briefing note.

What could arguably be described as ‘tinkering around the edges’ is not going to generate the kind of sizeable reduction in childcare costs that is needed to make a material difference to families’ budgets, parents’ labour supply decisions or children’s development. It is also crucial that any actions taken to reduce childcare costs for families are not delivered at the expense of reducing the quality of care on offer. For both reasons, as we highlighted in another recent briefing note reviewing the evidence on the links between childcare ratios and children’s outcomes, relaxing childcare ratios is unlikely to be a good solution to the current childcare conundrum either.

The evidence suggests that investing in high quality ECEC is likely to provide a triple whammy of benefits – improving children’s development, increasing family income and boosting productivity – and that the long-run benefits are highly likely to outweigh the short-term costs. We can only hope that current labour demands are sufficient to provide the short-term incentive needed for the Chancellor to look beyond the gloomy fiscal position and make the kind of long-term investment needed to deliver these benefits.

Are Ofsted inspections helpful for choosing secondary schools?

By Blog Editor, on 1 March 2023

By Sam Sims.

Each year, the parents and carers of around half a million pupils submit applications to attend their preferred secondary schools. This decision usually determines the quality of the education that their child receives between age 11 and 16, when they take their GCSEs.

A 2017 YouGov poll found that just under half of parents looked at Ofsted reports to inform their choice. Besides proximity to home, Ofsted inspection judgments were cited as the most important influence on their choices.

Parents put enough weight on these judgements that Rightmove displays the inspection judgements of nearby schools for each of the properties listed on their website. Indeed, research shows that when schools’ Ofsted ratings increase, the price of nearby houses also increase as parents move into the ‘catchment area’.

Ofsted also argue that inspections support school choice: “Inspection provides important information to parents… [who] should be able to make informed choices based on the information published in inspection reports.”

However, one reason to doubt whether Ofsted can inform school choice is that many schools haven’t been inspected for several years. Consider a parent who was choosing a school in October 2013. They might be reading a report from an inspection that was conducted in 2011, and their child wouldn’t start at the school until September 2014. Of course, parents are really interested in how good the school will be in the five years after that (2014 to 2019).

So are Ofsted inspections really helpful for informing secondary school choice? To find out, John Jerrim, Christian Bokhove, and I analysed all inspections of mainstream secondary schools conducted between 2005 and 2015. We focused on inspections from this period – under the old inspection framework – so that we could look up what really happened to the pupils who subsequently attended these schools.

For our hypothetical parent, we found that the most recent inspection report across our 2,538 secondary schools was on average 1,040 days (almost 3 years) old by the time their child would begin at the school. Reports were even older in many schools rated ‘Good’ or ‘Outstanding’, which are inspected less frequently (Figure 1). These long lags mean that half of the headteachers in place when the most recent inspection was conducted were no longer in post when the child would have started at the school.

Figure 1: lags between last inspection and school entry

But perhaps old inspection grades are still informative? To tests this, we looked at whether the most recent inspection judgement predicted pupil outcomes during the period in which our hypothetical child would have actually attended the school – between  2014 and 2018.

Figure 2 shows the results when we rank schools based on average pupil GCSE attainment across 8 subjects. Schools on the far left have the lowest average attainment in the country and those on the far right have the highest. It is clear from the graph that pupils who attend a school with higher Ofsted grades at the time of application go on to attain higher.

Figure 2: School average Attainment 8 ranking by Ofsted judgement, unconditional

However, Figure 2 may simply reflect a self-fulfilling prophecy in which rich parents of high achieving children buy houses near ‘Outstanding’ schools. In Figure 3, we therefore control for pupil prior attainment, admission type, and pupil deprivation. The difference in school rankings across the four Ofsted judgements now collapses – the ranking of schools is very similar across Ofsted judgements.

Indeed, once we control for the school exam results that were available to our hypothetical parents at the time they made their choice, there is no longer any detectable difference between schools rated ‘Good’, ‘Requires Improvement’, or ‘Inadequate’. Only an ‘Outstanding’ judgement was associated with a (0.1 standard deviation) increase in pupil results.

