Reaching the limits: a geographic approach for

Reaching the limits: a geographic approachfor understanding food insecurity and household hungermitigation strategies in Minneapolis-Saint Paul, USA

Joel Larson • William G. Moseley

� Springer Science+Business Media B.V. 2010

Abstract Research on hunger and food security in

the Global South and the Global North has often

emphasized different factors and scales of analyses.

Unlike newer monitoring systems in the Global

South, which evolved substantially following cri-

tiques by Amartya Sen, US food security research has

rarely combined the two dimensions of food avail-

ability and food access. Furthermore, this research

has paid scant attention to household coping strate-

gies. This study responds to this lacuna in US hunger

research by developing a spatial model for predicting

risk to food insecurity based on proxy measures for

access (three demographic variables) and availability

(grocery store density). The study then employs

qualitative methodologies (surveys and semi-struc-

tured interviews) to understand household coping

strategies in two ethnically distinct areas in Minne-

apolis-Saint Paul at risk to food insecurity. One

neighborhood is dominated by Southeast Asian and

East African immigrants and the other by African-

Americans. This approach should allow for better

targeting of food aid and programs that help alleviate

food insecurity.

Keywords Coping strategies � Food security �Hunger � Spatial modeling

Introduction

Food security as a field of research has been

dominated by work done in the Global South. Work

on this theme in the US is more recent, even lacking a

measurable definition until the mid-1990s (Curtis and

McClellan 1995). The term ‘food security’ in the US

context is defined as ‘‘all people obtaining a cultur-

ally acceptable, nutritionally adequate diet, through

nonemergency food sources at all times’’ (US House

Select Committee on Hunger 1989, p. 4). Hunger

research in the US has also evolved more or less

independently from the more sizeable and older body

of food security literature focused on the Global

South. As a result, US hunger research may not be

capitalizing on all of the advances made in its

developing world corollary.

In creating models to predict areas of food

insecurity in the US, most research up to this point

has focused on one of two issues: availability of food

for sale within certain geographic areas (referred to as

accessibility in the US food security literature) and

people’s ability to access food in terms of ability to

pay (often referenced as demographic variables in the

US literature). As of yet, however, little, if any, US

based research has attempted to combine these two

aspects of food insecurity. Furthermore, this research

J. Larson (&)

University of Minnesota, Minneapolis, MN, USA

e-mail: [email protected]

W. G. Moseley

Macalester College, Saint Paul, MN, USA

123

GeoJournal

DOI 10.1007/s10708-010-9371-9

has paid scant attention to household coping strate-

gies (for which there is a well developed literature in

the field of developing world food security). This

study seeks to learn from advances in the Global

South and apply them to the study of hunger patterns

and dynamics in the US.

This study has four main objectives. First, to

develop a model that combines (1) measures of food

availability through access modeling techniques with

(2) people’s ability to acquire food as shown by

demographic proxy variables in Minneapolis-Saint

Paul. Second, to apply this model to the Minneapolis-

Saint Paul area to determine those areas at greatest

risk of food insecurity. Third, to use qualitative

methods (surveys and interviews) to determine how

residents at risk of food insecurity cope with that

danger. Fourth, to offer policy and program recom-

mendations based on the results of this model and the

insights on coping gleaned from interviews and

surveys.

Food security research in the Global South

and North

Research on hunger and food security in the Global

South and the Global North has often emphasized

different factors and scales of analyses. Hunger

studies in the developing world, or Global South,

historically focused on assessments at the national

scale. A number of national scale monitoring systems

were established following catastrophic famines,

particularly in Africa after major droughts in the

early 1970s and mid-1980s (Babu and Quinn 1994;

Quinn and Kennedy 1994). The UN Food and

Agriculture Organization’s food balance sheet

approach is a classic example of such national scale

monitoring (Maxwell and Frankenberger 1992). This

approach compares national food supplies (locally

grown food ? imports - exports) to national food

needs (per capita caloric requirements x national

population). Subsequent critiques of this methodol-

ogy by Sen (1981) and others (e.g., Jenkins and

Scanlan 2001) pointed out that national scale anal-

yses often missed more localized pockets of hunger

and conflated the availability of food with people’s

ability to access it.

As a consequence, other food security models

began to be developed that allowed for monitoring at

the subnational level. Examples of these approaches

include those that tracked a set of key indicators (Eele

1994) while others focused on the modeling of

household food economies (Seaman 2000; Moseley

and Logan 2001). These new approaches often took

account of food availability (i.e., is there enough food

available at the local, provincial or national scales to

meet the food needs of a given population) as well as

people’s ability to access such food. ‘‘Access’’ in this

context was conceptualized as people’s ability to

acquire food—and this was often measured along

several dimensions. Examples of these dimensions

include: whether or not households grew sufficient

amounts of food in a given year to meet their annual

needs; levels of household food stocks; proximity and

access to food markets; household monetary

resources available to purchase sufficient calories

given prevailing food prices; the availability of assets

that might be sold in order to purchase food; or other

non-market strategies for acquiring food.

