\\jciprod01\productn\M\MIA\68-1\MIA103.txt unknown Seq: 1 19-NOV-13 17:04 Inconsistency in Antitrust RAMSI A. WOODCOCK* I. INTRODUCTION ...................................................... 106 R II. A FIRST LOOKAT INCONSISTENCYIN ANTITRUST.......................... 109 R A. Pro Monopoly and Anti Cartel, Without Basis........................ 109 R B. The Per Se Rule in Favor of Monopolization......................... 116 R C. What It Means for Refusal to Deal to Exempt Property-Based Exclusion.. 120 R D. The False Distinction Between Collusion and Exclusion................ 123 R E. Section Two as a Dead Letter...................................... 125 R III. NO JUSTIFICATIONFORA BLANKET EXEMPTION: MONOPOLY POWER IS STILLA PROBLEMINA WORLDOF PROGRESS.................................... 126 R A. An Antitrust and Product Improvement Model ........................ 126 R B. The Model Shows That There Is No Reason to Assume That Monopoly Pricing Is Efficient, Even When Firms Spend Their Profits on Improving Their Products .................................................. 133 R IV. INPUT CONTROLASA NECESSARY CONDITIONFOR ANY KINDOF MONOPOLY INEFFICIENCYANDTHE INCONSISTENT WAYIN WHICH ANTITRUST RESPONDS TO IT .............................................................. 136 R A. Types of Exclusion: Input Control and Product Transformation ......... 137 R B. The Old Types, the New, and the Law: A Patchwork .................. 142 R C. Collusion Excludes, so Exclusion Is Not More Fundamental Than Collusion ....................................................... 143 R D. Pro Monopoly and Anti Cartel, Starkly.............................. 146 R E. The Refusal to Deal Exemption as Modified by Merger and Transformation Law.............................................. 147 R F. No Reason to Think That Superior Product or IP-Based Exclusion Is Any More Likely to Result in Better Products Than Other Forms of Exclusion. 150 R G. Pro Property and Anti Contract, Starkly............................. 152 R H. Denial and Avoidance: Cartels Refuse to Deal ....................... 153 R V. AGAINST REFUSALTO DEAL........................................... 154 R A. The Cost of Breakup Is Not a Good Argument for the Refusal to Deal Exemption ...................................................... 154 R B. An Example of Unclear Thinking About Breakup...................... 162 R C. Imagining a World Without the Refusal to Deal Exemption............. 165 R D. Antitrust and Theft ............................................... 166 R E. Essential Facilities Cannot Be Reconciled with Refusal to Deal Because It Is Sworn to Destroy It ............................................ 167 R VI. NO EXEMPTIONFOR REFUSALSTO DEAL MEANS NO MORE PER SE RULEIN FAVOROF IP-BASED EXCLUSION ....................................... 168 R VII. MORE INCONSISTENCIES............................................... 171 R A. Pro Vertical and Anti Horizontal................................... 171 R B. Inconsistent Treatment of Natural Monopolization..................... 172 R C. Hard- and Soft-Core Cartels but No Hard- or Soft-Core Monopolies..... 175 R D. Inconsistent Application of Remedies: Input Control for Transformation but No Transformation for Input Control ............................ 177 R * Bruce R. Jacob Visiting Assistant Professor of Law, Stetson University College of Law. Ian Ayres, Jonathan Baker, Dennis S. Corgill, Robert Feinberg, Timothy S. Kaye, F.M. Scherer, Martha Starr, and Louis J. Virelli III provided helpful comments or criticism. 105 \\jciprod01\productn\M\MIA\68-1\MIA103.txt unknown Seq: 2 19-NOV-13 17:04 106 UNIVERSITY OF MIAMI LAW REVIEW [Vol.68:105 VIII. THE BALANCING TEST IS FLAWEDANDTHE MONOPOLY POWER REQUIREMENT IS REDUNDANTAND WORSE ........................................... 178 R IX. CONCLUSION ........................................................ 