Feeding Urban Communities: A Look at Disparities in Food Access and Nutritional Consumption

Large nutritional disparities exist across different socioeconomic groups in the United States. A commonly cited explanation for these nutritional disparities is spatial disparities in access to stores selling nutritious foods. The past decade has seen a growing number of federal and state programs providing loans, grants, and tax credits to stimulate supermarket development and to encourage retailers to offer healthful foods in “food deserts” (CDC, 2011). The hope is that improving access to healthy foods will improve the diets of households residing in currently underserved areas. Pennsylvania was one of the first states to dedicate substantial resources to this; between 2004 and 2011, it provided $73.2 million in loans and $12.1 million in grants to stimulate supermarket development in underserved areas through the Fresh Food Financing Initiative. Various states followed suit, and since 2011 such efforts have been supported federally through the Healthy Food Financing Initiative, which was recently authorized for $125 million in annual funding through the 2014 farm bill (Aussenberg, 2014).

Despite the growing popularity of such programs, little is known about the true extent of differences in access and the role that these differences play in generating nutritional disparities. While various studies have documented that socioeconomically disadvantaged neighborhoods have access to fewer stores, there is relatively little systematic evidence from national studies indicating that these neighborhoods have access to less nutritious foods (Bitler and Haider, 2011). If differences in access are in fact limited, or even if differential access is substantial but does not limit the consumption of local consumers, then policies which target access will have little impact on nutritional disparities.

In my recent work with Ilya Rahkovsky at the USDA Economic Research Service and Molly Schnell of Princeton University, my colleagues and I use novel data describing the nutritional quality of household food purchases and the retail landscapes in which these consumers are making decisions to shed some light on these issues. We first document significant differences in the nutritional quality of foods purchased by different socioeconomic groups across the United States. This generalizes the results of previous studies that have documented disparities in nutritional consumption by focusing on purchases of a few products, such as fruits or vegetables, or purchases in specific localities.[1]  (You can read our full report here: What drives nutritional disparities? Retail access and food purchases across the socioeconomic spectrum.)

In our study, we construct a dataset describing the full nutritional profile of the grocery purchases made by over 60,000 households from 52 markets across the U.S. between 2006 and 2011 and calculate two complementary household-level indexes, an “expenditure score’’ and a “nutrient score’’ which represent the healthfulness of the products purchased relative to USDA category-level expenditure recommendations and FDA recommendations for per calorie nutrient consumption, respectively.[2]

These nutritional indexes reveal significant disparities in the healthfulness of purchases across households with different income and education levels. The products purchased by households in the highest terciles for income and education are 40 percent closer to the USDA recommendations for product category expenditure shares and 40 percent closer to the FDA recommendations for per calorie nutrient consumption than the products purchased by households in the lowest terciles of income and education. In other words, the data suggests that households in the highest third of income and education levels are much more likely to purchase foods that are in line with the USDA’s health recommendations than are those in the lowest third of education and income levels. Interestingly, there is 50% as much variation in the nutritional quality of purchases across households with different education levels as there is across households with different incomes. In other words, the purchases of two households earning the same income but with different levels of educational attainment are likely to be far more different in terms of nutritional quality than the purchases of households that have similar levels of education but earn different incomes.

We then provide the most comprehensive picture of the healthfulness of products available at retail locations across the United States and the degree to which retail environments differ by socioeconomic status. Consistent with previous studies, we find that access to nutritious foods is much greater in wealthier and more educated neighborhoods (Beaulac et al. (2009); Ver Ploeg et al. (2009)). However, we also find that the differences in the number of stores in different neighborhoods are much greater than the differences in the types of products being offered in different neighborhoods.

Using geo-coded data on the location of over 200,000 retailers across the United States, we find large disparities in the concentration of stores across neighborhoods with different socioeconomic profiles. We then use weekly store-level sales data from Nielsen to identify the products that are available at over 30,000 participating retailers between 2006 and 2011. Analogous to the household-level analysis, we calculate two store-level healthfulness indexes that reflect the nutritional quality of the products available on the shelves. Our examination of these nutritional indexes reveals significant disparities in the healthfulness of products purchased in neighborhoods with different income and education levels, but only small differences the products offered by stores in these different neighborhoods.

Identifying the causal role that access plays in generating disparities in purchases is difficult. Since households sort into neighborhoods and retailers cater to local tastes, consumption disparities across locations with differential access may reflect not only the role of access but also demand-side factors. In a monopolistically-competitive retail industry, firms will cater to the prevalent tastes in the local market. If high-socioeconomic households have stronger tastes for healthful food products than low-socioeconomic households, it follows that more healthful food products will be sold in high-socioeconomic neighborhoods. On the other hand, supply-side factors, such as a complementarity that makes it relatively more profitable to distribute healthy food products in areas with high retail rents, could also imply that firms in wealthy neighborhoods will offer more healthful food products than stores in poor neighborhoods, even if the households in these neighborhoods have identical tastes.

The detailed nature of our household-level purchase data allows us to go beyond existing work and discern the highest amount of observed nutritional disparities that could be due to differences in access. The intuition behind our approach is simple. We expect that differences in access are limited within neighborhoods, so we first look at whether consumption disparities persist when we restrict our attention to households in the same small geographic area. While the correlation between income and the healthfulness of food purchases is reduced by half when we control for the household’s census tract, the more prominent relationship between education and healthfulness is more persistent - differential access explains less than 10% of the disparities across education groups.

While informative, our “within-location’’ approach has its limitations. Households living in the same neighborhood possibly have differential access, either because they live in different locations within the neighborhood or because of differences in mobility, such as car ownership. To eliminate the effect of access entirely, we look at purchases made within a given store. The results from the within-store analysis mirror those from the within-location analysis: the correlation between income and the healthfulness of food purchases is again cut in half when we look at purchases made within the same store, whereas the correlation between education and nutritional quality is only reduced by 10 percent.

Finally, we show that the impact of store entry on the healthfulness of food purchases made by local households is limited in the 3,087 store entries that we observe. We find that the elasticity of the healthfulness of household food purchases with respect to the density and nutritional quality of retailers in their vicinity is positive, but close to zero. Improving the concentration and nutritional quality of the stores in the average low-income and low-education neighborhood to match those of the average high income and high-education neighborhood would only close the gap in consumption by 1 to 3 percent.

If disparities in retail access do not account for the consumption disparities that we observe, then something else is to blame. We are agnostic as to the reasons why we observe systematic differences in the healthfulness of purchases between households either living in the same location or shopping in the same store. Further work is necessary to determine which factors are most important in explaining the large disparities that persist when we look at households in the same location.

Jessie Handbury is a Penn IUR Faculty Fellow and Assistant Professor of Real Estate at the Wharton School.