Close
The CEPEO Blog
By IOE Editor, on 2 February 2020
Welcome to the UCL Centre for Education Policy and Equalising Opportunities (CEPEO) blog. This blog is a forum for CEPEO members, affiliates, alumni and guests to write about research on our four research themes.
Our Research Themes
CEPEO concentrates on four research themes, each underpinned by the aim to improve the education system and equalise opportunities for all. These are:
We also recently published our evidence-based Policy Priorities, and have been blogging about each of these. You can see all of the posts about this topic here.
Testing, Testing – Why are Black applicants not getting graduate jobs?
By Blog editor, on 11 March 2025
By Dr Claire Tyler, Professor Lindsey Macmillan, and Dr Catherine Dilnot
Last Thursday we released our Nuffield-funded ‘Inequalities in Access to Professional Occupations’ report, and an accompanying blog post showing that Black undergraduates are well represented in application pools for professional graduate roles but are almost half as likely (45% less likely) to receive to receive job offers than white applicants (Figure 1 – blue dot).
Even when we compare candidates who ‘look the same on paper’, including attending a similar university, studying a similar subject and applying for similar jobs, Black applicants are still a third (34%) less likely to receive a job offer than white applicants (yellow dot).
This analysis is based on a large sample of 117,043 applicants (of which 9,859 reported being of Black ethnicity) to graduate schemes at 17 large UK employers, across a range of sectors including accounting, law and public sector.
Figure 1: Offer rates to graduate programmes, conditional on observable differences across applicants, by ethnicity
Recruitment stages
In this blog post, we delve deeper into the low success rates for Black applicants to explore which stage of the recruitment process appears to be causing the largest barriers for these undergraduates.
For our analysis, we firstly needed to create some alignment in the recruitment data we received from 17 employers, as recruitment stages varied both by employer, and over time for the same employer. For consistent reporting, we grouped recruitment activity into two key stages:
1) Screening and testing, which includes all application sifting, screening based on educational credentials and online testing. Online tests assess skills such as numerical reasoning, verbal reasoning, situational judgement, behaviours, preferences and strengths.
2) Face to Face which includes interviews (video and in person), and assessment centres (case studies, group exercises).
In our sample, only around a quarter of candidates applied to a recruitment process which included screening based on prior educational qualifications (GCSE, A-levels and/or 2:1 degree) as these criteria are becoming less popular with employers. Whereas most participating employers used online tests, interviews and assessment centres as selection methods (80-100% of candidates applied to employers who used these methods).
Screening and testing matters
Exploring success rates for each of the two key stages revealed that Black applicants are 45% less likely to pass the upfront screening and online testing stage than white applicants and 8% less likely to pass the face to face stage (Figure 2 and 3 – blue dots). Even comparing applicants on a ‘like for like’ basis , Black applicants are 37% less likely to pass the screening and online testing stage than their comparable peers of white ethnicity, and 5% less likely to pass the face-to-face stage (Figure 2 and 3 – yellow dots).
Figure 2 – Relative likelihood of passing screening and online tests for the graduate programmes, conditional on observable differences across applicants, by ethnicity
Figure 3 – Relative likelihood of passing face to face stages for the graduate programmes, conditional on observable differences across applicants, by ethnicity
A recent report by Black Talent Charter also shows that ethnic minorities are disproportionately rejected in the early stages of recruitment, which further supports our findings. Their analysis of 400 students shows that the representation of Black applicants falls markedly at the ‘screening & testing stage’ in financial services recruitment (Figure 4) from around 16% of the applicant pool to 10% of the candidates invited for job interviews. For professional services the representation drops from 13% to 7% and in law from 10% to 7%. In each case, more Black applicants are rejected from the process at the ‘screening and online testing stage’ than at the following ‘face to face’ stage, both in terms of raw numbers of applicants and percentage point representation. The authors of the report quite rightly highlight the drop in representation at the ‘face to face’ stage between job interview and job offer but do not draw out the implications of the larger barriers at the earlier screening and testing stages.
Figure 4 – Percentage representation of Black talent throughout the candidate journey, from university application to job acceptance, across financial services (FS), law and professional services (PS)
Note: Figure reproduced from the Black Talent Charter 2204 report ‘Why We Need To Fix The Broken Pipeline for Graduate Black Talent in the UK’ (Figure 6, page 12). Red oval added by UCL to highlight figures of interest
Widen the focus?
Our findings suggest that employers may be overlooking racial inequalities at screening and testing stages of the recruitment process. Racial inequalities may still be seen as an ‘in person’ issue, whereas our evidence highlights other significant barriers in the recruitment process. This may explain why limited progress has been made to date in removing these inequalities in access to professional careers.
These findings mirror patterns of prior education and skills inequality via online tests such as numerical and verbal reasoning, situational judgement and critical thinking (we are hoping to be working on a new project about this soon). But the fact that these gaps are still so large after accounting for university attended suggest they may also relate to lack of preparation for such tests if these applicants are less likely to have parents or networks with experience of these types of tests.
This evidence also suggests a potential conflict between outreach activities of employers to widen the diversity of application pools and the initial screening and testing processes which disproportionally reject ethnic minority applicants.
Given the huge increase in graduate job applications and a record high ratio of applications to vacancies (140:1), it is also important to recognise that recruiters are currently dealing with unprecedented volumes of applications in narrower recruitment windows. The screening and online testing stages of recruitment act as a substantial filter for recruiters, with only 15% of candidates in our sample passing this stage, with the remaining 85% being rejected before any face to face interactions with the employer. With the increasing use of AI by job applicants, ‘getting it right’ at the screening and testing stage is becoming ever more important for employers – balancing the need to process huge volumes of applications while ensuring the best talent is recruited.
Recommendations
To improve the success rates of Black and other ethnic minority applicants we therefore recommend:
- Employers broaden the focus of ethnic diversity interventions/strategies to include the screening and online testing stages of recruitment in addition to any existing focus on face to face assessments.
- Employers review the success rates of under-represented groups at each stage of the recruitment process to see where the largest barriers occur.
- University career teams and employers collaborate to share insights about which under-represented groups may need additional information or guidance to be prepared for online recruitment assessments.
- Employers share these findings with your ‘race equality’ leads to ensure these issues are considered when planning new initiatives.
- Employers challenge external providers of recruitment assessments to provide data on success rates of candidates from under-represented groups on each type of test conducted.
