The second session of the workshop focused on contextual factors related to child food insecurity and hunger. In addition to geography, the session considered the political environment, or how local and state governments use existing resources to address child hunger; the food environment, such as food costs, food deserts, and alternative food outlets and the role they play in potentially moderating impacts of poverty and other factors; and economic context, such as the role of wages, unemployment, housing costs, transportation, and the broader safety net. The session provided an overview of what is known and focused on identifying important research questions and data needs going forward. The moderator was Susan Parish, Brandeis University. The main speaker was Scott Allard, School of Social Service Administration at the University of Chicago. Lucia Kaiser, a cooperative extensive specialist in the Department of Nutrition at the University of California, Davis, served as first discussant. The second discussant was Bruce Weber, professor of Agricultural and Resource Economics, Extension Economics, and director of the Rural Studies Program at Oregon State University.
STATEMENT OF SCOTT ALLARD1
Allard said the purpose of his presentation was to describe why place matters in the context of food security. He reviewed key terms and defini-
1Allard (2013) prepared a commissioned paper for the workshop, which formed the basis of his presentation.
tions and discussed possible causal pathways related to place, along with a summary review of the literature. He concluded his presentation by describing methodological challenges, prioritizing next steps, and suggesting discussion questions that he said were intended to seed later conversations.
Allard noted considering how place and contextual factors relate to food security and hunger is similar to an issue of supply and demand. Supply issues relate to where stores, retailers, and food assistance providers locate. Demand issues relate to how food needs and preferences are spatially distributed within a community. One open question is whether food security is the right outcome variable for thinking about place effects, as he said place and contextual factors might matter in both direct and indirect ways. Most of the studies in the literature do not address child food insecurity and place effects, which he termed a data limitation for two reasons. First, there are very few, if any, studies that can locate child food insecurity in space and model place effects. Second, the lack of data on food security that can locate individuals or households in space is a major limitation confronting all research on food security and place effects. Part of the lack is sample size and part of it is how the problem has been conceptualized.
The poverty literature has addressed why place matters, but it is difficult to unravel place effects from self-selection issues. Food behavior and food outcomes have a spatial component, and spatially correlated phenomena may have a causal component. Food resources and food assistance are located and embedded in space. Where a person lives affects the kinds of grocery stores and kinds of food assistance programs that might be nearby.
Although much of the research has been focused on food deserts, other aspects of place matter as well, as he detailed in his background paper (Allard, 2013:1). Understanding how spatial context shapes food security can mean a number of things, he suggested, and could mean an improved understanding of household and child food security. For example, around the same time that the food security measure was developed, there was a discussion about community food security measures, with a community food security concept embodied in a tool developed by the Economic Research Service (ERS).2 Understanding how communities are food secure might help establish long-term household and child food security with the idea that food assistance programs are short-term solutions to these issues. Understanding place could also provide better insight into how individuals and households cope, as well as provide
insight into better models, which might result in better allocation of public and private resources. Allard said better conceptual and empirical understandings of place might support the development of better interventions and more dynamic solutions.
Allard provided context for commonly used terms when looking at the relationship between food insecurity and place. First he described the food insecurity measure (see Chapter 2). He went on to say that research on place is usually specific to a given city or community, such as a study of New Orleans or a study of New York City. The unit of analysis may be a state, county, municipality, tract, or neighborhood. Some studies focus on state differences or county differences. Much research uses tracts as proxies for neighborhoods or block groups. He went on to describe some of the measurement issues related to the terms distance, buffers or catchment areas, food resources, access, and nutrition.
Allard noted four possible pathways through which place matters: (1) spatial proximity to food retailers; (2) safety net programs; (3) political and policy environment; and (4) food prices and economic conditions. Different studies tend to emphasize one of these areas, although there are not strict boundaries between them. He encouraged thinking about the role that self-selection plays: how people get to where they are, and how spatial measures of food access or food resources can help resolve the endogeneity issues that emerge from self-selection as to the neighborhoods people live in and the programs they participate in.