A key aspect of this new breed of (or post-Sen)

monitoring strategies was attention to so-called

coping strategies. Coping referred to strategies

adopted by households in lean years to make up the

difference from food shortfalls. Many of these

strategies are included in the list of the monitoring

dimensions listed above (such as assets that could be

sold in order to purchase food), but also included

such factors as a household’s ability to collect wild

food, additional off-season employment, etc. A key

distinction between coping and regular food procure-

ment strategies is that the former are (typically) only

employed in years of food shortage. It should finally

be noted that most hunger monitoring efforts in the

Global South focused on the rural milieu as, histor-

ically, food insecurity was greatest in this realm.

More recently, attention has been given to monitoring

hunger in urban areas (Moseley 2001).

In contrast, research on hunger in the US has

typically been undertaken at more local scales as it is

assumed that sufficient food supplies exist at the

national level. Furthermore, there is no regular

hunger monitoring in the US as hunger is thought

to be a low level, chronic problem rather than an

episodic problem related to poor harvests (which is

reflective of the largely urban character of the US).

As such, most data on hunger in the US are collected

during yearly Current Population Surveys (CPS) and

the decadal national census. Academic researchers

GeoJournal

123

and the US Department of Agriculture (USDA) then

use this and supplemental data to understand hunger.

US food insecurity studies typically have focused

on two sets of factors. First, some studies have

concentrated on the availability of food in certain

neighborhoods by looking at the distribution and

density of governmental food programs and grocery

stores. Several scholars, including Guy (1983), Fra-

zier et al. (2003), and Clarke et al. (2002), use various

spatial accessibility models to predict communities

within urban areas that have fewer opportunities to

reach consumer goods. While not typically phrased as

such, what such studies are attempting to do is

understand the availability of food at different scales

of analyses (which is similar to market studies in the

Global South). Spatial units in the US with a paucity

of grocery stores have been referred to as ‘‘food

deserts’’ (e.g., Morton et al. 2005; see Wrigley 2002

for an example of this problem in the UK).

Second, the more common approach has been to

attempt to identify populations at risk of food

insecurity by focusing on demographic factors that

are often highly correlated with this problem, such as

race, household type, age and income. Here again,

while typically not phrased as such, these are proxy

indicators for people’s ability to access food through

purchase (which is similar to certain lines of research

in the Global South). Research to determine what

socio-economic groups are most at risk for food

insecurity have included efforts by the USDA’s

Economic Research Service (ERS) (e.g., Nord et al.

2002; Curtis and McClellan 1995). Results from this

work indicate that minorities, particularly African-

Americans and Hispanics, single females with chil-

dren, and households below the poverty line are more

at risk for food insecurity. While food insecurity

research has been conducted in urban and rural areas

in the United States, the majority of studies focus on

urban areas. In addition, there has been some research

modeling geographic availability to grocery stores

(US Department of Agriculture Economic Research

Service (USDA-ERS) 2009), but what US food

security research typically has not done is combine

these two dimensions: food availability and food

access.

Furthermore, very little attention has been paid to

coping in the United States. As these methods and

approaches are now frequently combined in the

developing world (Moseley and Logan 2005) and

the UK (Whalen et al. 2002; Wrigley et al. 2003;

Wrigley et al. 2004), it only seems appropriate that

these conceptual advances be applied to hunger

monitoring in the US. We finally note that, in this

very brief review of food security research in the

Global South and North, we clearly have omitted

large bodies of related scholarship. Perhaps the

greatest omission is the large body of literature

which examines the political economy of hunger.

Spatial access modeling

Related to the broad body of scholarship on food

security is a more specialized literature on spatial

access modeling. Spatial accessibility research on the

availability of food complements demographic stud-

ies by examining the physical environment, deter-

mining what areas of a city could be classified as

‘‘food deserts,’’ lacking in grocery opportunities

(Frazier et al. 2003; Guy 1983). Recent studies,

including Morton et al. (2005) and Gallagher (2007)

have that found residents living in food deserts not

only have to expend more time and energy to obtain

adequate nutrition, but they also have higher risks of

contracting diet-related diseases and other health

problems.

Geographic accessibility refers to ‘‘the inherent

characteristic (or advantage) of a place with respect

to overcoming some form of spatially operating

source of friction (for example, time and/or dis-

tance)’’ (Ingram 1971, p. 101). Put more simply, it

‘‘measures the potential interaction between places,’’

i.e., the attraction that one point places on another

(Frazier et al. 2003, p. 217). With regard to retail

activities, this attractiveness can be based on vari-

ables such as store size or population and can be

measured by time, distance, or cost (Song 1996).

General geographic or spatial accessibility can be

broken down further into two distinct types. The first

is relative accessibility, ‘‘the degree to which two

places (or points) on the same surface are connected’’

(Ingram 1971, p. 101). The second is integral

accessibility, the effort needed to overcome spatial

separation between many places (Frazier et al. 2003).

Each form has its own advantages and drawbacks and

must be considered together for each particular point

of interest. For example, relative accessibility can be

used to compare the shortest distance needed to travel

GeoJournal

123

from a house or block group centroid to obtain some

convenience good, such as gas, milk, or produce,

while integral accessibility can aid in viewing

systems as a whole, creating a larger picture of

access for a particular community.

