The Public Health Effects of Food Deserts: Workshop Summary (2009)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

2 Determining the Extent of Food Deserts The food environment is a dynamic one and can change rapidly due to many factors, such as prices and preferences. To understand why food deserts are a problem and what they impact, the first session of the work- shop featured presentations on the multiple dimensions used to define the food environment and the various cross-cutting ways to measure impact on both the macro and the micro levels. Lisa Powell provided a national overview and discussed price and outlet availability as aspects of access to healthy food. Mari Gallagher focused on the urban environments in Chicago and Detroit, while Joseph Sharkey pointed out the changing food retail environment in rural Brazos Valley, Texas. Lastly, Ephraim Leibtag discussed the dynamics of the food shopping environment and how it affects access to affordable and healthy foods. NATIONAL OVERVIEW OF FOOD DESERTS BY DEMOGRAPHICS AND SOCIOECONOMIC STATUS Lisa M. Powell, of the University of Illinois at Chicago, presented national data categorized by U.S. zip codes to provide a bird’s eye view of areas that do not have access to a supermarket or a grocery store. In doing so, she acknowledged that there is a trade-off between using data available on a national level versus the greater detail available from onsite data collection across smaller geographic areas. Access to healthy foods means that the food is available and afford- able. Powell defined availability as the number of food-related outlets 11 

12 THE PUBLIC HEALTH EFFECTS OF FOOD DESERTS within a measured geographic area assessed on a per capita and/or a per land area basis, with healthy foods associated with grocery store and supermarket availability and less-healthy foods associated with conve- nience store and fast-food restaurant availability. Availability gets at the time costs associated with food shopping (e.g., a convenience store that is a five-minute walk away versus a supermarket that is a half-hour bus ride away), whereas affordability is the monetary cost or purchase price of various items. Based on their available price data, the affordability of healthy foods is represented by the prices of fruits and vegetables and of less-healthy foods by the prices of fast food and soft drinks. Taken together, Powell mentioned that availability and affordability determine the total cost of food, or its accessibility. To provide an overview of accessibility, Powell and her team use data from the American Chamber of Commerce Researchers Association (ACCRA) for food prices and from Dun and Bradstreet (D&B) for outlet density. Drawing on D&B data, supermarkets and grocery stores are distinguished from convenience stores by the assumption that access to a convenience store alone does not provide access to quality food. Super- markets are substantially larger food stores than grocery stores and are more likely to have onsite food preparation such as a butcher, a baker, and a deli. Chain stores are studied because they often benefit from economies of scale in terms of purchasing power, distribution, and other factors that contribute to lower prices. A validation study is under way to ensure that the outlet data available from D&B and infoUSA, another proprietary business database, do not contain biases across neighborhoods of different socioeconomic status and racial or ethnic characteristics. Availability Using D&B data, 29 percent of zip codes nationwide do not have a grocery store or supermarket, and 74 percent do not have a chain super- market. Powell stressed that using zip codes alone is misleading, given that some zip codes contain no or very few people, and therefore she nar- rowed in on more densely populated urban areas. Of these urban areas, 7 percent have no grocery store or supermarket and 53 percent do not have a chain supermarket. When her team looked at food availability by linking D&B data in 28,050 zip codes with U.S. Census data on race, ethnicity, income, popu- lation, and degree of urbanization for the year 2000 (Powell et al., 2007), based on multivariate models, quite significant differences emerged: •  frican-American populations had half as much access to chain A supermarkets as Caucasians, controlling for other factors; 

