Be smart, live long: the relationship between cognitive and non-cognitive abilities and mortality

I study the association between cognitive and non-cognitive abilities and mortality, and investigate how well income and education act as proxy measures for ability. The risk of premature mortality is estimated using Cox proportional hazard models with a dataset of 692,303 Swedish men aged 18-20 years, enlisted between the years 1969-1983, and deaths between the years 1969 and 2009. Results suggest that both cognitive and non-cognitive abilities are strongly associated with mortality, independently and through income and education. Non-cognitive ability is a stronger predictor of the risk of mortality than cognitive ability. For middle and high income earners, and individuals with a college education, there are no associations between the abilities and mortality. However, for low income earners and individuals without a college education, cognitive and non-cognitive ability have strong associations with mortality. Results are mainly driven by the bottom of the measured ability distributions.