Download The Role of Supermarkets in Economic Geography: Market Power and Competition and more Schemes and Mind Maps Economic policy in PDF only on Docsity!
The Evolution of the Supermarket Industry:
From A&P to Walmart*
Paul B. Ellickson University of Rochester March 15, 2015 Abstract This chapter identifies important economic features of the supermarket industry and highlights their connection to market structure and economic policy. Starting with a historical overview the industry’s evolution, I then discuss the broad determinants of market structure and review several empirical studies that quantify their importance. I then consider the various empirical studies that quantify the extent to which supermarket firms exploit this structure to prevent entry or maintain high margins. Finally, I discuss the growing literature on the competitive impact of Walmart, emphasizing the connection to market structure and the mechanisms of supermarket competition.
- (^) This chapter draws on an historical overview prepared for the 2007 Grocery Store Anti-Trust Conference organized by the Federal Trade Commission. I thank Michael Salinger and Christopher Adams for helpful comments and suggestions on that draft and Emek Basker for excellent comments on the current version. All remaining errors and omissions are my own. All correspondence to: Paul B. Ellickson, University of Rochester, Rochester, NY 14627. Email: paul.ellickson@simon.rochester.edu.
1. Introduction
According to the Food Marketing Institute (FMI), Americans spent 620 billion dollars in U.S. supermarkets in 2013, accounting for 5.6% of their total disposable income. There are 37 459 supermarkets operating in the U.S. and the average store now carries almost 44 000 products in roughly 46 500 square feet of space. The average customer visits a store just under twice a week, spending just over $30 per trip (FMI^1 ). As the primary channel for sales for food at home, supermarkets play a central role in ensuring access to affordable and nutritious food. Consequently, the competitive structure of this industry is tracked closely by government agencies both in the U.S. and abroad. Between 1998 and 2007, the Federal Trade Commission (FTC) challenged mergers affecting 134 anti-trust markets and investigated an additional 19. The corresponding agencies in the U.K. and E.U. have taken a similarly active role. Understanding how supermarket firms compete is important for economic policy. More broadly, given the large number of stores, the frequency with which consumers visit them, and the extent to which urban shopping centers cluster around them, supermarkets also play a key role in economic geography and urban planning. In the U.S., there is growing concern that poor and minority consumers are underserved by chain supermarkets and therefore lack sufficient access to fresh and nutritious food. In the U.K., environmental planning authorities have placed restrictions on the development of “big box” supermarkets aimed at protecting town centers. Understanding how supermarket firms compete is important for social policy. Finally, supermarkets offer a staggering variety of differentiated products in outlets that are themselves differentiated in both product and geographic space. They invest strategically and heavily in information and distribution technology aimed at reducing cost. They wield significant buyer power vis a vis their upstream suppliers and control the scarce shelf space across which the vast majority of consumer packaged goods are sold. Understanding how supermarkets compete is important for evaluating broader economic frameworks. The aim of this chapter is to highlight important economic features of this industry and examine their connection to market structure and economic policy. The chapter begins with an historical overview of the evolution of the grocery industry. I turn next to a discussion of the broad determinants of market structure and related empirical studies, highlighting the mechanisms by which supermarket firms differentiate themselves either geographically or in product space. I then discuss various empirical studies quantifying the extent to which supermarket firms exploit this structure to prevent entry or maintain high margins. Next, I discuss the (^1) Food Marketing Institute (2015), ‘Supermarket Facts’, available at http://www.fmi.org/research-resources/supermarket- facts (accessed 15 March 2015).
