Association for Public Policy Analysis & Management

Limited access to affordable and nutritious food, such as a wide array for fruits and vegetables, whole grains, and low fat dairy, may be correlated with poor diet and health outcomes.  Individuals who live in food deserts—areas where access to sources of these foods may be limited—may face greater time and travel costs to access retailers that sell these foods at affordable prices and thus, consume less of them.  While there is fairly consistent evidence that poor access to supermarkets is correlated with poorer diets, the causal evidence is mixed,  perhaps at least partially because standard estimates of the effects of the food environment may be biased because retailer and consumer location choices are endogenous.

In this paper, we use two rounds of data on local food environments from the Economic Research Service’s (ERS) Food Access Research Atlas (FARA) and two rounds of small area estimates of fruit and vegetable intake and obesity rates derived from the Behavioral Risk Factor Surveillance System (BRFSS). These data allow us to estimate the effect of changes in access to healthy food retailers on changes in diet quality, the share of population that is overweight and the share of the population that is obese.

Data on store locations from 2006 and 2010 are derived from two independent directories—TDLinx and stores authorized to accept Supplemental Nutrition Assistance Program (SNAP) benefits. Data on census tract characteristics—share of population below the poverty rate-- come from the 2005-09 (2007) and 2008-12 (2010) American Community Survey (ACS) data. The same years of ACS data are used in conjunction with 2007 and 2010 BRFSS data to generate small area estimates of obesity rates and the share of adults consuming at least one serving of fruit or one vegetable a day. We first estimate the cross-sectional relationship between access and diet/health. We then use the two years of data on food access, population characteristics, obesity rates and food consumption and first-difference estimators to measure the effect that changes in food access has on changes in diet quality and obesity rates.

Comparing cross-sectional estimates to first-difference estimates shows the extent to which cross-sectional analysis exaggerates the relationship between access and diet quality or health outcomes. We then use our first-difference estimates to simulate the extent to which improving food access may lower obesity rates and improve consumption of fruits and vegetables. We also simulate how changing poverty rates alone, without changing access, could impact obesity and diet quality and explore whether changes in other neighborhood characteristics, such as education, interact with changes in access. Results will provide insights into which types of policies may be better-suited to improve diet quality and body weight—those that address food store access, such as the Healthy Food Finance Initiative, or those that address demand side factors such as income and education.

Disclaimer: Findings and conclusions in this paper are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.