Spatial Access to Food Retailers
The most prominent component of the literature, and the dominant way that scholars, policy makers, and advocates seem to think about place and food security, is spatial access to food retailers, access to community food resources, the food resource infrastructure, or food deserts. Allard said there are many reasons to be concerned about spatial access to food retailers. The closer a person lives to a store, the lower the commute costs, and the lower the time costs. Supermarkets may be better than grocery and other types of smaller stores in that they are more affordable, have greater selection of food, longer hours of operation, and healthier food. He noted that community food security is considered a long-term solution to food insecurity, with food assistance being a short-term way to help families make ends meet.
Food retailer data come from proprietary sources, including Dun and Bradstreet and InfoUSA. Supplemental Nutrition Assistance Program (SNAP) retailer data are also available from the U.S. Department of Agriculture (USDA). Unfortunately, Allard noted that these datasets are neither comprehensive nor consistent. Some studies have compared food
retailer databases, with the finding that 30 percent of food retailers are missing from one list versus another. The source of data used to model food access matters a lot, and he said a substantial effort is needed to clean the data.
Allard said there is no agreement about what is adequate or inadequate in terms of spatial accessibility and a lot of variation in how people measure accessibility. A number of studies, for example, look at which retailers are located in a given census tract or ZIP code. One result is that there can be many different measures that might not compare very well across studies. Sparks et al. (2011) developed a number of different measures in Portland and found that different measures within the same community are highly correlated. Hence, depending on the questions being asked, some of this variation in measurement of access might not matter as much in the context of a single community or city. However, it certainly makes it difficult to compare New York to New Orleans or Chicago.
A large number of studies find race and class differences in access to food retailers. Much of this is based on distance to the nearest supermarket, the average distance to the nearest three supermarkets, or the number of stores in a census track or a ZIP code. Gallagher (2006) found that African American neighborhoods in Chicago are 40 percent farther from the nearest chain grocery store than are white neighborhoods. This is a very common result in comparisons between black and white neighborhoods, poor and nonpoor neighborhoods. However, other studies show small or no differences by race and class. In an ERS report about this literature, Ver Ploeg et al. (2012) found some evidence of race and class differences, but also evidence with mixed results. Some studies find low-income and minority communities have greater access to supermarkets, while there are also some studies that show that nonchain grocery stores or specialty stores, particularly ethnic groceries, are more accessible to low-income and minority communities.
Bartfeld, Ryu, and Wang (2011) showed that only very long distances to grocery stores are related to food insecurity. Few studies can compare access measures to food security outcomes, in part because most of the surveys that provide the food security measures are national in scope. Thus, noted Allard, it is unclear how access affects behavior, program participation, or diet.
There are many assumptions about distance, mode of commuting, and type of store. The vast majority of households are within one mile of a supermarket. Most poor households use a car to get to the grocery store, whether it is their own car or a borrowed car. This is true of food pantries as well, especially outside central city areas. Allard urged challenging what he called common assumptions that all people in poor neighborhoods are poor, all people in poor neighborhoods do not have a
car, and that people can only walk to the grocery store. Allard said that as poverty in the suburbs has increased, there are more poor people in suburbs now than in cities. As a result, the issues of food access and food security have become a big unknown because much of the research focus has been on urban centers or rural places. Little research has been done in suburban areas.
Areas that are deeply segregated by race and class often have the biggest gaps, Allard said. A mix of approaches, a mix of data, and a mix of sites lead to a mix of results. However, this is not yet being linked to the food security questions. Assumptions are made about poverty rates, and what it means to live in a high-poverty tract. A lot of the research is cross-sectional, so establishing causality is difficult. Much of the research looks like asset maps, without connections to behavior.
Another question raised by Allard is how to assess statistical significance to decide if a result is substantively significant. For example, a 40 percent difference in access between black and white neighborhoods in Chicago is a difference between being six-tenths of a mile to the nearest supermarket and four-tenths of a mile to the nearest supermarket. He said research is needed concerning the meaningfulness of such significant differences observed in terms of their impact on people.
Allard asked how measurement could be improved, noting that ready access to GIS software may support better measures that can also be linked to individual and household outcomes. There are important at-risk areas and groups that deserve focus. Spatial access to food is a phenomenon that matters, he said, but it may matter more acutely in certain places.
