Competition effects of supermarket services.(Report)

Supermarket chains are significantly increasing their scope of services well beyond traditional food distribution by adding food services (such as deli, bakery departments, and eateries) that capitalize on the growth of food away from home (Park 1998) as well as nonfood services (such as pharmacies and banks) that combine one-stop shopping convenience and time-saving features (Kinsey and Senauer 1996). (1) The latter is also a strategic response to increasing competition from mass merchandisers and big-box retailers that are branching out into the grocery business. (2) Supermarket service levels can affect demand, costs, market power and, therefore, prices. Although theoretical work has recognized the role of these elements in the behavior of retail firms (Betancourt and Gautschi 1988), to date there is no empirical study that measures and ties together these effects via a structural analysis.

On the demand side, services are likely to directly expand demand for all items sold by attracting more consumers and behaving as strict complements to physical products sold (Betancourt and Gautschi 1988, 1990, 1993). The existing literature has not taken into consideration the strategic interactions among retail firms in their service-setting behavior and has neglected the fact that an increase in services is also likely to affect consumers' price responsiveness. Food retailers investing in services are more likely to differentiate themselves from the competition, segmenting the market in a more effective way and capturing consumers with low price sensitivity.

On the cost side, the process of enhancing store quality through services generates an increase in costs. The linkage between retailing quality and fixed costs has been strongly established by Ellickson (2006), based on the endogenous cost model (Shaked and Sutton 1987). Ellickson found that the increase in quality (measured via average store size) for the supermarket industry comes primarily from an increase in fixed costs, which establishes this industry as a natural oligopoly with a two-tiered structure: high-quality supermarkets and a fringe of lower-quality grocery stores. In terms of variable costs, it is customary in the business marketing literature to assume that an increase in services increases marginal cost (e.g., Lal and Rao 1997) although, conceptually, an increase in services may also generate economies of scale and scope (Betancourt and Gautschi 1988).

Given the effects of services on demand and cost, the impact of retail services on food prices is expected to be positive, although some studies have found weak evidence. Richards and Hamilton (2006), for example, used a nested constant elasticity of substitution model to find that, while supermarkets compete in product variety, the depth of product line has an unclear impact on prices. Cotterill (1999), using a traditional structure-performance model relating industrial concentration to store quality, found that some in-store services are positively related to food prices while concentration is not related to the level of supermarket services. (3) Bonanno and Lopez (2004) found that in-store services are positively associated with the price of fluid milk across three U.S. cities. Besides models of price discrimination in nonfood case studies (Shepard 1991; Barron, Taylor, and Umbeck 2001), the impact of services on pricing power has not been addressed in the empirical literature other than by relating profit margins to the level of services (Betancourt and Gautschi 1993; Messinger and Narasimhan 1997).

The purpose of this article is to estimate the impact of supermarket services on cost, demand, prices, and quantity sold as well as the extent of service competition among supermarkets, using fluid milk as a case study. Milk, being a relatively homogeneous good, allows the identification of cost and price differences resulting mainly from differentials in store services and pricing conduct. A two-stage conceptual model for supermarket services competition and monopolistic milk pricing is developed and estimated with scanner data for fluid milk sales from fifteen supermarket chains located in six U.S. cities. Empirical results corroborate strong effects of both in-store services (grouped into food services such as bakery, seafood, and prepared food departments, and nonfood services such as pharmacies and full-service banking) and average store size (as proxy for product assortment and other unobservable store characteristics) on milk retailing cost and supermarket-level demand for milk as well as on price responsiveness, market power, and milk prices paid by consumers.

The Conceptual Model

Consider the marketing decisions of supermarkets as consisting of two stages: prices are set in the short run while services are set in the long run (Betancourt and Gautschi 1993; Ellickson 2006). In this setting, the major goals of increasing services are to attract consumers with higher willingness to pay, to increase loyalty of existing ones (resulting in a positive spillover effect for all products), and to exploit economies of scale and scope (if any) generated by the provision of additional distribution services (Betancourt and Gautschi 1988).