183 R I. INTRODUCTION When the price of a good is too high, consumers who can afford to pay cost, including enough profit to make production worth the manu- facturer’s while, but cannot pay enough to meet the high price, are forced to do without. Economics teaches that efficiency would increase if price were to fall to cost because at cost the manufacturer would still be glad to produce and consumers could now afford to purchase more of the good. Efficiency requires that where more for less is possible, more must be had for less. If the point of antitrust is to maximize efficiency, then the point of antitrust is to drive prices down when they are too high. It follows that antitrust should make the charging of a monopoly price illegal, at least where such a price is not necessary to cover costs. Strangely, antitrust does this for cartels but not for monopolies. Cartels that fix inefficiently high prices are illegal (often they are illegal regardless even of the prices they charge), but courts and commentators have long been at pains to stress that a monopoly does not violate the law simply by charging an inefficiently high price. Antitrust has typically explained this hole in the law in three ways. First, it has argued that market forces compensate for the absence of antitrust in this area (the “entry argument”). Antitrust currently prohibits monopolies from taking affirmative steps to exclude competitors, such as tying, exclusive dealing, or predatory pricing. So long as a monopoly does not engage in such prohibited conduct, competitors will enter whenever the monopoly charges an inefficiently high price and drive price back down. Antitrust therefore does not need to intervene. Second, antitrust has argued that monopolies are more efficient than cartels (the “efficiency argument”). Monopolies have higher costs because they spend more on research and development (“R&D”), so they need to be able to charge higher prices. More cannot be had for less. Third, antitrust has argued that even if the hole is a real problem, there is nothing that antitrust can do about it (the “breakup argument”). Breaking up cartels is cheap (just dissolve the cartel agreement); breaking up monopolies is expensive. This article shows that these are flawed arguments. The efficiency argument is flawed because, among other reasons, even if we accept that monopoly profits finance R&D that leads to innovation, it does not fol- \\jciprod01\productn\M\MIA\68-1\MIA103.txt unknown Seq: 3 19-NOV-13 17:04 2013] INCONSISTENCY IN ANTITRUST 107 low that this relationship does not hold for cartels as well1 or that there is no longer such a thing as too high a price.2 The breakup argument is flawed because, among other reasons, breaking up monopolies means breaking up essential inputs, not the more daunting task of trying to reallocate all of a firm’s inputs between competitors by fiat. The market sorts that out.3 The entry argument fails because it does not recognize that monop- olies can exclude competitors passively, simply by using property rights to deny competitors access to essential inputs, without having to engage in affirmative acts of exclusion. There is no reason to think that in the absence of affirmative acts of exclusion, the market will always natu- rally respond to inefficiently high pricing by ramping up output and driving price back down. This article proposes input control as a unifying concept in antitrust and shows how existing categories in the field can be mapped onto vari- ous legal and extra-legal forms of input control.4 A necessary condition for monopoly power, whether exercised by a cartel, monopoly, or other entity, is control by the entity over inputs that competitors might other- wise use to enter the market. Without such control, competitors can use the inputs to enter and drive price down. From the perspective of input control, cartels and monopolies differ only in the legal form of their control over inputs: cartels use contract to control inputs (by using con- tract to control the inputs’ ultimate owners), whereas monopolies use property to control them. When a cartel forms, it uses contract to central- ize control over inputs in the hands of a single competitor; when a monopoly forms, it uses property to do the same thing. In this sense, collusion and exclusion are the same. You might say that collusion excludes, or exclusion colludes. This article argues that antitrust breaks up overcharging cartels but not overcharging monopolies because it harbors a bias in favor of prop- erty-based exclusion. A cartel can charge a monopoly price because, via contract law, it controls enough inputs to prevent competitors from entering to drive price down. In breaking up cartels, antitrust in effect recognizes that this contract-based control is exclusionary. A monopolist can charge a monopoly price because, via property law, it controls enough inputs to prevent competitors from entering to drive price down. It is a major blindspot in antitrust that it fails to recognize that such property-based control is equally exclusionary. 