which a small number of firms - but no single firm - compete to provide the widest array of products at the lowest possible prices. The following historical overview focuses on the United States. A brief comparison with Europe and the rest of the world follows. A&P and the Chain Store Revolution (1913-1930) This section draws heavily on Chapter 4 from Tedlow (1990), which charts the rise and fall of A&P. Before 1900, American shoppers purchased their groceries through a wide array of specialty shops and general stores. Meat was purchased from a butcher, fish from a fishmonger, bread from a baker, and produce from a vegetable stand. Mostly sole proprietorships, these stores were often run in a haphazard manner with little use of modern accounting practices or scientific management principles. There were certainly many stores, likely well over half a million, although accurate historical statistics do not exist for this period. Because most people arrived on foot, grocers needed to be close to their customers, so the stores were small and ubiquitous. They often delivered what was purchased and sold many goods on credit. The small sales volume of these tiny shops led to high costs and sizable markups. Furthermore, the shop owners purchased their own supplies from a Byzantine collection of jobbers and middle-men that was rife with corruption, adding additional costs to an already expensive distribution system. The Great Atlantic & Pacific Tea Company changed all of this. Although A&P began as a mail order tea business in 1859, it was the move to grocery stores in the late 1800s that changed the nature of retailing. The brainchild of brothers John and George Hartford, A&P's “economy” store format did for retailing what Henry Ford's Model-T did for automobiles, introducing both standardization and scale. The economy format was a standardized store, selling branded goods produced in A&P factories and delivered through a vertically integrated supply chain of factories, warehouses, and trucks. A&P quickly abandoned customer delivery and scaled back on credit, converting groceries to a cash and carry business. This move alone yielded significant cost savings (Lebhar, 1952). They also introduced modern accounting practices and scientific management principles such as Taylorism, yielding efficiencies in both back and front-end operations.^3 Their investments quickly paid off; from 1914 to 1919 A&P went from operating 650 to 4 224 outlets (Lebhar, 1952). This number would double again by 1923. As cataloged by Tedlow (1990), A&P introduced several key innovations. It switched to a cash and carry model, standardized both store layouts and product offerings, and integrated backwards into both distribution and manufacturing. Like the modern supermarket firms we observe today, A&P operated its own network of warehouses and delivery trucks, bypassing the middle men and independent jobbers that supplied its rivals and eliminating a prime source of double marginalization. It also produced many of their own products, specializing in what (^3) Taylorism, which was pioneered by Frederick Winslow Taylor in the late 1800s, is a management theory aimed at improving efficiency and labor productivity. One of the earliest attempts to apply scientific principles to management, it championed the use of time and motion studies, the standardization of best practices, and the efficiency of mass production.
would later come to be known as “store brands” and “private labels.” A&P conducted careful traffic studies to aid in site selection, studied efficient store design, and constantly streamlined their logistical operations. Investments in quality control and inventory management meant that their offerings were not only cheaper, but fresher, higher quality, and less apt to be out of stock. Moreover, its massive scale meant they could exploit buying power with respect to other manufacturers and input suppliers, providing yet another cost advantage over their, typically, single-unit rivals. Of course, A&P was not the only firm to exploit the chain format - Kroger, American Stores, and Safeway were all among the early adopters of this new business model. Not surprisingly (with the benefit of hindsight), chain stores quickly came to dominate the grocery business. Between 1919 and 1932, the share of the top 5 firms in the U.S. increased from 4.2% to 28.8% (see the column labeled “C5” Table 1). Table 1 : The Chain Store Revolution Year A&P Kroger Am.-Stores Safeway F.-National C 1919 4,224 1,175 4.2% 1920 4,600 799 1,243 5.6% 1921 5,200 947 1,274 6.3% 1922 7,300 1,224 1,375 118 7.1% 1923 9,300 1,641 1,474 193 8.0% 1924 11,400 1,973 1,629 263 9.3% 1925 14,000 2,599 1,792 330 11.5% 1926 14,800 3,100 1,982 673 13.6% 1927 15,600 3,564 2,122 840 1,681 16.9% 1928 15,100 4,307 2,548 1,191 1,717 20.4% 1929 15,400 5,575 2,644 2,340 2,002 24.5% 1930 15,700 5,165 2,728 2,675 2,549 27.6% 1931 15,670 4,884 2,806 3,264 2,548 29.3% 1932 15,427 4,737 2,977 3,411 2,546 28.8% 1933 15,131 4,400 2,882 3,306 2, 1934 15,035 4,352 2,859 3,228 2, 1935 14,926 4,250 2,826 3,330 2,623 25.7% 1936 14,746 4,212 2,816 3,370 2, 1937 13,314 4,108 2,620 3,327 2, Source:-Store-counts-and-concentration-estimates-drawn-from-various-tables-in-Tedlow-(1990)-and-Lebhar-(1952).