What’s next?
We are working on launching a new project soon to delve deeper into the barriers for applicants from ethnic minority and lower socio-economic backgrounds – please get in touch if you would like to hear more about how to participate in our research collaborations. Projects are anonymous, free of charge and full of insights for participating employers.
The Early Bird Catches The Worm – Why Applying Early Matters for Social Mobility
By Blog editor, on 10 March 2025
By Dr Claire Tyler, Professor Lindsey Macmillan, and Dr Catherine Dilnot
Last Thursday we released our Nuffield-funded ‘Inequalities in Access to Professional Occupations’ report, and an accompanying blog post showing that working class students are well represented in application pools for professional graduate roles but are 32% less likely to receive job offers than applicants from professional backgrounds.
On Friday, our blog post revealed that date of application accounts for 11% of the gap in offer rates between applicants from working class and professional backgrounds in the accounting and professional services sector. This means that even if an applicant from a working class background obtained the same A-level grades, attended a similar university and studied the same undergraduate subject (also with the same gender, ethnicity, UK region of origin and office location and applied to the same employer), they are still less likely to obtain a job offer because they are more likely to apply later.
Our findings suggest that employers who can reduce or remove the delay in applications from working class applicants, will likely see an improvement in the offer rates to working class applicants to their graduate schemes. We therefore argue that improving the ‘application readiness’ of working class applicants may be low hanging fruit amongst the many challenges which exist in improving social mobility into professional careers.
Even applying a few weeks earlier can make a difference to a candidate’s chance of obtaining a job offer. We found that applying one month earlier is associated with around a 6% increased chance of obtaining a job offer (0.4ppt benefit per month, mean offer rate 6.3ppt), even among otherwise similar candidates.
Closing windows
Highlighting the importance of application readiness for working class applicants is particularly timely as data from the Institute of Student Employers (ISE) Recruitment Surveys 2021 to 2024 shows that recruitment windows have been narrowing over recent years. In the last year of available data from the 2023/2024 recruitment cycle collected in July 2024, 38% of employers closed graduate recruitment between October to December (2023), compared to 24% the previous year and 14% in the year before that (Figure 1). In our own data from employers, we show that by the start of an undergraduate’s final year of study (early October), the accountancy firms in our sample have already received 30% of applications and made 50% of job offers for graduate schemes starting the following September.
Figure 1: Recruitment campaign closing date trends
The ISE suggest the decision of graduate employers to close earlier this year may have been a response to being overwhelmed by the recent 59% increase in graduate job applications and the record high ratio of applications to vacancies (140:1), the highest recorded in over 30 years. It is likely this huge increase in application volumes is due to the increasing use of AI by applicants and the ongoing removal of academic attainment criteria . Our findings suggest that if this early closure trend continues it risks further disadvantaging lower SEB candidates.
The work experience ‘treadmill’
Initial feedback from discussing our findings with universities and employers highlighted that lower SEB candidates may apply later to graduate schemes for many reasons including less awareness of application deadlines; being unaware that employers may close recruitment early; less confidence in their application so they may take longer to prepare; and being more likely to be balancing competing priorities of study and graduate job applications with part time employment.
This issue of ‘application readiness’ is also far from confined to graduate scheme applications. Internships, undertaken a year earlier, also face a similar issue as undergraduates must be ready during their second year of university to apply to these programmes. Figure 2 shows that state school applicants ‘under’ apply to internships, whereas private school applicants ‘over’ apply, relative to the graduate talent pool. In other words, applicants to internship routes from independent schools are hugely overrepresented relative to the proportion of graduates from our national benchmarks attending these types of schools, with double the proportion applying to internship routes (27% compared to 13% nationally). State school applicants are therefore underrepresented in applicant pools for internships, but crucially when they do apply, they perform equally as well as independent school students during the recruitment process (Figure 3). So the key here is encouraging undergraduates from state school to be ready to apply for internships during their penultimate year at university – and as early as possible within that year.
Figure 2 : Proportion of interns at various stages of the recruitment process, by school type
Figure 3: Offer rates to internship programmes, conditional on observable differences across applicants, by school type
Given the drive for employers to lock in talented undergraduates early, awareness of application deadlines and requirements can also be important for students as early as first year where undergraduates scramble to apply for spring internship opportunities as soon as they enrol in first term. The ‘treadmill’ of work experience schemes is therefore starting earlier in an undergraduate’s journey, with high conversion rates to competitive graduate jobs for the lucky few. As one young person told us this week, “having the right information makes it easier to be lucky”.
Recommendations
To improve the ‘application readiness’ of underrepresented applicants we therefore recommend:
- Employers clearly communicate planned recruitment windows and emphasise to university careers teams and students that you may close early if vacancies are filled or if you’re managing the pipeline due to a high volume of applications.
- University careers teams reach students early (in first or early second year) and where possible embed awareness of recruitment demands into the curriculum as early as possible. For undergraduates from underrepresented groups to be successful they need to know the ‘rules of the game’ they’re playing.
- Universities make clear to disadvantaged students that applying early to graduate and internship schemes is likely to improve their chances of success.
- Employers and university careers teams collaborate to build the confidence of underrepresented groups to apply earlier, factoring in the competing demands these students may be juggling. This includes providing guidance on testing requirements and practice tests. This has the double benefit of giving students who might not have a support network familiar with recruitment processes the chance to overcome a lack of confidence with testing through repeated practice and increasing the chance of their making early applications to graduate schemes.
- Employers consider opening spring week schemes to undergraduate students from all years and internships to finalists (as well as penultimate year students) to enable talented students from underrepresented groups a better chance of participating in these schemes.
- Employers expand outreach efforts to attract state-educated applicants to internships, given their high conversion to graduate roles.
Why are working class applicants less likely to be hired? Spotlight on the accounting sector
By Blog editor, on 7 March 2025
By Dr Claire Tyler, Professor Lindsey Macmillan, and Dr Catherine Dilnot
Yesterday we released our Nuffield-funded ‘Inequalities in Access to Professional Occupations’ report, and an accompanying blog post showing that working class students are well represented in application pools for professional graduate roles but are 32% less likely to receive job offers than applicants from professional backgrounds.
