Abstract:
This paper exploits molecular genetic data to quantify genetic confounding in parent-child educational outcomes. We develop a model of the intergenerational transmission of education based on insights from the literature on social science genetics. The model distinguishes between two types of genetic confounding. First, narrow genetic confounding reflects the direct transmission of genetic predisposition towards education. Second, broad genetic confounding captures direct genetic transmission as well as genetic nurture, i.e., an influence of parental genes on children's outcome through the family environment. Next, we use the Avon Longitudinal Study of Parents and Children (ALSPAC) data to decompose the association between parental years of education and their offspring's grades on Key Stage 4 national exams. To proxy genetic endowments, we construct Educational Attainment Polygenic Indices (EA PGIs) for parents and children. To correct for measurement error, we use Obviously-Related Instrumental Variables (ORIV) based on two independent PGIs. The results suggest that 'broad genetic confounding' explains 30-45% of the parent-child educational association, and 'narrow genetic confounding' 18-33%. We find no meaningful differences between mothers and fathers. Using our model, we compare our estimates to twin and adoptee designs, and show how molecular genetic approaches can recover both broad and narrow genetic confounding under plausibly weaker assumptions and with arguably greater external validity.