- Due at least in part to decreases in transportation costs, the chains were able to create large networks of stores that could take advantage of quantity discounts on the products they did not produce themselves and economies of scale on those they did. The chain stores also benefited from the network externalities associated with information processing. The large number of stores and intricate distribution network allowed the chains to improve demand forecasts and thus plan inventories and site selection more effectively. They were also able to centralize accounting. The resulting cost savings were passed on to consumers in the form of lower prices. Various price studies performed in the late 1920s and early 1930s found that chain store prices were 4.5-14% lower than their independent counterparts (Tedlow, 1990). While the distribution system they employed was novel, the physical stores operated by many of the chains were not much different from their independent counterparts: delivery and credit were still common in many locations and consumers continued to be served by a clerk who would retrieve items and suggest others. The chains also did not significantly advertise or build physically
Although this figure is small by today's standards, it represented a substantial outlay in 1930. Cullen's plan was to operate on low margins and low expenses, making up the difference in volume. This was not unlike the formula favored by the chains, only Cullen was taking advantage of both scale and scope economies at the store rather than the distribution level, essentially turning warehouses into stores and mitigating one of the main advantages of the national chains. Among the most notable changes Cullen proposed were increased store size (five to ten times larger), low-cost warehouse district locations, the shift to self- service, and the emphasis on advertising. Supermarkets also benefitted from the growth in nationally advertised brands that the incumbent chains, which were heavily invested in their own brands, often refused to carry. The shift in consumer tastes toward branded products sharply reduced the cost advantages to retailers of vertically integrating into manufacturing. Falling transportation and storage costs were also key - the spread of the automobile and paved highways facilitated the supermarkets' strategy of locating on the outskirts of town, while advances in refrigeration allowed shoppers to make fewer trips and stores to hold larger inventories. The invention of the shopping cart helped shoppers to buy in bulk. Interestingly, the existing chains (including Cullen’s employer, Kroger) were reluctant to adopt Cullen's proposal, so he struck out on his own and formed King Kullen supermarkets. Before long, other independent retailers followed suit. It is important to note that these early supermarkets were quite crude by today's standards. Dismissively referred to as “cheapies”, the early supermarkets occupied abandoned warehouses or factories and were located in low-rent commercial warehouse districts. They featured primitive shelving (often just crudely stacked pallets) and required consumers to serve themselves, which was quite a shock at the time.^5 However, they were very cheap, offering prices that were on average 13% below the conventional chains (Markin, 1968), strikingly similar to the advantage that Walmart offers today (Basker and Noel, 2009). From a current perspective, these early supermarkets were part club store, part supercenter, and part dollar store. In particular, they did not just carry groceries. King Kullen also sold tires and vacuum cleaners. Big Bear, one of the early success stories, made 34% of its sales on non-food items (Charvat, 1961), right in line with the Walmart supercenters we see now. Moreover, the supermarkets generated a lot more revenue per store than the incumbent chain outlets, typically 10 to 20 times as much. King Kullen stores sold over $1 million in groceries per outlet in 1933 (at about $18M in 2014 dollars, this puts them right in line with the typical modern supermarket). Big Bear, on the other hand, made about $3.8 million per store ($69M in 2014), squarely in line with a modern Walmart supercenter. While some are quick to credit the supercenter model to Walmart, it clearly dates to a much earlier era. An interesting point to emphasize here is that the it was the smaller firms that initially championed the supermarket format, since it did not require the type of scale that the large incumbent chains relied upon and thus did not play to their perceived advantage. (^5) The shortage of labor brought on by World War II helped hasten the spread of self-service formats (Charvat, 1961).