DETERMINING THE EXTENT OF FOOD DESERTS 13 •  ispanic populations had one-third the access to chain supermar- H kets as non-Hispanics, controlling for other factors; •  ower-income neighborhoods overall had less access to chain L stores than middle- and upper-income neighborhoods; and •  ndependent, non-chain stores were more prevalent in predomi- I nantly African-American and Hispanic communities than in pre- dominantly Caucasian communities. As observed throughout the workshop, longitudinal data on various aspects of research on food deserts are scarce. However, Powell has done some national-level comparisons of changes in food availability in 1997 versus 2008. Looking at predominantly African-American (defined as 70 percent and higher), predominantly Caucasian (70 percent and higher), and mixed neighborhoods, the predominantly African-American neigh- borhoods had the smallest increase in overall availability and the larg- est decrease in number of grocery stores since 1997. When looked at by income, lower-income neighborhoods had the smallest growth in overall access to food stores and the largest decrease in number of grocery stores (Figure 2-1). 50.00 40.00 30.00 Percent Change Low Income 20.00 Medium Income 10.00 High Income 0.00 –10.00 –20.00 Convenience Grocery Supermarket Supermarket Store Store Chain Other FIGURE 2-1  Change in food store availability by income, 1997-2008. SOURCE: Lisa M. Powell, ImpacTeen Project, University of Illinois at Chicago. Figure 2-1.eps Data drawn from Dun & Bradstreet, 1997 and 2008. 

14 THE PUBLIC HEALTH EFFECTS OF FOOD DESERTS Affordability Real (inflation-adjusted) prices for fruits and vegetables, dairy, and meat were generally flat from 1990 through 2007, based on ACCRA data. Yet, over the same period, real prices for soft drinks and fast foods—as noted earlier, the proxies for less-healthy foods—declined significantly (soft drinks by one-third, fast food by 12 percent), making them seem increasingly less expensive relative to healthier alternatives. Powell noted that people operate in terms of relative prices, so it has become relatively cheaper over time to purchase energy-dense foods such as fast food items like burgers and fries. Powell and Bao (in press) recently linked price and outlet density data with longitudinal data from 1998 through 2002 on the children of mothers from the 1979 National Longitudinal Survey of Youth. They found that food pricing is likely to have modest but measurable effects on weight outcomes of children ages 6 to 17 and that greater access to supermarkets, when defined on a per land area basis, is associated with a reduction in weight. Price elasticities were stronger for children in families of lower socioeconomic status: for example, among children in the bottom quintile of the income distribution, a 10 percent reduction in the prices of fruits and vegetables was associated with a 1.4 percent reduction in body mass index and a 10 percent increase in fast food prices was associated with a 2.6 percent reduction. This evidence, she concluded, suggests a multi- pronged approach of changing relative prices by both subsidizing fruits and vegetables and taxing fast food to improve weight outcomes among children and adolescents. MEASURING FOOD DESERTS: FOCUSING ON URBAN AREAS Mari Gallagher, of Mari Gallagher Research & Consulting Group, helped create the nonprofit National Center for Public Research. Through that group, she and others focus their research efforts on food deserts, and she presented findings from work in Chicago and Detroit. Gallagher pointed out that there are untrue stereotypes about food deserts—that food deserts only affect inner-city, African-American, or poor people— and these false notions may discourage investment by retailers. Gallagher differentiated between two different types of food ­venues: mainstream and fringe. Mainstream food venues are grocery stores or supermarkets, both small and large, where healthy foods can be pur- chased. On the other hand, Gallagher mentioned that fringe food ­venues— including fast-food restaurants, gas stations, and convenience and liquor stores—do not have healthy food options available on a regular basis. A community will usually have both types of food venues. The key, how- ever, is a term Gallagher calls food balance, so that consumers can easily 

DETERMINING THE EXTENT OF FOOD DESERTS 15 choose between a mainstream or a fringe food store. A food balance score is a ratio of the distance to the closest grocer versus distance to the clos- est fringe food establishment. When fringe food venues are handy but mainstream stores are not, a community is out of balance. Food Balance Effect Gallagher and her associates derive a food balance score for a neigh- borhood, then pair a food balance score with health-outcome data. To determine food balance, they conduct block-level analyses to find the location of the closest grocer and fast-food (since modified to fringe food) establishment, which they then pair with tract-level data on diet- related deaths. A food balance score reflects the fact that there is no perfect distance to a grocery store, given the different characteristics of different markets such as transportation options and geography. Rather, a food balance score gets around these characteristics to provide a relative measurement. Using these methods in Chicago, three key food deserts became evi- dent, comprising about 500,000 people and the city’s highest concentra- tion of single mothers and children. These areas were the most “out of balance,” using the definition above, with fringe food far closer than mainstream venues, especially for majority African-American areas (see Figure 2-2). Residents of food deserts face nutritional challenges evident in diet- related community health outcomes. Gallagher stated that her unpub- lished research shows communities with out-of-balance food environ- ments having statistically significant higher rates of residents’ dying prematurely from diabetes, when income, education, and race are con- trolled for. Gallagher found that African-American communities are most likely to experience the greatest total years of life lost from diabetes as a result. Furthermore, through an analysis of body mass index (BMI) based on drivers’ license data in Chicago, the areas with the highest BMI are roughly the same areas indicated as food deserts. Gallagher’s unpub- lished regression analysis shows that opening in a grocery store has a better impact on reducing obesity than closing a fast-food restaurant. In Detroit, the team studied 50,000 blocks and found that very few had mainstream grocery stores. They more likely had various types of convenience or fringe stores, particularly liquor or party stores that sell a few food items along with cigarettes, alcohol, and soft drinks. In some cases, some of the fringe outlets were classified as grocery stores in USDA data, which the team recoded after visiting the establishments. After this recoding of 1,300 retailers that accept USDA Food Stamps in the Detroit neighborhoods studied, 92 percent were fringe retailers, including liquor 

