Association for Public Policy Analysis & Management

Participation in the Supplemental Nutrition Assistance Program (SNAP) is correlated with higher body weight. In an effort to curb obesity in the U.S., a number of states have proposed to restrict the use of SNAP benefits to purchase sugar-sweetened beverages and junk foods. However, the casual relationship between SNAP and body weight has not been established. Among the few causal studies in the area, scholars have not reached a consensus (Ver Poleg & Ralston, 2008; Andreyeva et al., 2014).

Almada and Echernis (2016) employ an instrumental variable approach treating the proportion of children eligible for free or reduced price meals as an external income shock that is uncorrelated with other household characteristics. They found that additional SNAP benefits reduce adult’s body mass index (BMI) and the probability of being obese. However, the study lacked a direct measure of receipt of free/reduced price meals instead relying on a proxy the proportion of school-age children in the family for the instrument. This proxy may confuse the effect of children growing into school age and children receiving additional welfare benefit from school. Without specific information of whether children received free or reduced price school meal, this proxy may generate confusing or misleading results.

Building upon this prior work, the present study uses the National Household Food Acquisition and Purchase Survey (FoodAPS) data, which enable computation of number of free school meals that each household receives. In addition, FoodAPS uses multiple approaches, including matching with administrative data, to improve the estimation of the participants’ status of being a SNAP participant or SNAP-eligible-nonparticipant, addressing the concern of misreporting.

This study uses and extends Almada and Echernis’s (2016) approach by using the number of children’s receipt of free/reduced price meals as an instrumental variable to assess whether an exogenous increase in household income is associated with higher body weight and probability of being obese. The preliminary finding shows that number of free/reduced priced meals has no significant effect on adults’ BMI or probability of being obese among the SNAP participating families, eligible non-participating families, or combined. To address the endogeneity concern of self-selection into having free/reduced school meals, sensitivity analysis is performed. By only looking at the households with at least one free/reduced school meal, there is still no significant effect of free/reduced price school meals. This finding is inconsistent with Almada and Echernis’s, which shows that increase in proportion of school-aged children predicts decrease of adult BMI and probability of being obese. The inconsistency suggests a possibility of different result from different measures and datasets.

Further investigation will analyze the direct food purchasing behavior of SNAP participants and eligible non-participants using the same instrumental variable, including the soda and healthy food purchases.