Moreover, the incumbent chains were sitting on a large portfolio of existing stores that were suddenly outdated by the changing demographic landscape of the United States. It bears repeating that the basic business models behind both the supermarket and supercenter formats date back over 50 years, and the anti-chain sentiment of the 1930s was at least as strong as the movement against big box stores that we see today. Post War Boom & Malaise (1950-1970) This section draws on material from Charvat (1961). Supermarket growth was slow at first, but the format really took off after World War II and supermarkets quickly came to dominate the retail landscape over the next three decades. While the overall number of food stores decreased from about 400 000 to 162 000 from 1935 to 1982 (Tedlow, 1990), the number of supermarkets increased from 386 to 2 6 640, and the share of overall grocery sales accounted for by supermarket firms expanded from 3.2% to 74.5%, roughly comparable to what it is today (see Table 2; note that all sales figures are in 1972 dollars). The incumbent chains were initially slow to adopt the supermarket format, for fear of cannibalizing their own sales, and often rolled out a second brand (e.g. Kroger's Pay ’n Takit line) to mitigate the perceived risk. However, by the late 1930s, most of the dominant chains had at least begun converting to the supermarket format. Nonetheless, the balance of power had shifted, at least temporarily, to more regional firms. Table 2 : Supermarket Expansion Share&of&Overall&Grocery Year Sales&Cutoff Supermarkets Sales&($M) Stores Sales 1935 302.9 386 202 0.1 3. 1939 287.5 1,699 772 0.4 10 1948 635.6 5,600 5,654 1.6 22. 1954 703.4 10,506 14,214 3.8 41. 1958 747.0 15,282 23,562 5.9 53. 1963 762.9 21,167 31,484 8.6 59. 1967 825.7 23,808 43,433 10.9 66. 1972 1,000.0 27,231 64,960 14 69. 1977 1,515.0 30,831 113,111 17.2 75 1982 2,265.6 26,640 175,655 14.4 74. Source:&Manchester&(1992).&Sales&cutoff&is&annual&sales&in&$1000s&required&to&be&classified&as&a&supermarket. The post war boom was a period of steady growth for the supermarket industry. There was plenty of suburban real estate on which to build stores and ample markets to convert from chain grocery store to supermarket. Although the smaller chains were the earliest adopters of the supermarket format, even A&P started converting over by the late 1930s. More importantly, the “cheapies” began to disappear as firms moved closer to the suburbs and “traded up” for less price conscious consumers. In keeping with their increasingly upscale clientele, stores started adding services, while shopping center locations replaced freestanding units. By the 1950s, firms were rolling out stores we would recognize as supermarkets today. Regional, sectional, and local chains led the postwar supermarket boom. These firms were able to exploit local trends and expand through a mixture of
The Information Age: Brandwidth, Store Size & IT (1980-1995) While the 1970s introduced a host of new store formats, the most significant innovations were the introductions of the UPC code and the scanning register, which would transform back end operations and radically expand the number of products carried in each store. The first bar code scanner was installed in a Marsh supermarket in Troy, Ohio in 1974. By 1986, scanning registers were installed in half the existing stores, and by the early 1990s adoption was essentially universal (Progressive Grocer, various April issues). A tremendous labor saving device, scanning registers also gave retailers access to the same information as manufacturers, and a newfound ability to engage in market research and data based marketing (Messinger and Narasimhan, 1995). Basker discusses the adoption and impact of scanning registers in chapter 2 of this handbook. A new industry sprang up to support the processing of information. Information Resources Incorporated (IRI) was founded in 1978, beginning the era of the test marketing of new brands and laying the groundwork for an explosion of new products. By the mid 1980s, both IRI and Nielsen were running extensive consumer panels and integrating purchase and sales information with couponing, price, display and advertising data. From 1974 to 1990, the number of products carried per store went from 9 000 to 30 000 (Messinger and Narasimhan, 1995), while store size grew steadily at 1 000 square feet per year (Progressive Grocer, various April issues). Table 4: Format Evolution 1980 1982 1984 1986 1988 1990 1992 1994 Conventional 73.1 47.9 49.7 47.4 42.9 35.3 30.3 27. Superstore 17.7 28.9 28.3 27.5 30.2 33.5 34.3 37. Food/Drug?Combo 4.0 8.3 8.0 8.0 8.6 11.2 15.5 17. Warehouse?or?L.A. 4.2 14.9 11.9 12.3 12.2 12.6 12.2 9. Superwarehouse 1.0 1.7 3.2 3.9 4.8 5.1 5. Hypermarket 0.4 1.6 2.2 2.6 2.5 2. Source:?Progressive?Grocer?Marketing?Guidebook,?selected?issues.?L.A.?is?limited?assortment,?a?store?carrying?a?reduced?set?of?products?(e.g.?Aldi). Requiring greater space in which to stack all these new products, supermarkets increasingly turned to superstore and warehouse formats (see Table 4). However, the radical increase in product variety (brandwidth) also led to a renewed focus on logistics, since firms needed to crowd an ever-expanding product line efficiently onto their shelves. The increasing reliance on computerized inventory management systems and sophisticated logistical systems shifted the comparative advantage back to the larger chains. The diffusion of scanners meant access to scanner data, but created a greater need for coordination. Advanced back- end information technologies, such as Electronic Data Interchange and just in time delivery, required increased coordination between upstream warehouses and downstream outlets. Finally, already an established expert in retail logistics,