A research project’s purpose and design should be

critically examined to determine what type of acces-

sibility measure to use for its particular question.

Utilizing an inappropriate method can lead to prob-

lematic and inaccurate results. Most problems arise

from accessibility measures’ nature as a process

versus outcome indicator (Guy 1983). Process indi-

cators are measures of a supply (or availability) in a

system, and are the goal and end-product of model-

ing. Outcome indicators show actual use and level of

satisfaction, which is not easily measurable through

theoretical modeling. Because of the lack of shopping

behavior data available in many locations, outcome

measures are rarely used and process indicators are

often substituted in their place. With the latter,

behavior and decisions made by the shoppers are

assumed and fixed, side stepping the need for such

information.

Because of accessibility’s nature as a process

indicator, there are dangers associated with treating it

as an indicator of actual use (Frazier et al. 2003). For

example, in a gravity model, the assignment of an

‘‘attractiveness’’ value based on size is not the

equivalent of an individual’s ability to travel to that

particular store. In other words, a store’s size and

attractiveness does not mean that all consumers can

afford to reach it (which relates to Sen’s (1981)

critique in the broader food security literature of

equating availability with access). Another problem

that presents itself when determining the integral

accessibility of a particular place develops from the

averaging of distances from it to other points of

interest. While this calculation is often done to allow

comparison between places (Ingram 1971), it can be

erroneous due to the presence of outliers. These

conditions aside, there are circumstances where

accessibility can measure equality of opportunity,

but not (as mentioned) the ability of households to

pay or their choices made under financial or time

constraints (Frazier et al. 2003).

Calculations of equality of spatial access can be

useful for comparing various neighborhoods within

the same urban area. To this end, empirical measures

of accessibility have been developed and range from

the simple to complex. Perhaps the most easily

understood is straight-line, or Euclidian, distance.

Most often calculated using Pythagoras’ theorem,

straight line distance can either be relative, covering

only two points, or integral, averaging the relative

accessibilities of that point (Frazier et al. 2003;

Ingram 1971). In 2009, a report by the USDA ERS

modeled geographic access based on straight line

distance to compare how income, race, ethnicity, and

several other factors related to distance to a major

grocery store or retailer. They found that low-income

and ethnic and racial minorities have better geo-

graphic access than the general public, but those

populations tend to live in denser areas than high-

income, white residents.

Another relatively simple measure is rectangular

distance, or the distance from the origin along a

rectangular, right-angle pattern (e.g., a road system).

This method may be more appropriate than straight

line distances when the rectangular nature of the

travel network causes significant differences from

straight line distances (Ingram 1971). A third simple

measure is time–cost distance, which determines

either the potential time taken to travel a specific path

or the equivalent in other monetary or non-monetary

means.

Several studies, including Ingram (1971) and

Clarke et al. (2002) have compared various geo-

graphic or spatial accessibility measures and found

that a Gaussian equation developed by Guy (1983)

was the most appropriate index when examining

grocery stores. Frazier et al. (2003) used this equation

to examine accessibility to grocery stores in two

counties in the eastern United States. Their findings

indicate that areas of minority concentration

(AOMCs), where more than half of the population

is non-white, had relatively poor access. In addition,

differential access within AOMCs was found, with

Hispanic concentrations having better access than

African American concentrations.

These previous studies use several methods of

measuring accessibility and have demonstrated that

the choice of index and parameters greatly affects the

end values. Keeping with these findings, it is

important that assumptions made through the choice

of the index and parameters be clearly understood

and stated. A thorough understanding of the study

area, including its population characteristics, and the

nature of consumer behavior with regards to food

GeoJournal

123

shopping will help in determining the best method for

calculating accessibility.

Study area

The study area is encompassed by the seven-county

metro area of Minneapolis-St. Paul, USA (hereafter

known as the Twin Cities area). The seven counties

represented are Anoka, Carver, Dakota, Hennepin,

Ramsey, Scott, and Washington (Fig. 1). The larger

region was selected because of the expansive urban

development of the Twin Cities, which has moved

out of the of Ramsey and Hennepin counties with the

growth of suburbs such as Chanhassen, Coon Rapids,

and Eagan. The grocery store industry has changed

with the growth of these suburbs, causing food

security throughout the region to change as well.

Demographically, the metro area has some char-

acteristics that are higher and some lower than US

national averages. With a population of slightly over

2.6 million, ranging from 70,000 in Carver County to

1.1 million in Hennepin, the region encompasses

over half of the population of the state of Minnesota.

Racially, the area is not as diverse as the US as a

whole, with 86.4% of the population identifying as

White alone or in combination in the 2000 Census,

compared to 77.1% nationally. When broken down,

the levels of Black and African-American and

Hispanic members of the population follow this

trend, with percentages of 6.9 and 3.6, respectively,

compared to 12.9 and 12.5% for the US as a whole.

The Asian population follows a different pattern,

however, with 5.2% of the population identifying as

such against an average of 4.2% for the entire

country. This change is due to the large population of

Hmong immigrants that have arrived in Minneapolis-

St. Paul in the past several decades (US Census

Bureau 2000).

