|
Article Excerpt
This article considers vertical restraints in a setting in which duopoly retailers each sell more than one manufactured good. Vertical restraints by a dominant manufacturer enable the firm to acquire horizontal control over a competitively supplied retail good. The equilibrium contracts produce symptoms that are consistent with a variety of observed retail practices, including slotting fees paid to retailers by competitive suppliers, loss leadership, and predatory accommodation with below-cost manufacturer pricing for the dominant brand(s). Applications are developed for supermarket retailing, where the manufacturer of a national brand seeks to control the retail pricing of a supermarket's private label, and for convenience stores, where a gasoline provider seeks to control the retail pricing of an in-store composite consumption good.
1. Introduction
Vertical restraints by manufacturers on the retailers of their products continue to be a source of legal and policy debate. Indeed, in June of 2007, the U.S. Supreme Court reversed course on the legal treatment of resale price maintenance (RPM), overturning its earlier decision on the per se illegality of the practice in favor of a reasoned approach. (1) This ruling is in line with the traditional explanation for vertical restraints that the practice serves to align private incentives between manufacturers and retailers in the sale of manufactured goods. Absent restraint, intensive price competition among retailers can lead to an inadequate level of pre-sale retail services (Telser, 1960; Mathewson and Winter, 1984; Marvel and McCafferty, 1984; Rey and Tirole, 1986; Klein and Murphy, 1988; Winter, 1993) or promote excessive post-sale quality differentiation (Bolton and Bonanno, 1988). Vertical restraints can correct these distortions. Doing so generally produces efficiency benefits, a point that has been argued by many economists following Bork (1966) and Posner (1976). (2)
This article considers vertical restraints in a multi-product retail environment. In this setting, we identify more pernicious effects: a vertical restraint on a manufacturer's own good serves as a mechanism to control the retail pricing of another ("rival") manufactured good.
We consider a successive oligopoly market structure with two manufacturing industries, two retailers, and two goods. In the upstream market, one good is produced by a monopolist (or differentiated-product duopolists) and the second good is produced by a competitive fringe. In the downstream market, the two goods are bundled together in the sense that each retailer carries both goods and each consumer buys both goods from one of the two retailers. Examples of such a vertical structure include supermarkets that carry both a national brand and a store brand (private label), convenience stores that sell gasoline and in-store consumption goods, and computer retailers that bundle essential components (such as processors and operating systems) with a set of commoditized components (such as DRAM, hard drives, and flat-panel displays).
Our analysis builds on several recent papers. Winter (1993) considers a single manufacturer that imposes vertical restraints on duopoly retailers to elicit the optimal mix of priced and non-priced retail services. Absent contracts, the retailers compete excessively in prices and fail to provide a sufficient level of service; RPM combined with a wholesale price elevated above marginal cost simultaneously corrects both distortions. Rey and Verge (2004) consider how RPM can be used by duopoly manufacturers to control the retail pricing of duopoly retailers. The manufacturers' use of vertical restraints frees their wholesale prices to be set at marginal cost, and this circumvents retail-manufacturer contract externalities that would otherwise result in disadvantageous price competition. (3)
In the present setting, vertical restraints likewise serve to resolve retail market externalities; however, the essential difference is our focus on the joint pricing decisions of multi-product retailers who sell both a dominant manufacturer's good and a product produced by a competitive fringe. In this regard, our arguments are related to the substantial literature on the extension of monopoly power to other products through the use of tying arrangements (e.g., Whinston, 1990; Carbajo, de Meza, and Seidmann, 1990; Shaffer, 1991b) or commodity bundling (e.g., Nalebuff, 2004; Mathewson and Winter, 1997; DeGraba and Mohammed, 1999; McAfee, McMillan, and Whinston, 1989). Innes and Hamilton (2006), for example, show how a dominant firm can use explicit cross-market controls on retailers' contracts to achieve an integrated multi-good monopoly outcome; the monopoly manufacturer dictates that suppliers of the rival good pay lump-sum transfers that serve to elevate the good's wholesale price and thereby prompt retailers to charge monopoly prices. In practice, such explicit cross-market controls are likely to be infeasible, whether due to proscriptions of antitrust law or due to an inherent inability of the monopolist to observe and verify retailers' cross-market conduct. Here we focus on how a vertical restraint imposed on a manufacturer's own product can be used to extract rents from the market for another product, without stipulating any cross-market tying, bundling, pricing, or contract terms.