Safety Net Programs: Public and Private
Allard discussed safety net programs and providers, noting mismatches and gaps between the location of providers and the location of people in need. Safety net assistance matters because it increases household food budgets. Social service programs can help improve well-being and work earnings. They provide emergency assistance that helps families navigate job loss or periods when they might risk becoming food insecure. He characterized food pantries and religious congregations as first responders that can provide food as well as legal aid, employment, and other services. Having such an organization in a community instead of living in a place without that kind of asset potentially matters for many reasons, he said. Proximity to assistance programs is likely to be related to increased take-up. If so, then proximity to assistance programs is likely related to food security—whether a direct or indirect pathway.
He noted how different services and programs are bundled is also important, referring to a discussion in the literature that density of pro-
grams can lead to greater collaboration among community-based organizations. He noted a threshold or tipping point concerning the density of providers: When there are enough providers, the supply of advocacy and intervention or referrals and awareness is greater, and households are better served.
There is also reason to think that formal childcare centers and access to formal childcare matter, especially as they relate to what children eat during a day. He cited research by Kissane (2010) and Neckerman et al. (2009), who considered how people navigate place in the context of safety net programs as it relates to straight-line distance versus street grid. Both of these studies talked about the built environment, how people engage their environment, and concerns about stigma. A person might not go to a food pantry in her neighborhood because she does not want to have people see her in line or might have concerns about safety and violence or about issues of race and ethnicity. A person might not go into a neighborhood because people are different or because it is a different gang’s territory. Qualitative studies can educate about these types of conceptual pathways. He said there is not much work focused on the presence or absence of safety net programs or the role of providers.
There is reason to think that if food-insecure households are enrolled in food assistance programs, they will do better over time. Bartfeld and Dunifon (2006) looked at access to food assistance and found it is positively related to food security, and Nord and Golla (2009) looked at SNAP and food insecurity correlation over time.
Not much research has been done concerning food pantries, although they play a big role in areas of high poverty or with individuals who have persistent detachment from the labor market. Allard said he has found that, consistent with some supermarket access research, low-income neighborhoods have about half as much access to emergency assistance food providers as other neighborhoods (Allard, 2009a). He observed empirical research on the spatial location of food programs and food assistance lags behind what people on the ground know, and he noted that talking with someone who is running a grocery store or a food pantry in a poor neighborhood will provide a story that is not present in the literature. He pointed to some promising use of SNAP administrative data to study what people buy and where they buy it.
Allard suggested that research is needed into how receipt and bundling of assistance shapes food budgets and food shopping behaviors, how social service programs can be integrated more explicitly, how food security and program participation are related, and the causes/consequences of exiting from programs. Because store locations are driven by supply and demand and by the built environment, he said, economic conditions and how they shape the location of food resources are important.
He went on to say that there is not a good understanding of why or whether food deserts exist, but researchers are interested in food prices and cost-of-living measures by geography. Labor markets, housing costs, and energy prices all vary by geography. He said that researchers have expressed some interest in informal social support and increasing interest in civic structure and social capital. Where stronger, he suggested, a civic community with less crime and violence and more safety might lead to better health or food outcomes. However, without local-level data, the literature focuses mostly on economic issues.
Prices tend to be lower in supermarkets and supercenters, but there is much regional variation. Some studies find that ethnic grocery stores provide affordable healthy food options in low-income neighborhoods, but these stores are not widely accessible to all low-income neighborhoods. There is evidence that unemployment rates and wage rates affect food security. Allard noted that—in one study—response to the question “how strong do you think the community organizations around you are?” was positively related to food security. He observed that social support is a more prominent coping strategy in rural areas.
Many economic and social factors are highly correlated with each other and with other important variables, and teasing out what matters is difficult. He suggested that the next steps are to understand what creates gaps and mismatches in access, answering these questions: Why do food deserts exist? Can prices and price variation by place be connected to shopping behavior? Can effects of context over time be modeled? And, can measures of civic community and safety concerns be included?
Political and Policy Environment
Allard noted important political and policy variables might matter, but this is a relatively underdeveloped area of the literature. He recommended identifying key causal pathways, working on developing better measures, and establishing a direct connection to household or individual food security, and he pointed to some promising ideas to examine how policy varies at the state and local level and the effect of these variables.