1. See infra Part II. 2. See infra Part III. 3. See infra Part V. 4. See infra Part IV. Chart 2 summarizes this mapping. \\jciprod01\productn\M\MIA\68-1\MIA103.txt unknown Seq: 4 19-NOV-13 17:04 108 UNIVERSITY OF MIAMI LAW REVIEW [Vol.68:105 The doctrinal expression of antitrust’s bias in favor of property- based exclusion is the refusal to deal doctrine, which, with some small exceptions, guarantees the right of firms to refuse to share their resources with competitors, regardless whether this leads to inefficiently high prices. Plugging the hole in antitrust means doing away with the general rule that firms have a right to refuse to deal. This has important consequences for intellectual property (“IP”) law. The bias in favor of property-based exclusion has prevented antitrust from coming into direct conflict with IP in cases in which firms use IP to exclude. Without the bias, antitrust can take on the role of preventing inefficiently high pric- ing by IP-based monopolies.5 The unjustified favor antitrust accords monopolies relative to other forms of input control is expressed in other areas of antitrust in addition to refusal to deal. Antitrust regulates monopolies only when they form through merger or are created or maintained by affirmative acts of exclusion other than the mere exercise of property-based control over inputs. In the case of affirmative acts (what antitrust calls “exclusionary conduct”), antitrust applies a balancing test that approves all acts that increase efficiency, regardless whether they maximize it. But many acts that increase efficiency do not maximize it. As a result, current law approves of many more affirmative acts of exclusion than would be per- mitted under an efficient antitrust regime.6 In the case of merger, anti- trust scrutinizes merger to natural monopoly for efficiency. But it fails to break up preexisting natural monopolies even though the efficiency rationale for doing so is identical to that associated with preventing mergers to natural monopoly.7 The article is organized as follows. Part II introduces the notion that there is a hole in antitrust enforcement, calling it a per se rule in favor of monopolization. It argues that the hole is expressed in the doctrines of the conduct requirement and refusal to deal and that refusal to deal amounts to an exemption for property-based exclusion. Part II also introduces the notion that collusion and exclusion are conceptually iden- tical for antitrust purposes and argues that there is no basis for treating cartels and monopolies differently. Part III argues that the need for rewards does not eliminate the need for antitrust to constrain monopoly power. Part IV introduces the metaphysics of input control and uses it to give a more precise statement of some of the claims in Part II. Part IV also maps current law onto the input control paradigm. Part V attacks a pillar of the refusal to deal doctrine by arguing that the cost of breakup 5. See infra Part VI. 6. See infra Part VIII. 7. Part VII discusses this and other inconsistencies. \\jciprod01\productn\M\MIA\68-1\MIA103.txt unknown Seq: 5 19-NOV-13 17:04 2013] INCONSISTENCY IN ANTITRUST 109 need not be excessive as a rule. Part VI explores the consequences of the demise of refusal to deal for IP. Part VII identifies additional inconsis- tencies in antitrust enforcement, including its approach to natural monopoly.8 Part VIII critiques the balancing test. II. A FIRST LOOK AT INCONSISTENCY IN ANTITRUST A. Pro Monopoly and Anti Cartel, Without Basis It is apparent to any careful student that, as far as textbook econom- ics is concerned, the organizational structure of firms should be irrele- vant to antitrust. Textbook antitrust tells us that the goal of an efficiency-oriented antitrust is to eliminate the black triangle in Diagram 1 on page 110.9 The socially optimal price is C. If the price in the market is P, then consumers cannot afford goods corresponding to