Vertical contracts that extract cross-market rents produce several notable symptoms. Irrespective of the relationship between the products in utility (complements, substitutes, or independent goods), vertical restraints on the monopolist's product induce retailers to sign contracts with suppliers in the competitive fringe that involve fixed fees paid to the retailer ("slotting allowances"). Slotting allowances are prevalent in practice (Federal Trade Commission, 2003), and highly controversial. We show that their effects are often anti-competitive in a multiproduct context.
For weakly substitutable, independent, or weakly complementary goods, vertical restraints result in negative retail margins on the dominant manufactured good. Hence, our analysis offers a new explanation for loss-leader retail pricing that does not rely on coordination failures (Bagwell and Ramey, 1994), imperfect information (Lal and Matutes, 1994), heterogeneous consumers (DeGraba, 2006), or product complementarities in multi-product monopoly markets (Bliss, 1988). For strongly substitutable goods, the dominant manufacturer selects a wholesale price below marginal cost, producing a type of "predatory accommodation" similar to that derived by Marx and Shaffer (1999), albeit for different reasons.
In Marx and Shaffer (1999), below-cost wholesale prices arise because a monopoly retailer can thereby extract rents from competing suppliers. In our setting, a dominant manufacturer distorts her wholesale price from marginal cost to counter the retailers' incentive to discount the rival retail good. For independent goods, a dominant manufacturer sets a wholesale price above the (maintained) retail price to decrease retailers' per-customer rents. Retailer losses on the dominant good, in turn, temper interretailer incentives to attract customers from rivals by discounting the price of the competitively supplied good. For substitute goods, raising the wholesale price above marginal cost has two effects on the retailers' pricing incentives. Narrowing the retailers' margin on the dominant good under the restraint decreases the return to attracting customers (favoring a higher retail price for the rival good), but also reduces the opportunity cost of shifting consumption from the dominant good to the competitively supplied substitute good (favoring a lower retail price for the rival good). With strong substitutes, the second effect dominates and a below-cost wholesale price by the dominant manufacturer is needed to prompt the retailers to raise the retail price of the rival good.
The remainder of the article is organized as follows. In Sections 2-5, we frame our analysis around duopoly retailers who each sell a monopoly-produced good and a second, competitively supplied good. Section 2 develops the model and Section 3 compares the collective optimum that maximizes joint manufacturer-retailer profits to the outcome without vertical restraints. In Sections 4 and 5, we consider how vertical restraints can reconcile these outcomes in an environment without and with retailer-fringe contracts, respectively. We develop applications to gasoline stations/convenience stores and supermarket retailing, and derive some implications for antitrust policy.
In Section 6, we consider retail formats without product bundling and extend our observations to oligopoly settings in which multiple manufacturers impose vertical restraints on retailers. Under oligopoly, each manufacturer ignores the profits of rival manufacturers when selecting a target retail price and, as a result, incentives for predatory accommodation are particularly harmful. Indeed, in the case of highly substitutable goods, vertical restraints lead to higher equilibrium retail prices than those which would emerge in a multi-product monopoly market. The reason is that each manufacturer selects a below-cost wholesale price to control the retail pricing of the fringe good, and this motivates her rival to solicit higher retail prices (both for her own product and for the fringe good) to capitalize on the retailers' ability to acquire artificially high rents on sales of the rival manufacturer's good.
2. The model
* Consider a "2 x 2 x 2" successive oligopoly structure with two goods, two manufacturers, and two retailers. Good 1 is a "name brand" produced by a monopoly manufacturer and good 2 is a "generic brand" supplied to the retailers by a competitive fringe. Production of each good involves constant unit cost, denoted [c.sup.1] and [c.sup.2], and for simplicity, retail costs are suppressed. The retailers engage in pure intermediation, procuring goods from manufacturers at wholesale prices [w.sup.1] and [w.sup.2] in the upstream market and selling goods to consumers in the downstream market.
Due to economies of "one-stop" shopping that pervade retail environments (Bliss, 1988), each consumer purchases a consumption bundle ([y.sup.1], [y.sup.2]) from a single retailer. Given her choice of retailer j [member of] {1,2} and consumption bundle, a consumer obtains the utility
u([y.sup.1], [y.sup.2]) - [summation over (i=1,2)] [p.sup.i][y.sup.i], (1)
where [y.sup.i] is the quantity of good i purchased, and [p.sup.i.sub.j] is the price of good i at retail location j.
