Supporting analysis
Analysis and supporting evidence: wealth, education and social mobility
Methods, additional results and sources underlying the Ageing World article on wealth, education and social mobility.
This page contains the methodology, additional results and sources underlying the Ageing World article “How do you stop people at the bottom of the wealth scale staying there?”. The analysis is exploratory and is presented separately so that the main article can remain focused on the argument while the evidence can be examined in greater detail here.
What was analysed? The analysis uses Understanding Society: the UK Household Longitudinal Study, which follows individuals and households over time. The dataset used is the End User Licence version of Waves 1-15, covering 2009-2024 (University of Essex, Institute for Social and Economic Research, 2026).
The first analysis asks a relatively simple question: if somebody occupies a particular position in the observed wealth distribution at Wave 4, where are they by Wave 8? An observed-wealth measure was constructed using personal net financial wealth together with the respondent's share of the gross value of their main home. Respondents were ranked within the analytical sample and grouped into five parts of the distribution.
This is a deliberately pragmatic measure rather than a complete estimate of net worth. In particular, the housing component used for the Wave 4-to-Wave 8 comparison is a gross home-value share and does not provide a consistently comparable deduction for outstanding mortgages across both waves. Pension wealth, business assets and some household-level investments are also outside the measure. The resulting groups should therefore be read as positions within the observed wealth measure used for this analysis, rather than definitive estimates of each respondent's total net wealth.
The longer analysis follows respondents from Wave 4 to Wave 15 and focuses particularly on people starting in the bottom fifth of the Wave 4 observed-wealth distribution. Later housing assets, homeownership, degree-level qualification status and labour-market histories were then examined. Employment was represented by the share of observed intervening waves in paid or self-employment, while earnings were represented by the respondent's average position in the earnings distribution during working waves.
All models are exploratory and associational. They do not establish that education, employment or earnings caused the subsequent wealth outcome, and changes in coefficients after additional variables are included should not be interpreted as formal evidence of mediation.
The models are also unweighted because the standard Wave 15 longitudinal weight does not automatically correspond to an analysis beginning at Wave 4 while allowing intermittent non-response. The estimates should therefore be understood as relationships within the analytical samples rather than population estimates for the UK.
Finally, the analysis does not observe wealth at birth. References in the main article to people born into families with little or no wealth draw on the wider intergenerational literature rather than on the original analysis itself.
How persistent was wealth position? The Wave 4-to-Wave 8 analysis included 5,146 respondents with sufficient information for the comparison. The table shows the percentage of each starting wealth group found in each later group.
| Starting position | Bottom fifth later | Second fifth later | Middle fifth later | Fourth fifth later | Top fifth later |
|---|---|---|---|---|---|
| Bottom fifth | 49.5% | 25.9% | 12.6% | 5.5% | 6.4% |
| Second fifth | 26.9% | 52.1% | 15.4% | 2.8% | 2.8% |
| Middle fifth | 12.7% | 13.6% | 56.1% | 13.9% | 3.7% |
| Fourth fifth | 5.0% | 3.5% | 13.8% | 61.4% | 16.3% |
| Top fifth | 1.8% | 2.0% | 3.9% | 20.1% | 72.1% |
The concentration along the diagonal shows that remaining in or close to the same part of the distribution was common. Movement did occur, particularly between neighbouring groups, but movement across the full distribution was much less common.
Among the 999 respondents starting in the bottom fifth, 64 - 6.4 per cent - reached the top fifth. Among the 1,028 starting in the top fifth, 741 - 72.1 per cent - remained there.
Education, earnings and later assets The longer Wave 4-to-Wave 15 analysis was designed to separate two questions that can easily become conflated: whether people who are degree-qualified have better later asset outcomes, and whether moving into degree-level qualification status during adulthood is associated with subsequent upward movement.
Among respondents starting with low observed wealth, being degree-qualified at Wave 15 was associated with a higher later housing-asset position. After accounting for starting wealth, age and sex, degree-qualified respondents were around 10.9 centile positions higher. After intervening employment and earnings were added, the difference was around 7.3 positions. Earnings also showed a strong association with later housing-asset position, while the proportion of time observed in employment did not show a clear independent relationship.
That result concerns degree status, however, and cannot show what would happen if an otherwise similar person acquired a degree. A separate analysis therefore concentrated on people who began both in the bottom wealth fifth and without a degree.
Only 24 of the 443 people in this group moved into degree-level qualification status by Wave 15. Eight of those 24 - 33.3 per cent - were homeowners at the later observation, compared with 62 of the 419 people who remained below degree level - 14.8 per cent.
| Qualification status | Homeowner at later observation |
|---|---|
| Remained below degree level | 14.8% |
| Moved into degree-level qualification status | 33.3% |
The average later housing-asset rank was also higher among those who moved into degree-level status.
In regression analysis, the adult transition into degree-level status was associated with an estimated 8.3-place increase in later housing-asset rank before employment and earnings were added. Once those labour-market characteristics were included, the estimated difference was 6.9 places and sufficiently imprecise that the analysis could not confidently distinguish it from no independent difference. Mean earnings position remained clearly associated with the later asset outcome.