QQ′ even though they are willing to pay more than their cost of production in order to obtain them. The black triangle represents the value consumers lose from not being able to enjoy these goods, which also happens to be the loss to society generally from this underproduction. That is all text- book economics has to say about efficiency and antitrust.10 8. One inconsistency that is not discussed at length in this article is the exemption for collusion when organized as oligopoly but its condemnation when organized as cartel. This inconsistency has been much discussed elsewhere. See, e.g., RICHARD A. POSNER, ANTITRUST LAW ch. 3 (2d ed. 2001). 9. The simple model described by Diagram 1 ignores economies of scale and product and process improvements. Improvements are dealt with in Part III. Economies of scale and the problem of natural monopoly are dealt with in Parts II.B, IV.A (including, in particular, note 87 R therein), and VII.B. 10. See William J. Baumol, Horizontal Collusion and Innovation, 102 ECON. J. 129, 129 (1992) (“The familiar textbook description of the social cost of monopoly applies also to horizontal collusion.”). Textbooks do not generally make this point explicitly. Instead, it is to be inferred from the fact that textbooks usually only provide a single model of market inefficiency (e.g., Diagram 1). The reader is left to infer that any inefficient organizational form must be inefficient in the style of that model. This is starkest in the way textbooks generally treat monopolies and cartels. The textbook will explain inefficiency in the context of a monopoly, and then subsequently remark that cartels have power over price as well. The reader is left to infer that cartels are therefore inefficient in the same way. See, e.g., DENNIS W. CARLTON & JEFFREY M. PERLOFF, MODERN INDUSTRIAL ORGANIZATION 95–96, 122 (4th ed. 2005) (“A cartel that includes all firms in a market is in effect a monopoly . . . .”); F.M. SCHERER & DAVID ROSS, INDUSTRIAL MARKET STRUCTUREAND ECONOMIC PERFORMANCE 235, 661–67 (1990); HERBERT HOVENKAMP, FEDERAL ANTITRUST POLICY 19–21, 158 (4th ed. 2011) [hereinafter HOVENKAMP, FEDERAL ANTITRUST POLICY]; HAL R. VARIAN, INTERMEDIATE MICROECONOMICS 431–33, 438 (7th ed. 2006). The point is also captured in Richard Posner’s call for the repeal of all antitrust laws other than Sherman Act Section 2 because “[i]f by ‘monopolizing’ we were to mean simply conduct that unjustifiably promotes supracompetitive pricing, we would cover all practices . . . that pose a threat to the maintenance of competition in the economic sense.” POSNER, ANTITRUST LAW, supra note 8, at 260. R \\jciprod01\productn\M\MIA\68-1\MIA103.txt unknown Seq: 6 19-NOV-13 17:04 110 UNIVERSITY OF MIAMI LAW REVIEW [Vol.68:105 DIAGRAM 1 P C Q Q' Textbook economics has no interest at all in how price might have ended up at P instead of C. Price might get there because independently owned firms agree to charge the same price. Or because for whatever reason there is only one firm in the industry and it chooses to charge P. Or because gangsters threaten to kill anyone who does not sell at P. Or because all the firms in the industry have delegated pricing authority to an independent trustee who has chosen P.11 Or because all the firms in the industry are community-oriented folk interested in being good neigh- bors to the other firms in the industry rather than competing hard with them on price.12 Or it might get there by accident. The point is that if the goal of antitrust is to make the black rectan- gle go away,13 antitrust does not care whether it got into Diagram 1 through cartelization, monopolization, gangsterism, neighborliness, or 11. See HERBERT HOVENKAMP, ENTERPRISE AND AMERICAN LAW, 1836–1937, at 249–50 (1991) (describing the operation of a stock-transfer trust). 