For over a decade, researchers have noted persistent disparities in access to healthy food in urban areas of the U.S. [1
]. In census tracts with high proportions of minority or low-income residents, supermarkets are scarce and convenience stores and fast-food outlets are abundant [7
]. Beyond the presence of different types of stores, low-income neighborhoods have demonstrated lower fruit and vegetable availability [10
] and lower overall healthy food availability [12
]. A growing concern about the impact of ‘food deserts’ on dietary quality and obesity has sparked nationwide efforts to identify priority areas for food environment interventions. According to the U.S. Department of Agriculture (USDA), food deserts are defined as census tracts that are low-income (a specified proportion of residents fall below an income threshold) and low-access (a specified proportion of residents have poor access to a supermarket) [13
]. Residents of food deserts face constrained food choices by the concurrence of both economic and physical obstacles [14
]. Residing in a USDA-defined food desert has been associated with indicators of poor health, including higher rates of obesity [15
], lower levels of serum carotenoids, and higher blood pressure [16
While the presence of food deserts across the U.S. has been well established, few studies have examined potential differences in food prices in various food retail spaces [17
]. This is important to understand, as pricing differences influence food purchasing decisions [17
]. In general, supermarkets obtain food at wholesale prices and have more efficient economies of scale than smaller food stores, resulting in lower prices [9
]. In a recent national study, Rimkus et al. examined milk prices across a number of store types and found prices were significantly lower at supermarkets than at limited service stores [22
]. Nevertheless, price variability for the same product does not always follow expected patterns [23
], as several studies have demonstrated lower prices for fresh foods at small food stores than at supermarkets [25
]. If food deserts lack supermarkets, residents of food deserts might face higher prices simply because of the types of stores available in low-income neighborhood.
Looking beyond store type, a store’s geographic location is relevant to pricing. Stores that are geographically isolated from other food stores may carry more expensive inventory. Although recent U.S. studies have generally not found food to be more expensive in low-income neighborhoods [22
], food stores located far from a competitor may set prices higher than food stores located near supermarkets, which may need to offer more competitive pricing [9
]. This relationship has not been well studied in the U.S. context. In an Australian study, having a greater number of major supermarket chains within 5 km of small stores was associated with a lower cost of healthy food [30
]. These patterns could have particular implications for residents of food desert. Already economically and geographically constrained, they may also face higher prices due to store isolation.
The purpose of this study was to assess differences in staple food item prices in small and non-traditional food stores (hereafter “smaller food stores,” including corner stores, gas-marts, pharmacies, and dollar stores) compared to supermarket prices, and how pricing patterns could exacerbate the problem of food deserts. Our aims were to: (1) compare prices of staple foods at smaller food stores with prices at their closest supermarkets; (2) compare prices of staple foods in smaller food stores more isolated (>1 mile) versus less isolated (≤1 mile) from supermarkets; and (3) examine prices of isolated smaller food stores inside and outside of food deserts. We hypothesized that prices would be higher in smaller food stores compared to supermarkets, while prices in isolated stores would be higher than in non-isolated stores. We had no reason to expect that higher prices would be evident in isolated food deserts versus non-food deserts, but because even moderate price differences may have significant consequences for low-income residents, we examined store isolation pricing patterns specifically in food deserts.
The analysis was limited to items carried by at least 40% of smaller stores, so that item-specific comparisons could be made between smaller stores and their closest supermarkets. Consequently, some healthy items for which we obtained data (e.g., whole wheat bread, brown rice) were excluded from the analysis. The final list of 15 items included three varieties of fruits and vegetables, four varieties of milk, four varieties of protein staples, and four varieties of grain staples. All four varieties of milk (skim, 1%, 2% and whole) were common in both gallon (3.78 L) and 1/2-gallon (1.89 L) containers; since results were similar across sizes, we present only gallon prices for the milk varieties.
Availability of food and beverage items varied by store type (Table 1
). There was considerable variability of items across smaller store type. Across all smaller stores, the most common items were peanut butter (80%), eggs (79%), white bread (78%), tuna (76%), and 2% milk (75%).
3.1. Aim 1: Comparing Prices at Supermarkets versus Smaller Stores
presents the average price differential between smaller stores and supermarkets for each item. Price differentials are presented relative to the supermarket price. For example, results show that the price of milk was 11–14% higher in small food stores relative to supermarkets, depending on the milk variety.
On average, all items were more expensive in smaller stores than in supermarkets except white bread, which was more expensive in supermarkets. The items with the highest price differential were CheeriosTM cereal (54%), bananas (53%), and white rice (50%).
3.2. Aim 2: Comparing Prices at Smaller Stores Isolated from Supermarkets versus Those That Are Not Isolated
presents the price differentials between smaller stores and supermarkets, according to smaller store isolation status. For all items except eggs, dry beans, and Rice KrispiesTM
, the direction of these comparisons was the same, reflecting greater price differences in isolated stores than non-isolated stores. However, apples were the only item for which the price differential was statistically significantly greater in isolated stores compared with non-isolated stores (p
3.3. Aim 3: Examining Prices of Isolated Smaller Stores Inside and Outside of Food Deserts
presents the price differential for isolated smaller stores located inside and outside of food deserts. There was no statistically significant difference in price for any item. Of the 15 items, price differentials were higher inside food deserts for 7 items, the same for 1 item, and lower for 7 items.