16 THE PUBLIC HEALTH EFFECTS OF FOOD DESERTS FIGURE 2-2 Relative distance to grocers and fast food in Chicago. SOURCE: M. Gallagher, 2009. Figure 2-2.eps bitmap image stores, party stores, gas stations, and buy-and-fry shops, and only 8 per- cent, or fewer than 100, were grocery stores of any size. The Role of Convenience Gallagher asserted that people generally buy food at the places clos- est to them, even if the stores do not have the foods they may prefer or need. Thus, concentrating on the price, quality, and range of healthy food choices in existing locations is important. In her opinion, a store that “crosses” from a mainstream to a fringe store, or vice versa, has a big impact on a community, since nearby shoppers may rely on it for their primary food source. Similarly, the addition or removal of a new store is significant, especially a mainstream grocery store. For example, after the Chicago study was completed, a Food 4 Less opened in the community of Englewood, one of the food deserts identified. Returning to the block-by- 

DETERMINING THE EXTENT OF FOOD DESERTS 17 block analysis, the addition of the new store improved the food balance effect for more than 40,000 people, including almost 14,000 children. Gallagher said her data raised interest among Chicago officials and supermarket executives. The data showed where in the city the impact of a new store would be greatest and, thus, helped the city prioritize six sites for grocer recruitment and incentives. The grocery representatives she has spoken with were intrigued by the possibilities of locating in underserved areas. However, they were also worried about the spotlight on food deserts because they were concerned about being swayed by political concerns for the poor. Data on food deserts could ease some of those concerns and be helpful for decision making and useful in develop- ing policy to alleviate food access disparities. MEASURING FOOD DESERTS: FOCUSING ON RURAL AREAS Joseph R. Sharkey, of Texas A&M University, suggested that while each rural area is different, a study of food access in the six rural counties of Texas’s Brazos Valley can be helpful to understand food deserts in rural America. The sparser population and lack of public transportation mean that low income and lack of a vehicle—or in the case of some seniors, the inability to drive their vehicles—complicate access to a store with healthy foods. Rural Shopping Options Borrowing from healthcare access literature (Khan and Bhardwaj, 1994; Guagliardo, 2004), Sharkey said potential consumers make decisions about where and what food to buy based on many factors. He introduced a conceptual model of food access that considers the food environment (including stores’ location, price, quality, and availability) combined with consumer variables (including their own food preferences, income, trans- portation options, and other factors) that together determine the barriers or facilitators to healthful eating. In many rural areas, including the six Texas counties, the food environment is rapidly changing. In addition to more traditional supermarkets and grocery stores, supercenters (such as Super Wal-Mart and Super Kmart) are expanding into rural areas. These supercenters are very large stores that engage in retailing a general line of groceries in combination with general lines of merchandise. Convenience stores, or food marts, are including more food items in their product selec- tion. Nontraditional food stores, such as mass merchandisers (including Wal-Mart, Target, and Kmart), dollar stores, and chain pharmacies are entering the food business. In Brazos Valley and other rural counties, 

