Association for Public Policy Analysis & Management

Introduction: Prior research on food deserts has found associations between the food retail environment and nutrition and health outcomes, implicitly assuming that intermediary differences in food shopping patterns account for these relationships. This paper investigates whether food shopping and the food retail environment are, in fact, associated.

Methods: Using electronic benefit transfer (EBT) card transactions of 40,593 Supplemental Nutrition Assistance Program (SNAP) participants, we investigate the relationship between SNAP households’ continuous distance to supermarkets, as well as the proximity of other types of SNAP retailers, and (1) the percentage of SNAP benefits spent at supermarkets, (2) the frequency of supermarket shopping trips, and (3) the percentage of supermarket spending spent on fruits and vegetables. Three months of data - from March, July, and October 2012 - are pooled together to account for seasonality in food shopping patterns.

Results: We find that distance to the nearest supermarket and supermarket shopping outcomes correlate only modestly, and that the relationship between the two shifts from negative to null at a relatively low threshold distance. While supermarket shopping patterns also appear to vary based on the presence or absence of a local, non-supermarket food retailer, the magnitudes of these relationships are small. On average (controlling for income, demographics, and social service program participation), SNAP participants living immediately adjacent to a supermarket spent 81.0% of their benefits at supermarkets and took 4.5 supermarket shopping trips per month. SNAP participants living both far from a supermarket and close to a non-supermarket, however – the food retail environment least conducive to supermarket shopping – still spent an average of 76.6% of their benefits at supermarkets and took 3.8 supermarket shopping trips per month, a substantively similar quantity. Moreover, Hampden County SNAP participants tended to travel outside of their neighborhoods to spend their SNAP benefits, traveling an average of 3.2 miles to the retailer at which they spent the largest portion of their benefits despite living an average of only 1.0 mile from the supermarket nearest to their homes.

Discussion: These results suggest that SNAP households’ inability to reach healthy food retailers at which they can spend their benefits is at most a minor driver of geographic disparities in nutrition and health outcomes. Poverty itself is likely more relevant. Distance to the nearest supermarket may not be the most substantive measure of food retail access.