This article considers whether or not the bundling of two identical products is usefully treated as a tying arrangement. Following Prof. Williams’ article in this issue, I discuss three tying cases in which the defendants bundled two products that were arguably identical. Of the three, I focus on the EpiPen case. Bundling two identical products might appear to be a tying arrangement because the purchase of one product in the bundle is conditioned on the purchase of a second product, identical to the first. Notwithstanding this, I conclude that the bundling of two identical products is not usefully analyzed by tying analysis. Doing so adds no useful incremental economic insight as to the anti-competitive effects, if any, of this type of bundling arrangement.
By David S. Sibley1
I. INTRODUCTION
Elsewhere in this issue, Professor Melanie Williams discusses the fact that there have been tying claims that have centered on the bundling of two products that are, arguably, identical. Judicial reactions to such claims have been inconsistent. In Paul v. Pulitzer and Metromedia v. MGM/UA, courts ruled that the products in question were identical and bundling them was not an illegal tie.
In EpiPen, however, the court was open to a different view2 in a motion to dismiss. EpiPen is a class action case filed against Mylan, the exclusive distributor of the EpiPen, the most widely used Epinephrine auto-injector device (“EAI”). Although EpiPens had been sold one at a time since they were introduced in 1987, in 2011 Mylan announced that would only sell EpiPen in packets of two (the “2-Pak”).
The plaintiffs argued that the 2-Pak constituted an illegal tie. Mylan filed a motion to dismiss, based, inter alia, on the claim that the 2-Pak could not be a tying arrangement because each EpiPen unit in the bundle was identical, and not a separate product.
However, the court found that the plaintiffs had pled a tying cause of action sufficiently well to overcome a motion to dismiss. The basis of the plaintiffs’ argument was that users sometimes wanted one EpiPen to serve as a backup to another EpiPen. This fact allegedly differentiated the two EpiPens.
Professor Williams has addressed the legal issues involved. In this article, I describe the economic issues. I will focus on the EpiPen case, because it comes the closest to one in which a firm sells two literally identical products as a bundle. The other two are similar, but the bundled products in those cases are less obviously identical than in the EpiPen case. The EpiPen case is ongoing, so my discussion, like that of Professor Williams, is based on the court’s memorandum rejecting Mylan’s motion to dismiss. I will proceed, as did the court, by accepting the facts asserted by the plaintiffs as true.
I conclude that if a firm with market power bundles two identical products, the competitive effects of doing so are not usefully addressed by tying analysis. This is true even if one of the identical items in the bundle were used on different occasions or otherwise differently than the other. Rather, the bundling of two identical products is better viewed as one way to exploit market power by increasing prices in a single market. Based on the facts cited by the court in EpiPen, characterizing identical product bundling as tying does not add economic insight to the analysis of its economic effects.
II. TRADITIONAL TYING ANALYSIS
It is well known that tying and bundling are innocuous in many situations, and may even be pro-competitive in some circumstances. For example, tying can be a method of price discrimination that a firm without market power might use as part of its ordinary competitive strategy. In other cases, tying may enable a seller to assure quality control in aftermarkets. Tying may also solve certain types of free rider problems.3
However, in cases in which the firm doing the bundling (call it Firm 1) possesses market power in one product in the bundle, tying may have anti-competitive effects. Suppose that Firm 1 has market power in the sale of a product A and competes in the market for another product, B, where it is not the only seller.
The products A and B must be shown to be separate products, based on accepted market definition analysis. In many cases, this is done by focusing on the functional characteristics of the two products. For example, A and B may be complements. As noted, the EpiPen plaintiffs argue that even if A and B are perfect substitutes in a functional sense, they could be considered different products if they are used on different occasions for different purposes.
Given separate products A and B, an anti-competitive tie must involve coercion.4 That is, a consumer buys B from Firm 1 only because that is required in order to buy A from Firm 1.
Coercion leads to competitive harm because Firm 1 will typically set a price for B higher than those set by its rivals in the market for B. Consumers are willing to pay this higher price only because of the coercion. Typically, output in the B market declines, due to Firm 1’s tying arrangement.
In practice, A and B are often complements. Examples are: (1) Microsoft bundling the Internet Explorer browser with its Windows operating system;5 and (2) Jefferson Parish Hospital tying surgical facilities to the use of its own anesthesia6 services. This fact makes the three cases discussed herein atypical.
III. THE EPIPEN CASE
In 2007, Mylan acquired exclusive distribution rights to EpiPen. Since 1987, most users only had bought a single EpiPen at a time. However, a relatively small percentage of EpiPen users7 also needed a second EpiPen, to serve as a backup.