In today’s blog post, we focus in on analysis from the accounting and professional services sector to ask why we see these differences in offer rates for working class applicants. This sector is broadly representative of the full sample of employers we work with but were able to share more detailed data on applicants. Among applicants to the accountancy sector, 7.0% of applicants from professional/managerial backgrounds receive job offers compared to 4.8% of applicants from working class backgrounds (31% less likely). But which specific characteristics explain the socio-economic background (SEB) gap?
Figure 1 decomposes the SEB gap into three elements – unfavourable, favourable and unexplained parts. Unfavourable characteristics (in orange) reduce the chances of working class applicants obtaining a job offer, whereas favourable ones (in green) increase these chances.
Figure 1: Decomposition of the working class disadvantage in graduate offers in the accounting sector
Education does not equalise opportunities
Starting first with the unfavourable characteristics, we know from existing research that working class applicants are likely to have lower prior attainment and attend lower ranked universities than their more privileged peers.
Working class applicants in our sample achieved one grade lower at A-level than their peers from professional backgrounds (124 v 132 UCAS tariff points). Each grade lower reduces the chances of being an offered a job by around 10% (0.6ppt penalty, mean offer rate 6.3%) even when applicants are comparable on a range of other characteristics. This accounts for one third (34%) of the difference in offer rates by socio-economic background.
Working class applicants are also more likely to attend lower ranked universities than their peers from professional backgrounds. This accounts for a quarter (27%) of the gap in offers. Degree class also makes a small contribution (2.8%) due to small differences in university attainment which are magnified by the importance of educational attainment for these careers. This indicates that the university attended accounts for a significant part of the SEB gap over and above attainment prior to university.
But yet even if a working class applicant gets the same prior attainment, attends a similar university courses, and performs similarly well as someone from a professional background, the working class applicant is still less likely to get an offer. Education therefore is not equalising opportunities for young people to enter professional careers.
Applications matter
‘Application readiness’ is an important driving factor of the SEB gap as working class applicants apply later on average to graduate schemes than their peers from professional backgrounds. This accounts for 11% of the difference in offer rates. By the start of an undergraduate’s final year of study (early October), the accountancy firms in our sample have already received 30% of applications and made 50% of job offers for graduate schemes starting the following September.
But in contrast, applicants from working class backgrounds made favourable application decisions relating to their choice of employer and line of service. These applicants were more likely to apply to the ‘least competitive’ accountancy firm and service line (although all are very competitive) which improved their chances of obtaining a job offer. For example, applying to audit rather than consulting roles can double an applicant’s chance of obtaining a job offer.
Double disadvantage
Working class applicants are also more likely to be of Asian or Black ethnicity than their peers from professional backgrounds. Barriers for these ethnic minorities, which are not explained by the other characteristics in our data, account for 12% and 3% of the SEB gap respectively. This highlights the importance of intersectional analysis to identify further barriers in the recruitment process for ethnic minorities from working class backgrounds.
Together these elements explain almost two-thirds (62%) of the SEB gap in entry-level access to the accountancy profession.
What’s left?
Over one third (38%) of the SEB gap in entry level access to the accountancy profession remains unexplained by the detailed data, including a range of factors which are not significant drivers of the SEB gap here and therefore do not account for this unexplained gap. These include subject choice, use of networks, postgraduate qualifications, UK region of origin and UK region of office applied to. So this SEB gap remains even when taking all these factors into account. Other potential unexplained barriers may include performance on online tests (only those uncorrelated with educational attainment as those correlated will already be captured by our analysis), quality and duration of work experience, quality of networks used, commercial awareness, cultural capital and private schooling.
The firms in our sample have proactive social mobility strategies, are strong performers in the Social Mobility Employer Index, and are open and generous with their data for research purposes, yet still barriers for working class applicants remain. We suggest these barriers may be even larger for employers who are at the beginning of their social mobility journey.
Recommendations
Employers from all industries should review the extent to which they are rewarding potential in their recruitment processes and whether prior educational attainment criteria (or tests correlated with these) may be a barrier for high potential working class applicants entering their organisation.
Policy makers should improve the chances of high potential young people from all backgrounds being able to achieve the highest levels of academic success which are required for entry to competitive professional careers.
Greater information, advice and guidance should be provided to working class applicants relating to timelines and requirements of graduate scheme applications to improve their ‘application readiness’. This should include improving readiness for internship applications due to their high conversion rates to graduate roles. Guidance about which sectors, employers and roles are most or least competitive may also be useful for working class students to partly offset their educational disadvantage.
Why do we see so few working class and ethnic minority young people in top jobs?
By Blog editor, on 6 March 2025
By Professor Lindsey Macmillan, Dr Claire Tyler, and Dr Catherine Dilnot
We know that people from working class backgrounds and ethnic minority groups have worse labour market outcomes than their more affluent and white peers. We also know that they are less likely to work in professional occupations early in their careers. These are the types of roles that offer the first rung of the ladder to careers with higher levels of job security, and better longer term earnings prospects, so this ultimately limits their social mobility. Yet to date we have not been able to distinguish between whether a) working class and ethnic minority groups are not applying to these roles, or whether b) they are applying but are not being hired. Our new Nuffield-funded report provides the first large-scale multi-sector evidence that these disadvantaged groups are applying to these roles but are not being offered positions at the same rate as their more advantaged peers.
Recruitment processes rather than outreach?
Using unique individual-level information on over 250,000 applicants to 17 employers’ entry-level roles across accountancy, law, and the public sector, we can show that most of the socio-economic background (SEB) and ethnic inequalities that we observe in entry to professional roles appear to be driven by employer decisions made during the recruitment process rather than a lack of aspiration or applications from underrepresented groups. .
While there has been a lot of emphasis on outreach from employers to diversify workforces, we find that the applicant pools to graduate programmes are representative of the available pool of talent from linked-administrative data records. While rates of withdrawal from programmes are equally spread across applicants from different backgrounds, final job offer rates are heavily skewed in favour of more advantaged applicants (with parents from professional backgrounds) and white applicants. Working class applicants are 32% less likely to get a job offer than their more advantaged counterparts. Black and Asian applicants are 45% and 29% less likely to get a job offer than their white counterparts respectively.
Figure 1: Proportion of graduates at various stages of the recruitment process, by parental NS-SEC
Figure 1 illustrates this point for working class compared to professional background applicants. The first column shows the proportion of all graduates that come from each social class, based on linked-administrative data (National Pupil Database linked to Higher Education Statistics Agency data) breakdowns. The second column shows our total applicant pool to graduate-level entry routes across all employers. As we can see, working class applicants (green part of the bar) are well represented in the pool of applicants. But as we move to the third and fourth bar, the proportion of working class applicants drops while the proportion of professional background applicants increases through the recruitment process.