In terms of monetary resources, all seven counties

have higher median household incomes than the US

as a whole, ranging from US $45,722 in Ramsey

County to US $66,612 in Scott County. Accordingly,

the region has a lower percentage of people in

poverty (6.9%) than the nation as a whole (12.4%),

although there are substantial differences between the

counties. The final demographic at a higher risk for

food insecurity, single female headed households, is

lower in the metro area (9.5%) than for the entire US

(12.2%).

Methodology

This research was divided into two major phases; the

first focused on using quantitative methods to deter-

mine which areas of the Minneapolis-St. Paul were

most at risk for food insecurity and the second relied

on qualitative techniques to help describe and under-

stand coping strategies among those groups at risk of

food insecurity. To determine which areas of the

Twin Cities were most at risk for food insecurity, it

was necessary to compile data from a variety of

sources and model what neighborhoods were most

vulnerable.

Fig. 1 Map of study area

GeoJournal

123

Food insecurity risk modeling

The two parts of the food insecurity risk model,

demographic risk factors (as a proxy for people’s

ability to acquire food) and accessibility, were

calculated and combined using ESRI’s ArcMap 9.1

software. The first and simpler half, the demographic

model, was calculated through the creation of a

demographic food insecurity risk index. The output

of this index calculates the magnitude of three

combined characteristics found to be positively

correlated with food insecurity: (1) poverty status,

(2) single female headed households, and (3) minor-

ity status (Nord et al. 2002). These data were

obtained from the 2000 US Census’s Summary File

3 data (US Census Bureau 2000). This information

was aggregated by tract level and expanded to the

county level for summary statistics. The scale of the

tract level was used for two reasons: (1) they are

small enough to identify various neighborhoods in the

Twin Cities, and (2) using divisions any smaller

would make it difficult to determine if any patterns

exist within the data.

The percentage of the population falling into each

of the three aforementioned groups was determined

as relative amounts (in the form of percentages) to

allow for comparison between tracts, eliminating

problems posed by varying tract populations. Each of

the above variables were then converted into an

index, which allows for the identification of the most

vulnerable populations, enabling their needs to be

addressed first. The three demographic groups were

then combined into an demographic risk index, which

constituted one half of the food insecurity risk index.

Once the demographic model was obtained, the

more complicated calculations required for the

accessibility index were performed. Spatial accessi-

bility to food resources was calculated by using

Guy’s (1983) Gaussian measure, which can be

written as:

Ai ¼X

Sj exp 1=2dij�d�

� �h i;

where Sj is the size of store j, dij is the distance

between origin i and opportunity j, and d* is the

distance from origin i at which accessibility declines

at the most rapid rate. For Guy’s (1983) study of

Reading, England, d* was set at 0.6 and 1.5 km, two

values between the maximum walking distance and

the distance most easily traveled by public transit.

Frazier et al. (2003) set d* at one mile due to the

automobile-oriented nature of the United States and

the minimum radii of primary trade areas for grocery

stores. Since this study was performed in the US, d*

was also set at one mile.

The three pieces of information needed to calcu-

late the above spatial accessibility equation were

grocery store addresses, property information (to

determine store size), and distance between origin

points and groceries. The addresses of grocery stores

in the Twin Cities area were found in online business

directories. The property data, containing square foot

data for the groceries, was provided by the Metro-

politan Council, an administrative body overseeing

some aspects of the governance of the Twin Cities

area. The final data required to determine accessibil-

ity around the Twin Cities were origin points to

determine distances to the groceries. A systematic

sample of block group centroids was taken, which

provided a distribution of points that correspond with

the population density of the Twin Cities area.

The last piece of information needed to calculate

spatial accessibility, the distance between the origin

and destination points, was was then determined. A

cutoff radius of one mile around each of the sample

points was used, to represent the maximum distance

that most people travel to obtain groceries. This

distance was chosen based on previous research (Guy

1983; Frazier et al. 2003) that determined that one

mile was the approximate minimum radius for

grocery store trading area and was a distance slightly

more than that which is normal walked but less than

that which is easily accessed through public trans-

portation. Spatial accessibility was then calculated for

the centroid sample using the Gaussian measure

above. Once the accessibility had been calculated for

each point, zonal analysis was conducted to deter-

mine the average accessibility value for each tract.

The accessibility results were then divided into an

index, to allow combination with the demographic

risk index.

The final food insecurity risk model involved

multiplying the two index values, demographic and

spatial accessibility, together. Creating a multiplica-

tive index caused the tracts with both high demo-

graphic risk and low accessibility to be magnified. In

this way, the areas at greatest risk for food insecurity

were highlighted, while still indicating those with

GeoJournal

123

moderate or low risk. The final index was then

mapped on the tract level, allowing for further

research to be conducted to determine how individ-

uals and families coped with being food insecure.

