Over the last twenty years the goals of SNAP have been expanded to include improving the health and nutrition of low-income Americans over other dimensions than food security.  One nutrition-related issue that SNAP has been asked to address, especially in recent years, is the extent of obesity among SNAP eligible households.  As such, some have urged a restructuring of aspects of SNAP to help lead to reductions in obesity among low-income households.  In particular, some have suggested the imposition of restrictions on what food items can be purchased with SNAP benefits.

In this paper I begin with a theoretical consideration of the effect of an increase in resources (e.g., through the receipt of SNAP) available to households on food purchases and other items in the health production function.  As for any household with an increase in purchasing power (through, e.g., a raise at a job), whether the receipt of SNAP leads to purchases that have the potential to lead to better or worse health outcomes is ambiguous.

While the theoretical effect of increasing resources is ambiguous, empirically it is generally thought that those with higher incomes are less likely to be obese than those with lower incomes.  To illustrate this I use data on BMI taken from the 1999 to 2008 waves of the National Health and Nutrition Examination Survey (NHANES) to construct non-parametric representations of the relationship between the income-to-poverty ratio and the probability of obesity and overweight.  Given the large sample size, along with the full population I present these results for girls, boys, women, and men and for African-Americans, Hispanics, and non-Hispanic whites within each of these categories.  The main contribution here will be to see if there are differences across the income spectrum that can be uncovered by such non-parametric representations.

If income is inversely related to probabilities of obesity, since SNAP increases resources for low-income households, we would imagine that it also leads to decreases in obesity.  I therefore consider findings from the literature regarding the effect of SNAP on obesity.  In this review, I will especially emphasize studies that have explicitly considered selection into SNAP (due to non-random distribution of eligible households into SNAP) and that use measured heights and weights (due to well-known non-random biases that emerge with self-reports of height weight).

I conclude with a consideration of the advantages and disadvantages associated recent proposals to restrict purchases using SNAP benefits.   This will be structured around the following questions:  Does increasing the resources available to low-income households lead to increased probabilities of obesity?  Under the existing structure of the program, are SNAP recipients more likely to be obese than eligible non-recipients?  Given the other goals of SNAP,  how might purchase restrictions influence the ability to meet these goals?