The Geography of Poverty and Nutrition: Food Deserts and Food Choices Across the United States

We study and quantify the variation in healthful consumption habits across income brackets in the United States. Using a detailed grocery shopping database, we document a positive nutrition-income gradient: the highest-income quartile of households buy groceries that are 0.25 standard deviations more healthful than the lowest-income quartile. We then study the underlying mechanisms leading to the gap. On the supply side, stores in low-income neighborhoods offer fewer produce items and overall less-healthful groceries. However, we find no evidence to support the hypothesis that the low availability of healthful groceries causes unhealthful eating. Using an event study framework, we find that nearby supermarket entry does not increase households’ produce purchases or the overall healthfulness of grocery purchases. Instead, we observe substitution in household expenditures away from incumbent supermarkets. On the demand side, We estimate a formal demand model and use the estimated preferences to decompose differences in purchasing patterns between lowand high-income households. We find that 91% of the nutrition-income gradient is driven by differences in demand across products, while only 9% can be attributed to differences in supply. The demand impact can be broken down as follows: 16% due to product group demand differences, and 75% due to nutrient preferences. The 9% supply effect is entirely driven by different product offerings, with prices actually mitigating the nutrition income relationship by 2%. ——————————————————————————–————————∗Allcott: New York University and NBER. [email protected]. Diamond: Stanford GSB and NBER. [email protected]. Dubé: Chicago Booth and NBER. [email protected]. We thank Catherine Wright for exceptional research assistance, and we thank Charles Courtemanche and Sungho Park for sharing data. We also thank seminar participants at Duke, the 2015 and 2016 NBER Summer Institutes, New York University, Princeton, Stanford, and the University of Pennsylvania for helpful comments. We are grateful for funding from the Chicago Booth Initiative on Global Markets.