Do Food Stamps Cause Obesity? A Generalised Bayesian Instrumental Variable Approach in the Presence of Heteroscedasticity

The impact of covariates on obesity in the US is investigated, with particular attention given to the role of the Supplemental Nutrition Assistance Program. The potential endogeneity of participation in SNAP is considered as a potential problem in investigating its causal influence on obesity using instrumental variable (IV) approaches. Due to the presence of heteroscedasticity in the errors, the approach for dealing with heteroscedastic errors in Geweke (1993) is extended to the Bayesian instrumental variable estimator outlined in Rossi et al. (2005). This approach leads to substantively different findings to a standard classical IV approach to correcting for heteroscedasticity. Although findings support the contention that the SNAP participation rate is associated with a greater prevalence of obesity, the evidence for this impact is substantially weakened when using the methods introduced in the paper.