Food Prices and Economic Conditions
Although there is no robust theory of place effects and food behavior, he said that careful thought about causal mechanisms, adequate access, access shifts over time, and the relation of those shifts to food outcomes is important. A lot is known about how poverty varies by place, but available data do not allow very well for the study of food insecurity by place. Moreover, food deserts can be identified but not always why they exist.
Allard said there are serious data limitations in the grocery store data and individual household-level data. He recommended including the food security questions in more panel surveys to obtain geographic coverage.
He reminded the audience that the processes are different in urban, rural, and suburban areas, and models and theories are unlikely to work well in all places. There are endogeneity and self-selection issues with individual and household outcomes, and place variables might help to account for them.
Research on how to link data on place to different food behaviors is important: how people buy food, where they buy it, and for how much. He said this information needs to be connected to stores and the environment around them. Allard suggested natural experiments and behavioral economic experiments, using technology and “big data” to the extent possible to generate new ideas. He noted that as technology is able to scrape data and pull data together from different and unique places, there might be some opportunities to find new sources of data and new measures that would not require panel surveys or that would allow 80 percent of the answer obtained with 20 percent of the investment.
STATEMENT OF LUCIA KAISER
Kaiser began her comments by asking if the concept of place is the “silver bullet.” She said she would re-visit the concept of causal pathways, take a closer look at food deserts, and provide her thoughts about reaching hard-to-reach populations.
Kaiser reminded the audience about Allard’s four causal pathways: (1) spatial proximity to food retailers; (2) safety net programs; (3) political and policy environment; and (4) food prices and economic conditions. Related to spatial proximity, she encouraged research on mixed methods that combine GIS data with other measures, as well as perceptions of proximity. In terms of the safety net, she said analysis needs to decouple the location of the application process from the location of receipt of the service, or “program delivery.” She reminded the audience that immigrant populations in particular have many linguistic and trust barriers to the application process, noting that these can be managed by community organizations. She also noted that application for and receipt of benefits may be in different places. She said that in the political and policy environment, state and local policies have a significant impact on local access and food availability and new methods should be piloted before being
rolled out. Finally, concerning food prices and economic conditions, she observed that seasonality is a key issue for immigrant populations. Many immigrants are farm workers, implying both seasonality in work and in availability and prices of food. Some of her qualitative work has demonstrated that these fluctuations may mean an abundance of food at certain times and, at other times, not enough, which affects parenting strategies and the way children are nourished. She said that other types of populations engaged in seasonal work might have similar issues.
Kaiser questioned the focus on food deserts. She noted that few studies have compared perceived and objective measures of distance to grocery stores and supermarkets. From a local perspective, this is very important and opens the door to mixed-methods studies. To implement strategies that will help prevent food insecurity, she said it is important to understand how people view the distance to stores.
She cited a study (Caspi et al., 2012) conducted among 20 low-income housing projects in Boston, which found a mismatch of 31 percent between actual measured distance by GIS and the perception (from survey data) that the supermarket was within walking distance. Perhaps, she said, those two different measures tap into different constructs or there is an unmeasured individual or place-based characteristic. This study suggests that GIS data need to be augmented from another point of view. To describe the relationship between measured and perceived distance to the supermarket, Caspi et al. (2012) used as an outcome variable the number of servings of fruits and vegetables. The authors found that, after controlling for food insecurity, income, age, gender, and country of origin, the measured distance from the housing sites to the supermarket was not significantly related to servings of fruits and vegetables, but that the perception that the supermarket was within walking distance was very much related to it.
In 2006, Kaiser and colleagues collected 24-hour recalls of food receipts from 117 low-income women living in four counties in California. In a recent analysis of that data,3 they found educational and motivational factors play a role in diet quality as measured by the Healthy Eating Index.
Andreyeva et al. (2012) looked at the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) stores and non-WIC stores in Connecticut before and after the 2009 implementation of the revised WIC food packages that provided online vouchers to purchase fruits and vegetables, whole grains, and low-fat milk products. The study
3See Kaiser et al. (2006).
considered the question “Can food assistance policies change the local availability of healthy foods?” The study found that WIC stores showed greater improvement in availability of healthy foods after implementation of the WIC food packages. Stores that were non-WIC had some changes, probably because there were more providers (e.g., of whole grain bread) going into those areas, but there was a much greater effect on the WIC venders in terms of healthy supplies of foods at their stores. There was also a greater positive effect on food supply for stores that were farther from supermarkets. This study demonstrates that changes in food assistance programs can have a local impact on the variety and the quality of foods that are available at stores. Kaiser observed pilot studies of policy changes prior to implementation may be valuable. She speculated about what a pilot study would have demonstrated if it were conducted before the new WIC food packages were rolled out.