To make the problem tractable, demand is assumed to be separable for all products except for services offered, meaning that consumers cannot separate retailing services from the physical products that they purchase (Betancourt and Gautschi 1988, 1993). Thus, let the demand for milk faced by the ith supermarket chain be given by

(1) [q.sub.i] ([p.sub.i], [s.sub.i], [s.sub.j], [Z.sub.i])

where [q.sub.i] is the quantity of milk sold by supermarket chain i, [p.sub.i] is its retail price, [s.sub.i] and [s.sub.j] are, respectively, the services offered by supermarket chain i and the other supermarket chains in the market (indexed by j), and [Z.sub.i] is a vector of other demand shifters.

Assuming nonjointness of production and following the technology specification of Roller and Sickles (2000), each supermarket faces the same long-run cost structure ([C.sup.LR.sub.i]) for milk retailing specified by

(2) [C.sup.LR.sub.i]([q.sub.i], [s.sub.i] | [omega], w) = [C.sub.i]([q.sub.i] | [s.sub.i], w) + [h.sub.i]([s.sub.i] | [omega])

which is decomposed into a short-run component given by [C.sub.i](*) and a long-run component indicated by [h.sub.i](*). The vectors [omega] and to are short- and long-run input prices, respectively. Equation (2) implies that services are given in the short run but variable in the long run.

Modeling explicitly the mechanism through which costs are affected by long-run investments (services, in this case) is crucial to understanding the pricing strategy of firms operating in industries where such investments are considerable (Fudenberg and Tirole 1984; Roller and Sickles 2000). The fact that retailing cost is affected by the level of services is an inherent condition of retail firms' operations (Betancourt and Gautschi 1988, 1990); as it is true that not all product categories are likely to be affected in the same way, any a priori decision to disregard the effect of services on milk retailing cost (short and/or long run) may result in a bias of the empirical measures.

In this model, food retailers maximize profits by competing in services in the first stage (long run) and setting prices in the second stage (short run). In modeling the second stage, supermarkets are assumed to act as local monopolies (Slade 1995; Besanko, Gupta, and Jain 1998), based on the fact that consumers value the overall convenience and attributes of a store more than the price charged for a single product or category. (4)

In the short run, supermarkets are assumed to choose milk prices monopolistically, taking the level of services as given, so that

(3) [p.sub.i] = - 1/[[eta].sub.i] + [mc.sub.i]

where [[eta].sub.i](= [partial derivative] ln [q.sub.i]/[partial derivative][p.sub.i]) is the semielasticity of demand and [mc.sub.i](= [partial derivative]C(*)[partial derivative][q.sub.i]) is the short-run marginal cost of selling milk. Differentiating (3) with regard to services, one obtains

(4) [partial derivative][p.sub.i]/[partial derivative][s.sub.i] = [partial derivative][[eta].sub.i]/[partial derivative][s.sub.i]/[[eta].sup.2.sub.i] + [partial derivative][mc.sub.i]/[partial derivative][s.sub.i].

The first term on the right-hand side, the market power effect, is expected to be positive ([partial derivative][eta]/.[partial derivative] s > 0) since services are expected to increase store loyalty and to attract higher-income customers, decreasing price responsiveness. The second term, the shift in short-run marginal cost due to a change in services, is left unsigned: although marginal cost may increase with services, a combination of both economies of scope and size may also take place. Economies of scope in the retailing industry are generated from the use of inputs that are common to both physical products and services such as selling area as well as logistic tools (such as computers) that can be used to assure product delivery and manage product assortment (Betancourt and Gautschi 1988), consistent with the classical definition of economies of scope (Panzar and Willig 1981). Despite the fact that the two terms may be of opposite sign, the sign of (4) is expected to be positive.

In the long run, supermarkets engage in competition in services to attract more customers and to capitalize on the eventual gain in pricing power over all product categories. Assuming that the proportion of long-run profits coming from a category is constant, maximizing total supermarket profits is equivalent to the following (5)

(5) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

The first-order condition for setting the services level is

Print Get the Mag Weekly Updates