Working Paper

Abstract

The authors examine the relationship between body mass index (BMI) and mental health for young adults and elderly individuals using data from the National Longitudinal Study of Adolescent Health and the Health & Retirement Study. While ordinary least squares (OLS) estimates show that BMI is significantly associated with worse mental health in both young adulthood and old age, they are likely to be confounded by (i) unobserved factors that affect both BMI and mental health and (ii) reverse causality.

To tackle confounding, the authors take two complementary approaches. First, they use a polygenic score for BMI as an instrumental variable (IV) and adjust for polygenic scores for other factors that may invalidate this IV. The IV estimates indicate that there is no statistically significant relationship between BMI and mental health for young adults, whereas there is a positive and statistically significant relationship for the elderly. Moreover, the authors show that IV estimates likely have to be interpreted as identifying a weighted average of effects of BMI on mental health mostly for individuals on the upper quantiles of the BMI distribution. Given potential remaining concerns about the validity of the IV, their second approach is to consider it an “imperfect” IV and estimate an upper bound on the average treatment effect for the corresponding population following Nevo & Rosen (2012).

The estimated upper bounds reinforce the conclusions from the IV estimates: they show little evidence of a detrimental effect of BMI on mental health for young adults while being consistent with an economically meaningful effect for elderly individuals. Lastly, the authors explore some of the potential channels through which BMI may affect mental health for the elderly.