Reaching the Hard to Reach
Kaiser discussed the role of community-based participatory research studies in better understanding hard-to-reach populations, drawing on her experience as a member of the Niños Santos, Familia Sana (Healthy Children, Healthy Family) project, a five-year project funded by USDA to prevent childhood obesity. The project involves, among other things, administering the 18-item Household Food Security Survey Module (HFSSM) twice a year over five years in a very high-risk population.
About 90 percent of this population is immigrants, and many of them are undocumented. Her team is working with community advisory groups to reach this population. The group uses lay people from the communities as recruiters, who have helped the team in many ways, including testing questionnaires and techniques. She characterized the project as a long-term commitment that involves building trust. She showed some preliminary data that indicate that in this population about 13 percent have very low food security and only 36 percent are food secure but underscored they are preliminary data and the figures may change.
In conclusion, Kaiser stated that mixed-methods studies, not just GIS, are very important to understand place-based issues. She added that pilot studies of innovations in food assistance programs might help researchers to understand program impacts. Community-based participatory approaches to working with the hard-to-serve require a long-term commitment and may involve new research teams but will lead to a much better understanding of these populations. Her final suggestion was to consider small case studies to look at possible mechanisms, in addition to larger panel data studies.
STATEMENT OF BRUCE WEBER
Weber said he wanted to make the concept of “place” more concrete and consider it in the rural/urban continuum as a set of places. A person can insert his or her own place in this continuum to get a sense of why it matters. He said he would suggest research opportunities along two of the pathways that Allard identified and end with an argument to include measures of place or indicators of place in research that involves individuals to support understanding about how place affects individual outcomes.
He observed that a map of the United States produced by Feeding America, displaying child food insecurity estimates by place,4 shows how food insecurity among children varies by place, with high concentrations of childhood food insecurity in the Southwest, in the South, in some of the southern counties along the Mississippi River, and in Appalachia. What is more surprising, he said, is that some western states have high levels of child food insecurity in places that do not have particularly high poverty rates.
Pointing to Table 4-1, which displays aggregate measures of food insecurity in all counties, metropolitan counties (split into principal cities and outside principal cities), and nonmetropolitan counties, Weber described how food insecurity and child food insecurity vary across the rural/urban continuum. Food insecurity rates are higher in nonmetropolitan than in metropolitan counties across all of these variables. However, the places that have the highest food insecurity and the highest child food insecurity are the principal cities of metropolitan areas. Weber asked why these differences exist and what characteristics of these places and the people in these places explain the differences in food insecurity and child food insecurity.
Weber stated one of the roles of policy in reducing food insecurity is to change the economic context so households can develop their capacities and earn sufficient income so that they are not food insecure. The major policy player, he said, is macroeconomic policy and all other policies that affect the ability to earn income. Some policies, including local policies, make place-specific investments. These local investments, which are usually physical investments, affect the opportunities that exist in those places.
In addition to changing the economic context, he said policy also needs to provide a safety net, including many of the safety net programs discussed later in the workshop by David Ribar (Chapter 6), for those for
4Available: http://feedingamerica.org/hunger-in-america/hunger-studies/map-the-mealgap.aspx [August 29, 2013].
TABLE 4-1 Child Food Insecurity Across the Rural/Urban Continuum: 2011
|Percent of Households That Are Food Insecure||Percent of Households with Children That Are Food Insecure||Percent of Households with Food Insecure Children|
|Outside principal cities||13.2||17.9||8.8|
SOURCE: Coleman-Jensen et al. (2012). Reprinted with permission.
whom general economic conditions do not allow them to earn enough income to feed themselves. Food, housing, energy, and childcare assistance all affect the ability of a household to feed itself, and need to be considered as part of food assistance policy and its impact.