Walmart started rolling out supercenters (combination grocery/discount store outlets) in 1988.^7 Walmart and the Supercenter Era (19 88 - ) A virtual non-entity in the grocery business in the early 1990s, Walmart is now the largest supermarket firm in the United States by sales volume. Starting in 1988, Walmart has averaged more than 100 supercenters openings per year and currently operates more than 3 200 outlets throughout the United States. Not surprisingly, the impact of Walmart’s entry has been dramatic. I will postpone a full discussion of Walmart’s impact until section 4. However, as shown in Table 5, Walmart’s expansion coincides with an almost one for one contraction of the number of supermarkets, and an overall 1.5 for one expansion in the number of gourmet and limited assortment stores, suggesting a possible push toward differentiation. Walmart has also been cited in the bankruptcy proceedings of at least 26, mostly smaller, regional chains (Lambert, 2008) as well as several high profile merger cases. A spate of high profile mega-mergers in the late 1990s and early 2000s has sharply increased concentration at the national level and created several national chains (see Table 6, which shows the expansion of Walmart supercenters alongside the national share of the top 4, 8 and 20 grocery firms, including Walmart). At the same time, alternative formats like limited assortment (Aldi and Trader Joe’s) and gourmet (Whole Foods) have thrived in this new retail environment. Table 5: Evolving Structure in the Walmart Era Year Gourmet/L.A. Supermarket Supercenter Independent Chain8Store Total 1996 707 29,742 705 8,691 22,698 31, 1997 722 28,168 821 7,688 22,260 29, 1998 866 28,282 899 7,773 22,595 30, 1999 1,165 27,616 1,060 7,370 22,856 30, 2000 1,807 27,913 1,263 7,696 23,750 31, 2001 2,041 27,826 1,509 7,780 24,076 31, 2002 2,169 27,831 1,720 7,939 24,273 32, 2003 2,881 28,187 1,885 8,664 24,891 33, 2004 3,285 28,085 2,114 8,662 25,247 33, 2005 3,356 27,846 2,382 8,509 25,503 34, 2006 3,527 27,201 2,659 8,468 25,355 33, Source:8Author's8calculation8from8Trade8Dimensions8TDLinx8data. Comparison with Europe and the U.K. Due in large part to the constant attention of government competition authorities, the evolution of the supermarket industry is very well documented in the United States. The other regions for which this is true are the U.K. (due to the work of the (^7) Foster et al. discuss the increasing role of chains throughout the retail sector in Chapter 1 of this handbook.
U.S. is much larger geographically than either the U.K. or the nations of western Europe. The share of the top five retailers in California, Texas, and Florida were 62%, 64% and 49%, respectively. In France and Spain, the top five firms control 56% and 52% of sales, whereas in Germany and Italy the figures are more modest, namely 36% and 26% respectively (Dobson, 2005). Dobson notes that Germany and Italy have much larger fractions of limited assortment stores and consumers visit stores almost twice as frequently as those in the U.K., France and the U.S., where large footprint stores (e.g. supercenters and hypermarkets) are much more prevalent. As in the U.S., concentration is a cause for concern to European competition authorities, who have also taken an active approach to regulation.
3. The Determinants of Market Structure
Supermarkets, like other retailers, are distinguished by the fact that they primarily sell other firms’ products. Therefore, their uniqueness (and market power) mainly stems from other forms of differentiation than brand, at least in terms of access to unique products. There are few significant proprietary technologies and most innovations are both observable and easily replicated. Spatial location is clearly a key factor that distinguishes one store from another, as is the particular set (and quantity) of products they choose to carry. Scale is also critical. Larger chains are able to obtain greater quantity discounts and take advantage of density economies arising from operating a network of stores. Firms can also distinguish themselves based on format (gourmet, limited assortment) or the frequency with which they have sales.^8 Each of these aspects, tackled in various academic studies, play a key role in driving structure. I turn to these topics now. Scale and Scope Given the escalation in both store and chain size in the 1980s and 1990s and the constant movement toward consolidation, it seems natural to start by discussing the role of scale and scope in driving competitive structure. This aspect of differentiation also has the benefit of being fairly straightforward to quantify. In a series of papers, Ellickson (2006, 2007, 2013) examines the role of endogenous fixed investment in determining the equilibrium structure of the supermarket industry. The endogenous fixed cost (EFC) model, originally developed by Sutton (1991) in the context of advertising, posits a minimum efficient scale that increases with the extent of the market – rather than inviting additional entry, larger markets simply invite additional sunk investment by the same set of incumbent firms. The relevance of the EFC model to grocery competition stems from the stylized fact that geographic markets in the U.S. (as well as the U.K., Europe, and Latin America) are consistently dominated by a handful of large chains, but never one chain. The fact that it is always more than one suggests that the mechanism of (^8) Many of these features of stores and chains are also discussed in Betancourt’s chapter on distribution services; see Chapter 4 of this handbook.
