Ensuring Sound Antitrust Analysis: Two Examples
DEBORAH PLATT MAJORAS
July 3, 2003
Thank you, Dick, and many thanks to NERA for inviting me back to participate this year. I am particularly pleased to speak today with my good friends, Joe Simons and Bob Joseph. The remarks I delivered here two years ago were my first in my new position as Deputy Assistant Attorney General, and I could do little more than look ahead, because we had not done much yet. I talked about the new Front Office, and I guess the more things change, the more they stay the same, because I can talk once again about the new Front Office! New Assistant Attorney General R. Hewitt Pate was sworn in two weeks ago; Dr. David Sibley joined us two months ago as our new Economics Deputy; Bruce McDonald, formerly a partner at Baker & Botts, joined us this week as the new Regulatory Deputy; and we expect to announce the appointment of our new International Deputy soon. I remain the Principal Deputy (when last I spoke with you I was the new kid; now, apparently, I am the grizzled veteran.) (In an unfortunate development for the Division, but an exciting development for her, long-serving Director of Operations and Civil Enforcement, Connie Robinson, has announced that she will be retiring from government service in September and becoming a partner at Kilpatrick & Stockton.)
Despite the economic situation in which we have been working, we have remained busy, and antitrust continues to generate fascinating and important issues. I had much to choose from in thinking about my remarks today, and I settled on two topics of importance to the Division: (1) coordinated effects analysis in merger review, and (2) the Section 2 framework we have urged courts to apply, including in our recent amicus brief in Verizon v. Trinko.
Coordinated effects analysis, long a core analytical tool in merger review, examines the potential for mutual accommodation among the market players post-merger. Section 2.1 of the Horizontal Merger Guidelines defines coordinated interaction as "actions by a group of firms that are profitable for each of them only as a result of the accommodating reactions of the others." Coordinated interaction can take several forms, including government-supported cartels, explicit collusion, tacit collusion, and conscious parallelism.(1)
Since 1996, the Division has filed successful challenges to 28 proposed mergers on the ground that they created an appreciable danger of coordinated interaction among suppliers.(2) And last summer, then-AAG Charles James announced a new initiative at the Division to refine our analysis of coordinated effects.(3) That undertaking is well underway, with a dedicated team of economists and lawyers who have contributed to a comprehensive coordinated effects manual that we have rolled out internally and which is currently under review. Although the manual is of finished-product quality, it is still a work in progress from a policy perspective, and, for that reason, I will discuss its conclusions and analyses only in broad terms. To illustrate some of those themes in practice, I will then discuss two of our recent cases in which we sought to enjoin acquisitions on the basis of coordinated effects analysis.
II. Analytical Framework
The ultimate inquiry in any merger review is whether the proposed merger will affect the state of competition: why does this merger matter? Under a coordinated effects analysis, that inquiry focuses on whether, post-merger, suppliers will have a greater incentive and ability to coordinate. That is, will the merger make coordination more likely, more perfect, more complete, or more durable? A merger could make coordination more likely simply by reducing the number of players needed for consensus or by removing an industry "maverick" with a history of undercutting or resistance to market leadership. A merger could make coordination more perfect by creating a price leader or by enabling the now fewer number of firms to support a higher price or otherwise more stringent terms of coordination. In much the same way, a merger could make coordination more complete by, for example, extending the duration, product coverage, or geographical reach of coordination. Finally, a merger could make coordination more durable by enabling firms to better monitor compliance.
These observations are not novel. The key is to use market facts to predict whether these generalities are likely to result in a given case, but without simply applying a checklist approach.(4) Like any antitrust analysis, a coordinated effects merger analysis necessarily will be a fact-driven inquiry tailored to the important characteristics of the market(s) under review. To assist in analyzing how a proposed merger might affect firms' incentives or ability to coordinate, we engage in a two-step inquiry: (1) what constraints exist, pre-merger, on the incentive and ability of suppliers to coordinate? And (2) how will the proposed merger change those constraints?(5)
Although the types of restraints on coordination can be parsed further, we have found it useful to organize them into three broad categories: (i) presence of a large number of rivals; (ii) asymmetry among rivals; and (iii) opportunities for industry "mavericks."