Most of the staple food items examined in this study (14 of 15) were priced higher at smaller stores than supermarkets (range 6% to 54% higher in price). Smaller stores have limited distribution of perishable items and smaller economies of scale than supermarkets, which manifests itself in higher food supply costs [20
]. For instance, Minneapolis and St. Paul smaller stores mostly acquire produce through independent purchases from a larger store/wholesaler versus a distributor [42
], which may increase costs to consumers. Smaller stores also tend to sell items in smaller package sizes than supermarkets, so these items generally have a greater price per unit. Historically, chain supermarkets have been more likely to carry generic items than other store types [20
], although this is possibly an area where the food retail landscape is changing, as non-traditional food retailers like pharmacies have begun offering their own brands of foods. The observable price differences between smaller stores and supermarkets may contribute to customer purchasing decisions. Research has indicated that price elasticity for specific food products can be high [43
], assuming stability of other product categories. By one estimate, a 10% decrease in the price of fruit was estimated to increase purchasing by 7% [44
]. However, it would be prudent to interpret this estimate as an upper bound, as the elasticity for any particular price point does not reflect the entire demand curve.
Descriptive findings in this study showed that isolated smaller stores had higher prices than non-isolated stores for 12 out of 15 staple food items. However, only one item (apples) was statistically significantly more expensive. Given our small sample size, which limits statistical power, more research is needed to understand pricing of healthy and staple foods in isolated areas. Another area for future research is whether price differentials are more pronounced for more healthful compared to less healthful foods. If prices of less healthful food are also elevated in smaller stores, higher prices for staple items might have no net effect on the nutritional quality of foods purchased due to reduced purchasing power for all food types.
The prices of items at smaller isolated food stores were similar inside and outside of food deserts. While no objective pricing disparity was observed by food desert status, the public health relevance of price differentials is particularly salient in food deserts. By definition, in food deserts, a high proportion of residents are low-income, making them more price sensitive [17
]. As discussed in several qualitative and mixed-methods studies, decisions around food shopping may be particularly complex in low-income households, incorporating economic and geographic realities as well as other personal, cultural, or demographic considerations [14
]. Households with young children may be particularly attuned to prices when making shopping decisions [14
]. In a qualitative study of low and mixed-income residents in Philadelphia, price was the top consideration in deciding where to purchase staple food; distance to the store was noted as an important constraint only among residents who did not have a car [19
]. Residents without vehicle access are either limited to small local food stores for grocery shopping or rely on alternate transportation, which adds both travel costs and time [48
]. In one study of food access in Bridgeport, CT, Leclair and Aksan contend the true cost of traveling from a food desert to a supermarket (including both travel and opportunity costs) is $
]. Research in the Minneapolis and St. Paul, MN metropolitan area has shown that low-income residents frequently pay for transportation to more remote food establishments, including taxicabs and informal taxis, because of limited personal vehicle access [48
]. While proximity is only one factor influencing shopping decisions [50
], geographic constraints in low-income and low-access areas are not as relevant for higher-income households with vehicle access.
Reliance on smaller food stores for household food supplies may also contribute to food insecurity in food deserts. Higher prices in smaller stores mean people are able to buy less food with their limited resources. Many smaller food stores accept Supplemental Nutrition Assistance Program (SNAP) benefits (95% in our smaller food store sample), and unlike WIC-authorized stores, in which prices for WIC-allowed foods are capped at 115% of the state average costs [52
], SNAP-authorized stores have no such limits. Thus, when SNAP benefits are used on staple foods in smaller food stores, they may not stretch as far. To compensate for reduced purchasing power in smaller food stores, policymakers could consider capping prices for designated staple foods in SNAP-authorized retailers, or increasing SNAP incentives for families residing in food deserts.
Results are also relevant in determining the success of the Minneapolis Staple Foods Ordinance, which requires small food stores to stock minimum amounts and varieties of staple foods, but makes no reference to prices. The study presents a unique opportunity to examine possible price differences in staple food as they change over time. As availability of healthy food expands to a new retail spaces, policy success will require that the expanded inventory in small stores is affordable. While availability is likely to increase, it remains unseen how prices will be affected, particularly in low-income/low-access areas of the city.
In interpreting results, one consideration is that when pre-specified sizes were not available, data collectors gathered data on the closest package size instead, so sizes needed to be standardized for comparison. This was common for grain items and dried beans. For example, in smaller stores the average package size for CheeriosTM
cereal was 22% smaller than in supermarkets. Comparison of non-like package sizes is a less conservative approach than if we had only compared precisely equivalent sizes. Non-like comparisons reflect a difference in product availability across two stores; thus, we cannot assume that, if the stores hypothetically stocked the exact same brand or size item, that the prices would be different. Nevertheless, as noted by Kaufman et al. (1997), our method is a better reflection of the lowest price available to the consumer than an approach that compares only precisely equivalent items [20
]. When stores stock similar (but not identical) items, the difference in availability is manifested in a difference in price, which is relevant to the consumer and the likelihood of purchase. Furthermore, differences in package sizes were not always as expected. For example, dry beans and white rice had average package sizes that were larger in the smaller stores than supermarkets.
The study has several other limitations. The focus of this research was on the price differentials in different store types, but the effect of those differentials on food purchase decisions must be explored in future work. The study also did not include WIC stores, which may be source of competition for smaller stores in this study sample. However, inclusion of WIC stores in pricing studies presents unique challenges, as WIC products are capped at 115% of the state average cost, and vendors may set prices differently when customer are likely to pay with WIC vouchers. Finally, USDA food deserts do not align perfectly with other measures of healthy food access. Food desert status can be defined by many criteria, which may yield different boundaries [53
]. Thus, results could vary if alternate criteria were used to define low-income/low-access areas.