Determining coping strategies

In order to provide a more qualitative perspective on

the results of food insecurity and hunger, the second

research approach for this project included a series of

surveys (designed to spark informal conversations),

semi-structured interviews, and focus groups. Over

the course of a three-month period, 58 surveys and

interviews were administered to community mem-

bers, two food shelf coordinators were interviewed,

and a focus group of neighborhood residents was

convened. The surveys were designed similarly to the

US government-administered Food Security Supple-

ment of the Current Population Survey (CPS), which

is given every year to measure change between the

Censuses. By utilizing the same questions, it was then

possible to determine if the survey respondent was

‘‘food insecure’’ as defined by the CPS, and then

determine what coping strategies they used to miti-

gate their food insecurity risk. The supplement

includes questions about some coping strategies,

such as the use of government aid programs, while

the authors added questions regarding other strate-

gies, such as relying on friends and family. The in-

depth interviews were conducted with food shelf

coordinators to see if any larger trends were occurring

in the communities, and the focus group was utilized

to provide area residents a chance to compare

situations and provide information about area-wide

trends over longer periods of time.

Quantitative results

Combining the demographic risk and spatial acces-

sibility indices results in Fig. 2, showing which areas

of Minneapolis-St. Paul are at the greatest risk of

food insecurity. The highest areas of vulnerability

were in north and south Minneapolis as well as tracts

along the southern edge of St. Paul and several north

and east of the downtown. Several tracts that had high

demographic risk had their composite vulnerability

lowered due to their relatively high levels of access,

such as those in downtown Minneapolis. Without

examining the results of this model, intuition might

lead one to the conclusion that the central city

neighborhoods would have the highest risk for food

insecurity, but this is not the case here, where the first

ring neighborhoods surrounding the central business

districts are the areas most at risk. Another reason for

Fig. 2 Food insecurity risk in Minneapolis-St. Paul

GeoJournal

123

this result may be the increasing development drive

in downtown Minneapolis and St. Paul. Gentrification

and condominium construction is on the rise, leading

to a larger market for high-end grocery chains that

might not have located in these neighborhoods

previously.

More areas that have their composite risk reduced

by high spatial accessibility to grocery stores are the

neighborhoods directly south of downtown Minne-

apolis. In these predominantly minority tracts there

are high numbers of groceries able to serve the

population, increasing their accessibility result

(Fig. 3). The tracts with few groceries easily stand

out as having the highest risk, particularly in south

Minneapolis and north and southeast St. Paul.

The results of the food insecurity risk model are

not immediately intuitive, but they can be explained

through careful examination of the data that led up to

the index. The downtown Minneapolis area has

relatively high percentages of all three demographics

at risk for food insecurity, but it is also densely

populated. This large population leads to markets for

groceries that have established themselves in those

neighborhoods, increasing their accessibility.

In addition, many of the predominantly minority

neighborhoods have significant numbers of ethnic

groceries run by minority businessmen and women.

The exception to this trend is the Afro-American

dominated areas, particularly north and south Min-

neapolis. There are fewer groceries in these neigh-

borhoods, and few local business owners opening

new ones, so accessibility is not able to decrease the

aggregate insecurity risk like it does in other demo-

graphically vulnerable communities.

Initially, several areas of the Minneapolis-St. Paul

Metro were seen to have higher risks for food

insecurity. Upon greater examination of these areas,

it was discovered that, while being similar in their

high levels of risk for food insecurity, differences did

exist, particularly among the different ethnic groups

represented. While many of the areas of highest risk

were populated by African-Americans (North and

South Minneapolis, East St. Paul), some areas also at

risk were populated by immigrants, particularly

Southeast Asian and East African in the Midway-

Frogtown area.

Qualitative results

While two areas may have similar food insecurity

risk levels, their respective situations and coping

strategies may vary, particularly among different

ethnic groups. For that reason, the two areas that were

focused on (which were not necessarily the most food

insecure) were North Minneapolis, a traditionally

Fig. 3 Accessibility index and grocery stores

GeoJournal

123

African American community, and the Frogtown-

Midway neighborhoods of St. Paul, a destination for

many of the Southeast Asian, Somali, and Ethiopian

immigrants to the Twin Cities area. The initial set of

49 surveys and 19 interviews were conducted in the

Midway-Frogtown neighborhood in conjunction with

English Language Learner (ELL) classes taught by

the St. Paul Public Schools’ Adult Education Depart-

ment and by grassroots community organizations.

Out of all of the respondents surveyed, 23.7% were

determined to be food insecure, using the same

criteria as the CPS Food Security Supplement. While

this is more than double the proportion of the US

population that is food insecure (10.7%), it is

important to note that this is not a representative

sample and is likely biased toward the more food

insecure (Nord et al. 2002).

Respondents in the study area who were food

insecure tended to utilize both governmental and

private aid programs more often than food insecure

households in the entire US (see Table 1). The largest

differences were for free or reduced-cost lunch

programs, Women, Infants, and Children, and the

use of emergency kitchens. One potential reason for

this finding is that demand for such programs is

higher among urban minority households. If so, these

results provide evidence for those who wish to

increase both emergency aid and long-term hunger

reduction programs in these areas.

