Does compulsory schooling reduce physiological wear and tear? Evidence from a regression discontinuity design

A large body of evidence documents a positive association between education and health, yet establishing causality remains challenging. This study uses the 1972 Raising of the School Leaving Age (ROSLA) reform in England and Wales as a natural experiment to estimate the causal effect of an additional year of compulsory schooling on allostatic load (AL), a composite biomarker index capturing cumulative physiological dysregulation across multiple body systems.
We employ a fuzzy regression discontinuity design exploiting the fact that individuals born on or after 1 September 1957 were required to remain in school until age 16, rather than 15. Our running variable is month-year of birth, measured in months from the September 1957 cutoff. Using data from the Understanding Society (UKHLS) nurse health assessment, we construct a five-item AL score following the consensus definition proposed by McCrory et al. (2023). This index sums binary indicators of high-risk status across C-reactive protein, resting heart rate, HDL cholesterol, waist-to-height ratio, and glycated haemoglobin (HbA1c), with high risk defined as scoring above the 75th percentile of the sample distribution for each biomarker.
Our first stage is strong: the reform reduced the probability of leaving school at age 15 by approximately 20 percentage points (p<0.001). Fuzzy RD estimates suggest that the additional year of compulsory schooling reduced AL by roughly 1.2 points on a 0–5 scale among compliers (p=0.060). While the direction and magnitude of this effect are substantively meaningful — equivalent to shifting more than one biomarker from high-risk to low-risk status — confidence intervals remain wide, reflecting the limited sample size of nurse-assessed respondents.
In forthcoming analyses, we plan to replicate these findings using biomarker data from the English Longitudinal Study of Ageing (ELSA), which offers a larger sample of older adults with comparable biomarker measurements, with the aim of increasing statistical power and assessing the robustness of results across cohorts.