Weber discussed two of the four causal pathways identified by Allard: economic and social conditions and safety net programs. Weber noted these different pathways imply and require different kinds of polices. Related to economic and social conditions, Weber suggested a couple research questions: (1) How much, if at all, do local economic conditions, including job opportunities, wage rates, unemployment, housing costs, and other things, affect participation in food assistance programs? and (2) How much do these local conditions influence the effectiveness of food assistance programs in reducing food insecurity? He concurred with other speakers that answering these questions requires mixed-methods research and would help in program design and, in particular, government investment decisions. Household data need to have geographic identifiers in order to link the household to the economic conditions in which that household lives.
Related to safety net programs, he suggested a need to look at cross-program participation dynamics (food assistance, energy assistance, housing assistance) along with employment dynamics to see how they vary with impact. He said he wondered how these dynamics vary across the rural/urban continuum, and if they do, how program and work dynamics are affected by personal demographics, local economic conditions, and program design.
He also considered how well matched food security needs in different places are to federal food assistance programs and emergency food
programs. He suggested one way to start to answer that question would be to look at the map from Feeding America5 to identify places with high food insecurity and evaluate food security programs in those areas.
Weber also identified two aspects of measurement of local economic conditions that are important for understanding food insecurity. While there are often good measures of some local conditions (unemployment, for example), there are not often good local measures of food costs, housing costs, jobs available to low-income people, and other conditions. He went on to say that the other aspect of measurement is geographic detail. There are some county-level data, but often no good measures of conditions at the subcounty level, which may be more relevant for studies of food insecurity. He reminded the audience of Allard’s question: “Why are so many poor households food secure?” Weber argued that part of the answer to that question may be where they live and the personal and household economic conditions of the place they live.
He said another research question is stimulated in part by Wilde (2013) about the “new normal.” He asked if a “new normal” regarding food assistance program participation exists since the recession, querying whether the way people view these programs has shifted so that their entry and exit differs from historic patterns. If it is different, forecast models for food assistance programs will no longer be on target because they use models based on an old set of relationships. Data show that people are leaving SNAP at slower rates, which may be related to their job opportunities, food prices, changes in policy, and/or a change in the norms regarding participation. Mixed-methods research would help in understanding this question, and answers may be important in designing programs as well as in making forecasts.
Weber said more policy research has not looked at place for at least three reasons. First, many people can and will move to other places to improve their opportunities. Second, getting people to move can be a very costly way to achieve better food security. Finally, migration of low-income people does not always improve low-income job holding. However, he observed, some people will not or cannot move, and hence the place they live shapes the outcomes affecting their household’s resources and decisions. In order to develop and implement policies to address food insecurity, he concluded, it is necessary to understand the spatial context.
5See http://feedingamerica.org/hunger-in-america/hunger-studies/map-the-meal-gap.aspx [September 6, 2013].
Hilary Seligman (University of California, San Francisco) noted that the presentations indicate that many barriers result in the same food insecurity outcomes. Food insecurity is conceptualized very broadly in terms of spatial access, physical access, and financial access to food. She said this is very important because the outcomes are all the same. However, the HFSSM in the Current Population Survey, in every question, asks respondents to parse out responses, asking “is this because you can’t afford the food item?” A research question related to a measurement issue may be to what extent people are able to cognitively parse out a response, while another is whether there is a reason to more broadly conceptualize food insecurity to include all other considerations that decrease access to food.
Allard asked whether food security is the right outcome for the conversation, observing that having enough food is different than having the right food. A conversation about access to supermarkets or places to get fresh produce is a different question than a conversation about having enough to eat. He stated that better research looks at health outcomes, such as obesity, noting that it is a more seasoned body of literature. He also suggested that it may be worthwhile to think about multiple outcomes, not just food security.
Kaiser described her past research on food insecurity and the interpretation of the food security questions among a Latino population. She said that there may be some issues in understanding the questions and possibly even some overreporting. Children may skip meals for a variety of reasons. She said that, in some of the populations she works with, it is difficult for people to cognitively follow the questions—although she said she is not sure that adding words would make the questions easier to understand. She noted that Pérez-Escamilla has researched issues of food insecurity and understandability of the questions. While the issue of respondent understanding of questions has been addressed in research, she stressed it should always be considered.