In addition to utilizing local, state, and federal

resources, the majority of those deemed food insecure

(50.0%) relied on friends or family for aid when they

did not have enough food. A question about this

behavior is not asked on the CPS food security

supplement, but was an addition of the authors, and

this high level of response indicates that it is an area

that would benefit from further research at the state

and federal level. Another common experience

among the Midway-Frogtown residents surveyed

was difficulty understanding aid programs when they

first arrived in the United States. Several respondents

indicated that they did not go hungry now, but during

the first few months after their arrival they had severe

problems affording food. They attributed this diffi-

culty to a lack of knowledge about federal programs,

little knowledge of English, and no local resources

such as friends for family to rely upon.

In the neighborhood of domestic minorities, North

Minneapolis, the nine interviewees indicated that the

greater concern revolved around the changing gro-

cery industry, instead of difficulties obtaining either

grassroots or government aid. As one resident

expressed, ‘‘There’s not a lot of possibilities, we

only have one grocery store [in the entire north

side].’’ Occasionally, a new store would move into

the area, but generally it would not last very long

before it closed down or moved out of the neighbor-

hood. These results are concerning, given that many

low-income households that participate in food stamp

programs shop only at supermarkets (US Department

of Agriculture Economic Research Service (USDA-

ERS) 2009).

An additional concern in North Minneapolis was

the lack of public transportation for residents. At one

grocery store ‘‘first, they took the bus shelter away

and now they are building townhomes there [after the

store was torn down].’’ Finally, there was significant

concern about the lack of grocery stores in the

neighborhood. Of the few grocery store points in

North Minneapolis, only one is a larger-size retailer;

the rest are convenience stores or small corner

markets. With only one stable grocery store in the

neighborhood and an inadequate public transportation

infrastructure, North Minneapolis could easily be

classified as a ‘‘food desert’’ lacking in any oppor-

tunity for its residents to purchase safe, healthy food.

Overall, the residents in the North Minneapolis

neighborhood seemed more concerned with physical

access to grocery stores and supermarkets, while

respondents in Frogtown-Midway were more often

worried about economic access to adequate food. These

findings echo those of Whalen et al. (2002), where

focus-group participants also differed in their percep-

tion of economic versus geographic accessibility.

Table 1 Coping strategies among the food insecure

Study areas

(%)

United Statesa

(%)

Food stamps 50.0 25.1

Free or reduced-cost lunch 71.4 33.4

Women, infants, and children 50.0 13.4

Food pantry 28.6 18.6

Emergency kitchen 12.5 2.8

Friends and family 50.0 N/A

a Nord et al. (2002)

GeoJournal

123

Coping strategies

Unfortunately, the small number of surveys con-

ducted did not allow for the separate study areas to be

broken out by type of coping strategy, but there were

several findings from the interviews that pointed to

both similarities and differences among the varying

ethnic groups. For example, many of the Hmong

families in Frogtown had personal gardens, a trend

that was also found among the African-American

residents of North Minneapolis. Additionally, farmers

markets were mentioned as places that people had

shopped at in the week before the interview.

There were also several unique strategies that were

utilized in one study area but not the other. For

example, co-ops had been a part of North Minneapolis

in the past, but had closed their doors due to lack of

funding. As one resident stated, ‘‘We had a co-op for

12 years and we moved it three times … the commu-

nity was not prepared because it was expensive, unless

you were a worker.’’ Another kind of organization

mentioned in North Minneapolis was food buying

clubs, which grouped people together to buy groceries

in bulk at reduced prices. ‘‘Years ago, I had friends that

were part of a food buying club,’’ said one Minneapolis

woman. ‘‘That is who the co-ops got started.’’ These

strategies match the increase in community food

projects noted by the US Department of Agriculture

Economic Research Service (USDA-ERS) (2009).

In the Midway-Frogtown neighborhood, however,

these two coping strategies were not mentioned. Instead,

several interviewees stated that they had family mem-

bers shop for them. ‘‘I don’t know how much we spend

on food. My brother does all of the shopping for me.’’

While one reason for this choice was cultural expecta-

tions, another that was mentioned was a better knowl-

edge of English and the grocery system in the United

States. A few Frogtown residents did most of their

shopping at local, immigrant-owned markets, due both

to similar language and food choices. As found in

Whalen et al. (2002), the proximity of extended family

was an important mechanism for reducing food insecu-

rity risk due to both economic and geographic barriers.

Food shelves

To try to develop a more comprehensive perspective

of those in need of food assistance, it was logical to

turn to food shelves themselves. An overarching view

of the shelves was provided through interviews with

two coordinators, one in North Minneapolis and

another in Midway-Frogtown. The information that

they provided was instrumental in creating a picture

about those who turned to emergency food shelves as

one of their coping strategies. By tracking the

participants in their programs, the managers were

able to develop a sense of who their clients were and

how to best serve them.

Economically, clients in both neighborhoods

tended to have fixed incomes, often through Social

Security or disability payments. One major reason

that was stated for coming to the county-funded shelf

was that the clients did not qualify for national

government programs such as Women, Infants, and

Children (WIC) or Temporary Assistance for Needy

Families (TANF). In both neighborhoods, the largest

ethnic group seen was African-Americans, although

both were seeing a rise in the number of immigrants

utilizing their resources. The food shelf in Midway-

Frogtown saw a disproportionately small number of

Hmong, Ethiopian, and Somali immigrants, but this

may be expected given that those interviewed seldom

indicated that they used these resources.