The presumption that fewer rivals facilitates coordination by increasing the incentive and ability of rivals to coordinate dates back at least to the Supreme Court's decision in Philadelphia National Bank in 1963.(6) Conversely, the larger the number of rivals, the lesser will be the incentive and ability to coordinate, for several reasons. First, the greater the number of rivals, the harder it is to reach consensus on terms of coordination e.g., price, output, market allocations. Second, the greater the number of rivals, the smaller the incentive each has to forego short term profits in favor of the longer-term benefits of coordination. Third, cheating is more likely, and detection and punishment are less likely, when there are numerous players.
All else being equal, fewer suppliers means that: (i) the terms of coordination should be easier to reach; (ii) cheating is more difficult because there is a smaller pool of customers to be stolen from rivals, and cheating is less profitable because each firm will have a greater stake in the long-term collusive profit, making the short-term profits of cheating less attractive; and (iii) punishment is more effective because larger rivals have the ability to absorb the costs of punishment through undercutting. For these reasons, an increase in concentration tends to make collusion incrementally easier to negotiate and sustain. The important question, of course, is: when does that incremental increase raise concerns?. The case law and empirical evidence tell us that the quantitative importance of this effect varies with the premerger structure of the market, such that presumption that a merger will facilitate coordination is stronger in a 3-to-2 merger than in a 6-to-5 or a 5-to-4.
Asymmetries among rival suppliers e.g., varying product lines or cost structures can constrain coordination by creating conflicts of interest among suppliers that cannot be overcome to reach a collusive consensus. At the outset, it is important to note that the categories of constraints we identify do not work in isolation, can point in opposite directions, and will vary in their application depending on the market characteristics. While often a merger might narrow asymmetries between firms and thus make coordination more likely, that is not always the case. In a market with fairly symmetrically positioned suppliers, a merger could create asymmetry in the post-merger landscape and, thus, introduce a potential constraint on coordination. Alternatively, a merger could introduce such drastic asymmetry that a market leader emerges, thereby facilitating coordination through price leadership. I will briefly discuss asymmetries in product attributes and cost structures.
In addition to the asymmetries noted above, other market attributes can lead to opportunities for maverick firms that will act as constraints on collusion. Both opportunities for innovation and lack of transparency in supplier transactions are industry features that can disrupt coordination.
A merger can eliminate the destabilizing influence of innovation by removing a potential entrant or nascent competitor that would have introduced more efficient production and or an improved product into the market. This consideration can also point in the opposite direction, however: a merger can lead to the absorption of a nascent technology by an existing player who will be able to bring it to market more quickly or more broadly.
A merger can facilitate coordination by reducing transparency. By necessity, a horizontal or vertical merger puts the market information previously in the hands of one firm into the hands of another. An expansion up- or downstream may expose the merged firm to information that its horizontal competitors do not have. But the opposite may also be true: a vertical merger can create the ability to conceal sales to a downstream affiliate, because intra-firm transfers are harder to monitor.
III. Case Applications
As I noted at the outset, the Division has pursued a coordinated effects theory of harm in numerous cases over the years, but I will highlight two of our recent cases:
In April, the Division filed suit in the U.S. District Court in Pittsburgh to block SGL Carbon AG and its U.S. subsidiary, SGL Carbon L.L.C., from acquiring certain assets of Carbide/Graphite Group in a bankruptcy court auction. The assets at issue included a plant for manufacturing graphite electrodes, which are used as electrical conductors in the production of steel and are a critical input. SGL Carbon and Carbide/Graphite were two of the only four producers capable of manufacturing quality 18-inch diameter and larger graphite electrodes for sale in the United States. (C/G Electrodes Acquisition LLC, a financial bidder that intended to independently operate the plant, also bid on the assets.)
