Education, social mobility, and earnings outcomes for students receiving free school meals in England
Author: Aidan Tolland, James Tierney, and Holly Bathgate (Office for National Statistics)
Date: January and August 2022
A research study using secure data has found that 23% of free school meals (FSM) recipients recorded earnings above the Living Wage compared to 43.5% of non-recipients. Further research determined that at age 30, FSM recipients earned around 20% less than independent school attendees with similar characteristics and attainment. This research demonstrates the potential of using linked administrative data to understand labour market outcomes and provide analysis of complex policy questions, such as social mobility.
This project used the Department for Education’s (DfE) Longitudinal Education Outcomes (LEO) dataset. This dataset enabled the researchers to explore whether people who grew up in low-income households experienced earnings mobility in young adulthood. The initial findings explore demographic and geographic factors which provide valuable insight into the patterns of social mobility across England.
This project used the DfE’s LEO standard extract which was accessed through the ONS Secure Research Service.
The DfE has collaborated with other areas of government to create the LEO de-identified linked administrative dataset. LEO is a major database that as of 2021 comprised 38 million individuals. LEO brings together data from:
- the National Pupil Database
- Individualised Learner Records
- Higher Education Statistics Agency (HESA)
- Department for Work and Pensions (DWP), which includes income, employment, and benefit data.
This project focused on early adulthood earnings for pupils who had been in education in England at Key Stage 4. In total, over 5 million individuals who turned 25 during tax years 2011/12 to 2018/19 were included in the living wage analysis, and around 1 million who turned 30 during tax years 2016/17 to 2018/19 were included in the free school meals earning gap analysis.
The researchers used free school meals eligibility at age 15 as a proxy measure for individual level socio-economic disadvantage during childhood. This is a common approach to identifying whether an individual grew up in a household with a low income, as details of family income and background are not yet available in the LEO dataset.
For the living wage analysis, the research team used earnings data from employment Pay As You Earn (PAYE) records and self-employment self-assessment records to calculate annual ‘combined earnings’ at age 25. These were then compared with the Living Wage for the tax year they turned 25, an age by which most have completed formal full-time education and entered the labour market.
Individuals were then categorised based on matching criteria, employment status and combined annual earnings:
- "Below Living Wage" - matched records that have recorded earnings at age 25 years below the annual equivalent of the Living Wage
- "Above Living Wage" - matched records that have recorded earnings at age 25 years above the annual equivalent of the Living Wage
- "No earnings record" - matched records that do not have earnings records at age 25 years but do appear elsewhere in the outcomes data
- "Unmatched" - those whose education records were not matched to outcomes records at any age.
The free school meals earnings gap analysis looked at PAYE earnings only. It estimated the raw earnings gap at age 30 between FSM recipients, non-recipients at state schools, and independent school students. The researchers then used ordinary least squares (OLS) regression (a statistical method that estimates the relationship between one or more independent variables and a dependent variable) to estimate an adjusted earnings gap, taking into account attainment, experience, region, ethnicity, and gender.
Finally, Blinder-Oaxaca decomposition (a type of methodology used to study labour market outcomes by groups) was used to divide the difference in average pay between FSM recipients and non-recipients into the part which is explained by the differences in observed characteristics and the part which is not.
This project found that at age 25, 23% of FSM recipients had recorded earnings above the Living Wage and 42% had earnings below the Living Wage. Almost a third (29.2%) of FSM recipients were in the no recorded earnings group, with the remainder unmatched. In comparison, 43.5% of non-recipients had earnings above the Living Wage. There was a 13.8 percentage point difference between the proportion of FSM recipients who had no recorded earnings at age 25, compared with non-recipients. However, some of the individuals could have been in full time education, others in receipt of benefits.
Earnings status at age 25 years by free school meals status, tax year ending 2012 to tax year ending 2019
Free school meals earnings gap analysis showed that at all levels of qualification, those eligible for free school meals were earning less at age 30 years than their peers who had the same highest level of qualification.
The earnings gap between FSM students and independently educated students was around 20% after adjustment. Therefore, if a person who was eligible for free school meals as a child had the same level of education, KS4 attainment, years of labour market experience, ethnicity, and went to secondary school in the same region as a person who attended an independent school, the model predicts that on average they would still earn around 20% less than that of an independent school student.
Insights provided by this research will supply evidence for policy decisions to address societal inequalities within the low-income population. Outputs from this research have been widely disseminated in the Department for Education. The researchers have also spoken to organisations such as Teach First who are using the findings to inform their policy recommendations.
This research project demonstrates the value of data linkage and collaboration which enables analysis of complex policy questions. The LEO dataset has enabled analysis at a level of detail that would not have been possible using previously available data. The LEO dataset comes from population administrative records meaning it can provide greater population coverage, greater accuracy, and longer reporting periods than other approaches to understanding labour market outcomes.
The variables within the LEO dataset will enable the researchers to investigate additional educational and demographic factors associated with earnings. This will lead to a series of publications covering:
- highest level of qualification achieved
- qualification type
- route through education
- educational attainment at each stage of qualification
- type(s) of educational establishment attended
- subject(s) of study
- special educational needs and disability
- English as a first language
- outcomes at age 30.
Publications and reports
- Office for National Statistics Initial findings, January 2022: Education, social mobility and outcomes for students receiving free school meals in England: initial findings on earnings outcomes by demographic and regional factors
- Office for National Statistics Article, August 2022: Why free school meal recipients earn less than their peers
Blogs, news posts, and videos
- Youth Employment UK News article, January 2022: Education, Social Mobility And Outcomes For Students Receiving Free School Meals In England
- Teaching Times News article, February 2022: Education, Social Mobility and Outcomes for Students Receiving Free School Meals in England
- National Education Union Press Release, January 2022: ONS social mobility report
- The Guardian News article, August 2022: Children who get free school meals in England earn less as adults, study finds
- FE News article, August 2022: Why free school meal recipients earn less than their peers
Presentations and awards
- ADR UK and ONS Education Data Symposium, June 2022
About the ONS Secure Research Service
The ONS Secure Reseach Service (SRS) is an accredited trusted research environment, using the Five Safes Framework to provide secure access to de-identified, unpublished data. If you would like to discuss writing a future case study with us, please get in touch: IDS.firstname.lastname@example.org.