Figure 3: School average Attainment 8 ranking by Ofsted judgement, conditional

But perhaps it’s misplaced to expect inspection judgements to add value to predicting exam results? It might be argued that inspections inform parents primarily by gathering evidence of what it is really like ‘on the ground’ in a school, which may not be adequately captured by exam results?

We tested this by looking at a range of quality metrics that would be easier for inspectors to observe than for parents. We found small differences on pupil absence rates, no differences in parental satisfaction, and (again) no detectable difference between the bottom three judgments in terms of parent reported behaviour standards.

By and large, inspection reports are not particularly useful for parents choosing secondary schools. The same could be achieved by simple applying intuitive labels like ‘Good’ or ‘Requires Improvement’ to schools based on their Progress 8 scores. The one exception to this is the ‘Outstanding’ judgement, which is informative. However, we recommend that parents think twice before paying more money for a house because it is near a ‘Good’ school.

Is there any way that inspections could better inform school choice? We found that there is a stronger relationship between Ofsted judgements and pupil progress if inspections were less than two years old. However, given that 88% of schools are currently inspected only once every five years, moving to inspect every school biennially would be a costly reform.

Young people’s physical health during the COVID-19 pandemic

By Blog Editor, on 30 January 2023

Jake Anders is Principal Investigator of the COVID Social Mobility & Opportunities study (COSMO), and an Associate Professor and Deputy Director at UCL Centre for Education Policy & Equalising Opportunities.

Although young people were among those least likely to be directly affected by severe effects of COVID-19, they were not immune from its immediate effects on health. We are better able to understand implications of this using data from the COVID Social Mobility & Opportunities study (COSMO). The study includes a representative sample of over 13,000 young people across England, who were aged 14–15 at the onset of the pandemic, and 16–17 during the academic year 2020/21 when our first data were collected.

COSMO’s purpose is also wider than the direct health impacts of COVID-19. As such, this blog post — drawing on our latest COSMO briefing published today — also takes a wider look at young people’s health behaviours during this period.

Young people’s experiences of COVID-19

When first taking part in COSMO (between October 2021 and March 2022) almost half of the cohort reported having had COVID-19 (of whom 28% had had this confirmed by a test, likely with many having been affected before widespread testing was available).

Many of these young people will have experienced relative mild — or at least, transient — symptoms. But this was not the case for all. 1 in 5 of young people who reported having had COVID-19 (just under 10% of the cohort as a whole) reported that they continued to experience symptoms more than 4 weeks after first catching the virus: the accepted definition of Long COVID. Among this group, a quarter described their case of Long COVID as limiting their ability to carry out daily activities a lot.

Figure 1: Proportion reporting a case of severe long COVID by deprivation quintile group

Furthermore, there were inequalities in young people’s chances of experiencing long COVID. Among those who reported having caught COVID-19, those from less socio-economically advantaged backgrounds were more likely to report that this had persisted into a case of long COVID. This is despite no difference in reporting having contracted COVID-19 between these groups. What, then, explains differences in this persistence? Reasons for these inequalities are rather uncertain. One hypothesis is that getting sufficient rest is important for COVID-19 recovery, and this may be easier to achieve for some groups than others. Another is that pre-existing differences in overall health levels associated with socio-economic status pre-infection explain risk of prolonged disease.

Educational consequences of physical health during the pandemic

Young people who reported having experienced severe Long COVID received lower teacher-assessed GCSEs than their peers who had never had COVID-19. Given the differences discussed above between young people who experienced severe Long COVID, we wanted to ensure that this was not simply an artefact of these differences. After adjusting statistically for differences in demographics, socio-economic status, and prior attainment, it was still the case that young people who had experienced severe long COVID performed worse than comparable peers. The size of this difference is very roughly equivalent to two months of learning.

Figure 2: Differences in GCSE teacher assessed grades by shielding status and severe long COVID

Notes: Differences reported have been standardised. Severe long COVID analysis is compared to those who reported not having had COVID at all.

