Tomeu López-Nieto Veitch (University of Bologna)
Pietro Biroli, Elisabetta De Cao and Rodrigo Pinto
This paper studies how the causal returns to college vary with genetic predisposition to educational attainment. We extend the Marginal Treatment Effect (MTE) framework to settings in which an observed categorical moderator shifts the distribution of latent resistance to treatment. Our approach introduces a monotone warp function that maps group-specific resistance ranks onto a common pooled scale, allowing valid comparisons of marginal returns across groups with different selection patterns. We also show how differences in treatment-effect parameters across groups can be decomposed into a returns component and selection-related components. Empirically, we apply this framework to the post-war expansion of university access in the United Kingdom, combining UK Biobank data with a newly constructed geocoded dataset of higher-education institutions. Using the reduction in distance to the nearest university between birth and adolescence as an instrument for college attendance, we find substantial heterogeneity in returns across Educational Attainment Polygenic Index (EA PGI) groups. Returns to college are markedly larger for individuals with low genetic predisposition to education than for those with high predisposition. Our decomposition shows that this gap is driven primarily by heterogeneity in returns rather than by differential sorting into college. Overall, the evidence suggests that the UK university expansion may have played a compensatory role by reducing inequalities associated with genetic predisposition to education.