Food Deserts in Baltimore

Feast or Famine:  Exploring Food Deserts in Baltimore, MD

Abstract: An increasing amount of literature reveal that the suburbanization of food retailers in North America is creating urban ‘food deserts’, or areas of the cities that have relatively poor access to affordable and healthy foods.  This study explores the current ‘food desert’ condition in the city of Baltimore, MD using large scale grocery chains.  Geographic Information Systems (GIS) was used to map the precise location of supermarkets and analyze where the food deserts were with the network analyst extension.   We aim to discover if spatial inequalities in access to supermarkets exist.  We also gathered a list of vacant sites in order to remedy the situation and determine the two best plots to open up a grocery store.

Key Words:  Food Deserts, Baltimore, Food Security, Food Insecurity

Though society has become more effective in growing, storing , and distributing food from all over the world, there are still  people who do not have access to fresh, healthy food.  Food deserts are studied in areas, often with low socioeconomic status, that experience a shortage in food security.  These areas do not have sufficient access to large grocery stores to meet their common shopping needs. According to Larsen & Gilliland  (2009), a food desert is “a socially distressed neighborhood with relatively low average household incomes and poor access to healthy and affordable food.”
In areas where food deserts are present, the goal is to increase food security.  Food security has three components: accessibility, good quality, and reasonable prices.  All residents should be able to access a grocery store within proximity to their homes.  Not everyone has access to cars, especially the poor and the elderly.  Therefore, accessibility of supermarkets by walking and public transport is extremely important.  The food should be nutritious  so that people can live healthy lifestyles. Furthermore, the grocery stores also need to have affordable food.  Many  upper-end grocers have healthy food, but it is often expensive (Carlson, 2003).
This study addresses where food deserts are located in Baltimore, MD, and assesses where two new grocery stores are needed.  We focused on the city of Baltimore because it offered socioeconomic variability that is often associated with  food deserts.  For our study, we characterized  food deserts as neighborhoods that do not have a supermarket within a 20 minute walking distance.  Baltimore is considered one of the top walkable cities in the nation, and therefore aligns with our study parameters well.  In our search of two new grocery stores within a food insecure area, we were able to locate one in an area of high poverty and one in an area of wealth.

Literature Review:
The term desertification first arose in the 1990s and came to describe the effect on any area that was deprived of public and private service. Desertification usually occurs from retail consolidation as economies of size become ever more profitable and thus prevalent (Bitto et al., 2003, 35). Food deserts represent one such type of desertification. As larger supermarkets, wholesale clubs and discount merchandisers become more common, the distance between grocers’ increases. Longer distances require new methods of transportation, and disproportionately disadvantage those people without access to new transportation requirements (Bitto et al., 2003, 26). Examples of people disadvantaged by increasing distances to grocers include: the elderly, people with handicaps and people of low socio-economic status who are unable to afford increasing travel expenses. Though typically associated with rural areas, the prevalence of urban food deserts has recently gained attention in literature and research. Often synonymous with food insecurity, both rural and urban food deserts have become increasing associated with diabetes and obesity – as the lack of accessibility to grocers leads to unhealthy food alternatives such as vending machines, fast food, and convenience stores (Bitto et al., 2003).

Since its coining, research and literature on food deserts has become ever-present in a variety of journals. In particular, the passing of the 2008 Farm Bill by the U.S. government, which “directed the USDA to assess the extent of the problem of limited access”, has spurred research on food deserts (United States Department of Agriculture, 1). The USDA paper asserts the need to measure both distance and density in regards to food access.  Distance to closest facility is important in determining the time and travel cost associated with access, while density measures the diversity of grocer options and offerings – discount grocers, nutrition value, etc (USDA, 4).

In the Journal of Applied Geography, McEntee and Agymen (2010) introduce the use of GIS in identifying rural food deserts in Vermont. They assert that GIS has been integral in transforming the study of food deserts from a primarily qualitative approach to an empirically quantitative approach (166). GIS has been key in studying accessibility – categorized in this article as informational access, economic access and geographic access – especially in regards to saving-time in comparison to traditional hand and ruler mapping techniques (ibid). Though quantitative analysis is helpful in determining food deserts, McEntee and Agymen (2010) recognize the need for qualitative analysis in understanding the nuances of food deserts.

