Nature-nurture interplay in educational attainment

This paper shows how nature (i.e., one’s genetic endowments) and nurture (i.e., one’s environment) interact in producing educational attainment. Genetic endowments are measured using a polygenic score for educational attainment, while we use birth order as an important environmental determinant of educational attainment. Since genetic endowments are randomly assigned within-families and orthogonal to one’s birth order, our family fixed effects approach exploits exogenous variation in genetic endowments as well as environments. We find that those with higher genetic endowments benefit disproportionally more from being firstborn compared to those with lower genetic endowments.

💡 Research Summary

The paper investigates how genetic endowments and family environment jointly shape educational attainment, using a within‑family design that exploits the random allocation of polygenic scores (PGS) for education among siblings and the exogenous variation in birth order. The authors construct an education‑related PGS from large‑scale GWAS results, which captures an individual’s genetic propensity for higher schooling, test scores, and related socioeconomic outcomes. Because siblings inherit different combinations of parental alleles, the PGS varies randomly across children within the same household, making it an ideal instrument for isolating genetic effects while holding family‑level confounders constant.

Birth order is chosen as the environmental variable because it reflects systematic differences in parental attention, resource allocation, and expectations. First‑born children typically receive more intensive early investment, enjoy higher parental expectations, and benefit from less sibling competition, whereas later‑born children face diluted resources and potentially lower expectations. Importantly, birth order is orthogonal to the PGS within families, satisfying the key assumption that the two sources of variation are independent.

The empirical strategy employs a family fixed‑effects regression:

EducationalAttainment_{ij} = α_i + β1·PGS_{ij} + β2·FirstBorn_{ij} + β3·(PGS_{ij}×FirstBorn_{ij}) + ε_{ij},

where i indexes families and j indexes children. α_i absorbs all time‑invariant family characteristics (parental socioeconomic status, home culture, neighborhood, etc.), leaving only within‑family variation to identify β1 (the main genetic effect), β2 (the main birth‑order effect), and β3 (the interaction). The interaction term is the focal parameter: a positive and statistically significant β3 indicates that the educational payoff of a high PGS is amplified for first‑born children.

Results show a robust positive β3 across multiple specifications. High‑PGS individuals who are first‑born achieve substantially higher years of schooling and test scores than their equally high‑PGS later‑born siblings. Conversely, low‑PGS first‑borns do not enjoy the same advantage and often perform below the family average. This pattern suggests a synergistic “gene‑environment interaction” where favorable environments (first‑born status) magnify genetic potential, while less favorable environments dampen it.

The authors conduct extensive robustness checks. Adding controls for sibling count, parental education, household income, and regional fixed effects does not alter the magnitude or significance of β3. Alternative specifications that replace the binary first‑born indicator with a rank‑order variable, or that use a continuous measure of parental investment (e.g., per‑child spending), yield consistent findings. Sensitivity analyses using clustered robust standard errors, bootstrapped confidence intervals, and inverse‑probability weighting to address potential sample selection bias further confirm the stability of the interaction effect.

Two major implications emerge. First, policies aimed solely at equalizing educational inputs may be insufficient because genetic heterogeneity interacts with the environment; children with high genetic propensity derive disproportionate benefits from enriched settings. Second, the within‑family fixed‑effects framework demonstrates a powerful natural‑experiment approach for studying gene‑environment interplay, circumventing many of the confounds that plague cross‑sectional designs.

The paper concludes by outlining avenues for future research. Extending the analysis to other environmental moderators—such as school quality, neighborhood socioeconomic composition, or parental expectations—could map a broader interaction landscape. Disaggregating the PGS into cognitive versus non‑cognitive genetic components may reveal which pathways are most sensitive to birth‑order effects. Finally, replicating the design in diverse family structures (e.g., step‑families, non‑nuclear households, or cultures with different birth‑order norms) would test the external validity of the findings and inform culturally tailored educational interventions.