تجزیه و تحلیل آنتی تراست از تخفیف هایی وفاداری همراه

Consider a monopolist in one market that faces competition in a second market. Bundled loyalty discounts, in which customers receive a price break on the monopoly good in exchange for making all purchases from the monopolist, have ambiguous welfare effects. Such discounts should not always be treated as a form of predatory pricing. In some settings, they act as tie-in sales. Existing tests for whether such discounts violate competition laws do not track changes in consumer surplus or total surplus. We apply a new test to an illustrative example based on SmithKline that assumes the “tied” market has homogeneous goods. If the tied market is characterized by Hotelling competition, bundling by the monopolist causes the rival firm to reduce its price. In numerical examples, we find that this can deter entry or induce exit.

Bundled loyalty discounts provide price breaks on one or more products to buyers that remain sufficiently loyal to a supplier. They resemble volume discounts in that shifting incremental purchases from a rival supplier may allow a buyer to satisfy a loyalty requirement and thereby enjoy lower prices. They resemble tying when affordable prices for one (tying) product are available only to consumers that are completely loyal to the supplier in their purchases of a second (tied) product. Like volume discounts and tying, bundled loyalty discounts can enhance or diminish consumer surplus and total welfare by improving the ability to price discriminate, and they can change the long run incentives for single-product firms to enter/exit. Open questions include: under what conditions should competition authorities treat bundled loyalty discounts like tying or predatory pricing, and when do bundled rebates enhance welfare? Bundled loyalty discounts have recently received considerable attention in antitrust circles in light of LePage's, Inc. et al. v. 3M Company, 1Michelin v. Commission, 2 and a number of pending court cases. 3 Until recently, the practice of bundled rebates has received comparatively little scholarly and judicial scrutiny. Indeed, only a small number of cases litigated in the U.S. have focused squarely on these practices, including SmithKline Corp. v. Eli Lilly & Co. 4, Ortho Diagnostics Sys., Inc. v. Abbott Lab, Inc. 5, and Cascade Health Solutions v. PeaceHealth. 6Virgin Atlantic Airways Ltd. v. British Airways PLC 7 involved similar practices, but the district court found allegations of anticompetitive behavior to be unsupported by fact. LePage's provides a good background for the issues involved. The facts in LePage's begin with the very strong brand name of Scotch tape, a 3M product. Most retail merchants believed that they had to offer Scotch tape. Until the early 1990's, 3M's share of the U.S. market for transparent and invisible tape exceeded 90%. Starting in the early 1990's, however, 3M's share began to erode with the rise of office superstores such as Staples and Office Depot. These retailers sold products, including tape, under their own names as private labels. LePage's dominated the growing private label segment, with an 88% market share in 1992. Its share of the overall tape market, however, was only 14%. 3M responded by adding a private label of its own, under the Highland name. 3M's entry involved the use of a bundled rebate program that “offered discounts to certain customers conditioned on purchases spanning” multiple product lines, with the size of the rebate depending on the extent to which the customer met growth targets established by 3M. 3M's bundled discounts apparently proved successful in shifting a large fraction of private label tape business to 3M. In 1992, LePage's filed a four-count antitrust suit against 3M that was ultimately narrowed to a monopolization claim under Section 2 of the Sherman Act.8 LePage's argued that it was foreclosed from selling private label tape because it could not cover its costs and still compensate customers for the rebates lost on other products in the discount program when the customer bought private label tape from LePage's instead of from 3M. The legal strategy of the defendant, 3M, was to argue that as long as the price of the bundle exceeded its cost, the pricing scheme was legal. In other words, 3M argued that the relevant case law was that of predatory pricing, as presented in Brooke Group. 9 In its appellate filings following an initial jury verdict in favor of LePage's, 3M used a different approach, the Ortho test. In Ortho, the defendant offered bundled discounts on products, 10 some of which were competitive and some of which were monopoly products due to patents. The District Court in Ortho allocated the total bundle rebate to the competitive product in question and compared that globally discounted price to the defendant's cost. This approach takes the idea behind the Brooke Group test, and for bundled rebates, applies it incrementally to the competitive product. According to this test, if the discount-adjusted price of the competitive product exceeds its cost, then the bundled discount pricing strategy is deemed not anti-competitive. In LePage's, a circuit court panel reversed the original verdict and found that the plaintiff had to show that 3M prices were below some measure of cost (where costs might include rebates provided on other products). LePage's provided no such price-cost proof, yet when the case was reheard en banc, the Third Circuit restored the original jury verdict and rejected the necessity of showing that 3M had failed a price test. To quote the U.S. Department of Justice: …The court of appeals was unclear as to what aspect of bundled rebates constituted exclusionary conduct, and neither it nor other courts have definitively resolved what legal principles and economic analyses should control.11 The great interest in LePage's probably stems from the fact that bundled rebates are a widespread business practice, yet the courts have provided little guidance about when they may violate the antitrust laws. This dissatisfaction with the LePage's decision has been echoed in many commentaries, including Antitrust Modernization Commission (2007): Because the court failed to evaluate whether 3M's program of bundled rebates represented competition on the merits, its decision offers no clear standards by which firms can assess whether their bundled rebates are likely to pass antitrust muster.12 Employing a bundled loyalty discount imposes switching costs because a customer that shifts some purchases to a rival seller pays a higher unit price for the quantity still purchased from the incumbent. In VAA/BA, Virgin Atlantic proposed an incremental sales test similar to the Ortho test to judge the legality of such pricing plans. This test compares the competitor's incremental revenue from making those new sales (e.g. from adding a frequency on a trans-Atlantic route) to the competitor's incremental cost. Because the competitor must compensate the customer for the switching cost, net incremental revenue may be forced below incremental cost. In such a setting, “incremental” competition is unprofitable and is deterred by the loyalty discount plan. The VAA/BA test simply checks whether net incremental revenue exceeds incremental cost, and if it does, the loyalty discount plan is presumed legal. Like Brooke Group and Ortho, this test analyzes bundled rebates using predation case law. The three tests (Brooke Group, Ortho, VAA/BA) are price/cost tests that differ in the “competitive increment” over which revenue and cost are compared—in Brooke Group the increment is all sales of all goods, in Ortho the increment is all sales of the competitive product (e.g. transparent tape), and in VAA/BA, the increment may be smaller (not necessarily all sales of a competitive product). 