We assume u(*) is increasing and concave with bounded first derivatives and that own product effects dominate cross-effects, [absolute value of [dlnu.sub.i] / [dlny.sub.i]] > [absolute value of [dlnu.sub.i] / [dlny.sub.i]] for j [not equal to] i, where [u.sub.i] = [equivalent to] [partial derivative]u(*)/[partial derivative]u(*)/[partial derivative][y.sup.i]. The products can be substitutes, [u.sub.12] [equivalent to]u(*)/[partial derivative][y.sup.1][partial derivative][y.sup.2] 0, or independent goods, [u.sub.12] = 0. The optimal consumption choice at retailer j yields the indirect utility
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)
A consumer's choice between retailers is determined in a standard Hotelling framework, with the two retailers located at either end of the unit interval. (4) Consumers are uniformly distributed on this interval and incur preference (travel) cost of t per unit distance. The location [theta] [member of] [0,1] represents a consumer's distance from retailer 1, and (1 - [theta]) her distance from retailer 2. This, a [theta] -type consumer, obtains net utility [u.sup.*.sub.1] - t[theta] if purchasing from retailer 1 and [u.sup.*.sub.2]-t(1 - [theta]) if purchasing from retailer 2. Given retail prices for each good at each retailer, a consumer of type
[0.sup.*] ([u.sup.*.sub.1], [u.sup.*.sub.2]) = (1/2) + [([u.sup.*.sub.1] - [u.sup.*.sub.2])/(2t)]
is indifferent between the retailers, and the market is partitioned into consumer types [theta] [less than or equal to] [[theta].sup.*] ([u.sup.*.sub.1], [u.sup.*.sub.2]), who purchase both goods from retailer 1, and consumer types [theta] > [[theta].sup.*] ([u.sup.*.sub.1], [u.sup.*.sub.2]), who purchase both goods from retailer 2.
3. Collective optimum and no contract outcomes
* In this section, we examine how the retail pricing outcome departs from the collective optimum in an environment without contracts. We then characterize the role of vertical restraints in aligning incentives between a dominant manufacturer and the retailers of her product. We assume throughout the article that retailers cannot implement exclusive territories that would split the consumers between them, void any competition for customers, and thereby lead trivially to multi-product monopoly prices. (5) For most examples in practice, exclusive territories are infeasible because consumers cannot be compelled to buy from a given retailer.
A vertically integrated monopolist solves:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (6)
where [y.sup.i](*) [equivalent to] argmax {u([y.sup.1], [y.sup.2]) - [[summation].sub.i=1.2] [p.sup.i][y.sup.i]}. (6) The solution to (3) yields the maximum profit available in the market, [[PI].sup.*] = [PI]([p.sup.1*], [p.sup.2*]; [c.sup.1] [c.sup.2]).
Next consider the choice problem of retailer 1. (7) Absent contracts, the dominant firm selects a wholesale price [w.sup.1] and the competitive fringe prices at cost, [w.sup.2] = [c.sup.2]. Given these wholesale prices, the duopoly retailers compete in retail prices. Retailer 1 solves
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)
where [PI](*) is defined in (3). Normalizing the number of consumers to one, [phi]([p.sup.1.sub.1], [p.sup.2.sub.1]; [[bar.u].sub.2]) = [[theta].sup.*]([u.sup.*]([p.sup.1], [p.sup.2.sub.1]), [[bar.u].sub.2]) is the market demand for retailer 1, given the prices set by retailer 2 and the attendant utility level [[bar.u].sub.2]. The first-order necessary conditions for a solution to (4) are
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (5)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (6)
where, using Roy's identity,
[partial derivative][phi]/[partial derivative][p.sup.i.sub.1] = ([partial derivative][u.sup.*]/[partial derivative][p.sup.i.sub.1])2t = -[y.sup.i]([p.sup.1.sub.i], [p.sup.2.sub.1]/2t < 0. (7)
Notice that the collective optimum ([p.sup.1*], [p.sup.2*]) is achieved when the first term in each of equations (5) and (6) is equal to zero. Hence, the individual incentives of a retailer are compatible with the collective interest only when the remaining terms sum to zero. These terms correspond to two distortions. First, on the interretailer margin, higher prices by retailer 1 prompt consumers to switch to the rival retailer (the business-stealing effect). This loss of store traffic is costly to the retailer, but of no concern to the vertically integrated chain. The business-stealing effect provides the retailer with an incentive to set each retail price below the level which maximizes joint profits. Second is an effect on the intraretailer margin. To the extent that the retailer pays above-cost wholesale prices to its suppliers ([w.sujp.i] > [c.sup.i]), retail price effects on demand have a smaller impact on retailer profit than on the profit of the vertically integrated chain, which faces true cost [c.sup.i]. This double-marginalization effect generally induces the retailer to set prices above the level which maximizes joint profits.
Now, following Winter (1993) (and recalling that [w.sup.2] = [c.sup.2]), suppose that the wholesale price of good 1 is selected so that the business-stealing and double-marginalization effects exactly offset for the good 1 retail price,
[MATHEMATICAL...
|