competition is more complicated than simple (exogenous) scale economies, which should instead yield monopoly, at least in those markets that are well below the overall minimum efficient scale. Ellickson (2007) adapts Sutton’s model to retail competition by interpreting the outcome of those investments as product variety. In particular, firms make sunk investment in IT and distribution systems aimed at efficiently stocking and replenishing an ever-increasing breadth of products, facilitating one-stop shopping and increasing the likelihood of providing a consumer’s preferred set of products. Ellickson interprets product variety as a “vertical” model of quality meaning that consumers always prefer an increase in its level, holding price fixed. The empirical test of the model focuses on its implications for market structure: markets both large and small should be served by roughly the same number of firms. A key challenge is to define markets in such a way that sunk investments are distinct across markets. Ellickson does so by focusing on distribution markets, which he argues are both geographically distinct from each other and coincide for spatially proximate firms. Exploiting a detailed, store-level census for 1998, he argues and empirically demonstrates that supermarkets in the U.S. are indeed a natural oligopoly in which a small number of firms (between four and six) capture the majority of sales, regardless of market size. He finds that the number of firms does scale up with the size of the market, but the expansion consists solely of low quality stores. The two-tiered structure uncovered in Ellickson (2007) suggested a second method of testing the EFC model of retailing: the contrast between different types of players, firms that invest heavily in distribution and vertical product quality and those who do not, with product variety as the key positioning aspect by which the chains differentiate themselves from their smaller scale competitors. Ellickson (2006) focuses explicitly on this proposed mechanism of competition. He first formulates an EFC model with two segments of consumers, only one of which cares about variety, giving rise to a two-tiered structure in which a small set of high- quality chains competes with an expanding fringe of “mom and pops” offering low variety at minimal price. He then tests this model empirically, first documenting a large quality wedge between the tiers and then directly demonstrating the expansion in quality that the theoretical model posits. He argues that variety is indeed a key principle of differentiation for these firms, and plays a central role in explaining market structure. The importance of product variety (and store size) is also a key aspect of Messinger and Narasimhan’s (1997) model of retail formats. The authors are motivated by the rapid increase in assortment and store size that occurred in the 1980s and 1990s (as documented in Messinger and Narasimhan (1995)). To explain this, they build a model of endogenous assortment in which consumers choose where to shop based on the trade-off between time-saving shopping convenience and price. In their model, larger stores with greater assortment offer the convenience of one-stop shopping, which lowers the cost of shopping. They calibrate the model using aggregate U.S. supermarket data spanning 1961 to 1986. They find that a major driver of the growth in one-stop shopping formats is a contemporaneous increase in the value of consumers’ time (via higher wages and
resulting economic geography? Like scale, physical location is easy to quantify, which helps facilitate detailed empirical work. There are several datasets containing the physical location of stores both in the U.S. and abroad, and various authors have examined them in great detail. Aguirregabiria and Suzuki provide a general discussion of location and entry games played by retailers in Chapter 9 of this handbook; my discussion focuses on research specific to supermarkets. Smith (2006) uses the same data and a similar consumer model as Smith (2004) to analyze the location and size decisions of the firms, rather than prices. He is motivated by the tension between two regulating motives in the U.K.: the desire to promote competition (on the part of the Competition Commission) and the desire to protect town centers (on the part of various local interest groups lobbying the government for greater protections). In principle, government regulation could be beneficial if firms fail to internalize the effects of their decisions on consumers and rival retailers. To explore these issues, Smith estimates a consumer choice model and performs counterfactuals in which stores are opened, resized and relocated. He then computes the benefits to consumers and producers of various alternative configurations in an experimental fashion. Smith finds that overall industry profits are maximized by opening or expanding large stores (thereby increasing overall expenditures) whereas the benefits to individual firms are maximized by attracting consumers from rival stores, which would lead to better located, medium sized stores. This latter outcome coincides closely with consumer interests, suggesting limited consumer-protection justification for imposing regulations on size and location. He finds that the stricter planning regulations put into place in 1993 led to lower consumer and firm benefits than the earlier regime would have, due in part to the greater proportion of small stores that were opened ex post. Schiraldi et al. (2012) revisit the issue of site selection from a dynamic perspective. Yang (2012) develops a similar model that focuses on entry in small Dutch municipalities and considers counterfactuals that decrease the sunk cost of entry.