18 THE PUBLIC HEALTH EFFECTS OF FOOD DESERTS these places are often in central locations, such as along a highway, to reach the maximum number of shoppers. The counties studied by Sharkey and his team span 4,500 square miles and 101 census block groups. There are five small urban clusters and no public transportation. To observe the location of food stores and the availability of fresh or processed fruits and vegetables, they drove all the major roads and conducted in-store surveys of food items available in all traditional, convenience, and nontraditional food stores. Residents’ mean distance to a supermarket is 9.9 miles, with grocery, convenience, and nontraditional stores somewhat closer. Most neighborhoods do not have any type of food store within a mile, but for those that do, the store is most likely a convenience store. Rural Access to Fruits and Vegetables: Fresh and Processed Different types of stores offer a range of fresh and processed (canned or frozen) fruits and vegetables. Supermarkets, supercenters, and grocery stores offer fresh produce, while convenience and nontraditional food stores, with few exceptions, offer only canned fruits and vegetables. To access fresh fruits and vegetables, about one-third of the population must travel 10 miles or more, although that percentage is halved if processed food is included (see Figures 2-3 and 2-4). The residents who live in urban clusters—about 25 percent of the total population in five clusters of between 3,500 and 11,950 people—are among the most socioeconomically deprived households in the Brazos Valley. Even in these five towns, there is limited access by walking and there is no public transportation. Thus, these residents have difficulty accessing healthy food. Sharkey summed up what his research in the Brazos Valley shows about food deserts in rural areas: Access is particularly problematic for rural residents without vehicles or sufficient financial resources. Most rural neighborhoods are not near supermarkets or even smaller stores that stock fruits and vegetables. As elsewhere, store formats are changing, with superstores, convenience stores, dollar stores, and even pharmacies getting into the food business. The older population, with more limited mobility, is increasing. Sharkey said that these factors from his model of food access suggest focusing efforts on stores that people currently use, the frequency with which they shop, and the types of products they purchase. 

DETERMINING THE EXTENT OF FOOD DESERTS 19 Access to Fresh Fruits N = 1 mile = 1 to 3 miles 0 10 20 Miles = 3 to 5 miles = 5 to 10 miles > 10 miles Urban Cluster Figure 2-3.eps FIGURE 2-3  Access to fresh fruits by distance to nearest vendor. map composed of bitmap wedges SOURCE: Sharkey, 2009. 

20 THE PUBLIC HEALTH EFFECTS OF FOOD DESERTS Access to Fresh Vegetables N = 1 mile = 1 to 3 miles = 3 to 5 miles 0 10 20 Miles = 5 to 10 miles > 10 miles Urban Cluster Figure 2-4.eps FIGURE 2-4  Access to fresh vegetables by distance to nearest vendor. SOURCE: Sharkey, 2009. composed of bitmap wedges map 

DETERMINING THE EXTENT OF FOOD DESERTS 21 DYNAMICS OF THE FOOD SHOPPING ENVIRONMENT To provide a national perspective, Ephraim Leibtag, of the U.S. Department of Agriculture’s Economic Research Service, summarized government and proprietary data to discuss current trends in the retail food environment. The government data include producer and consumer price information from the Bureau of Labor Statistics and geocoding of store locations from the Census of Retail Trade. Consumer-based data include the Consumer Expenditure Survey, American Time Use Survey, and National Health and Nutrition Examination Survey. Nielsen’s Scan­ track and Information Resources, Inc.’s Infoscan databases track store sales for major grocery store chains, while consumer shopping and eat- ing information can be analyzed using the NPD Group’s National Eating Trends and Consumer Reports on Eating Share Trends data along with Nielsen’s Homescan data. Neighborhood and local economies were not discussed in this presentation on the national overview. Price Stability and Volatility After a relatively stable period of 20 years, when prices were flat or even falling in real terms, commodity prices spiked in 2008, with most basic food crops and energy costs at record highs. However, with the recession, prices fell in late 2008-early 2009. One legacy of price stability was the advantage of mass production, in which suppliers centralized operations to set up large distribution chains. Three of the main factors that determine retail food prices are the costs of goods sold and operating costs, the dynamics of competition in the market, and consumer demand. Almost 40 percent of every dollar that a consumer spends on food goes for labor, with less than 20 percent going to the agricultural sector and the rest for expenses that range from advertising to rent to energy (see Figure 2-5). Until the late 1980s, the Consumer Price Index for food and the Pro- ducer Price Index for finished consumer foods tracked closely, and tradi- tional food retailers were by far the dominant players in the marketplace. Comparing these indices in the past two decades explains why the retail environment has become more crowded. As the indices diverged, with consumer prices rising faster than producer prices, new types of retailers saw a business opportunity—and joined traditional grocery stores and supermarkets in selling food to consumers. The Rise of Nontraditional Retailers As recently as 1998, grocery stores and supermarkets accounted for 80 percent of the consumer food dollar. Nontraditional stores now get 40 