13 The academic economics literature on bundled discounts has been rather sparse until recently. Bundling for the purpose of price discrimination has a large literature, but usually in a monopoly setting. An important exception is Nalebuff (2004a), who has written extensively on how bundling can be used to manipulate competition and deter entry. Discounting that applies to every unit purchased has a smaller literature. Kolay, Shaffer, and Ordover (2004) examine a monopolist's use of all-units discounts on a single product, and compare them to menus of two-part tariffs. They find that all-units discounts eliminate double marginalization in full information settings (as two-part tariffs do), and extract surplus more efficiently than a menu of two-part tariffs when there is asymmetric information regarding the state of demand. Greenlee and Reitman (2006) analyze loyalty discounts, both with and without bundling. In a duopoly differentiated products setting in which both firms can offer a loyalty program to a retailer that sells to consumers with varying tastes, they show that firms pass the VAA/BA test in equilibrium. Thus failing the VAA/BA test suggests that a firm has goals other than short-term profit maximization. In a setting similar to LePage's, they show that introducing bundled discounts causes standalone prices to rise. The welfare effects of bundled rebates in Greenlee and Reitman (2006) are ambiguous, since their models allow for price discrimination across consumer types. 14 Nalebuff, 2004b and Nalebuff, 2005 focuses on the use of bundled discounts to foreclose competition, and establishes results and intuitions similar to our Theorem 1 under general demand conditions. 15 Caminal and Claici (2007), Carlton and Waldman (2006), and Ordover and Shaffer (2007) consider the use of loyalty discounts in dynamic settings. In a setting with many similarly-situated firms, Caminal and Claici (2007) illustrate how intertemporal loyalty discounts (e.g. frequent flyer programs) operate as business-stealing devices that ultimately benefit consumers. When there are intertemporal switching costs for consumers, Carlton and Waldman (2006) and Ordover and Shaffer (2007) show how an incumbent firm's use of a loyalty discount can profitably (and inefficiently) exclude by denying scale to a disadvantaged potential entrant. From the standpoint of antitrust enforcement, it is unclear what body of case law, if any, best applies to bundled discounts. The approach of the Ortho and VAA/BA incremental profit tests suggest that bundled rebates should be analyzed as a method of predation. In LePage's, 3M argued vigorously in its appellate briefs that its discounts were legal under Brooke Group because even its discounted bundle price exceeded cost, an assertion not challenged by the plaintiff. However, because the intent of bundling is to induce consumers to buy multiple products from a single seller, perhaps tying is a more appropriate way to look at bundled discounts. Indeed, the Third Circuit suggested this in LePage's, but went no further. It is worth noting that in LePage's and in Ortho, the firm offering the loyalty program included in the bundle a product over which it had either a monopoly (in Ortho, due to patents) or a very strong brand name (in LePage's, Scotch tape). Monopoly goods within the bundle, then, may play the role of tying products. As we show below, however, bundled discounts are not always equivalent to tying. Although our focus is on the short term competitive effects of bundled loyalty discounts, a full analysis would also consider the longer term implications, as new competitors contemplate entry and/or incumbent competitors consider exit. Similar to tying, bundled rebates can change the terms on which firms compete in the linked market. As we will discuss, however, the manner in which bundled rebates influence the dynamics of competition differ from the effects of tying. To summarize, this paper provides preliminary answers to two main questions. First, what is the correct legal framework to use in analyzing bundled loyalty discounts? Second, how can one distinguish between bundled discounts that benefit consumers from ones that do not? We reach the following conclusions. • When the tied market is perfectly competitive, bundled discounts resemble tying and are not a form of predatory pricing. • Bundled discounts can raise or lower consumer welfare. In either case, equally efficient competitors may be foreclosed. Profit-maximizing bundled discounts lower consumer surplus when the tied market is perfectly competitive. • If products in the tied market are homogeneous, simple price comparison tests exist that can distinguish bundled rebates that raise consumer surplus from those that do not. • These tests suggest that the bundled discounts at issue in SmithKline lowered consumer surplus and total surplus. • If the rivalrous market is a differentiated product duopoly, bundled discounts are not equivalent either to tying or to predation. In such cases, profit-maximizing loyalty programs have ambiguous welfare effects, and aggregate consumer surplus can rise or fall due to bundling. Rather than tying, bundled discounts can resemble mixed bundling. • In the differentiated product setting, bundled discounts reduce rents for the non-bundling firm. Numerical examples indicate that bundled discounts can deter entry or induce exit. The remainder of this article is organized in the following manner. Section 2 introduces our base model of a monopolist in one market that faces perfect competition in a second market. 3 and 4 discuss existing tests of bundled discounts and introduce a generalized test. Section 5 applies this test to SmithKline. Section 6 briefly looks at nonlinear pricing and establishes that a monopolist benefits from offering a menu of bundled “three-part tariffs” that is not feasible without bundling. Section 7 modifies the initial model by making the second market a differentiated duopoly. Section 8 concludes.