These translate to offer gaps of over 30% for working class applicants relative to professional background applicants when we model offer rates for those who have not chosen to withdraw from the recruitment process – focusing then on only decisions made by employers rather than individual applicants. Of course, it could be the case that working class applicants have different prior attainment or live in different places to more advantaged applicants, and this could be leading to them being less likely to receive job offers to these programmes. Our analysis allows us to take this into account by looking at offer rates between applicants from different backgrounds with otherwise similar characteristics, comparing like-for-like. When we do this, as shown in Figure 2, we still find that working class applicants are 18% less likely to get a job offer than applicants from professional backgrounds, even comparing applicants who attended similar universities, studied similar subjects, and applied to similar roles.
Figure 2: Offer rates to graduate programmes, conditional on observable differences across applicants, by parental NS-SEC
Note: Raw models contain employer fixed effects only; + Demographics adds controls for region of origin, and visa status; + University adds controls for university category and subject studies; + Application adds controls for networks, region of office, and job role applied to. Points represent % estimates while lines represent 95% confidence intervals.
For ethnic minority applicants, we see a similar story – ethnic minority applicants are actually overrepresented in the applicant pools for these graduate training programmes, but are less likely to receive job offers, relative to white applicants. This is true even when we take into account differences in prior attainment and other factors that might lead to differential offer rates. Figure 3 shows that Black applicants are still 33% less likely to receive a job offer than a white applicant, even when comparing applicants that look the same on paper. Asian applicants are 25% less likely to receive a job offer than their similar white counterparts.
When considering the intersection between socio-economic background and ethnicity, we observe a double disadvantage for working class ethnic minority groups, who are 37% less likely to get an offer than otherwise similar white applicants from professional backgrounds.
Figure 3: Offer rates to graduate programmes, conditional on observable differences across applicants, by ethnicity
Note: Raw models contain employer fixed effects only; + Demographics adds controls for region of origin, and visa status; + University adds controls for university category and subject studies; + Application adds controls for networks, region of office, and job role applied to. Points represent % estimates while lines represent 95% confidence intervals.
Recommendations
The analysis and discussions with employers and universities has led us to make a number of recommendations. For employers, there’s a clear need for more tracking of diversity data and reviewing of recruitment processes. Employers who performed particularly well at maintaining diverse applicant pools proactively monitored their recruitment processes and their applicant pool. The report goes into more detail about particular stages of the recruitment process that are problematic and close monitoring of both online testing and design of face-to-face assessments could lead to wider recruitment of potential talent. Employers should also look to broaden the group of universities they work with and ensure there is relevant assistance for careers teams to support their students in applying to roles. Universities can also play a key role in ensuring that information is readily passed on to those who need it most, including the benefits of applying early to graduate schemes, and providing more assistance with tests and assessments.
Why do teachers in some schools keep getting better for longer?
By Blog Editor, on 27 February 2025
Research shows that teachers rapidly become more effective over the first few years of their career. After that, the rate of growth tends to slow down. However, in certain schools, teachers keep getting better for longer. What is different about these schools? And can it be emulated by other schools?
Jackson & Bruegmann and Sun et al. use data on teachers moving between schools to show that teachers improve more when they work with more effective peers. Papay et al. and Ronfeldt et al. use random assignment studies to show that this also applies to mentors and mentees.
Ronfeldt et al. present survey data showing that teachers who report better collaborative working arrangements are more effective at raising test scores. Sun et al. get more precise by using detailed information on teachers’ social networks. They show that when a subset of teachers in a school attend training their colleagues who did not attend the training also become slightly more effective. Moreover, the extent to which the non-trainees improved was proportional to the extent that they interact with trainees.
Taken together, this evidence suggests an answer to our first question: some schools help teachers to keep getting better by providing the conditions in which teachers can continually learn from each other.
By contrast, this research provides less insight on the second question: how might other schools replicate that? Researchers might advise school leaders to create regular opportunities for productive collaboration. But school leaders could be forgiven for asking: how do we ensure that the interactions are productive, rather than a contrived imposition on teachers’ precious time?
Social capital theory provides a helpful way of thinking about this question. Putnam defines social capital as “features of social organization such as networks, norms, and social trust that facilitate coordination and cooperation for mutual benefit.” For our purposes, the “mutual benefit” would refer to teachers learning from one another. In line with this definition, social capital is typically thought of as having three dimensions: structural, cognitive and relational.
The structural dimension of social capital refers to the networks of interactions between people. This is important because it provides teachers with access to other teachers from whom they might learn. Looking back at the empirical research above, it is noteworthy that all of it focuses on structural aspects of collaboration. For example, Ronfeldt measures the collaborative activities in which teachers engage and Sun maps the interactions between individual teachers and their colleagues.
What social capital theory adds to this picture is the cognitive and relational dimensions. Cognitive social capital relates to shared language, interpretations and understandings. This is important because it allows teachers to productively engage with other teachers. Without it, conversations about the complicated work of teaching might not generate much insight. The relational dimension of social capital captures shared norms and values. This facilitates reciprocal exchanges of ideas. Without it, the conversations may be short-lived or mis-aligned.
For a powerful illustration of the cognitive and relational aspects of social capital, see the recent Steplab documentary, which unpacks teaching in two high performing schools in England: Reach Feltham and Ark Soane. The cognitive social capital can be seen clearly as different teachers fluently use shared language to refer to and analyse teaching practice in their schools. The relational social capital can be seen in the way that teachers coordinate their work, reinforcing shared norms around, e.g., expectations for pupil behaviour.
Social capital theory provides a better account of what is required for productive collaboration by specifying that relationships have to be underpinned by common understanding and trusting relationships. There is also a growing empirical literature suggesting that workplace social capital is correlated with improved organisational performance. However, social capital theory has yet to be subjected to a careful empirical test in schools.
As part of our Nuffield funded TIDE project, we are now developing the first validated measure of teachers’ workplace human capital. This summer, we will be working with teachers in the founding MATs of the National Institute of Teaching to measure social capital in their schools. We will then be linking this data to measures of teacher test score value added, as captured in the new Teacher Education Dataset.
