Sponsored by:
It is increasingly accepted that important individual outcomes, such as educational attainment, are influenced by a complex combination and interaction of environmental circumstances and genes. Empirically estimating the contribution of the two components, however, is complicated by the endogeneity of both factors. In this paper, we exploit the exogenous variation in both genetic propensity for education and the family socio-economic environment. We measure the genetic effect using a highly predictive polygenic score for educational attainment in siblings. Our measure of environment is an individual's birth order, which has been consistently negatively correlated with educational attainment within and between family. The literature suggests that this is driven by parental preferences for fairness and equality in investments in children at a time and, thus, reduced parental investments, which are naturally more constrained for later born children. We overcome endogeneity issues in the estimation of GxE by using only within-family variation in our measures of genes and the environment. Indeed, genetic variants are randomly assigned across siblings within a family, and, conditional on family size, birth order is random within families. Our empirical analysis exploits around 19,000 siblings from the UK Biobank, a population-based sample from the United Kingdom. We construct genetic scores for educational attainment based on our own tailor-made Genome-Wide Association Study (GWAS) that uses the UK Biobank but excludes all siblings and their relatives. Moreover, we apply Obviously-Related Instrumental Variables to reduce measurement error in the genetic score. Our empirical strategy is a family fixed effects approach, exploiting within-family differences in genetic scores and birth order to study their main effects and interactions in educational attainment. We focus mainly on first-borns versus later-borns since the literature suggests birth order effects are particularly salient for the first two children in a family. Our findings are as follows. We confirm earlier findings in the literature that later-borns have a lower level of education than first-borns. We also confirm the long-held conjecture that genetic endowments do not differ across birth order, and so by definition birth order effects must be due to environmental influences, e.g., greater parental time investments in the firstborn child. Our main finding is that birth order and genetic endowments interact: being first born and having relatively high genetic endowments for education exhibit a positive interaction, meaning that those with higher values for the genetic score benefit disproportionally more from being first born compared with those less privileged in terms of genetic scores. Hence, besides this finding being a clean case of how genes and the environment interact in producing important life outcomes such as educational attainment, our empirical results are consistent with the existence of dynamic complementarity in skill formation: the parental time investments associated with being first born are more "effective" among those siblings who randomly inherited a higher genetic score for educational attainment.