^10 Orhun (2013) examines the choice of spatial location of U.S. supermarkets using a static discrete game framework. Using the same census of supermarket locations as Ellickson (2007), she finds that supermarkets are shielded by geographic differentiation and that different firms target different consumer segments. Supermarkets of the same type exert higher competitive pressure on each other than competitors of different types. This type of localized competition is a feature of many geographic studies of the industry and represents one clear stylized fact. Ellickson et al. (2014) develop a model of individual store choice that exploits U.S. census tract level consumer demographics and a census of supermarkets revenues. The goal is to distinguish the role of location from the impact of brand/quality and the spatial distribution of tastes. They estimate their model using store level data on revenue, location, size and chain affiliation from 1996 and 2006. They find evidence of both spatial and brand level differentiation that has increased over time, possibly in response to Walmart. In particular, they (^10) See Chapter 11 by Schivardi and Pozzi for a general discussion of entry regulation in retail markets.
find that the share of revenue attracted from consumers in the wealthiest markets expanded markedly over this period and the role of brands became much stronger, consistent with a shift toward greater horizontal differentiation. Finally, it is interesting to consider the role of e-commerce in the grocery industry. The failure of Webvan in 2001, the purchase of Peapod by Royal Ahold (a “brick and mortar” supermarket firm) in 2000, and the fact that web-based grocery services seem to thrive in only the richest and densest markets suggest that a pure internet based grocery shopping service is not viable in most markets. However, many traditional grocery firms are experimenting with online/delivery offerings to complement physical sales. Pozzi (2013) examines the introduction of an online shopping service by a large incumbent (brick and mortar) supermarket chain that was already operating a large network of physical stores. He finds that the new online channel led to a 13% increase in revenue, with little encroachment on traditional sales. He attributes the increase in new business to a reduction in travel costs. He also finds that revenues increase more in markets in which the firm faces more rivals, consistent with an element of business stealing vis a vis competing chains. It seems likely that we will continue to see web-based shopping play a complementary role to traditional brick and mortar sales.^11 Price and Format Having discussed the role of scale (in both firm and store size) as well as the importance of geographic location, I now turn to the other, less tangible, aspects of product differentiation. Since these features are less quantifiable, the existing literature is somewhat thinner than that of the previous topics. The two aspects that have received the most academic attention are pricing strategy and store format. In marketing parlance, “pricing strategy” is not the level of prices per se, but rather the extent to which goods are occasionally offered on deep discount. Discounts are clearly a mechanism with which to clear excess inventory, but perhaps more importantly a way to price discriminate between different consumer segments (e.g. cherry pickers and time constrained shoppers who are less sensitive to price). The segmentation motive has been studied extensively in the academic marketing literature. In that literature (as well as in practice), the choice is often framed as a simple dichotomy between “every day low pricing” (EDLP) and “promotional pricing” (PROMO). Lal and Rao (1997) view the choice of pricing strategy as part of an overall positioning strategy that also sets differing levels of service in order to better segment the market. This is the sense in which pricing format is a mechanism for differentiation. In their framework, PROMO pricing is paired with higher service (to target time constrained consumers) and EDLP with lower service (to target cherry pickers). Bell and Lattin (1998) instead frame the segmentation strategy around basket size (and shopping frequency), with EDLP aimed at large basket shoppers and PROMO at small basket shoppers, and present empirical evidence consistent with this prediction. More broadly, differing pricing (^11) See Chapter 19 by Smith and Zentner for a discussion of online and offline competition in other retail sectors.