This unique dataset will allow us to reveal the links between the three dimensions of social capital and the rate at which teachers improve across different schools and the departments within them. Our hope is that – by carefully measuring teachers’ workplace social capital – this research will provide a better answer to school leaders’ questions about how to create collaborative working relationships that help teachers to keep getting better.
Bouncing off the runway – recovery continues (for most) but stark inequalities remain
By Blog editor, on 15 August 2024
By Jake Anders, Lindsey Macmillan, Gill Wyness
Today’s level 3 results show a continuing post-pandemic ‘bounce back’, with A level grades largely improved since 2019 (i.e., before the pandemic) and since 2023 (which was the first year of ‘normal’ grading standards since the pandemic). The group of young people receiving their level 3 qualifications today were about half-way through year 9 when schools closed in March 2020 due to the Covid-19 pandemic, and were in the first year of their GCSE studies (year 10) in January 2021 during the second round of all-school closures. While they took their GCSEs under normal exam conditions, their results were adjusted according to the ‘glide path’ in 2022, with results adjusted to reflect the generous teacher assessed grades of 2021 and centre assessed grades of 2020.
A level performance by school type and region
First, looking by centre type, focusing on academies, independent schools, grammar schools and FE colleges (which the majority of pupils attend when sitting their level 3 qualifications), Figure 1 shows that, with the exception of further education (FE) colleges, the proportion of pupils achieving an A or higher at A level has increased across the board since last year, and since 2019. However, independent school pupils continue to be far more likely to achieve an A grade or higher than their state school pupil counterparts, and the gap between independent schools and academies has widened by 0.9 points between 2023 and 2024. Now, almost 50% of independent school pupils achieve at least an A at A-level, versus just 26.5% of those in academies – there is much work to do.
Also of concern are the results for those in FE – comprising typically more disadvantaged pupils. Despite their results improving somewhat since last year, they remain below pre-pandemic levels, with only 14.8% of A level entrants achieving at least an A this year. This is consistent with the evidence that those from disadvantaged backgrounds faced more disruption to their education during the pandemic.
Figure 1: A level results by school type
In previous years we’ve seen large disparities in A level grades across the country. Figure 2 shows results by region over time, again depicting substantial bounce-backs since last year, although clearly some regions have recovered faster than others. Pupils in London and the South East continue to be the highest achievers, with around 31% of pupils achieving top A level grades, but London in particular appears to be pulling away, with the biggest improvement among the regions since 2019. Meanwhile pupils in the East Midlands, the South West, and the East of England have seen their results largely flatline. Last year’s lowest performing region along this metric, the North East, has seen a large kick up in proportions achieving an A or above this year, which moves the region above the East Midlands.
Figure 2: A level results by region
T level expansion
Beyond A levels, it is somewhat more difficult to compare the results of other types of Level 3 qualifications over time. Figure 3 depicts the story for T levels, the relatively new technical-level qualification, which are broadly equivalent to 3 A levels, but provide a more “vocational” route for students. Having been developed in conjunction with employers. T-levels were launched in 2020, and 2023 was the first year where these qualifications were based fully on formal assessments. However, as they are relatively new qualifications, there are changes in the composition of students taking T Levels, and the types of T levels students are taking, meaning comparisons over time must be treated with caution.
As Figure 3 shows, the proportion of students achieving a ‘merit’ or above in their T-level has fallen by 10 percentage points since last year, with 62% of T-level students now achieving this grade. There are also marked inequalities by gender, with females being far more likely to achieve a merit than males.
Figure 3: T level results by gender
Note that this may reflect the different subjects taken by males and females. As Figure 4 shows, T levels are extremely gendered, with many of the subjects on offer being studied either mostly by males (e.g., Building Services Engineering, Onsite Construction and Digital Production, Design and Development – where over 90% of students are male), or mostly by females (e.g., Health and Education and Early Years – where over 90% of students are female). Indeed, Figure 4 highlights that there are currently many more options available currently for male-dominated subjects compared to A levels (Figure 5) which offer a more balanced mix of male- and female-dominated subjects.
Figure 4: % of males and females doing T levels, by subject
Figure 5: % males and females doing A levels, by subject
In all then, today’s results will be good news for many pupils, as results seem to have recovered since the pandemic related disruption. But they continue to paint a stark picture of educational inequality by school type, region, and gender. There is still a great deal of work to be done to reduce these inequalities in attainment and in qualifications offered.
The implications of Labour’s plans for VAT on private school fees
By Blog Editor, on 17 July 2024
A lot of attention has been paid to Labour’s commitment to removing the VAT exemption currently applied to private school fees. The stated aim is to raise just over £1.5billion to recruit new teachers in the state education sector. However, concerns have been raised that this will price families out of private schooling and lead many to move their children to state schools. There are reasons to think, however, that such concerns are overblown.
Those who attend private schools are already highly concentrated right at the top of the income distribution, as shown in Figure 1. Only 5% of those with incomes that are above 80% of the population send their children to private schools, rising to 10% of those only once inside the top 10% of the income distribution, only then rapidly increasing above this point. So there just aren’t many of those outside the top of the income distribution attending private schools.
Figure 1. Probability of attending a private school across the household income distribution
Source: Henseke, Anders, Green & Henderson (2021).
And, perhaps because those attending private schools are almost exclusively from such affluent families, there is little evidence that even rather significant rises in fees cause parents to withdraw their children from these schools. Between 1980 and 2016 average fees trebled in real terms (i.e., relative to other prices). Figure 2 shows this in more detail — calculating average private school fees as a share of household earnings at different points of the income distribution as each of these have changed over time — demonstrating they have more than doubled in terms of the share of earnings across most parts of the distribution. Yet the share of English-domiciled pupils attending private schools has remained essentially flat throughout the period.
Figure 2. Affordability of private schools for households at the median, 80th percentile and 95th percentile of household income
Source: Green, Anders, Henderson & Henseke (2017).
This is because evidence shows that demand for private schooling doesn’t respond much to price (other factors also seem to be important among those who can afford it). Estimates of what’s known by economists as the price elasticity of demand for private schooling, such as those summarised by the IFS, imply an upper-end estimate from the imposition of VAT of a 7% fall in private school attendance. That implies less than a one percentage point drop in the share of the overall school population that is privately schooled.
