Measuring Long-Term Healthcare Spending Attributable to Adult Obesity

While obesity is believed to be a major driver of poor health, there is little evidence about whether it leads to higher healthcare spending. To date, only one paper has produced a causal estimate. We build on the work of Cawley and Meyerhoefer (2012), proposing a new instrumental variable model and using up to 12 years of panel data. We exploit genetic variation in weight across siblings as a natural experiment to estimate the impact of obesity on healthcare spending using Panel Study of Income Dynamics data from 1999 and 2011. While the PSID has been used to study demographic trends in obesity and related health outcomes using cross-sectional descriptive statistics, this study is the first to apply the PSID in a causal study exploring the longitudinal impact of obesity on health expenditures. We estimate a two-stage residual inclusion model with a generalized linear model and find no significant differences in healthcare spending among obese versus non-obese adults.