12. In this article, “oligopoly” means not just textbook models in which price ends up somewhere above C and below P, but any kind of monopoly inefficiency due to the behavior of multiple firms that are not explicitly communicating to coordinate their actions (i.e., those that are not cartels). This includes oligopolies that can charge P. Economics knows that any price can be an oligopoly equilibrium. DAVID M. KREPS, A COURSEIN MICROECONOMIC THEORY 525 (1990). “[E]xogenous norms of cooperative behavior” determine which equilibrium obtains, and the right norms (neighborly norms) can get you to P, even in a relatively unconcentrated industry. Albert O. Hirschman, Rival Interpretations of Market Society: Civilizing, Destructive, or Feeble?, 20 J. ECON. LITERATURE 1463, 1470 (1982). In my view, “oligopoly” used in this way corresponds to what antitrust lawyers mean by “tacit collusion.” See generallyCARLTON & PERLOFF, supra note 10, at 127 n.8 (discussing collusion terminology). R 13. This article assumes that the goal is total welfare (producer plus consumer), as opposed to just consumer welfare. However, the consensus in antitrust is that consumer welfare is the standard. See Richard Schmalensee, Thoughts on the Chicago Legacy in U.S. Antitrust, in HOW THE CHICAGO SCHOOL OVERSHOT THE MARK: THE EFFECT OF CONSERVATIVE ECONOMIC ANALYSIS ON U.S. ANTITRUST 11, 13 (Robert Pitofsky ed., 2008). As a rhetorical matter, this article wishes to make the argument that even the strongest case for monopolization is flawed. Because the total approach makes the strongest case for the efficiency of monopolization, the article therefore assumes the total approach. \\jciprod01\productn\M\MIA\68-1\MIA103.txt unknown Seq: 7 19-NOV-13 17:04 2013] INCONSISTENCY IN ANTITRUST 111 accident. It just cares that it is there.14 It turns out, however, that antitrust law discriminates heavily based on how an industry gets to P. The bias is strongly in favor of punishing cartels and gangsters, and ignoring inefficiency associated with oligopo- lies (e.g., good neighbors) and monopolies. The bias is extraordinarily pronounced in enforcement practice, the case law, and scholarship.15 If textbook efficiency does not account for the difference, what does? The argument in the literature is that you often can have P without 14. Likening monopolization to the creation of an involuntary cartel, Jonathan Baker observes that “[i]t does not matter to buyers whether the cartel is voluntary or involuntary; either way, the same firms collectively reduce output and the price that buyers pay increases.” Jonathan B. Baker, Exclusion as a Core Competition Concern, 78 ANTITRUST L.J. 527, 557–58 (2013). 15. Gangsters. State criminal law prohibits the use of violence, threats, or theft, regardless whether used to restrict industry output. See, e.g., MODEL PENAL CODE §211.1 (assault); id. §223.2 (theft). Federal criminal law also reaches gangsterism relevant to monopoly inefficiency. See, e.g., 18 U.S.C. §1951 (2006) (imposing criminal penalties for use of threats or violence to obstruct commerce); Ray V. Hartwell III, Criminal RICO and Antitrust, 52 ANTITRUST L.J. 311, 312–19 (1983) (discussing application of RICO to prosecute attempts to raise prices). Cartels. The courts read the Sherman Act to prohibit per se all agreements between competitors with the purpose or effect of raising price, meaning that such agreements are prohibited regardless of efficiency. SeeHOVENKAMP, FEDERAL ANTITRUST POLICY, supra note 10, R at 279. Even though the criminal provisions of the Sherman Act may be applied to monopolization, the U.S. Department of Justice (“DOJ”) applies them exclusively to cartels that fall under the per se rule. See 2 ABA SECTION OF ANTITRUST LAW, ANTITRUST LAW DEVELOPMENTS 956 & n.9 (7th ed. 2012) (and sources cited therein). Cartel fines have exploded in recent years; not so for violators of the antitrust laws that have other organizational forms, because they are never prosecuted. See Criminal EnforcementFine and Jail Charts, U.S. DEP’TOF JUSTICE ANTITRUST DIV., http://www.justice.gov/atr/public/criminal/264101.html (last visited July 1, 2013) (showing that total criminal cartel fines increased tenfold to $1.1 billion in the ten years ending in 2012). DOJ filed one civil monopolization case in the ten years ending in 2012 and challenged 123 mergers over the same period; but it filed nearly three times as many criminal cases against cartels (345). See U.S. DEP’T OF JUSTICE, ANTITRUST DIVISION WORKLOAD STATISTICS FY 2003–2012, at 4–5, 7, available at http://www.justice.gov/atr/public/workload-st atistics.pdf. Despite a largely successful effort by antitrust opponents starting in the 1960s to reduce the scope of the antitrust laws across the board, the Supreme Court firmly rejected