And that’s assuming private schools just pass on the rise in fees in full without trying to cut costs. As with any organisation, if they are worried parents really will walk away, it seems more likely that they’ll try to minimise what they have to pass on — especially after so many years of being able to increase their prices above inflation. It is fair to say, however, that the private education sector is diverse and schools’ ability to do this will vary.
In any case, analysis by the Financial Times finds that across most local authorities there is sufficient capacity in the state sector to absorb all pupils currently attending private schools. That allows for a pretty dramatic scenario in terms of parents moving their children out of the private sector.
We should also set potential shifts from private to state sectors in a wider demographic context. Pupil numbers are falling (in primary at first, then secondary from 2028) by much more than the size of the private sector. This will also provide additional spaces over time if parents do decide not to send their children to private school. It seems considerably more likely that the challenges that our education system is going to face — in terms of the frictions for school finances and class sizes demographic shifts can cause — will be from declining, rather than rising, state school pupils in the coming years.
Frozen out – How English Higher Education has become a fragile sector
By Blog Editor, on 2 July 2024
By Richard Murphy and Gill Wyness
Of all the priorities the incoming government may have on 5th July, England’s higher education sector is likely to be low on the list. But this may be short sighted – our once world leading HE system has become increasingly fragile over the last decade, with both students and universities suffering from real financial hardship. This is in stark contrast to 2012 where the controversial tripling of tuition fees (to £9,000 per year) put universities in their strongest financial position in decades, and the income-contingent loan system meant students from any income background could attend university. So what has happened?
Myself and colleagues evaluated the series of reforms which took us from zero tuition fees to the highest average tuition fees in the world by 2012. (“The end of free college in England”). We evaluated these reforms in terms of: enrolments, equity of access and funding per student. We showed that enrolments had held up in the face of the significant fee increase, and the large participation gap that existed under free tuition marginally decreased, and the funding per head reached a 20 year high. The success of the system (by these measures at least) came down to i) the higher tuition fees injecting more cash into the system and allowing universities to expand, and ii) the well-designed income contingent loans system, which ensured that no student had to pay upfront fees, and had enough money to live on. The reformed system of 2012 protected against key market failures – credit constraints, risk and uncertainty and debt aversion – an economists’ dream.
So where did it all go wrong?
In short, with the governments’ decision not to index-link tuition fees and maintenance loans (along with the removal of maintenance grants). The tuition fee cap has only been allowed to increase once since 2012, by £250 . Tuition fees are now worth around 20% less than they were in 2012 (Ogden and Waltman, 2023). This is disastrous for universities, which rely on fees as a key component of their income. Things have been tough for students too – maintenance loans have increased over time, but have not kept up with the UK’s high rate of inflation over the last few years.
In the light of these issues, my co-author Richard Murphy and I decided to revisit today’s system, in terms of our three key measures of success – enrolments, equity of access and university funding.
Enrolments have continued to grow
For our first measure of success is easy to measure – have student numbers continued to grow since 2012? Here, there is good news. As figure 1 shows, enrolments have continued to increase at a steady pace since the 1990s, with a COVID-19 bump in 2020, due to more students receiving their required grades.
Figure 1: Student enrolments: number of full-time equivalent undergraduate students at UK HEIs
Notes: series break in 2000 due to change in sources. Source: 1994-2002, figures compiled by V. Carpentier. 2001-2021, Higher Education Information Database for Institutions (HEIDI)
Equity under threat as the value of maintenance loans decline
But have these increases in enrolment come from all sectors of society? Figure 2 shows the gap in participation between students from high and low SES backgrounds, Since 2005 there has been a slow general closing of the SES gap in enrolments between high and low SES students which has continued post 2012 (though it’s worth pointing out that the POLAR4 measure of SES used here has been criticised in the past – other measures are tricky to access but may show a different story). The exceptions are the 2011 and 2020/21 blips. The former likely caused by fewer SES taking gap years in order to enter under the lower fee scheme, and the latter the COVID-19 bump.
This general reduction in inequality is potentially surprising, given the deterioration in financial conditions experienced by lower income groups in recent years. Figure 3 shows the amount of money students from different income background have to live on while studying during each of the reforms. Since 2012, there has been little improvement in liquidity. An increase of about £1,000 per year in real terms for the poorest students, between 2012 and 2016 (as well as a gradual increase between 2016-2020), has now been eroded entirely by inflation, so that the maximum maintenance loan is now worth less, in real terms, than it did in 2016.
Figure 2: Higher education enrolment gap, POLAR 4 measure of deprivation
Source: UCAS End of cycle report, 2023
Figure 3: Net liquidity (grants+maintenance loans-upfront fees) by parental income for different fee regimes
Source: Authors’ calculations using data from Student Loans Company, 1991–2024. Figures expressed as amounts per year.
Funding per head hit hard as tuition fees frozen
Before 2012, a significant proportion of university funding came through the government teaching grant, with a smaller proportion from tuition fees. The 2012 reforms were designed in part to shift the burden of payment towards graduates. This was seen as preferable to relying on government to fund the system, which is invariably low-priority in times of austerity (Murphy et al, 2022).
Figure 4 plots university funding per full-time equivalent student, both for ‘domestic’ undergraduate students and all student types –postgraduate, undergraduate, UK, EU and overseas students (who typically pay higher fees). The funding per head for domestic undergraduates is closely tied with the tuition fee increases. Real funding per head spiked in 2012 then stabilised. But in recent years, funding per head has declined and now stands at the 2011 level.
By contrast, overall funding from all student types has not suffered the same decline. This reflects universities attempts to protect their balance sheets with income from students whose fees are unregulated – predominantly international students. Analysis from the IFS (Drayton et al, 2023) reports that in the 2021–22 financial year, tuition fees from non-UK students accounted for 42% of higher education course fees and 21% of all income for universities in England. Any reductions in the numbers of international students will translate into lower funding per head of domestic students.
Figure 4: Average funding per full-time equivalent student
Sources: Funding from all sources – statistics for 2000–2002 are taken from Carpentier (2004) and statistics for 2002–2021 taken from Higher Education Information Database for Institutions. These figures are not available for 2015,2016,2017. FTE enrolments used in the computation contain all student types (full-time, part-time, postgraduate, undergraduate, UK, EU, overseas); funding per head is for all students and comprises teaching grants and tuition fee income (the latter for all student types listed above). Funding from domestic students taken from IFS (2023)– total up-front public resources provided for teaching. This is effectively tuition fees for domestic students (minus any fee discounts) plus teaching grants. Figures expressed as funding per student per year. All figures expressed in constant 2023 pounds sterling.