Using a qualitative approach, Smith and Morton assess rural food deserts in Minnesota and Iowa using seven focus groups (176).  Through focus groups they are able to draw connections between poverty and a higher incidence of inaccessibility to nutritious and available food. Social and cultural environments, the authors assert, are just as important factors in determining why a food desert exists as geographical environment. Social and cultural standards define how food is accessed (walk, bus, drive, taxi) and influence attitudes in regards to food issues (Smith and Morton, 180). The importance of nutritious food, diversity of options and low prices varied among communities, and contributed to perceptions of food (in)security. Lack of access to nutritious food – namely produce – was often discussed. Especially as low quality and high cost produce often turn shoppers to the reliable quality and lower cost of processed foods (182). By including social analysis, the authors can begin to unravel both causes and implications of food deserts.

The connection of food deserts to health is one effect that has gained ever-increasing attention. In a study from Rural Sociology, the authors examine the linkage between rural food deserts and overweight school children (Schafft et al., 2009). Using GIS to identify food deserts and comparing the zones with students’ body mass indices (BMI), the authors found a positive relationship between the population inhabiting a food desert and increasing rates of overweight children – a finding which has serious implications about spatial inequality.  Schools with a higher percentage of students living in food deserts were more likely to not only have more overweight students, but these students were also more likely to be economically disadvantaged (Schafft et al., 2009, 172).  Issues concerning the type and quality of foods available are key in understanding food deserts.

Block and Kouba (2006) address the differences between corner grocers and chain supermarkets in regards to food security in their study of two neighborhoods in Chicago. In their study the authors compare Austin, a lower-class African-American neighborhood to Oak Park, a bordering upper-class mixed race neighborhood (837). Austin has many more grocers and fewer supermarkets than Oak Park – an occurrence that is not uncommon. Due to consolidation of retail businesses as well as the movement of upper-class households to the metropolitan periphery, businesses have followed their client base to the suburbs in order to capitalize on economies of size.  Whereas grocers often offered produce at prices competitive with supermarkets, quality was often inferior. Packaged foods were usually more expensive at independent grocers than those purchased from supermarkets. The diversity of offerings also made chain supermarkets a better solution for food insecurity than independent grocers. In conclusion, the authors assert that “food access is related more to store type than number” (Block and Kouba, 2006, 837). Number of grocers or convenient stores can be misleading, therefore, in determining the prevalence of food deserts. Furthermore, the urban occurrence of quantity of grocers over quality supports other studies which link obesity with food deserts. A USDA article from the Center for Nutrition Policy and Promotion is useful for understanding what constitutes a balanced diet with a low, moderate and liberal food budget. This article presents a nutritious food basket – one that falls within the food pyramid’s allotments for carbohydrates, proteins, produce and fats.

While there is significant literature on rural food deserts, urban food deserts have received considerably less focus. Larsen and Gilliland (2008), however, do research urban food deserts. They focus their study on London, Ontario and use GIS to not only determine if food deserts exists rather also to compare the distribution of supermarkets in 1961 to that of 2005. Larsen and Gilliland (2008) set a 1000-meter walking distance and a 10-minute bus ride as the parameter for an accessible supermarket (10). Using network analyst, closest facility and service area were calculated. In comparing the city of London from 1961 to that of 2005, the consolidation of supermarkets and their movement to the suburbs becomes particularly apparent. The authors conclude that the incidence of urban food deserts has increased in the last half decade (Larsen and Gilliland, 2008, 17).

The Johns Hopkins School of Public Health has released two studies that chronicle the occurrence of food deserts in Baltimore. One project, Maryland Food Mapping System, examines grocery store locations in comparison with impoverished populations – defined in their study as earning less than $25,000 annually. An important aspect included in their study is the differentiation of grocers. Supermarkets, small grocers, corner grocers, convenience stores and behind glass corner stores are classified separately and their differing presences were analyzed for underlying trends. For example, there is a higher concentration of behind glass corner stores and convenience stores in low-income neighborhoods – a trend that classifies some areas as food deserts because of the lack of nutritional and affordable food.

