The heterogenous relationship of food stamp participation with body mass: Quantile regression model

This study explores the heterogenous relationship between Food Stamp Program (FSP) participation and body mass index (BMI) for low income individuals whose per capita family income falls in the bottom two quintiles. We estimate changes in the dispersion of the entire distribution of BMI following FSP participation using a quantile regression model in both cross-sectional and longitudinal analyses. For women, we find that the mean OLS estimate of FSP participation underestimates the extent of the positive association with BMI at the top quantiles in both cross-sectional and longitudinal individual fixed effects models. Controlling for individual fixed effects in the quantile regression model reduces the estimated magnitude of the association by a factor of three to four, although the estimates remain statistically significant for women. For men, a marginally statistically significant positive association of FSP participation with BMI is found at the top quantiles compared to no statistically significant mean estimate although the association is no longer statistically significant when time-constant individual heterogeneity is adjusted for in the individual fixed effect models.