22 THE PUBLIC HEALTH EFFECTS OF FOOD DESERTS Profits, 4.0% Depreciation and Repairs, 5.0% Rent and Interest, 6.5% Taxes and Other Costs, 7.0% Labor, 38.5% Energy and Transportation, 8.0% Advertising and Packaging, 12.0% Farm Value, 19.0% FIGURE 2-5  Breakdown of a consumer dollar spent on food. Figure 2-5.eps SOURCE: Leibtag, 2009. percent of Americans’ food dollars, and the percentage has been increas- ing. For example, in 2003, about 1,000 Wal-Mart Supercenters operated around the country; by 2008, that number had more than doubled to 2,400. Traditional stores are squeezed in the middle and are trying to determine their niche. Wal-Mart and warehouse stores offer lower prices and large quantities. Dollar and other discount stores offer a more lim- ited item assortment, have lower expenses than some supermarkets, and oftentimes focus on selling to low-income households. Meanwhile, gour- met and organic stores target more upscale consumers. The traditional retailers must figure out their niche and determine optimum size in terms of scale and efficiency. As Leibtag observed, a “plain old grocery store does not really cut it anymore.” Retailers are looking to distinguish themselves in terms of price, special items such as organic produce or fresh meats, or other characteristics. Some are building smaller stores that are more amenable to urban locations. Overall consumer food prices rose 5.5 percent in 2008, which was the largest increase since 1990. Although prices have declined from their record highs, the situation remains uncertain, increasing the difficulty of making predictions about the retailing environment. Yet it seems evident 

DETERMINING THE EXTENT OF FOOD DESERTS 23 that the more competitive diversity of stores selling food and the wider range of consumer food options are not going to change any time soon. DISCUSSION: MEASURING FOOD DESERTS Heidi Blanck, of the Centers for Disease Control and Prevention, moderated the discussion period after the speakers’ presentations. The questions were submitted by workshop participants invited from the general public. Many questions centered on data collection and the most effective ways to interpret and use the available data. Prices and Child BMI Several questions centered on the effects of fruit and vegetable prices on children’s weight. Powell explained how her research linked price and outlet density data by geocodes with individual-level data to exam- ine relationships between these economic contextual factors and food consumption behavior and weight outcomes. Higher prices for fruits and vegetables are expected to decrease consumption of these foods and increase weight. She cautioned against assuming that the presence of a supermarket that sells healthier food will always lead to better diets, espe- cially without taking account of prices. Whereas cross-sectional models rely on comparisons across different people, a longitudinal framework examines the same individuals over time, which helps to disentangle causal relationships. Sharkey noted that the variety of items stocked in stores poses a challenge to research into the effect of price on consumption. Meaningful comparisons are difficult when stores carry different items, although agri- cultural economists at his institution are attempting to devise common measures. He also suggested that the access point for food shopping, in many cases, is not where people live, but where they work or engage in other daily activities. Gallagher shared findings from two focus groups of participants who live far from mainstream grocers, with one group slightly better off eco- nomically. She said both groups were distance-sensitive, but much more price-sensitive. They desired different fruits and vegetables, but, she said, “they were pretty much price shopping all the time.” Role of Qualitative Research Most panelists agreed that more qualitative research is needed, in which people living in food deserts describe their access and how it might impact their food choices. For example, Gallagher said even the 