A second Baltimore specific study was conducted by the Baltimore City Health Department. After mapping food availability and household vehicle access across the census tracts of the city, this study has led to the implementation of a Baltimarket, a virtual supermarket. In areas designated as food deserts, residents can use the internet at their local library branch to order nutritiously relevant food that is then picked-up from the library on the following day. (Baltimore CIty Health Department, 2010).

By reviewing this literature we were able to contextualize our own project and define our research question. We adopted many of the definitions, research parameters and GIS tools used in the aforementioned literature into our own tutorial and exercise.


GIS has been integral in transforming the study of food deserts from a primarily qualitative approach to an empirically quantitative approach (McEntee. & Agyeman, 2010).

In order to determine where food deserts were located in Baltimore and how they could be solved, data regarding grocery store locations, percent of poverty by census tract, and vacant lots over 40,000 square feet were obtained.  After manually gathering a list of large-chain supermarkets, grocery store addresses were georeferenced in ArcMap.  These supermarkets included Giant Foods, Mars, Whole Foods, Safeway, and Shoppers.  Then, poverty rates by census tracts were downloaded through the U.S. Census Bureau.  Baltimore census tracts were attained by ESRI and the poverty data was joined.   Lastly, we gathered vacant lot locations through various online sites and georeferenced their addresses to fit in the shapefile.

To determine the location of  food deserts, we used Network Analysis to measure distance from census tracts to supermarkets.    The “Service Area” function measured the area surrounding each supermarket that was within a 20 minute walking time. This function determines the scope of a facility’s accessibility. We used path analyses along the Baltimore street grid to account for walking along roads.  For our study we assumed that the average pedestrian walks at a speed of 80m/s.

We had to make a few decisions in order to determine the parameters of our study.  To quantify low income and high income areas, we chose to use the poverty rate, which is the percentage of people in the census tract living below the federally defined poverty line, to represent the socioeconomic status of the census tract.  This is not necessarily an inclusive representative of poverty.  Often previous methods have used a combination of factors, such as:  (1) low educational attainment (proportion of individuals that have not graduated from high school)  (2) lone parenthood (proportion of lone parent families versus the total number of families)  (3) unemployment (unemployment rate)  (4) incidence of low income (proportion of households that fall below the low income cut-off according to Statistics USA) (Larson & Gillilnad, 2009).  We felt that the federally defined poverty line would most accurately represent food deserts.

Additionally, we chose to use the centroid of a block group as a representation of each tract.  The Network Analyst is unable to operate compatibly with the polygons that represent the Baltimore census tracts.  Therefore, we created point data using the centroids of the tracts to run the “Service Area” function.  Block groups were considered to be food deserts if the center of the block group was more than a 20 minute walk from the nearest supermarket.

The parameters used for this study include food desert elements, which in this case is the distribution of poverty among census tracts, and walking speed.   Food deserts are described  as neighborhoods that do not have a supermarket within a 20 minute walking distance.  This must be calculated by walking along roads.  Census tracts that have above 40% of residents living in poverty are defined as low income areas, and tracts with less than 12% of residents living in poverty are specified as high income areas.

To determine appropriate locations for future locations within the food deserts, we searched online for available real estate.  We searched specifically for locations that were at least 40,000 square feet, the minimum suitable size for a supermarket.  The list we compiled contained the locations of nineteen different viable locations for supermarkets.  By using the “Select by Attribute” function in ArcMap we were able to select all of the supermarkets that were located in low income areas and high income areas.  The vacant lots located outside of the service areas of the current supermarkets were considered for the construction of a new supermarket.

Part I.