attacks on the per se rule against cartels and a consensus in favor of the rule now prevails among both opponents and defenders of antitrust. See Arizona v. Maricopa Cnty. Med. Soc’y, 457 U.S. 332, 347–48 (1982) (affirming continued vitality of the rule); Schmalensee, Thoughts on the Chicago Legacy in U.S. Antitrust, supra note 13 (identifying Chicago School victories); Eleanor M. Fox, R The Efficiency Paradox, inHOWTHE CHICAGO SCHOOL OVERSHOTTHE MARK, supra note 13, at R 77, 98 n.18 (describing consensus of left and right in favor of the per se rule against cartels). Oligopolies. Firms that raise price as a group without explicitly coordinating with each other are exempt from antitrust scrutiny. See Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 588 (1986) (insisting on evidence that conspirators did not act independently). Monopolies. As we shall see in Part II.B, much monopolization is exempt entirely from the antitrust laws. When it is not, courts impose no blanket rule but instead test for efficiency on a case-by-case basis. See HOVENKAMP, FEDERAL ANTITRUST POLICY, supra note 10, at 214 R (acknowledging that “unilateral conduct receives the lowest level of antitrust scrutiny”). (The emphasis of enforcers on prosecuting cartels instead of monopolies might be due to factors such as efficacy of deterrence, rather than bias. See Baker, supra note 14, at 577–78. But R taken together with the bias in the case law and scholarship, the unequal enforcement is certainly suggestive.) \\jciprod01\productn\M\MIA\68-1\MIA103.txt unknown Seq: 8 19-NOV-13 17:04 112 UNIVERSITY OF MIAMI LAW REVIEW [Vol.68:105 the black triangle for monopolies, so it does not make sense to discour- age monopolization.16 The logic goes like this. Most monopolies do not just have marginal costs; they also have fixed costs, which are the amounts you have to pay for factors of production regardless how much you actually produce or sell. Let us assume that you have a monopolist whose fixed costs exactly equal the monopoly profit (i.e., they equal the area of the box created by the y-axis, Q, P, and C in Diagram 1). Then the monopo- list will go out of business unless she can charge P and earn the monop- oly profit. That means that the extra goods amounting to QQ′ could not be feasibly produced at a price below P and therefore the black triangle can no longer correspond to lost social or consumer value. Now P is efficient. We can stop coloring the triangle black. The trouble with this argument is that there is no reason to think that cartels do not also have fixed costs.17 The notion that they do not, but monopolies do, seems to arise because antitrust participants are in the habit of telling themselves ex post stories about cartelization but ex ante stories about monopolization.18 The ex post story about cartelization is that initially you have a bunch of firms in a competitive industry, each of which fully expects to remain in a competitive industry in the future. This expectation means that the firms all made investment decisions that assumed that they would only need a price of C to cover their costs. They did not expect to be able to charge P and they ensured that they would be able to remain in the market without doing so. The price is at C. At some point, however, cartelization happens. Price jumps to P. We know that the black triangle must appear as a 16. See, e.g., CARLTON & PERLOFF, supra note 10, at 99 (stating that the benefit of monopoly R is that it allows innovative firms to cover R&D costs). Modern arguments for monopoly are traceable to the work of Joseph Schumpeter, who believed that monopoly is necessary for innovation and growth. See, e.g., JOSEPH A. SCHUMPETER, CAPITALISM, SOCIALISM AND DEMOCRACY (Harper Torchbooks 1975) (1942). The fixed cost argument described in this article is a generalization of one such Schumpeterian argument, which is that monopoly profit is needed to cover the costs associated with R&D. Id. at 87–89. Because this is the only Schumpeterian argument to have survived empirical review, it seems reasonable to focus on it in this article. WESLEY M. COHEN & RICHARD C. LEVIN, Empirical Studies of Innovation and Market Structure, in 2 HANDBOOK OF INDUSTRIAL ORGANIZATION 1059, 1060–61, 1070, 1074–75, 1078 (Richard Schmalensee & Robert D. Willig eds., 1989). 