In 2011, Mylan changed its pricing policy. From that point on, EpiPens could only be bought in “2-Pak” containers. Even those who did not want a second EpiPen were forced to buy one. At the same time, the price of an EpiPen went up drastically, from about $100 to about $600.
From an economic standpoint, the question arises: why did competitors not undercut these prices and offer the equivalent to the EpiPen 2-Pak, but sold one unit at a time? One answer is that the EpiPen patents were still in force at that time, although different methods of injecting epinephrine were available and had been approved by the FDA. Mylan sued four generic entrants for infringement, each of which had its own type of injection device. They were: Sandoz, Intelliject, Sanofi, and Teva. Sanofi, Intelliject, and Teva all made agreements with Mylan to delay market entry. Sandoz was also sued, and stopped its entry process.
During this time, Mylan’s market share was in excess of 80 percent. Much of this was due to exclusivity agreements that it had with pharmacy benefit managers (“PBMs”) and school systems. During the period of 2013-2015, Mylan’s market share was in the high 90-percent range.8 It was during this period that the conduct at issue is alleged to have taken place.
IV. ANALYTICAL ISSUES
As noted by Prof. Williams, the first issue concerns whether or not it makes sense to think of the two identical EpiPens in the Mylan bundle as separate products. From a purely functional standpoint, of course, they are the same product.
A tying claim also requires that one product be defined as the tying product, and the other as the tied product. This is important because the locus of anticompetitive harm in a tying case is the tied product market. A claim regarding the bundling of two identical products requires the analyst to identify which of the two identical EpiPens is the tying product and which is the tied product.
Finally, are the alleged competitive effects of the Mylan bundling strategy similar to those in a traditional tying case? A key concern in tying analysis is whether or not the market power that Firm 1 possesses in the tying product is extended, via tying, into the tied market.
Regarding the first issue, the defendants in EpiEpen argued that the Mylan bundle did not consist of two products, because the two EpiPens in each 2-Pak are identical. The plaintiffs argued that the two EpiPens in the bundle could be distinguished by differences in their intended use.
However, even accepting this “occasion of use” argument, the relevance to formal tying analysis is not clear. Which of the two identical EpiPens is the tying product? Which is the tied product? If a patient has the type of serious allergic reaction calling for the use of epinephrine, it is likely that the patient would simply reach for the first EpiPen that he or she could extract from the package.9 There is no meaningful distinction between the two EpiPens in the bundle.
This makes it hard to see how the EpiPen bundling constitutes a tying arrangement that makes a tied market less competitive than it would be absent the tie. It may well be that Mylan’s conduct caused the market for EAIs to be less competitive than it might otherwise have been. Even so, it is not clear that this harm to competition occurs in a separate market. If all EAIs constitute an antitrust product market, for example, then harm to competition may have occurred in that market, but not in another “tied” market.
However, it is hard to see how the Mylan bundling policy, by itself, could have caused harm to competition. EAI technology was well known by 2013, and the firms mentioned above had obtained regulatory approval to provide generic products that served the same purpose as the EpiPen did. If the other exclusionary conduct engaged in by Mylan had not taken place, any attempt by Mylan to use bundling to increase prices would likely have been undercut by generic entry in 2013.
This entry did not occur on large scale, however. Accepting the plaintiffs’ allegations as true, Mylan’s exclusivity arrangements with PBMs and school systems likely led to its high market share, reducing opportunities for generic competition.
Its aggressive litigation policy towards entrants almost certainly allowed it to deter entry for an appreciable period of time. The basis for this claim is that the Sanofi, Teva, and Intelliject lawsuits all led to pay-to-delay agreements.
Had entry occurred on a large scale in 2013-2015, the Mylan bundling arrangement could easily have been duplicated by generic entrants. Mylan would likely have faced downward pressure on prices and would likely have lost market share to generic entrants. Today, even Mylan offers a generic version of the EpiPen at prices equivalent to those of its competitors.10
I conclude from this that the Mylan bundling strategy, by itself, is not likely to have led to reduced competition in the market for EAIs, or in any other market. Rather, it may simply have been a way to increase prices in a single EAI market, something made possible by other types of entry-deterring conduct engaged in by Mylan.
So what were the competitive effects of Mylan’s conduct, assuming that the plaintiffs’ allegations were true? They would be much the same as those alleged by the plaintiffs. A substantial number of consumers were forced to buy two EpiPens, the great majority of whom only wanted one EpiPen. Others may have wanted a backup EAI, but would have preferred to buy a cheaper generic EAI. Accepting the plaintiffs’ factual allegations as true, prices were higher than they would have been absent the bundling.