focused on overall retail format (e.g. mass merchandiser versus supermarket) instead (since this is a firm, or at least nameplate, level distinction). Fox et al. (2004) examine store choice by an IRI panel of 96 consumers over a two-year period. The competitive set of stores includes grocery stores, drug stores and mass merchandisers and the research question involves consumers’ store and expenditure decisions. Interestingly, they find that price is the weakest predictor of shopping and spending, though this is likely due to a lack of relevant variation in the specific dataset they use. Instead, they find that assortment is the biggest driver of choice, particularly at grocery stores. With respect to cross format competition, they find little evidence of direct substitution. Instead, they find that substitution within the grocery format is much stronger than across formats. In fact, they find some evidence of complementarity between grocery and mass merchandisers. Courtemanche and Carden (2014) examines the competitive response by supermarkets to competition from club stores, focusing on store-level entry by Costco and Sam’s Club between 1994 and 2006. Controlling for a rich array of fixed effects (bolstered by clever falsification tests), they find that Costco entry results in higher prices amongst incumbent supermarkets, while competition from Sam’s Club has no statistically significant effect. Consistent with prior literature (discussed later), they find a negative impact from increased competition with Walmart supercenters. They argue that the Costco result is consistent with a model in which firms compete along non-price dimensions (such as service or variety) and the incumbent grocery firms are thereby shifting upmarket in response to Costco (and serving consumers with less elastic demand), while the null result for Sam’s Club is consistent with its focus on small business customers. Hanner et al. (2015) examine data comprising a census of supermarkets, supercenters, and club stores in operation between 2004 and 2009. They find a significant amount of turnover and churn. While the number of big-box grocery outlets remained relatively constant over this period, each year roughly 7% of these stores either open or close. While clubs and supercenters expanded over this period at the expense of traditional supermarkets, those traditional firms continued to open new stores each year. They also find that most of the churn amongst chains is due to expansion or contraction by incumbent brands, as opposed to de novo brand entry. The majority of brand entry is due to small chains and independents.
4. The Impact of Concentration
Two stylized facts that emerge from the study of market structure in the grocery industry are 1) markets are very concentrated, especially locally and 2) supermarkets enjoy a significant degree of spatial differentiation, at least within a small radius of a few miles. A natural follow-on question is whether these stylized facts translate into non-competitive behavior or conditions that otherwise harm consumers. Concentration and Prices
One concern, relevant to anti-trust authorities, is whether the increases in concentration that occurred at the regional level in the 1980s and the national level more recently are likely to facilitate the abuse of market power on the part of incumbent grocery retailers. In an early study of a single market in Vermont in 1981, Cotterill (1986) examined the relationship between prices and market structure, controlling for observable firm characteristics. He finds that grocery prices increase with retailer concentration, which is consistent with the exercise of market power. He also finds that independent supermarkets charge higher prices and that prices fall with store size, consistent with the importance of both scale and scope, and lending credence to the notion that heterogeneous firms serve distinct consumer niches (i.e. that differentiation is effective). Cotterill and Haller (1992) analyze entry by large supermarket chains in geographically defined U.S. markets (MSAs) between 1972 and 1981. They find that entry is positively related to growth, negatively related to market concentration and the total number of large chains already present in the market, and positively related to geographic proximity and the potential entrant’s return on capital. The impact of concentration and proximity are consistent with the presence of entry barriers, although the largest and most aggressive firms continued to enter distant and concentrated markets. These papers helped inform FTC policy in the 1980s and 1990s. More recently, Hosken et al. (2012) combine data from the ACCRA cost of living index with a census of store locations to examine the price impact of fourteen grocery mergers that occurred between 2004 and 2009. Consistent with the presumption underlying the Horizontal Merger Guidelines, namely that mergers in highly concentrated markets are presumed likely to enhance market power, they find that five mergers resulted in estimated price increases of more than 2%, and four of these were in highly concentrated markets. They also find that five mergers resulted in estimated price decreases of over 2%, and only one of these was in a highly concentrated market. The remaining four mergers had relatively little impact on prices. Together with the Smith (2004, 2006) papers documenting market power in the U.K., these studies are consistent with the notion that supermarket firms are indeed able to exploit market power in both setting higher prices and constraining entry. Buyer power A second concern is whether increased retailer concentration (as buyers) can lead to anti-competitive practices with respect to upstream suppliers and an unfair advantage with respect to smaller retailers. This has been a particular concern in the U.K., where retailers are especially concentrated relative to upstream suppliers. This was a key focus of the Competition Commission inquiry in 2000 and led to enhanced guidelines regarding conduct. Dobson (2005) notes that the problem is especially acute in the U.K. due to the relative strength of private labels and the high level of store loyalty exhibited by British consumers. The expansion in the number