17. Cf. MICHAEL D. WHINSTON, LECTURES ON ANTITRUST ECONOMICS 16–17 (2006) (sketching a model in which cartelization allows an industry to cover fixed costs and relating it to the 19th century “ruinous competition” argument made by a railroad cartel). 18. I am not aware of any clear instantiation of these stories in the antitrust literature (see infra note 19 for an example that hints at them). But I believe that they are what antitrust scholars R and practitioners think about when they think about the different treatment of cartels and monopolies. \\jciprod01\productn\M\MIA\68-1\MIA103.txt unknown Seq: 9 19-NOV-13 17:04 2013] INCONSISTENCY IN ANTITRUST 113 result of this price jump because we know that all the firms could have continued to charge C and produce Q′ going forward. The ex ante story about monopolization is that initially you have an entrepreneur who is deciding whether to create a superior product that no rivals are sophisticated enough to imitate. Making better products is expensive. But she realizes that if she can count on charging P for the product, then she will be able to afford to develop it. She invests in the project fully expecting, and on the condition, that she will be able to charge P when she enters the market. The fact that her willingness to enter the market is conditional on being able to charge P implies that she faces fixed costs equal to the size of the monopoly profit at P. When she enters the market, she charges P. No black triangle appears, however, because we know that if she had expected to be forced to charge C, she would not have entered the market in the first place and there would be no superior product for anyone to buy at any price.19 The ex post telling of the cartel story makes it sound like P is not necessary, while the ex ante telling of the monopoly story makes P essential. What this masks is that there is no reason to think that firms cannot take cartelization into account ex ante when deciding whether to take on high fixed costs. Similarly, there is no reason to think that monopolies always need P to cover their costs. It is perfectly reasonable to think that each member of a cartel might only produce a superior product (relative to the products of non-members) if it can expect to be able to work together with other members to fix the price at which it sells it at P.20,21,22 Similarly, it is perfectly reasonable to think that a 19. I read Herbert Hovenkamp’s argument that cartels deserve more scrutiny because they can be created “very quickly” whereas monopolies are the product of hard work to contain an element of this ex post/ex ante story. HOVENKAMP, FEDERAL ANTITRUST POLICY, supra note 10, at R 214. It seems obvious to Hovenkamp that cartels form quicker than monopolies because monopoly power “ordinarily takes many years of innovation and aggressive production and marketing.” Id. In other words, cartels do not need planning; firms that expect to face a competitive market can turn around and enjoy P instead, whenever they wish, by cartelizing. By contrast, to charge P, a single firm must plan and fight for it. But the fit is not precise. The argument also rests on the notion that speed of formation matters for efficiency. But Hovenkamp gives no explanation for why this should be so. What matters for efficiency is the existence of fixed costs; if cartels are as likely as monopolies to have them, then they should be protected, regardless whether they happen to form faster than monopolies. 20. The odd thing about antitrust’s treatment of cartels and innovation is that while it professes to ban price fixing per se, it permits it when the price fixing is used to fund fixed costs associated with cooperative R&D. There is no other way to understand the membership fees that open-membership R&D joint ventures are permitted to charge other than as a way of encouraging members not to compete price too low (there would be no incentive to join the joint venture if the fees were thought to be ineffective at encouraging pricing discipline). SeeHOVENKAMP, FEDERAL \\jciprod01\productn\M\MIA\68-1\MIA103.txt unknown Seq: 10 19-NOV-13 17:04 114 UNIVERSITY OF MIAMI LAW REVIEW [Vol.68:105 superior product monopolist might never have expected to win big and would have stayed in the market