Those who were asked to shield during the pandemic — because pre-existing medical conditions meant that they were at the highest risk of severe illness if they caught COVID-19 — also received lower grades in their teacher assessed GCSEs than their peers. 8% of the COSMO cohort reported being asked to shield and — after adjusting for demographics, socio-economic status, and prior attainment — had lower GCSE grades very roughly equivalent to four months of learning.

Young people’s wider health

COSMO also allows us to shine a light more broadly on the health of this generation of young people.

Continuing a trend seen across successive cohorts, this cohort of 16–17 year olds are less likely to drink alcohol than those who came before them. 63% say they have ever done so, compared to 85% of a comparable group of 16–17 year olds in 2007. Across various measures, pupils from advantaged backgrounds (such as those attending private schools, or in state schools with more advantaged intakes) are more likely to report ever having drunk alcohol, and reported drinking more frequently.

This group of young people are more likely to have ever used an e-cigarette (33%) than to have ever smoked a cigarette (23%). However, regular e-cigarette use is much lower than this might imply. Just 6% report using e-cigarettes once a day or more. In contrast to alcohol use, those from less advantaged backgrounds are more likely to use e-cigarettes than their more advantaged peers.

Just under 1 in 6 of this cohort of young people report having ever tried any illegal drugs. Among those who have, almost all had tried cannabis. Just 4% of the sample reported having tried any other illegal substance. And similar to the patterning seen with alcohol, those in more advantaged circumstances were more likely to have taken an illegal drug. Those identifying as White or Mixed ethnic background are far more likely to report having tried illegal drugs than any other ethnic groups.


With attention on issues around COVID-19 fading, we should not forget its continuing effect on those for whom it was a particularly debilitating illness to experience. Some continue to suffer from long COVID. Even those who have recovered have seen implications for their wider lives and life chances, such as lower academic attainment scores for those who experienced severe long COVID. These impacts also seem to have reinforced existing health and socioeconomic inequalities.

COSMO also helps us shine a wider light on the experiences of this cohort of young people, and quite how much this has changed compared to earlier generations. Substantially reduced drinking of alcohol compared to earlier generations, low rates of drug use, and the emergence of e-cigarette use all raise questions about implications for future health, too.

There is more detail on these findings, as well as a wider range of health impacts and behaviour of the COSMO cohort, in our latest briefing note. You can also find our earlier briefing notes on the study website.

Do contextual admissions hold back productivity?

By Blog Editor, on 20 December 2022

By Gill Wyness, Lindsey Macmillan, Claire Crawford and Richard Murphy.

There are substantial financial gains to attending university, and gains are especially high for the most selective institutions, from which many of the highest paying professions recruit. Recognising this, many parents send their children to independent schools, believing that this will dramatically increase their chances of gaining a place at a highly competitive institution. But a recent investigation by The Telegraph suggests the independent school stranglehold on Oxbridge is under threat, with the likelihood of their pupils receiving an offer dropping dramatically over the last five years. Commentators have blamed the practise of contextual admissions, which give pupils from state schools preferential entry – in some cases even when they have lower entry grades than those from private schools. Those against the practise argue it amounts to “class discrimination” and “social engineering”.

But the reality is that there is a clear economic rationale for contextual admissions. Students from disadvantaged backgrounds have typically received far less investment in their education than their more advantaged peers – including, but not limited to, the fact that they tend to attend lower quality schools. We would expect these disadvantages to translate into lower grades for a given level of underlying ‘ability’, reducing promising but disadvantaged students’ chances of gaining a place at a selective university. Grades are therefore contextualised for equity reasons, to “level the playing field” at this crucial life stage.

While there can be little doubt that contextual admissions improve equity, good economic policies should also be efficient. But many are concerned that contextual admissions may be inefficient. This would be the case if the lower attaining state school students who were given preferential entry to Oxbridge found themselves struggling academically, did poorly in their degrees and were less productive in the labour market. If their net gain from Oxbridge (versus how well they would have done at a lower quality institution) is lower than that which would have been experienced by the independent school students they displaced (taking account of how well the displaced student would have done at Oxbridge versus where they eventually ended up), this would constitute an efficiency loss. We should consider this equity-efficiency trade-off when we think about contextual admissions.