An increasingly fragile sector
The governments decision to freeze domestic tuition fees – apart from the 2017 increase of £250 – as well as its failure to protect the real-term value of maintenance loans is perhaps not surprising. Tuition fees (and student loans) are perennially unpopular with voters (though less so when they are shown how much fee abolition would cost the taxpayer – as a recent Public First report showed) and no government wishes to risk the wrath of the middle-classes by putting them up during a cost of living crisis. But freezing university and students’ incomes is a policy decision as much as increasing them is, and it comes at a cost.
So far, the sector seems to have coped. Enrolments have held up, access to HE has continued to gradually improve , and university funding per head is still above 2012 levels. However, this masks an increasingly worrying picture. Funding for domestic students is in serious decline, meaning that universities are more and more reliant on international students to keep themselves afloat. Whether this is likely to be sustainable for much longer is debatable, as the impact of student visa restrictions is felt. Meanwhile, students themselves are increasingly squeezed by falls in their income from maintenance loans.
So what can be done? The obvious solution would be to immediately reverse the erosion of tuition fees and maintenance loans in real terms (and to index link fees and finance going forward). This would amount to an increase of around £2000 per year in tuition fees for home students (Ogden and Waltmann, 2023) – likely to be extremely politically unpalatable for a new government and its already cash-strapped electorate. The major party manifestos have offered little detail on how they might tackle the crisis. The Conservatives plan to keep the 2022 finance regime in place, the Liberal Democrats propose to hold a review and Labour acknowledge the issues stating ‘The current higher education funding settlement does not work for the taxpayer, universities, staff, or students,’ suggesting a review is on the cards.
But action is urgently needed to prevent universities – especially those with less access to lucrative foreign students – from going under. Given the strength of our university sector, let’s hope the new government acts before it’s too late.
Sizing up Labour’s ambition to address the teacher shortage in England
By Blog editor, on 28 June 2024
by Dr Sam Sims
The Labour party have promised to “recruit over 6,500 new teachers” in England.[1] The 6,500 number was apparently chosen to reflect the best available estimate of the shortage in England at the point when the policy was first announced in 2023. This blog post looks at whether Labour are likely to achieve the target based on their recent manifesto commitments.
In Labour’s 2023 education ‘mission’ document, they promised two policies aimed at retaining more teachers.[2] First, introduce a new £2,400 early-career framework (ECF) retention payment paid to all teachers after they complete their first two years on the job.[3] Second, reform retention incentives targeted at specific subjects and phases. In their manifesto, Labour announced that they have set aside £450m per year to spend on hitting their teacher target. This is to be paid for from the application of VAT and business rates to private schools.
There are a couple of different ways that we might interpret the 6,500 target. A less ambitious version would be to recruit or retain 6,500 additional people. This version is defined in terms of the ‘flow’ of people into and out of teaching. Of course, recruiting or retaining an extra person in a given year doesn’t mean they will stick around indefinitely. A more ambitious version of this target is therefore to increase the total number (or ‘stock’) of teachers in the workforce by 6,500 (1.4%[4]) by the end of the parliament. Let’s evaluate the likelihood of achieving each of these versions of the target.
Achieving the less ambitious target
Approximately 25,000 teachers per year will finish their early career training during this parliament.[5] This means the ECF retention bonus policy would cost around £60m per year. The £2,400 payment amounts to about 7% of a third-year teachers’ salary.[6] Research suggests that this would decrease the number of teachers leaving in the year the ECF incentive is paid by about 14-21%.[7] Since approximately 2,000 per annum leave the workforce after their second year on the job, this amounts to about 300-400 additional teachers staying on after their ECF each year.[8]
However, the research on which this is based focuses on maths and science teachers. These are the subjects in which shortages are largest, in part because those with STEM qualifications can often get paid more outside of teaching.[9] Labour’s ECF bonus will be paid to everyone, regardless of whether they are in a shortage subject or their likely earnings outside of teaching. The actual number of additional teachers retained each year is therefore likely to be a bit lower than 300-400.
Spending £60m a year on the ECF bonus leaves £390m per year for retention incentives targeted on shortage subjects or geographic areas. Existing research has modelled the costs of a £2,000 retention payment paid after each of the first two years of teachers’ careers (£4,000 in total).[10] Such a policy prevents a teacher leaving (in the short run) at a cost of around £63,000 per teacher.[11] Spending the entire remaining £390m on such a policy would therefore buy about 6,190 additional teachers per year, in the short run.
Taken together, spending £60m a year on an ECF bonus and £390m on targeted bonuses would, therefore, retain approximately 6,500 additional teachers per year. Sustained across the parliament, this would mean that Labour would easily exceed the ‘less ambitious’ version of their target. This raises the prospect of meeting the more ambitious, and more important, version of the target: having 6,500 more teachers by the end of the parliament.
Achieving the more ambitious target
Since Labour seem to have set aside plenty of money, let’s think about a radical policy to pay certain teachers an additional £2,000 retention bonus for each of their first five years in the profession. This supercharges the policy because the additional teachers in the system compound over the five-year period.
To fix ideas, let’s apply this to a cohort of maths teachers. By extending the modelling exercise in Table 6 of Sims & Benhenda (2022), it’s possible to estimate that this would lead to 423 extra maths teachers working in England by the end of the next parliament. Applying this policy to all new cohorts of maths teachers across the next parliament would come at a total cost of £28m.
This £28m price tag is a fraction of the £2bn that Labour has set aside for spending on targeted recruitment incentives for new teachers over the course of the next parliament. Evidence suggests that increasing the value of the retention payments has a broadly linear relationship with retention, at least up to £7,500.[12] Doubling the retention payments should therefore at least double the number of additional maths teachers by the end of the parliament.[13] Doing that across seven shortage subjects, which is affordable within Labour’s budget, would therefore likely be enough to hit the more ambitious version of the target.
