رقابت کیفی در خرده فروشی: تجزیه و تحلیل ساختاری

This paper presents empirical evidence that endogenous fixed costs play a central role in determining the equilibrium structure of the retail food industry. Using the framework developed in Sutton [Sutton, J. (1991). Sunk Cost and Market Structure: Price Competition, Advertising, and the Evolution of Concentration (MIT Press, Cambridge).], I construct a structural model of retail competition in which escalating investment in firm level distribution systems yields a natural oligopoly of high quality supermarkets, while a low quality fringe of grocery stores serves consumers who do not value quality. Using a full census of the retail food industry to evaluate the model, I construct a structural prediction for the limiting number of supermarket firms and identify the quality escalation mechanism that sustains this oligopoly. Apart from the specific setting analyzed here, this model can help explain why certain retail industries remain highly concentrated as markets grow, while others quickly fragment.

In many retail industries, the most successful firms are the ones that offer the widest selection. For example, Wal-Mart rose to the top of the Fortune 500 by offering consumers a vast array of products at very competitive prices. The emphasis on product variety is particularly strong in the supermarket industry, where the introduction of computerized logistical and inventory management systems in the 1980s allowed firms to stock an ever expanding array of products. The explosion in both product variety and store size in the supermarket industry is striking. According to the Food Marketing Institute, the number of products offered per store increased from about 14,000 in 1980 to over 30,000 by 2004. To accommodate the greater selection, store size has increased an average of 1000 ft2 per year for the past three decades. Maintaining this variety requires substantial firm level investments. Every major supermarket firm invests in proprietary information technology and logistical systems aimed at increasing variety while minimizing storage and transportation costs. The emphasis on variety and the requisite fixed investments yield tightly contested markets among a handful of rival chains, a pattern that is repeated throughout much of retail. This is the second of two complementary papers that explain the industrial structure of the supermarket industry using an endogenous fixed cost (EFC) model of vertical product differentiation (VPD). The unifying theme of both papers is that escalating investments in variety enhancing distribution systems yield a natural oligopoly of high quality firms. The explanation for why we observe a natural oligopoly among supermarkets is based on Shaked and Sutton's (1987) claim that “entry in certain industries is limited to a small number of firms, not because fixed costs are so high relative to the size of the market, but rather because the possibility exists, primarily through incurring additional fixed costs, of shifting the technological frontier constantly forward towards more sophisticated products”. The tendency for larger markets to have better products instead of more firms reflects the dominance of vertical over horizontal differentiation; failure to match a rival's quality carries a severe penalty. To establish the relevance of the EFC mechanism to the market for groceries, my earlier paper adapted Sutton's (1991) model of advertising to account for some specific features of supermarket competition. In my version of Sutton's model, supermarkets compete for customers by offering a greater variety of products in every store, requiring a fixed investment in distribution. Serving a larger share of the market requires building more stores. Expanding variety requires building larger stores and more advanced distribution systems. Because variety is a purely vertical form of product differentiation, firms that fail to match the quality increases of their rivals cannot survive. Therefore, as markets grow, existing firms must incur higher costs if they are to remain in the industry, and this escalation in costs discourages entry by additional firms. Consequently, markets both large and small are served by roughly the same small number of high quality chains. As I demonstrate in Ellickson (2002), this simple model does not match what is observed in the data exactly: larger markets do have more firms. However, the expansion of firms is limited to a fringe of low quality stores that do not vertically integrate into distribution. In particular, there are two distinct tiers of firms in the food industry, supermarkets and grocery stores, but only one (supermarkets) is subject to endogenous investment. The current paper extends my earlier analysis by proposing and estimating a structural model of retailing in which each firm serves only one of these two submarkets, operating either supermarkets or grocery stores. This framework easily generalizes beyond the specific setting analyzed here. In particular, several retail industries such as book stores, video rental outlets, and pharmacies feature two distinct tiers of firms: large regional (or national) chains and local “mom and pop” stores. While the dominant chains build large stores (or exploit the advantages of the internet), stock a vast array of products, and invest heavily in distribution and advertising, firms in the fringe offer a narrower, more specialized selection and build smaller stores that require little or no investment in distribution or advertising. The central claim of this paper is that retail industries can be viewed as containing two distinct submarkets, only one of which (the chain store segment) is subject to endogenous investment. Moreover, the equilibrium market structures that characterize these two submarkets are polar opposites: one fragments while the other remains concentrated indefinitely. By estimating a structural model of competition, I am able to test these implications directly. The main empirical challenge to applying the EFC model to any two-tiered retail industry lies in determining which firms belong in each tier. While it is clearly possible to rely on trade publications or personal judgment, a formal model that can be applied to several settings is clearly preferred. Therefore, to distinguish the submarkets, I propose a formal mechanism for predicting the size of the fringe, based on Sutton's (1998) stochastic growth model for firms that do not invest in endogenous fixed costs. Having identified these two submarkets, I empirically estimate a structural model for each industry, using a full census of both supermarkets and grocery stores. Consistent with the implications of the EFC framework, I find that the number of grocery firms expands with the size of the market, while the supermarket industry is served by an oligopoly whose size is largely independent of market size. I also find that supermarket quality expands with the market, validating the mechanism that sustains this supermarket oligopoly. This paper contributes to a growing literature that applies Sutton's framework to retail competition. While my earlier paper was the first study to use an EFC model to explain the structure of a retail industry, subsequent authors have adopted a similar approach in a number of settings. In particular, Dick (2003) applies the EFC framework to the banking industry, Berry and Waldfogel (2003) to the newspaper and restaurant industries, and Latcovich and Smith (2001) to the market for online books. However, this is the first paper to propose a structural model of EFC competition and directly estimate the parameters governing the asymptotic number of firms. It is also the first study to directly establish the quality escalation mechanism that sustains this natural oligopoly. In addition, I provide empirical evidence that supermarkets do not differentiate themselves spatially, reflecting the primacy of the vertical dimension in driving market structure. This paper also contributes to a smaller literature focusing specifically on supermarket competition. Most closely related to my analysis is the work of Smith, 2004 and Smith, in press, who analyzes supermarket competition in the United Kingdom using a differentiated products discrete choice framework. While his focus is mainly on demand estimation and the impact of potential mergers, he also finds that the U.K. market is split between the “top 5” chains and a fringe of “high street emporiums” who build much smaller stores. Consistent with what I will argue here, he finds that consumers who switch from a top 5 firm are most likely to switch to another top 5 firm rather than a firm in the fringe, and that the characteristic that most influences their decision over where to shop is the size of the store. The paper is organized as follows. Section 2 presents a brief description of the retail food industry, highlighting the role of distribution networks in defining local markets. Section 3 contains a theoretical model that adapts the EFC framework to the retail food industry, providing a structural model of equilibrium quality choice in two distinct submarkets. Section 4 documents the existence of the two submarkets, and proposes a formal mechanism for distinguishing them. Using the two sets of firms, the structural model of competition is estimated in Section 5 using a market level census. Section 6 addresses the role of spatial differentiation and Section 7 concludes.

Traditional models of retailing have focused on horizontal differentiation, a natural choice given the importance of spatial location as a determinant of where one shops. This paper stresses the role of vertical differentiation. Obviously, supermarkets and grocery stores are differentiated both horizontally and vertically. However, as this paper has demonstrated, it is the vertical dimension that is key to understanding why this industry is dominated at the regional level by a natural oligopoly, a set of firms that invest heavily in logistics and distribution to offer a much wider variety of products in larger stores at lower prices than do grocery stores. These features seem common to many forms of retailing.