So, does it matter which students go to which universities? Are there higher returns for higher achieving students attending highly selective courses? Could the practice of contextual admissions be said to be damaging productivity or hampering economic growth?

One area of research we can look to, to inform this debate, is the area of student to university match. Research in this area considers the types of courses that students of different achievement levels attend, asking which is the most important for earnings: student achievement, course quality, or the interaction – or match – between the two. This helps address the question of whether students have better outcomes at universities to which they are better ‘matched’.

The evidence from this relatively small area of literature is mixed, however, with much of the evidence based on the effects of affirmative action bans in the United States. This has strong parallels with the use of contextualised admissions, in that it concerns the use of race (and other factors) to judge university applicants in order to improve the diversity of student populations, and is just as controversial in the US as it is here, with ongoing legal action currently being brought against two top universities – Harvard and University of North Carolina – which has now made its way to the Supreme Court.

One paper (Arcidiacono et al, 2016) examines the impact of the ban on affirmative action in California in 1998. After the ban, minority students no longer received preferential admissions to highly selective universities like Berkeley. This study argues that lower achieving minority students at top-ranked campuses would have higher science graduation rates had they attended lower-ranked campuses – in other words it was economically inefficient to send them to these universities.

However, a more recent paper studying the same affirmative action ban (Bleemer, 2022) found that it harmed underrepresented minority students by lowering their degree attainment and wages, as after the ban they were cascaded into lower quality universities. Indeed, the study suggested that that affirmative action’s net benefits for underrepresented minority applicants exceed its net costs for the white and Asian applicants at most risk of being displaced. The two papers use quite different approaches to study the question which may provide some explanation of why they find opposing results (with the former paper using a structural model, and the latter using nonexperimental methods). Other papers (Dillon and Smith, 2020; Light and Strayer, 2000) have found some evidence of student-university match effects, but have also found that these are far outweighed by course quality effects, suggesting students of all ability levels should try to get onto the best quality course they can to maximise their earnings. In other words, disadvantaged students are just as likely to benefit from going to a high quality course as those with higher attainment on entry, meaning contextual admissions are unlikely to harm them individually – while any efficiency losses are likely to be small.

A related strand of literature from the UK compares degree outcomes for students from different backgrounds or different schools with the same achievement on entry. These studies (e.g. Crawford, 2014a,b) find that students from private schools are, on average, more likely to drop-out, less likely to complete their degree and less likely to graduate with a 1st or 2:1 than state school students entering the same course with the same grades. This hints at the fact that achievement in school does not capture ‘ability’ or ‘potential to succeed’ in the same way for those from different school types, providing support for the rationale underlying contextual admissions.

Given the scarcity of evidence, and its mixed results, it is hard to argue that there are likely to be substantial efficiency losses to society from contextual admissions – while there are undoubtedly equity gains. There are still vast socio-economic disparities in entry rates to highly selective institutions in the UK. These disparities occur because those from lower socio-economic backgrounds typically achieve lower grades than those from richer backgrounds, who have had the benefit of a lifetime of high-quality education and parental input.

Contextualising admissions is one of the few tools we have at our disposal to help students with great potential to gain a place at a selective institution. We should not outlaw this practice without good evidence that it significantly harms productivity.

How can education and skills contribute to levelling up?

By Blog Editor, on 7 December 2022

Education and skills contribute to differences in earnings and opportunities across generations, both at the national level and between local authorities in England. Targeted policies can help to reduce the gaps between individuals and contribute to efforts to level up the UK economy.

The UK is very unequal, in terms of both outcomes and opportunities. Among developed nations, it has one of the highest rates of income inequality – unequal outcomes – and one of the lowest rates of intergenerational mobility (unequal opportunities).

This is illustrated in the Great Gatsby Curve, a coinage of the late, great economist and US presidential adviser Alan Krueger, based on work by Miles Corak, which highlights how greater inequality is associated with less mobility across generations (see Figure 1).