An ambitious reform
The back-of-the-envelope estimates presented here contain many assumptions about how retention incentives will be implemented and play out. There is of course also considerable uncertainty about what will happen in the wider economy, and therefore to graduate job choices, over the next five years. However, what should be clear from the above is that Labour is committing a very serious sum of money to addressing the teacher shortage. To further put this in perspective, consider that the Department for Education budgeted a total of £450m for all spending on the ‘Teaching Workforce’ (including spending on teacher retention payments) in 2023/24.[14] Labour are planning to spend that much again, purely on addressing teacher shortages. If sustained, this investment is likely to go a long way toward solving teacher shortages by the end of the parliament.
Notes
[1] “Labour will use money raised from ending private school tax breaks to: Recruit over 6500 new teachers to fill vacancies and skills gaps across the profession” Page 10 here: https://labour.org.uk/wp-content/uploads/2023/07/Mission-breaking-down-barriers.pdf
[2][2] “Labour will restructure teacher retention payments into one payment scale incorporating different factors such as subject and geography, based on evidence showing incentive payments are an effective means of retaining teachers with knowledge and expertise. On top of this, Labour will introduce a new Early Career Framework retention payment upon completion of the updated Framework recognising the professional development staff have undertaken.” Page 11 here: https://labour.org.uk/wp-content/uploads/2023/07/Mission-breaking-down-barriers.pdf
[3] https://schoolsweek.co.uk/labours-school-policy-blitz-what-we-know-so-far-and-what-we-dont/
[4] https://www.tes.com/magazine/analysis/general/how-many-teachers-are-there-uk-england-scotland-wales-northern-ireland#:~:text=In%20England%20data%20from%20the,216%2C000%20in%20secondary%20schools.
[5] https://explore-education-statistics.service.gov.uk/find-statistics/teacher-and-leader-development-ecf-and-npqs
[6] https://www.nasuwt.org.uk/advice/pay-pensions/pay-scales/pay-scales-england.html
[7] https://www.gatsby.org.uk/uploads/education/datalab-simulating-the-effect-of-early-career-salary-supplements-on-teacher-supply-in-england.pdf
[8] About 2000 teachers per year leave after their NQT+1 year i.e. after ECF https://department-for-education.shinyapps.io/turnover-and-retention-grids/
[9] http://www.gatsby.org.uk/uploads/education/increasingscienceteachers-web.pdf
[10] Sims, S., & Benhenda, A. (2022). The effect of financial incentives on the retention of shortage-subject teachers: evidence from England. CEPEO working paper. Table 6. https://www.gatsby.org.uk/uploads/education/reports/pdf/the-effect-of-financial-incentives-on-the-retention-of-shortage-subject-teachers-evidence-from-england.pdf
[11] It costs more than £4000 per teacher because most of the teachers that receive the bonus would not have left anyway.
[12] https://assets.publishing.service.gov.uk/media/656761b65936bb000d3166ea/Evaluation_of_the_phased_maths_bursaries_pilot_-_final_report_November-2023.pdf
[13] Depending on how the compounding plays out, it should more than double it.
[14] https://committees.parliament.uk/publications/40066/documents/195550/default/
More detail needed on what’s next for education policy
By Blog Editor, on 25 June 2024
The manifestos are out and the general election is looming. We at CEPEO, as the name suggests, are particularly interested in education policy and equalising opportunities. So, what did the manifestos actually tell us about potential plans in these areas? We focus here on plans set out by the Conservative, Labour and Liberal Democrat parties.
While the details differed, there were some agreed areas of importance across the parties: the need to recruit, train and retain high quality teachers; to do more to support children with special educational needs; and to incentivise more adult learning. These issues would probably appear quite high up most lists of pressing issues likely to be facing the Department for Education over the next few years.
There was also an emphasis on support for disadvantaged students. The Conservatives highlighted £3bn of spending via the pupil premium – helpful, of course, but only so high because of the very large proportion of pupils eligible now – 25% as of January 2024. Some of this rise is driven by transitional arrangements for Universal Credit, but I’m sure we could all agree that it would be better for far fewer children to be experiencing low family income, even temporarily.
The Liberal Democrats promised to go further on this front by tripling the early years premium to bring it closer to the amount allocated to school children – and extending the pupil premium it to those aged 16-18 as well – both very welcome ambitions. There were no specific details on this issue in the Labour manifesto, but one of their five missions is to break down barriers to opportunity, so there may be more specific announcements to come if they are the ones in office come 5th July.
There were also some clear omissions though. There was very little detail on what might be done to shore up HE funding. Labour and the Liberal Democrats were clear that ‘something’ should be done, but unclear what that would look like. Certainly no-one was brave enough to say that tuition fees might need to go up substantially. We can probably expect another independent review in the coming months to spell out the unappealing choices. The Lib Dems did commit to reintroducing maintenance grants, though, while the Conservatives re-emphasised their plan to close down “poor quality” degrees, which of course are challenging to identify and may disproportionately affect those from lower socio-economic backgrounds, as we have discussed previously.
In contrast to the strong focus on the importance of high-quality staff for schools, there was also very little in the way of detail on a potential workforce strategy to help deliver the extended early education entitlements to children in working families from 9 months, which both the Conservatives and Labour have committed to delivering. The Conservatives are presumably relying on the market to deliver the places (and staff), incentivised by the higher funding rates they have committed to over the coming years. Labour are planning to re-use freed-up space in primary schools to deliver more places, but have not provided any specific details of their plans for the workforce.
While it is possible that providers will use some of the higher government funding to pay staff more, the market does not provide a strong incentive to invest in high quality staff – or in quality more generally. It is challenging for parents to identify setting quality (beyond Ofsted ratings) and many also weigh other considerations – such as availability and convenience – more highly when choosing a place for their child, suggesting that more will need to be done if we are serious about delivering high quality early education.
There is also a lot more that could be done to distribute this funding more equitably, as I spell out in this companion piece. And, of course, it would have been great to see consideration of some of our more ambitious policy priorities to equalise opportunities, including reforming school admissions, introducing a post-qualification admissions system and greater commitment to funding for further education (although credit to the Liberal Democrats for making an explicit commitment on this).
Of the three, the Liberal Democrat manifesto was the most ambitious in terms of the number and scope of specific policy ideas to equalise opportunities – but it certainly wasn’t radical. We must therefore hope that the next government is going to over-deliver on its manifesto commitments. For, as my colleagues so eloquently put it in a recent blog post, there can be no economic growth without education and skills. Ensuring that the benefits of this growth are distributed equitably starts with the education system. We here at CEPEO therefore hope that bolder action to equalise opportunities in education is just around the corner, whichever party is in power next month.