“Using Genes to Explore the Effects of Cognitive and Non-Cognitive Skills on Education and Labor Market Outcomes”, 2021-09 (; similar):
A large literature establishes that cognitive and non-cognitive skills are strongly correlated with educational attainment and professional achievement. Isolating the causal effects of these traits on career outcomes is complicated by reverse causality and selection issues.
We suggest a new approach: using within-family differences in the genetic tendency to exhibit the relevant traits as a source of exogenous variation. Genes are fixed over the life cycle and genetic differences between full siblings are random, making it possible to establish the causal effects of within-family variation in genetic tendencies.
We link genetic data from individuals in the Swedish Twin Registry to government registry data and find evidence for causal effects of the genetic predispositions towards cognitive skills, personality traits, and economic preferences on professional achievement and educational attainment. Our results also demonstrate that education and labor market outcomes are partially the result of a genetic lottery.
[Keywords: personality traits, economic preferences, cognitive skills, labor markets, education]
…We find strong evidence for a causal effect of the predisposition toward stronger cognitive skills on income, occupational status, and educational outcomes. We also find evidence for statistically-significant effects of the predispositions toward several non-cognitive traits: individuals who tend to be more risk seeking, mentally stable, and open tend to work in more prestigious occupations. The opposite is true for individuals with a tendency towards narcissism or discounting the future. A tendency towards being open and forward-looking also increases educational attainment (EA). Finally, we document large causal effects of the general genetic tendency towards higher EA on all the outcomes we study. This illustrates that success in education and professional careers is in part down to “genetic luck”. We also investigate heterogeneity in these effects by gender and socioeconomic status (SES) of the parents. We find some evidence of a stronger effect of the predisposition toward cognitive skills for high-SES individuals, in particular on educational outcomes. We also find that the effects of the genetic tendencies on income tend to be stronger for women, implying that gender differences in labor market outcomes are generally larger for less skilled individuals. The exception is the link between genetic tendencies and management positions: our results suggest that cognitive and non-cognitive skills strongly increase the likelihood for men to work in a management position but that effects are much weaker for women.
…The polygenic indices we use stem from the work of the Social Science Genetic Association Consortium (SSGAC) ( et al 2021).
…2.4 Sample: For the full-sample analyses looking at educational outcomes, we will limit the dataset to genotyped individuals born 1934–61199529ya (that is, individuals who have likely completed their education) whom we can link to their parents’ records for the construction of the socioeconomic controls.13 This subsample contains 29,393 individuals. For the analyses looking at labor market outcomes, we will limit the dataset to individuals born 1934–56199034ya (that is, individuals who have likely completed their education and worked for a few years). This subsample contains 25,515 individuals. For our causal analyses using within-family variation, we will limit the sample to complete sets of genotyped dizygotic twins. This sample contains 11,344 individuals (5,672 twin pairs) for the education analyses and 9,594 individuals (4,797 twin pairs) for the income analyses.
…The scaled estimates in Figure 2 show that the magnitudes of the effects are economically meaningful. A one-standard deviation difference in the cognitive performance PGI is associated with a roughly 10 percentage points increase in the likelihood of having graduated from university. The effect of math skills is roughly 5 percentage points. These 2 effects are estimated simultaneously, meaning that an individual with one-standard deviation higher cognitive performance and math skills is around 15 percentage points more likely to graduate from university. The effects of the statistically-significant non-cognitive traits (openness, narcissism, and time discounting as proxied by smoking) are similarly large. Finally, a one-standard deviation increase in the educational attainment PGI is associated with 0.4 to 0.6 additional years of education.
[Given the large sample size, it’d be better to skip the PGSes—which still capture so little of the genetics—and use sibling IBD or RDR to establish estimates of total causal effects.]