In addition to knowing about the clients that they

serve, the managers and organizers were well in

touch with both the problems that their patrons faced

and the larger issues present in trying to address food

insecurity and hunger. From both interviews and

internal materials distributed among food shelf

administrators, it was apparent that there is recog-

nition that there are more issues to solve than

meeting emergency food need. As Arnold (2004) has

stated:

The average community in the United States

already possesses and likely is already expend-

ing on hunger relief enough resources to end

hunger five times over, but likely is meeting

only about one-fifth of the need because of how

those resources are being utilized and

employed. (p. 1, emphasis added)

To address the issue of resource misallocation,

individual managers are moving away from a ‘‘band-

aid’’ model and attempting to address the root of the

problem. In their efforts, they are moving beyond

their own organizations by forming community

‘‘listening projects’’ to determine the needs of the

GeoJournal

123

neighborhood, talking with both state and federal

legislators, and reaching out to other food aid

programs to form networks and combine

resources.

With these changes being made within the food

shelf system, administrators realize that there are still

problems that are present for their clients. One

obstacle encountered by patrons even before they

reach the shelf is the difficulty in finding the facility

that serves where they live. In the 1970s and 1980s

food shelf funders required that their organizations

set up boundaries to prevent clients from jumping

from shelf to shelf and not being ‘‘productive’’

members of society. As shelves merged and split,

these boundaries changed, rarely reflecting neighbor-

hoods or taking public transportation networks into

account. This model is beginning to change, however,

and shelves are beginning to serve larger areas

without restrictive boundaries.

Another difficulty seen by food shelf coordinators

stems from cultural differences about food. As

mentioned before, the US House Select Committee

on Hunger includes in its definition of food security

as ‘‘all people obtaining a culturally acceptable,

nutritionally adequate diet’’ (1989, p. 4; emphasis

added). With a large immigrant community, the

food donated to shelves in the Twin Cities may not

be the food that is needed (or understood) by those

who are patrons of those shelves. One example of

this cultural misunderstanding occurred at a food

shelf in North Minneapolis, where a recently arrived

Hmong family left the facility with their box of

food, but, unsure of how to cook the boxed

macaroni and cheese, left it in the parking lot, only

taking the meat that they had been given. This

problem has begun to be addressed, as shelves are

employing native language speakers and holding

cooking classes to teach both native Americans and

recent immigrants how to properly store, cook, and

serve food that is available to them through shelves,

food stamps, and emergency kitchens. Other diffi-

culties faced by food shelf clients and administrators

include long waits for appointments and little

control of inventory, often resulting in surpluses of

unusable food. These problems are not discouraging

either patrons or organizers, however, and there

seems to be some progress made in how effectively

and efficiently hunger is reduced for the most

vulnerable households.

Conclusions and policy recommendations

This study demonstrates that both quantitative and

qualitative research methodologies may be combined

to determine where hunger exists, how food insecure

households cope with their situations, and what

strategies and programs are effective for individuals,

families and communities. Quantitative research,

such as the two-stage model used in this project,

can be used to determine what areas and populations

are at the greatest risk of hunger. Qualitative

methods, including surveys and interviews, help to

understand the causes and effects of food insecurity at

an individual and household level.

Strategies and behaviors that help households and

individuals mitigate the effects of food insecurity

included relying on friends and family, enrolling in

federal aid programs, and turning to local resources

such as food pantries and emergency kitchens. By

focusing on two ethnically distinct areas of Minne-

apolis-Saint Paul, it was also possible to determine

how different groups utilize coping strategies. Such

information could be used to design or ameliorate

programs for reducing hunger for these populations.

Based on the results of this research, we recom-

mend three strategies that could be used to identify

and address food insecurity in the US, and then

prevent these conditions from becoming chronic

barriers to healthy development and lifestyles. First,

combining two different ways of measuring food

insecurity (demographic risk and spatial accessibility)

into a single food insecurity risk index allows for

more efficient and accurate identification of areas and

populations susceptible to hunger. Second, encour-

aging food shelf and pantry programs to adapt to the

local cultural environment will make them more

successful both in the short and long-term. Adapting

to the changing demographics of US urban centers by

incorporating culturally appropriate food choices,

hiring staff who speak residents’ native languages,

and providing food storage and preparation classes

will allow existing food programs to better meet the

needs of those seeking their services.

Third, it is vital that administrators and policy-

makers at all levels of government understand that

hunger cannot be completely resolved through emer-

gency food programs. Government tax incentives to

encourage grocery stores to move into existing food

deserts may help alleviate spatial accessibility

GeoJournal

123

problems. In some case, food buying clubs (where

communities purchase basic food stuffs in bulk from

wholesalers and resell it to their membership) may

help reduce the price of food and thereby make it

more accessibly to low income families. Food

intervention programs that encourage supermarkets

to place new stores in food deserts have also been

found to change the behavior and accessibility of

residents in those areas (Wrigley et al. 2004). Finally,

job programs to improve household income will, by

extension, increase people’s abilities to purchase

food.