Our theory of harm was that the acquisition would facilitate coordination among the three remaining producers of large graphite electrodes for sale in the United States. We placed importance on the recent history of collusion among graphite electrode producers and the fact that the market structure which obviously had sustained collusion over a number of years had not significantly changed. SGL AG was one of the manufacturers that participated in a conspiracy to fix prices and allocate markets for graphite electrodes worldwide in the 1990s. During the conspiracy, prices of graphite electrodes increased by at least 40-50% in the United States through a series of collusive price increases. As a result of SGL AG's price-fixing and market-allocation agreements with its competitors, steel makers paid anticompetitive, higher prices for graphite electrodes, adversely affecting a variety of business and consumer products that are dependent on steel.
Our complaint alleged that SGL's acquisition of certain assets of Carbide/Graphite would further increase the opportunity for the remaining dominant players to coordinate to raise prices to the detriment of consumers. Given the small number of competitors, higher prices, lower quality and service, and less innovation were likely to result if the Carbide/Graphite assets were not operated by an independent, significant competitor.
The Division filed the complaint to ensure that we could seek to enjoin SGL from closing on the acquisition in the event SGL prevailed over the other bidder in the auction. On the day we filed, the bankruptcy court determined that SGL's bid was not the highest and best offer and awarded the assets to C/G Electrodes. After receiving notice that C/G Electrodes had closed on the assets, the Division voluntarily dismissed its complaint.
Also in April, the Division filed a lawsuit in the United States District Court for the Northern District of Illinois to block UPM Kymmene's Raflatac subsidiary from acquiring Bemis Corporation's MACtac subsidiary.(7) UPM, through Raflatac, and Bemis's MACtac are the second and third largest producers, respectively, of pressure-sensitive labelstock in North America. Labelstock is the base material for labels used in a variety of applications that American consumers encounter every day, including shipping labels and supermarket scale labels. UPM and MACtac both produce labelstock on a bulk basis, that is, at high-volume production and low-unit cost for high-demand applications (in contrast to specialty labelstock produced at low volume for low-demand applications). UPM, MACtac, and the largest North American labelstock producer (hereinafter "the leading producer") collectively account for over 70 percent of total sales of such labelstock in North America. UPM has been a particularly aggressive competitor, having made strategic commitments to substantially expand its North American labelstock sales. As a result of this vigorous competition, labelstock customers have enjoyed significantly lower prices and higher product and service quality than they would have otherwise received.
The Division concluded that the proposed merger would facilitate coordination between the merged company and the leading producer. UPM's acquisition of MACtac would leave two large producers, UPM and the leading producer, in a position to lead jointly and to coordinate generally a lessening of competition in the production and sale of bulk labelstock. By acquiring MACtac, UPM would more than double its current North American labelstock sales, achieve its strategic growth objectives, and begin to approach parity with the leading producer in sales volume and market share. UPM's incentive to compete for sales to the leading producer's customers would diminish (and its incentive to coordinate increase) because UPM would stand to lose proportionately more business than otherwise if the leading producer retaliated by competing for UPM customers.
Indeed, as alleged in our complaint, shortly after announcement of the transaction, MACtac's CEO, whom UPM had chosen to manage UPM's North American labelstock business after the transaction, advised a securities analyst that the transaction should bring pricing "discipline" to UPM. And senior UPM officials advised at least two labelstock customers about UPM plans to increase prices after the transaction.
Post-acquisition, the remaining smaller labelstock producers would have neither the capabilities nor incentives to prevent UPM and the Leading Producer from engaging in anticompetitive coordination. The transaction thus would likely substantially lessen competition in North American markets for the production and sale of bulk labelstock, leading to higher prices and lower quality products and services.
The district court rejected defendants' motion for a consolidated preliminary injunction hearing and trial on the merits, which the Department had opposed. Instead the court held a 10-day preliminary injunction hearing that concluded on June 20. The Division expects a decision by the end of July.