The findings indicate that there are sections of the city that are not served with large-scale supermarkets.  The southwestern half of the city experiences the most food insecurity.   Though these pockets may be served with smaller, convenience stores, the variety of healthy and affordable foods are not as prevalent without the large stores.   To find the percentage of Baltimore that is not served by grocery stores within a 20 minute walking distance, we added a field and used the calculate geometry function to find the area of the food desert census tracts.  After combining all of the food desert tracts, the combined area was divided by the total area of Baltimore.  Our study found that 86% of the city is not served by grocery stores within a 20 minute walking distance.

All of Baltimore’s large-scale grocery stores are located in middle to high income areas. Therefore, food security is positively correlated with middle to high income areas. This is a direct result of suburbanization that has taken place in Baltimore.  As white flight occurs, the capital follows, leaving the inner-city with disadvantages such as a lack of large-scale grocers.  Because grocery stores are businesses with the intention of making the largest profit, they do not have priorities to build stores within food deserts.     Residents of the suburbs usually have access to cars and can easily get to the supermarkets of their choice.  It is more common for inner-city residents to be without access to a car, leaving them at an even greater disadvantage.   However, the high rates of poverty and lack of grocery stores in the center of the city may be explained by the fact that this area is the business district and not many people live in these census tracts.

(Approximately 42% of low income areas and 62% of high income areas of the city experience food deserts.  This greater coverage of food deserts in wealthier areas of the city may be contributed to our limited methods, lack of smaller supermarkets or residents ability and reliance on cars. )

[insert poverty/ food deserts maps]

Part II.

Of the 19 vacant lots in Baltimore that have  a large enough area, over 40,000 square feet, to construct a new supermarket, thirteen of these are located outside the service area of a current supermarket.  There was a general equal distribution of the vacant lots throughout the city, though there were not many available vacant lots in the northeastern part of the city.  The thirteen lots are located in areas that do not have a large scale grocery store within a 20 minute walking distance from the census centroid.  Our research found that one vacant lot outside of supermarket service areas was in a high income area (2.5-12% poverty), two were in a moderate to high income area (12-18%), three were in a moderate income area (18-27%), five were in a moderate to low income area (27-40%), and two were in a low income area (40-66%).

[insert vacant lots map]


Food deserts provide information on inequality within urban and rural areas. Rural food deserts are studied to a greater extent than urban food deserts, however the problem of inner-city areas that are food-insecure are growing (Bitto et al., 2003).  In Baltimore, food insecurity is prevalent and policy makers are attempting to find solutions to ameliorate the situation.  In August 2008, researchers at the University of Maryland found that about 13.5% of low-income Baltimore families suffer from food insecurity.  In other words, due to financial constraints, these families have a lack of food access to live an active and healthy lifestyle.   Furthermore, recent studies by the Johns Hopkins Bloomberg School of Public Health have found that only 10% of Baltimore’s food stores are considered supermarkets.  In consequence, many Baltimore residents make major food purchases at neighborhood corner stores.  These stores often do not have healthy basics, such as fruits, vegetables, whole wheat bread, and low-fat milk available.  Even so, when these healthier items are offered at convenience stores, these food items are as much as 20% more expensive than they are at the nearest supermarket (John Hopkins, 2010).

Policies aimed at improving public health must recognize the socioeconomic inequalities with respect to convenient access to healthy and affordable foods.  The John Hopkins University is pursuing additional research on food desert in Baltimore.  Hopefully, their findings will be substantial and they can make an argument and resolution for how food deserts can be avoided.

Currently, one way in which the city is addressing these issue of food deserts is through the provision of online access to groceries.  Though it may be helpful to construct a store in one of the locations that we discovered, there is additional  hope as the city is attempting to solve food insecurity through a new strategy, the Virtual Supermarket Project.  This allows residents in and around East Baltimore and the Washington Village neighborhood to order groceries online at their local libraries and pick them up at the same locations the next day.  Mayor Stephanie Rawlings-Blake claims that “this program will make these neighborhoods stronger and healthier, allowing residents the same access to full-service, competitively priced grocery stores that much of the rest of the city enjoys” (Baltimore City Health Department, 2010).