In response to Seligman’s point, Frongillo said that his research (Wolfe, Frongillo, and Valois, 2003) on in-depth interviewing with older people (who have multiple causes for food insecurity) has tested a series of questions with different ending stems. Their subjects were able to cognitively distinguish the questions. However, a questionnaire can include only so many questions that one can answer where the main part of the question seems very similar to something just answered, with a different tag line. The challenge is not so much cognitive differentiation, he stated, but how to ask a question in a practical way in a survey format.
Maureen Black (University of Maryland, Baltimore) noted it is useful to have the issue of context on the table, including geographic diversity. She referred to the map from Feeding America and said that agricultural
differences, food availability, and cultural differences may explain some of the variability. For example, beans are more popular in some parts of the country (e.g., Nevada) than in others (e.g., Maryland). If beans are included in a food package in Maryland, they might not be used, whereas in Nevada they might be. She asked about research that considers reasons for geographic diversity. Weber said much research looks at economic differences, but he was not aware of research that looks at cultural differences between places as a way of explaining geographic differences.
Allard said he has noticed that different food pantries have different levels of sophistication, such as understanding what local people eat. He stated that really good pantries understand local contours and develop food programs that speak to these contours. Although difficult to take into account from a quantitative perspective, mixed-methods or qualitative research would allow a researcher to think about such causal pathways. More of this type of research might give food pantries or local food programs ideas about how they can do their work better.
Diane Schanzenbach (Northwestern University) asked how Allard’s presentation fit into policy, such as locating supermarkets in certain neighborhoods or providing money to households in those neighborhoods. Allard responded that the policy response in the literature is an effort to create incentives for stores to locate in underserved areas. He noted that though this is a supply-and-demand issue, local governments can help developers see opportunities in communities that are underserved, as is occurring in Chicago. He said leading institutions also need to take a role in planning of new store development.
Schanzenbach followed up by asking about the role of the profit-seeking entities in planning the locations of grocery stores. Allard replied that Bitler and Haider (2011) addressed this question. Although it is not known why food deserts exist or change over time, a supply-and-demand logic is known, which Allard suggested is due to the same factors that lead to disinvestment in high-poverty, racially segregated neighborhoods. There is an issue of race that matters, he said, and also not much awareness. He noted the profit margin in selling groceries is very low. He reported one of his former students helped open a nonprofit grocery store called Louis’ Groceries on the south side of Chicago, suggesting that this different model might serve as an alternative to a model with the big chain grocery stores. There are probably all kinds of ways to ensure that people have access to food resources or food assistance, and spatial dynamics might help expand household budgets. There are also transportation ideas, whether mobile markets, or delivery or transportation services for the disabled populations.
Kaiser referred to the WIC example in Connecticut as a change in policy that had an effect on local foods, and there are many different
avenues to look at. She noted, however, the unintended consequences of building a large supermarket could put smaller stores out of business. Perhaps, she suggested, a policy should also consider how to help smaller stores. She noted that after WIC implemented its new policy, it was important to work with some of the WIC-only stores to learn how to properly handle perishable foods. Using this example, she noted sometimes changes in policy require education. Some educational interventions would bolster resources that are already in the community and help them do a better job when an incentive is added to encourage participants to buy more.
Allard said there is room to experiment with low-cost interventions to see if they make a difference. He suggested solutions that are tailored to the specific nature of the need in a community and the specific dynamics in which place might matter. Rather than a one-size-fits-all operation, a portfolio of options would be relevant or tailored to specific settings.
James Weill (Food Research and Action Center) noted that his organization is working on strategies for encouraging supermarkets that are subsidized by foundations or by local, state, or federal governments to go into a neighborhood to also conduct aggressive outreach for SNAP, Earned Income Tax Credit, and child credit. He expressed hope that these strategies would grow participation in these neighborhoods.
Sonya Jones (University of South Carolina) encouraged Allard to consider the fact that a food store is the front for an entire food system. It is not surprising to characterize a local problem, but it is also a global marketplace that is dealing with local constraints. She said that she appreciated the example of the nonprofit grocery store, saying that an interesting policy question is how much local control communities have over the kinds of food available and where they are located, and whether this is a predictor of food insecurity in the community. Allard reported on some conceptually promising work on civic community and social capital that addresses those issues. He said the case can be made that areas with greater advocacy, whether political, economic, or policy related, are going to have better or different types of resources.