Characteristics and Influential Factors of Food Deserts
by Paula Dutko, Michele Ver Ploeg
, and Tracey Farrigan
Economic Research Report No. (ERR-140) 36 pp, August 2012
What Is the Issue?
USDA's Economic Research Service previously identified
approximately 6,500 food desert tracts in the United States based
on 2000 Census and 2006 data on locations of supermarkets,
supercenters, and large grocery stores. These food deserts are
areas where people have limited access to a variety of healthy and
affordable food. As policymakers consider interventions to increase
food access, it is important to understand the characteristics
associated with these areas, such as income, vehicle availability,
and access to public transportation. In this report, we examine the
socioeconomic and demographic characteristics of these census
tracts and also examine which of these characteristics distinguish
food desert tracts from other low-income census tracts.
What Did the Study Find?
• Areas with higher levels of poverty are more likely to be food
deserts, but for other factors, such as vehicle availability and
use of public transportation, the association with food desert
status varies across very dense urban areas, less dense urban
areas, and rural areas.
• Areas with higher poverty rates are more likely to be food
deserts regardless of rural or urban designation. This result is
especially true in very dense urban areas where other population
characteristics such as racial composition and unemployment rates
are not predictors of food desert status because they tend to be
similar across tracts.
• In all but very dense urban areas, the higher the percentage
of minority population, the more likely the area is to be a food
• Residents in the Northeast are less likely to live far from a
store than their counterparts in other regions of the country with
similar income levels.
• Rural areas experiencing population growth are less likely to
be food deserts.
How Was the Study Conducted?
To provide a consistent, national-level estimate of the number
of low-income areas in which a substantial number or share of
residents is far from a supermarket or large grocery store, USDA's
Economic Research Service applied a census tract-level definition
of food deserts-areas with limited access to affordable and healthy
food-to the contiguous United States using 2000 Census data. The
2000 Census data and 2006 store location data that were used for
this analysis were the most recent demographic and store data
available at the time this analysis was conducted.
This study uses data from the 1990 and 2000 Census, as well as
5-year average data from the 2005-2009 American Community Survey
(ACS), to describe changes in characteristics of the 6,529 food
desert census tracts over time, relative to changes in all other
tracts. We focus particularly on population density, poverty rates,
unemployment, education, race/ethnicity, income, and vehicle
We first provide a statistical description of tracts classified as
food deserts versus all other tracts to give a broad image of how
food desert tracts differ. We then conduct regression analysis to
determine which characteristics are most strongly associated with
whether a low-income census tract is also a food desert. We model
the probability that a census tract will be a food desert using a
multivariate logit model to assess the impact of factors such as
population and housing characteristics; racial and ethnic
composition; unemployment; poverty; and changes in these
characteristics from 1990 to 2000. Separate analyses are performed
for urban areas and rural areas in order to accommodate different
definitions of food deserts and systematic differences in tract
characteristics between rural and urban areas. We also further
distinguish very dense urban areas from less dense urban areas for
the multivariate analysis.