Research
The Role of Gene-Environment Interaction in the Formation of Risk Attitudes - R & R in Economics and Human Biology
Attitudes toward risk (risk attitudes) are an important feature of every individual's decision-making process. Recent empirical literature suggests that risk aversion is more than a simple parameter of economic models but it rather captures a more complex psychological primitive which is endogenous to many models describing decision-making process. However, the mixed evidence about the causes of risk attitudes that prevails in the literature suggests that models that consider only the environmental variation, when explaining attitudes toward risk, may not adequately capture the rather complex relationship between willingness to take risk, environmental exposures, and initial endowments. I propose a model that allows to test for individual heterogeneity in the responsiveness to environmental changes. I argue that the treatment heterogeneity may be one of the reasons for the inconclusive evidence in the literature.
To investigate the individual heterogeneity, I exploit recent advances in behavioral genetics and estimate a gene-environment model. To accomplish our goals, I use a representative sample of the US population born between 1929 and 1959 from the HRS data set. The data contain information about the risk aversion of the respondents and about their genetic endowment. As a measure of genetic risk the paper proposes to construct a polygenic score, a measure commonly used in the literature. I find that the responsiveness to adverse economic situations varies by genotype which implies that average treatment effect estimates of an environment depend on the genetic composition of the sample. Substantively, I find that experiences with adverse development on the labor market affect the risk attitudes only for the individuals with low genetic predispositions for risk taking. The results imply that adverse economic conditions can increase aversion towards risk taking for individuals with stronger genetic predispositions for risk aversion. At the same time I cannot reject the hypothesis that adverse economic conditions have any influence on people with higher genetic predispositions for risk taking.. This finding can ultimately lead to an increase in inequality in health, wealth, income and other outcomes related to risk attitudes.
This study investigates the role of the interplay between genetic endowment and education in health formation, with a focus on addressing health disparities in later life stages. Leveraging data from the UK's Raise of School Leaving Age policy as a natural experiment, coupled with innovative methodology in gene-environment model estimation, I explore the impact of genetic predispositions and educational attainment on the prevalence of chronic diseases and medical conditions among the elderly population.
The paper reveals that genetic endowment significantly influences the likelihood of developing heart attacks, strokes, cancer, and type 2 diabetes in later life. Furthermore, the paper shows that education serves as a protective factor against adverse health outcomes associated with genetic predispositions, particularly concerning heart attacks and cancer. This study contributes to advancing our understanding of health formation processes and highlights the significance of tailored interventions and policies in reducing health inequalities and promoting overall population health.
A New Method to Study Gene-Environment Interaction in Empirical Economics Models
Many socio-economic surveys have started to include genetic data about their respondents, which has lead to new studies that investigate how environments and choices interact with genetic endowments to form important economic, behavioral, or health outcomes. To cope with the high dimensionality of genetic data, researchers often summarize individual genetic information using an index for genetic predisposition called a polygenic score (PGS). The index exploits information from genome-wide association studies (GWAS), which establish robust correlations between genes and determinants of economic wellbeing, health, and inequality: including preferences, smoking, obesity, and education. The GWAS correlations are then used to construct a PGS for a given outcome, which then often serves as a variable in empirical economic models. This paper revisits the validity of the usage of PGSs in the framework of the widely used gene-environment models and in the non-interacted models. First, I demonstrate that gene-environment (GxE) interactions can severely distort the PGS index and thereby skew the results of important parameters of GxE studies. To correct the bias that stems from omitted GxE interaction in the GWAS, I propose a new two-step method to estimate GxE models and their non-interacted counterparts. The new method requires only information from a GWAS to select the relevant genetic variables in the first step, and in the second step it estimates the full GxE model jointly. Unlike the standard method, this procedure does not rely on the GWAS estimates, which are often derived from a different population than in the survey used for the main empirical specification. Hence, the new method does not suffer from biases that stem from using GWAS estimates in the PGS index. In the empirical application I show that measurement error bias can significantly distort inference based on the standard GxE modelling approach. By not relying on the GWAS estimates, the new method expands the scope of the current survey-based studies that aim to incorporate genetic data into social research. The new method allows the study of outcomes for which suitable GWAS are not yet available.