Characterizing the Genetic Influences on Risk Aversion

Abstract

Risk aversion has long been cited as an important factor in retirement decisions, investment behavior, and health. Some of the heterogeneity in individual risk tolerance is well understood, reflecting age gradients, wealth gradients, and similar effects, but much remains unexplained. This study explores genetic contributions to heterogeneity in risk aversion among older Americans. Using over 2 million genetic markers per individual from the U.S. Health and Retirement Study, I report results from a genome-wide association study (GWAS) on risk preferences using a sample of 10,455 adults. None of the single-nucleotide polymorphisms (SNPs) are found to be statistically significant determinants of risk preferences at levels stricter than 5 × 10(-8). These results suggest that risk aversion is a complex trait that is highly polygenic. The analysis leads to upper bounds on the number of genetic effects that could exceed certain thresholds of significance and still remain undetected at the current sample size. The findings suggest that the known heritability in risk aversion is likely to be driven by large numbers of genetic variants, each with a small effect size.

The Genetics of Risk Aversion: A Systematic Review

“…An article (only for risk aversion) [ 26 ], comprehensively addressed the possible association between more than 2 million single nucleotide polymorphisms (SNPs) extracted from the U.S. Health and Retirement Study (HRS), and the risk-aversion level, measured with hypothetical gambles on lifetime income, in a sample of 10,455 adults. The remaining articles included in the review dealt with more specific genetic measures, known as “candidate-gene” studies [ 41 ].…”

“…According to Harrati [ 26 ], none of the multiple SNPs included in the study revealed a significant causal effect on risk aversion, pointing to greater phenotype complexity, as well as their possible polygenic origin. No other measures of health or behavior were considered in the study.…”

“…Moreover, risk aversion is not exempt from individual variability. This heterogeneity seems to depend on multiple factors such as age, gender or level of education, although these only predict a small fraction of the variance in risk-taking [ 26 ]. Multiple studies with twins (homozygous and heterozygous) have shown that risk aversion could depend between 20% [ 27 , 28 ] and almost 60% [ 29 ] on genetic variations.…”

Does Sociology Have Any Choice but to Be Evolutionary?

“…"Simple" correlations (like the probability that parents and children will have the same attitude to sharing) are also routinely analyzed statistically in sociology (for example in modeling social mobility or voting behavior). The slightly trickier kind of data, which is currently at the frontiers of the social sciences, involves any attempt to characterize relationships between genetic phenomena and social behavior (Harrati, 2014). Such approaches are not necessarily intrinsically difficult to implement (though they do rely on genetic information that has only recently become reliably available in quantity) but they do raise important issues of research design (any such correlation we discover is mediated by large quantities of socialization and social interaction between birth and sampling and should therefore be interpreted with caution) and academic culture: Sociologists are not at all comfortable with the idea of "admitting" the ongoing role of genetics in human behavior in case it becomes a rationalization (probably spurious) for biological reductionism or inhumane kinds of social intervention.…”

Genetics of Risk-Taking Behavior: Current Knowledge, Challenges and Future Directions

“…By analyzing the frequencies of several hundred thousand genetic polymorphisms between these groups, researchers are able to find genetic polymorphisms that are more frequently observed in one group compared to the other. In relation to RTB, an earlier GWAS comparing quartiles of elderly in terms of risk aversion to hypothetical gambles failed to identify any polymorphisms that was different between the groups beyond the significance level (Harrati 2014). On the other hand, later GWAS studies with self-reports of RTB identified multiple genetic polymorphisms that pass the significance threshold.…”