     This study is a brief introduction to the food desert situation in Baltimore.  For further analysis, it would be useful to aggregate data to the census tract level for further analysis and compare socioeconomic characteristics of neighborhoods in more detail.  It would also be useful to use statistical analysis and further investigate specific neighborhoods.


“Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences”. USDA. pp. 1-10.

“Baltimarket: The Virtual Supermarket Project”. (2010). Baltimore City Health Department.

Bitto, E., Morton, L., Oakland, M. and M. Sand (2003). “Grocery Store Access Patterns in Rural Food Deserts.” Journal for the Study of Food and Society 6(2): 25-48.

Block, D. and Kouba, J. (2006). “A comparison of the availability and affordability of a market basket in two communities in the Chicago area.” Public Health Nutrition 9(7): 837-845.

Carlson, A., Lino, M., Gerrior, S., and P. Basiotis. (2003). “Revision of USDA’s Low-Cost Moderate-Cost, and Liberal Food Plans”. USDA, Center for Nutrition Policy and Promotion 15(2): 43-51.

Schafft, K., Jenson, E. and C. Hinrichs. (2009). “Food Deserts and Overweight Schoolchildren: Evidence from Pennsylvannia.” Rural Sociology 74(2): 153-177.

Center for a Livable Future. (2010). “Maryland Food Mapping System”. Johns Hopkins Bloomberg School of Public Health.

McEntee, J. and J. Agyeman (2010). “Towards the development of a GIS method for identifying rural food deserts: Geographic access in Vermont, USA”. Journal of Applied Geography 30(1): 165 – 176.

Additional Bibliography for Further Research of Sociology within GIS:

Crooks, A. T. “Constructing and implementing an agent-based model of residential segregation through vector GIS.” International Journal of Geographical Information Science 24.5 (2010): 661-675. Academic Search Premier. EBSCO. Web. 24 Oct. 2010.

Dennis Jr, Samuel F. “Prospects for qualitative GIS at the intersection of youth development and participatory urban planning.” Environment & Planning A 38.11 (2006): 2039-2054. Academic Search Premier. EBSCO. Web. 24 Oct. 2010.

Lohmann, Andrew, and Grant McMurran. “Resident-Defined Neighborhood Mapping: Using GIS to Analyze Phenomenological Neighborhoods.” Journal of Prevention & Intervention in the Community 37.1 (2009): 66-81. Academic Search Premier. EBSCO. Web. 24 Oct. 2010.

Kai A., Schafft, Eric B. Jensen, and C. Clare Hinrichs. “Food Deserts and Overweight Schoolchildren: Evidence from Pennsylvania.” Rural Sociology 74.2 (2009): 153-177. Academic Search Premier. EBSCO. Web. 24 Oct. 2010.

Perkins, Douglas D., Courtney Larsen, and Barbara B. Brown. “Mapping Urban Revitalization: Using GIS Spatial Analysis to Evaluate a New Housing Policy.” Journal of Prevention & Intervention in the Community 37.1 (2009): 48-65. Academic Search Premier. EBSCO. Web. 24 Oct. 2010.

Pritchard, H. W. (2010). Race, Class and Environmental Equity: A Study of Disparate Exposure to Toxic Chemicals in the Commonwealth of Massachusetts.70(07), 2738.

Stewart, Dona J., et al. “Assessing the Spatial Structure of Urban and Population Growth in the Greater Cairo Area, Egypt: A GIS and Imagery Analysis Approach.” Urban Studies (Routledge) 41.1 (2004): 95-116. Academic Search Premier. EBSCO. Web. 24 Oct. 2010.

Vaughan, L., Clark, D. L. C., Sahbaz, O., & Haklay, M. (2005). Space and Exclusion: Does Urban Morphology Play a Part in Social Deprivation? Area, 37(4), 402-412.

Yamashita, T., & Kunkel, S. R. (2010). The Association Between Heart Disease Mortality and Geographic Access to Hospitals: County Level Comparisons in Ohio, USA. Social Science & Medicine, 70(8), 1211-1218.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s