As the public expenditure for the Supplemental Nutrition Assistance Program (SNAP) - formerly the Food Stamp Program (FSP) - increases, improving the effectiveness of the policy becomes pivotal to limit further surges in public spending. Along with social stigma, transaction costs, associated in part to the accessibility and proximity to food outlets, are the main deterrent to program participation. This study presents an empirical assessment of the relationship between food access and FSP participation among eligible population. The analysis uses county-level data for the continental U.S., distinguished by different stores formats (grocery stores, convenience stores and a non-traditional, low-priced alternative, Wal-Mart Supercenters) accounting for the endogeneity of their location decision. To estimate the parameter of the model we use a relatively novel estimator: the spatial generalized two-stage least square (GS2SLS) estimator with heteroskedastic autoregressive disturbances of order (1,1) or SARAR(1,1), developed by Kelejian and Prucha (2010). Empirical results show that, among eligible individuals, the presence of small convenience stores, large grocery stores, and Walmart supercenters entice participation in the food stamp program. In sum, increasing the number of stores for which commuting by car is not strictly necessary (convenience stores) or the proximity to stores with wider assortment of low-priced items may act as catalysts to ability to the policy ability to reach the underprivileged.