Risk Tolerance and its Relation to Important Life Events

Abstract:     
We study panel data on three birth-year cohorts to understand the changes in risk tolerance across different age groups. Interval regression models on cross-sectional data for each birth cohort are constructed to investigate how time-invariant factors and macroeconomic events such as September 11, 2001, might influence individuals' risk tolerance. Well-established maximum likelihood estimation procedure is followed to obtain a parametric distribution of logarithm of relative risk tolerance that is potentially subject to noise, and to assess how this distribution shifts with time. Finally, we conduct interval regressions on first difference models to capture the reasons for change in the risk tolerance in each cohort. The main differences in the results as compared to previous studies are that risk tolerance does decrease with age, but there is much more within-subject heterogeneity among younger individuals, and that income lowers risk tolerance among younger Americans, whereas wealth increases risk tolerance among older Americans. The risk tolerance of the oldest cohort is negatively affected by late-life marriage and widowhood, whereas risk tolerance increases with greater wealth, and somewhat due to retirement. An increase in family size through marriage and more children lowers the risk tolerance of the youngest cohort, whereas unemployment has a positive association with tolerance.

Keywords: Risk tolerance, Latent variables, Interval Regression, Panel data

JEL Classifications: C01, C10, C23, D0, D81

Working Paper Series