A main point of this article is that bundled rebates are not usefully analyzed using predatory pricing case law. Not only do prices typically exceed marginal cost, but anticompetitive effects may not require a short term profit sacrifice or a period of recoupment. Bundled rebates can generate anticompetitive effects, but they do so by con- fronting consumers with the choice between a collection of tied discount prices and unattractive standalone prices, all above cost. The analogy to tying, however, does not fit bundled rebates especially well either. When the rivalrous mar- ket has homogeneous products (Model 1), equilibrium bundled rebates act as a tying arrangement (Theorem 2). When there is product differentiation in the rivalrous market (Model 2), bundled loyalty discount programs, like mixed bundling, also provide a means to price dis- criminate between heterogeneous A -and- B consumers. The standalone price of A has a price discrimination role quite apart from inducing consumers to buy the bundle. This is unlike tying. This brings us to the second main issue in this article: are there tests that distinguish good bundled rebates from bad? Although the Ortho and VAA/BA tests have their uses, the discussion above suggests that they should be used with care, especially if loyalty programs are tailored to individual consumers. In Model 1, where sellers in B produce homogenous products, there may be useful tests for consumer welfare changes based on a comparison of the monopoly price of A before and the standalone price of A after the institution of bundled rebates. Absent a reliable estimate of the unbundled monopoly price of A , they depend on there being a distinct date at which bundled discounts are introduced. For example, if B is perfectly competitive, the dis- counted price of A falls, and the standalone price equals the pre-bundle monopoly price, then consumer surplus and total surplus have increased, even if the bundle price of B exceeds marginal cost. Alternatively, if the stand- alone price of A exceeds the initial price of A , then we can infer that the bundled rebate reduces consumer welfare. This latter test does not apply to Model 2In Model 2, the pricing effects of bundling are com- plex and can raise or lower consumer and total welfare, even if the standalone price of A exceeds the pre-bund- ling monopoly price. The non-bundling firm is induced to lower its price, so in that sense a bundled loyalty discount introduced by the monopolist increases com- petition in the B market. On the other hand, the monopolist raises its price of B . Furthermore, as the share of A- and- B consumers increases, both the stand- alone and bundled prices of A fall, as competition intensifies for A -and- B consumers with a preference for firm 2. Given the potential for influencing the entry/exit decisions of competitors, there do not appear to be simple testsforwhetherbundlinghasreducedwelfareasthereare when the market for B is perfectly competitive. We conclude with a discussion of what our analysis implies about when bundled loyalty discounts should be held to violate antitrust laws. In Model 1, bundled discounts work because they act in a fashion similar to a two-part tariff. They do not affect the price that a B -only consumer pays, and their main impact is on the consumer surplus of those who buy both A and B . Furthermore, if the monopolist can extract all surplus in the A market without bundling, then bundled rebates do not raise profits. Intuitively, bundled rebates achieve what im- proved nonlinear pricing would accomplish in the monopoly market. Because the bundling firm is a monopolist in A , one might argue that doing a better job of extracting surplus from a rightly earned monopoly should not be an antitrust offence. On the other hand, using a competitive markettoextractmonopolyrentsmay raise price to consumers that do not purchase the mono- poly good, as well as generate deadweight losses in the competitive market that exceed the deadweight losses generated by a monopolist that does not link markets. Model 2 has more complex implications for antitrust. Because bundled discounts can induce exit or deter entry, they have the potential to be anti-competitive by most mainstream interpretations of Section 2 of the Sherman Act. Even absent exit/entry effects, bundled discounts affect the welfare of B -only consumers. The conditions that determine whether aggregate consumer welfare rises or falls, however, are subtle and likely hard to measure in practice. This suggests that prospective antitrust enforce- ment for bundled discounts should be conservative.