24 THE PUBLIC HEALTH EFFECTS OF FOOD DESERTS BOX 2-1 Listening and Learning In addition to understanding how socioeconomic and demographic determi- nants impact food deserts, it is important that qualitative research also consider other contextual factors that influence people and their food choices. For example, a low-income mother was concerned about her teenage son’s unscheduled time after school and contracted with him to come home after school to care for a younger sibling. He agreed if his friends could also come over. His mother soon found that an unexpected consequence was that she had to provide snacks to a group of hungry teenage boys. The economic imperative favored providing them with energy-dense snacks. In this case, a mother with limited resources needed to address child-care needs while balancing the household food budget. The bot- tom line is that it is imperative to link quantitative data with the context of people’s lives because socioeconomic factors are interconnected with other variables that may be overlooked. quickly organized focus groups mentioned above highlighted that the respondents would not welcome a grocery delivery service, because of bad ­ experiences with spoiled or rotten food purchased from local stores, whereas an outsider might have considered such a service a viable option. Powell said her team used focus groups to conduct qualitative research with adolescents in terms of when and what they ate during the day. ­Sharkey incorporates quantitative and qualitative methods, such as observing participants within their own homes. It is imperative, he said, especially when policy decisions are in the mix, to link quantitative data with an understanding of the context in which people live (Box 2-1). Agglomeration of Stores A topic touched on in this panel, and returned to throughout the workshop, is what Gallagher termed the agglomeration of stores: “Grocers do not go where grocers do not go already,” with the converse being true as well. One concern often expressed is that a supercenter will drive out other stores. In contrast, Gallagher’s research in urban areas has shown that businesses thrive when they are located near each other, and in fact, other retailers do well in locations near supercenters. The highest level of retail helps shape the environment of the commercial district, which is why a proliferation of fringe stores will discourage a more mainstream store from entering a market. 

DETERMINING THE EXTENT OF FOOD DESERTS 25 Consumer Shopping Patterns A member of the audience noted that the presence of a supermarket provides access to healthier options, but it also increases access to lower- cost, energy-dense or “junk” foods. Gallagher said the Food 4 Less that opened in Chicago has found it difficult to sell some healthier foods, such as low-fat versus whole milk. Powell reminded the group that the mere presence of a store may not change consumption patterns and related weight outcomes if unhealthy energy-dense foods are substantially cheaper than healthy less-energy-dense foods. Observing what is inside customers’ shopping carts confirms that they often buy less-healthy foods, both because of price and because they have tastes that have developed for many years. Supply and demand must match up: the food must be on the shelf, but people have to buy it. Blanck added that product placement within stores can also help influence decisions. In reply to a question about whether different racial or ethnic sub- groups have different kinds of stores they prefer, Leibtag said the work he has seen on this topic does not show a large difference—people usually shop close to where they live—although there are differences in shopping behavior by demographic group. Unless a tax or subsidy were extreme, he does not think it would alter food choice by much, because of taste preferences. Gallagher suggested there might be an opportunity to tighten the rules of inclusion for Supplemental Nutrition Assistance Program (SNAP; formerly known as the federal Food Stamp Program) retailers so that participating outlets will have to offer healthier foods. However, she warned about some pushback, as residents expressed concern that fringe food retailers would shut down or stop accepting SNAP vouchers if rules were too stringent. Using Data in Policy Making Gallagher mentioned that many organizations around the country, such as the Chicago Food Policy Advisory Council, have done excellent work providing data for policy formation. Regular tracking helps direct limited resources. As she noted, “We need to keep the data honest, keep it live, and make sure that we are looking at the full picture of what is going on and not just certain neighborhoods.” In terms of presenting data, maps have been most useful in informing policy makers and the public about the scope of the situation. Population and Economic Changes Referring to Powell’s presentation about store growth over the last decade, a participant asked whether population decline in low-income or 

26 THE PUBLIC HEALTH EFFECTS OF FOOD DESERTS underserved neighborhoods might be the reason behind the less vigorous growth of stores. Powell had used 2000 census data and will update the information when new census figures are available. However, she said she did not think this was a significant factor except perhaps for cases on the margin. As the recession deepens, local employment will be affected and cus- tomers may purchase more energy-dense, relatively cheaper foods. Some stores, including food outlets, will go out of business or curtail services. In that regard, said Gallagher, two food stores in close proximity to each other can keep the market competitive and this would benefit consum- ers. In contrast, a store without competition, faced with dwindling profit margins, may not maintain cleanliness and service, thus exacerbating its problems.