anyway at C. Of course, the argument for the cartel/monopoly distinction is not that all cartels have black triangles but all monopolies do not. The argu- ment is that cartels tend to have black triangles and monopolies tend not to.23,24 Because of enforcement costs, the argument goes, it is more effi- ANTITRUST POLICY, supra note 10, at 248 & n.53 (membership fees permissible for open- R membership joint ventures). The inconsistency here is that antitrust recognizes that R&D joint ventures will not come to pass unless antitrust allows them to fix prices necessary to cover fixed costs, but it does not recognize that independent innovation by cartel members may not come to pass unless antitrust allows the cartel to fix prices. The inconsistency might make sense if there were reason to believe that cooperative R&D by cartel members tends to be more efficient than independent R&D by cartel members. But I am not aware of any study that establishes this. 21. For example, the members of the DVD patent pools presumably incurred fixed costs in inventing the DVD technology for which they hold patents. See Complaint at 25–27, Florida v. Hitachi-LG Data Storage, Inc., No. 13-1877 (N.D. Cal. Apr. 24, 2013). Because this technology is superior to VHS or VCD technology, the members of the pool are able to exclude nonmember competitors from the market. See id. at 26. However, excluding rival firms is not sufficient to allow them to cover the fixed costs associated with inventing the technology. To do that, they must also avoid competing too hard with each other; otherwise they will drive prices too low to cover those costs. But in order to reduce the transaction costs associated with coordinating their pricing, it might be efficient for these firms explicitly to work together to fix prices. Cf.OLIVER E. WILLIAMSON, THE ECONOMIC INSTITUTIONS OF CAPITALISM: FIRMS, MARKETS, RELATIONAL CONTRACTING 159–60 (1985) (discussing progressive integration of members of telegraph cartels in the 19th century). The analogue of this in the single firm monopoly context is the monopoly that invents DVDs on its own. In order for it to cover its costs it too must (1) exclude rival firms from the market (that would otherwise be able to drive price down) and (2) charge a high enough price to cover costs. It is not clear why antitrust allows monopolies, but not cartels, to do this. (Perhaps an enlightened antitrust judge would not apply the per se rule to the cartel in this example on the ground that the price fixing is in service of an innovation joint venture; but imagine that the firms first agree to fix prices, and only later realize that they can maximize profit at the new higher prices by inventing the DVD. Their price-fixing agreement would certainly be per se illegal, even though this change of timing has no effect on the efficiency of the price fixing.) 22. There is some reason to think that members of an unstable cartel might be more likely than a single firm monopoly to incur fixed costs associated with R&D because each member feels compelled to prepare for the free-for-all that will accompany the demise of the cartel. Cf. POSNER, ANTITRUST LAW, supra note 8, at 21. R 23. See, e.g., SCHERER & ROSS, supra note 10, at 335. Scherer and Ross concede that they R have no “solid estimates” about the efficiency of cartels but conclude that it must be “modest” because, based on a review of railroad and soft coal mining studies, “price competition does seem to have done its job in forcing ‘sick’ industries to shed high-cost capacity . . . .” Id. They do not explain why, if this is so, price competition would not do a similarly good job in forcing sick single firm monopolies to shed high cost capacity. 24. There are three kinds of antitrust rule for any particular organizational form: per se illegal, case-by-case illegal, and per se legal. Under the first, nothing goes. Under the second, it goes only if there is no monopoly inefficiency. Under the third, everything goes. There are three corresponding justifications: respectively, that the form tends to be inefficient, has no tendency either way, or tends to be efficient. The argument for the per se rule against cartels is therefore that cartels tend to be inefficient. The argument against a per se rule against monopolies is that