The small number of educational transitions is crucial when interpreting these results. They provide suggestive evidence rather than a basis for a strong claim about the effect of obtaining a degree in adulthood.
Earnings, employment and becoming a homeowner The distinction between having a job and having sufficient earnings to accumulate assets also appears in the homeownership analysis.
Among respondents beginning in the low-wealth, non-degree group, 443 were available for the descriptive homeownership comparison and 336 had complete information for the main regression model.
In that model, higher average position in the earnings distribution across the intervening waves was clearly associated with being a homeowner at Wave 15. By contrast, the share of waves spent in paid or self-employment showed a much less precise relationship.
The regression is deliberately not translated into a claim that a particular increase in pay causes a particular increase in the probability of homeownership. Earnings rank is an analytical measure, the model is observational, and other characteristics may influence both earnings and asset acquisition.
The more defensible conclusion is simpler. Within this sample, sustained position higher in the earnings distribution was more consistently associated with later asset accumulation than simply spending a greater proportion of observed time in employment.
That helps explain the emphasis on capability, credibility and connection in the main article. If education is being considered as an instrument of economic mobility, labour-market entry alone may be too limited an outcome. Progression and earnings over the subsequent career are also relevant.
Why capability, credibility and connection? Capability, credibility and connection are an interpretive framework rather than three variables directly tested in the Understanding Society analysis.
Capability describes the knowledge, skills and judgement that enable somebody to perform valuable work. The labour-market returns associated with higher education vary substantially, including by course, institution and individual background. Belfield and van der Erve (2018), discussing linked education and earnings evidence, show considerable differences between graduate outcomes and highlight the continuing association between family background and earnings even among graduates following similar educational paths.
Credibility describes the evidence through which capability becomes legible to an employer: qualifications, prior work, portfolios, references, practical experience and demonstrated performance. It helps distinguish education as the acquisition of a credential from education that also gives learners persuasive evidence of what they can do.
Connection describes access to the people and organisations through which opportunities are discovered and careers develop. The Social Mobility Commission's employer guidance explicitly recommends formal network-building and mentor matching to reproduce some of the access that employees from more privileged backgrounds may obtain through informal personal networks. It similarly cautions that informal sponsorship, prestigious assignments and networks can create unequal progression (Social Mobility Commission, 2022).
The framework does not imply that every learner needs the same intervention. Its purpose is to broaden the question asked of education. A qualification may be valuable, but social mobility depends partly on what somebody can subsequently do, how convincingly that capability can be demonstrated, and whether there are realistic opportunities for it to be recognised.
Family wealth and the barrier to asset ownership The original analysis begins with adult wealth at Wave 4, so it cannot establish how much wealth respondents' families possessed when they were born. Wider evidence is needed to examine that part of the argument.
Davenport, Levell and Sturrock (2021) analysed linked measures of parental and adult children's wealth in the UK. Among people in their thirties, average non-pension wealth was £107,000 for those whose parents were in the wealthiest fifth, compared with £18,000 among those whose parents were in the poorest fifth.
Their analysis found that education and earnings accounted for a substantial proportion of the relationship between parental and children's wealth, while differences also remained through saving, housing and other channels. At age 30, people with parents in the wealthiest third were 17 percentage points more likely to be homeowners than those with parents in the poorest third at a given level of the adult child's earnings.
This provides an empirical basis for distinguishing the ability to earn from the ability to acquire an asset. Two people with similar earnings may approach the housing market with very different levels of financial support, accumulated savings and protection against economic shocks.
Levell and Sturrock (2026) extend this argument by examining the UK house-price boom. They estimate that by ages 28-37 around 15 per cent of differences in parental gross housing wealth were transmitted into differences in children's gross housing wealth.
Their evidence links part of the process to the ability of wealthier parents to help adult children overcome liquidity constraints, including those associated with moving to expensive parts of the country. Their counterfactual analysis suggests that the house-price boom doubled the intergenerational persistence of housing wealth.
This is the basis for the policy discussion in the main article. Education and labour-market progression can improve earnings prospects, while government policy can alter some of the barriers between earnings and asset accumulation. The appropriate policy response is a separate question, but any account of wealth mobility that ends at employment misses this additional stage of the life course.
References
Belfield, C. and van der Erve, L. (2018) ‘What determines graduates’ earnings?’, The Times, 11 June, reproduced by the Institute for Fiscal Studies.
Davenport, A., Levell, P. and Sturrock, D. (2021) Why do wealthy parents have wealthy children? London: Institute for Fiscal Studies. DOI: 10.1920/re.ifs.2021.0196.
Levell, P. and Sturrock, D. (2026) How do house prices affect social mobility? IFS Working Paper 26/02. London: Institute for Fiscal Studies. DOI: 10.1920/wp.ifs.2026.0226.
Social Mobility Commission (2022) The building blocks: an employer’s guide to improving social mobility in the workplace. London: Social Mobility Commission.
University of Essex, Institute for Social and Economic Research. (2026) Understanding Society: Waves 1-15, 2009-2024 and Harmonised BHPS: Waves 1-18, 1991-2009. [data collection]. 21st Edition. UK Data Service. SN: 6614. DOI: 10.5255/UKDA-SN-6614-22.