DOES SCHOOLING HAVE LASTING EFFECTS ON COGNITIVE FUNCTION? EVIDENCE FROM COMPULSORY SCHOOLING LAWS

Cognitive function exerts an important influence over many domains of social and economic life, and a positive association between schooling and levels of cognitive function in later life has been well documented. While these partial correlations raise the possibility that schooling could modify cognitive outcomes in older ages, they do not necessarily imply a causal link. Since educational attainment and later life outcomes are jointly determined by potentially unobserved factors - genetic endowments, childhood environment and acquired skills - estimates of association are likely to overstate the average causal effect. This study exploits changes to compulsory schooling laws to provide new estimates of the causal effect of schooling on cognitive function in later life, and an assessment of the potential mechanisms underlying this effect.

I employ exogenous variation in schooling induced by the 1972 raising of the secondary school leaving age (RoSLA) in England and Wales. This reform increased the minimum school leaving age from 15 to 16 years, and increased the probability of staying on until 16 by approximately 25 percentage points. To exploit this variation I use a Regression Discontinuity design, which compares outcomes among treated and control cohorts born in a very small window around the treatment cut-off. The data were drawn from the third wave of Understanding Society, the largest nationally representative panel dataset in the United Kingdom. Measures of cognitive function examined include episodic memory, numeric ability and verbal fluency, among 4,137 respondents aged between 49 and 59 years. These respondents are younger than considered in previous studies - while old enough to experience the onset of cognitive decline, concerns of selective mortality are reduced. A key finding is a local average treatment effect of schooling on episodic memory of 0.4 standard deviations. This is greater than the estimated partial correlation; given the RoSLA affected those at the lower end of the schooling distribution, this discrepancy may reflect diminishing returns to schooling. In line with previous studies, little evidence was detected for any effect of schooling on numerical ability or verbal fluency. Mechanisms explaining the large effects on memory were also explored, including the potential mediating effects of occupational history, social engagement and health behaviours.

This is the first study to exploit the 1972 RoSLA as natural experiment in the schooling-cognition nexus. The empirical approach is internally valid, providing reliable estimates of the cognition returns to extra secondary schooling for the affected group.  One caveat is that due to the local nature of the treatment effect, the results may lack external validity. On the other hand, the effects of this reform are of interest in their own right, as the increases in schooling induced by this RoSLA are now yielding pay-offs in improved cognitive outcomes among those currently in their middle-age. Consequently we may also expect this cohort to experience lower rates of memory disorders in older age, highlighting the potential role for schooling in reducing the burden of cognition-related disease.