Figure 1: The Great Gatsby curve

Source: Corak, 2013

Based on this evidence, for the past few decades, successive governments have been focused on improving social mobility, equalising opportunities or, more recently, levelling up. But these things are notoriously hard to measure, not least because policies enacted today can take decades to come to fruition in terms of future changes in inequality.

Research looking at human capital emphasises the importance of education and skills in driving productivity, wages and economic growth. Those that invest in education and training are more likely to have higher skill levels and be more productive, which means that they can earn higher wages.

Education and skills have also been shown to play a direct role in inequality, contributing to differences in earnings and opportunities across generations, both at the national level and across local authorities in England.

Studies show that differences in skills can explain at least a third of the variation in earnings within countries. Further, education and skills can account for 60-80% of the transmission of income across generations – or unequal opportunities.

Trends in inequalities in education and skills can therefore give us some useful insight into future patterns of social mobility. For example, given the improvements in the educational performance of pupils in London seen over the last two decades, it is little surprise that the capital now appears to be one of the most mobile areas in the country in terms of labour market outcomes.

What do recent trends in educational inequalities tell us about progress with levelling up?

The recent evidence on levelling up doesn’t look good. Trends in educational attainment across regions show that educational inequalities are widening across places. London continued to pull away from other regions across primary school, GCSE and A-levelachievement in 2022 compared with 2019, before the Covid-19 pandemic.

While attainment fell everywhere due to the impact of the pandemic and learning losses, key stage 2 results fell by six percentage points (ppts) in London compared with eight in the North West. Similarly, 65% of London pupils achieved the expected standard in reading, writing and maths in 2022, compared with 57% in the North West, and 56% in Yorkshire and Humber, and the East (Figure 2).

Figure 2: Map of England by key stage 2 attainment

Source: Department for Education, 2022

At key stage 4, 32.6% of GCSE entries were awarded grade seven or above in London in 2022. In the North East, the figure was 22.4%, a difference of 10.2 ppts. In 2019, the corresponding gap was 9.3 ppts.

Similarly, at A-level, the percentage of pupils achieving an A grade was 39% in London, compared with 30.8% in the North East, a gap of 8.2 ppts in 2022, relative to 3.9 ppts in 2019.

There are two points to make about this. First, the fact that attainment fell everywhere is really bad news for future productivity. If pupils are progressing with less knowledge than they had previously, they have fewer of the building blocks needed for their next stages of learning – a less solid foundation to build on moving forward. This could have very serious consequences for future labour market productivity – estimates on the cost of this ‘learning loss’ range from billions to trillions of pounds. Second, the widening of these gaps is the opposite of what we want to be seeing if we are to level up or minimise the differences between regions of the UK. These widening educational inequalities are likely to translate, in part, into widening labour market earnings and incomes unless action is taken to address the disparities.

Is there scope for this to improve?

The government has committed to a range of policiesto level up the economy, including 55 new educational investment areas (EIAs), provision of high-quality online curriculum through Oak National Academy and setting the target of 90% of primary school children achieving expected standards in reading, writing and maths by 2030. There is also talk of an overhaul of funding and governance in colleges, a skills audit, and local skill improvement plans to target specific regional needs.

Yet looking forward, it is hard to see how this picture will improve against a backdrop of increasingly squeezed school budgets. Even before the pandemic, work by the Institute for Fiscal Studies (IFS) showed that the gap in funding between state and private schools has grown steadily since 2010 (see Figure 3).

Figure 3: State school spending per pupil vs average private school fees over time (2021–22 prices)

Source: IFS

Indeed, spending per pupil in state schools declined by 9% in real terms between 2019 and 2020. This has disproportionately hit schools serving more deprived pupils and further education colleges.

Alongside the pressure of learning losses from the pandemic, the cost of living crisis and rising teachers’ wages to be paid out of existing budgets, it is hard to see how the education sector will be able to offer the support needed, particularly for the most deprived pupils.

Where can I find out more?

This post first appeared on Economics Observatory on 7 December 2022.