While the approach for studying food insecurity

adopted in this study is most appropriate for use in

urban areas of the US, it was informed by decades of

research on similar problems in the Global South.

The scourge of hunger in the Global South and the

Global North needn’t last any longer than it has to.

Scholars and policy makers working on this issue in

various parts of the world can only benefit from being

in dialogue.

References

US Census Bureau. (2000). Detailed data tables (SF-3): Min-

neapolis-St. Paul MSA. Available from: http://factfinder.

census.gov.

Arnold, J. (2004). Charity food programs that can end hunger

in America. Available from: http://www.endhungerin

america.org/EndHunger.pdf.

Babu, S. C., & Quinn, V. J. (1994). Food security and nutrition

monitoring in Africa: Introduction and historical back-

ground. Food Policy, 19(3), 211–217.

Clarke, G. P., Eyre, H., & Guy, C. (2002). Deriving indicators

of access to food retail provision in British cities: Studies

of Cardiff, Leeds, and Bradford. Urban Studies, 11(11),

2041–2060.

Curtis, K. A., & McClellan, S. (1995). Falling through the

safety net: Poverty, food assistance, and shopping con-

straints in an American city. Urban Anthropology,24(1–2), 93–135.

Eele, G. (1994). Indicators for food security and nutrition

monitoring: A review of experience from southern Africa.

Food Policy, 19(3), 314–328.

Frazier, J. W., Margai, F. M., & Tettey-Fio, E. (2003). Raceand place: Equity issues in urban America. Boulder:

Westview Press.

Gallagher, M. (2007). Examining the impact of food deserts on

public health in Detroit. Available from: http://www.mari

gallagher.com/site_media/dynamic/project_files/Detroit_

Food_Desert_Report.pdf.

Guy, C. (1983). The assessment of access to local shopping

opportunities: A comparison of accessibility measures.

Environment and Planning B, 10(2), 219–238.

Ingram, D. R. (1971). The concept of accessibility: The search

for an operational form. Regional Studies, 5(2), 101–107.

Jenkins, J. C., & Scanlan, S. J. (2001). Food security in less

developed countries, 1970 to 1990. American SociologicalReview, 66(5), 718–744.

Maxwell, S., & Frankenberger, T. (1992). Household foodsecurity: Concepts, indicators, measurements. A technicalreview. New York: UNICEF and IFAD.

Morton, L. W., Bitto, E. A., Oakland, M. J., & Sand, M.

(2005). Solving the problems of Iowa food deserts: Food

insecurity and civic structure. Rural Sociology, 70(1),

94–112.

Moseley, W. G. (2001). Monitoring urban food security in sub-

Saharan Africa. African Geographical Review, 21, 81–90.

Moseley, W. G., & Logan, B. I. (2001). Conceptualizing

hunger dynamics: A critical examination of two famine

early warning methodologies in Zimbabwe. AppliedGeography, 21(3), 223–248.

Moseley, W. G., & Logan, B. I. (2005). Food security. In B.

Wisner, C. Toulmin, & R. Chitiga (Eds.), Towards a newmap of Africa (pp. 133–152). London: Earthscan.

Nord, M., Andrews, M., & Carlson, S. (2002). Household foodsecurity in the United States, 2001 (Food Assistance and

Nutrition Research Report Number 29). Washington DC:

United States Department of Agriculture.

Quinn, V. J., & Kennedy, E. (1994). Food security and nutri-

tion monitoring systems in Africa: A review of country

experiences and lessons learned. Food Policy, 19(3),

234–254.

Seaman, J. (2000). Making exchange entitlements operational:

The food economy approach to famine prediction and the

RiskMap computer program. Disasters, 24(2), 133–152.

Sen, A. (1981). Poverty and famines. Oxford: Clarendon.

Song, S. (1996). Some tests of alternative accessibility mea-

sures: A population density approach. Land Economics,72(4), 474–482.

US Department of Agriculture Economic Research Service

(USDA-ERS). (2009). Access to affordable and nutritiousfood: Measuring and understanding food deserts and theirconsequences. Report to Congress.

US House Select Committee on Hunger. (1989). Food securityand methods of assessing hunger in the United States.

Washington DC: U.S. Government Printing Office.

Whalen, A., Wrigley, N., Warm, D., & Cannings, E. (2002).

Life in a ‘food desert’. Urban Studies, 39(11), 2083–2100.

Wrigley, N. (2002). ‘Food deserts’ in British cities: Policy

context and research priorities. Urban Studies, 39(11),

2029–2040.

Wrigley, N., Warm, D., & Margetts, B. (2003). Deprivation,

diet, and food retail access: Findings from the Leeds

‘‘Food deserts’’ study. Environment and Planning A,35(1), 151–188.

Wrigley, N., Warm, D., Margetts, B., & Lowe, M. (2004). The

Leeds ‘‘food deserts’’ intervention study: What the focus

groups reveal. International Journal of Retail & Distri-bution Management, 32(2), 123–136.

GeoJournal

123