The bundling policy of Mylan may, indeed, have been part of a pattern of conduct that violated antitrust law. However, based on the facts set out in by the court, it does not seem useful to analyze that conduct using the framework of tying. Rather, the economically significant conduct is likely to have been Mylan’s successful efforts to obtain exclusivity with large customers and to deter entry by means of pay-to-delay agreements with potential entrants. These exclusivity arrangements were needed for the bundling to be effective.
V. SIMILAR CASES
In the court’s analysis of EpiPen, it noted two other cases that seemed germane. In each, a plaintiff alleged an anticompetitive tying arrangement. In each case, the defendant countered that the products alleged to have been tied were not separate products.
In Paul v. Pulitzer Publishing Co., two distributors of the St. Louis Post-Dispatch desired either not to distribute the Saturday edition of the paper, or wanted to price the Saturday edition somewhat higher than the weekday editions. The newspaper had required them to deliver the Saturday edition as a condition of being able to distribute the other daily editions and not to price them any differently. The plaintiffs argued that this constituted an illegal tying arrangement.
This case looks somewhat different from EpiPen in the sense that the demand for the Saturday edition seems to have been different – and lower – than for the weekday editions.11 Arguably, the Saturday news is of a different type than weekday news, and was less in demand. This appears to have been why the plaintiffs did not want to distribute the Saturday paper. This contrasts with EpiPen, in which the two products in the bundle were literally identical.
In the court’s decision, there is no reference to any effort by either party to investigate the substitutability of the Saturday paper and weekday papers. By the standards of modern antitrust economics, this is a significant omission. Therefore, whether or not the alleged tie and tying products were in separate markets cannot be inferred from the decision.
Therefore, in purely economic terms, it is not clear whether the above discussion of EpiPen is relevant to Paul v. Pulitzer.
The other case is Metromedia Broad. Corp. v. MGM/UA Entm’t Co.12 In that matter, the issue concerned MGM/UA’s syndication of episodes of the television series FAME. MGM/UA required that Metromedia agree to show reruns of this series as a condition of being allowed to license first-run episodes.
Metromedia claimed that to require the licensing of reruns as a condition of licensing new episodes constituted an illegal tie. MGM/UA argued that all FAME episodes were the same product, covered by the same licensing rights.
The court sided with MGM/UA, ruling that both new episodes and re-runs were a single product. The court summarized MGM/UA’s reasoning as follows: “MGM/UA contends…that there is no unlawful tie-in because the bundle of rights subsumed under the FAME copyright are of a single product which distinguishes the packaging of FAME first runs and FAME reruns…”13
This approach seems to ignore the product substitutability approach normally taken in market definition. This is because the court’s reasoning did not focus on the consumer demand characteristics of each product. Rather, the court focused on the fact that they were both licensed under a single set of rights.
It is possible, hypothetically, that first runs and reruns appealed to very different viewer populations and did not compete with each other to a significant degree. A hypothetical monopolist test, therefore, would find them to be separate products. Under the court’s approach, though, they would still be found to be a single product.
VI. CONCLUSIONS
The three cases discussed above show that two courts have accepted the view that a bundle of two allegedly identical products cannot be a tying arrangement. The EpiPen case suggests that an “occasion of use” argument may be accepted at some future date. This might open the door to applying tying analysis of bundles containing identical products.
However, this seems incorrect. If there are no functional differences between two products bundled together, deciding which of two identical products is to be the tying product, and which the tied product, is problematic. Whatever the details of the conduct at issue, tying analysis may not add much insight.
1 John Michael Stuart Centennial Professor of Economics, The University of Texas at Austin.
2 In re EpiPen Mktg., Sales Practices & Antitrust Litig., 336 F. Supp. 3rd 1256 (D. Kan. 2018).
3 Carlton, D.W. and J.M. Perloff. 2000. Modern Industrial Organization, Addison Wesley, Longman at 303.
4 Jefferson Parish Hospital District No. 2 v. Hyde 466 U.S. 2(1984).
5 U.S. v. Microsoft Corporation, 253 F. 3d 34 (D.C. Cir. 2000).
6 In Re Jefferson Parish Hospital District No. 2 v. Hyde 466 U.S. 2 (1984).
7 Said to be “up to 20%.” In Re EpiPen at 1284.
8 In Re EpiPen at 1277.
9 Imagine trying to implement a SSNIP test to define a market for only a backup EAI. Because the “primary” EAI and the ”backup” EAI are only distinguished by labels of convenience how could the analyst even set up the SSNIP analysis?
10 “FDA Approves fist generic version of EpiPen,” FDA press release, August 16, 2018.
11 No. 74-327(CA) Dated May 24, 1974.
12 611 F.Supp. 415 (C.D. Cal 1985).
13 Ibid. at 422.