Now I will shift gears and briefly discuss single-firm conduct in the nonmerger context. Sound antitrust policy is based on faith in the competitive process, even when that process is not pretty. The difficulty in distinguishing healthy, even cutthroat, competitive conduct from conduct that is designed only to exclude competitors i.e., conduct without any procompetitive justifications is well-known. As Judge Easterbrook recently stated: "Aggressive, competitive conduct by any firm, even one with market power, is beneficial to consumers. Courts should prize and encourage it. Aggressive, exclusionary conduct is deleterious to consumers, and courts should condemn it. The big problem lies in this: competitive and exclusionary conduct look alike."(8)
As enforcers, we are committed to rooting out anticompetitive, monopolistic conduct, and do not suggest, as some have, that we should therefore ignore exclusionary conduct in our enforcement efforts. But we are equally mindful of our responsibility not to deter aggressive competition on the merits. Judge Learned Hand astutely noted nearly 60 years ago that "the successful competitor, having been urged to compete, must not be turned upon when he wins."(9) Rather, "a monopolist, no less than any other competitor, is permitted and indeed encouraged to compete aggressively on the merits."(10)
At the Antitrust Division, we have sought to apply standards of single-firm conduct that are transparent, objective, and administrable, so that antitrust laws do not unduly interfere with the competition they are meant to protect. To distinguish healthy competition from illegal attempts to monopolize or maintain monopoly, we look to a standard articulated by the Supreme Court.
The Court held in Aspen Skiing that "[i]f a firm has been 'attempting to exclude rivals on some basis other than efficiency,' it is fair to characterize its behavior as predatory."(11) In accordance with this standard, the Division has found it useful in determining whether conduct is predatory or exclusionary, to inquire whether the conduct would make economic sense for the defendant but for its elimination or, lessening of competition. It is a standard that we have advocated in our Microsoft and American Airlines cases, and in a recent Supreme Court brief, in Verizon v. Trinko. In advocating this standard, we do not mean to suggest that it necessarily encompasses every single type of conduct that may violate Section 2. We do believe, however, that this "economic sense" test sets forth an objective, transparent, and economically-based framework for assessing single-firm conduct, and perhaps more importantly, a standard to which firms may conform their conduct.
By now, you all are familiar with the Division's 1999 complaint against American Airlines, which alleged that American had violated Section 2 of the Sherman Act by engaging in successful predation that involved adding money-losing capacity to drive lower-cost carriers out of four of American's Dallas-Fort Worth ("DFW") Airport routes. Where low-cost entrants threatened significant expansion of low-cost service, American flooded the routes with additional capacity more flights, bigger planes, or both. By way of example, an American "DFW Vulnerability" study had concluded that any strategy to make a lower-cost carrier unprofitable at DFW "would definitely be very expensive in terms of American's short term profitability." And former CEO Bob Crandall had stated, "'If you're not going to get them out then [there is] no point to diminish profit.'" In addition, the Division's economic expert performed an extensive quantitative analysis that confirmed that the incremental costs American incurred as a result of the challenged capacity additions substantially exceeded the incremental revenues attributable to them.
Applying the standard I set forth above, we alleged that by providing this additional capacity, American lost money, and that it made no business sense except for the fact that American stood to gain much more in the long run by eliminating competition and then elevating prices once competition was driven away. The district court granted American's motion for summary judgment, and, as you know, we appealed. We argued on appeal that the proper inquiry in reviewing American's conduct is whether "the increases made business sense unless they excluded rivals," that is, "whether the conduct would be profitable, apart from any exclusionary effects."(12) And, as we argue on appeal, American's conduct made sense only as what Judge Bork has called "an investment in future monopoly profits."(13) The Tenth Circuit heard oral argument in September, and we await that decision.
The Microsoft case is, of course, the most notable recent enforcement action the Division has taken involving single-firm conduct. Some commentators, as well as Microsoft's competitors, have viewed Microsoft's dominant position in the market for Intel-compatible PC operating systems as a competitive problem regardless of whether Microsoft obtained that monopoly through lawful means, and as a result those of this view see the Division's case as a vehicle for re-engineering the operating system market.
The Division consistently has indicated, however, that "if Microsoft had confined itself to improving and promoting its products on their merits, it would have faced no antitrust liability, whatever the effect on its rivals."(14) We have argued that the proper standard of inquiry in the Microsoft case was whether the conduct "would not make economic sense unless it eliminated or softened competition and thus permitted the costs of the conduct to be recouped through higher profits resulting from the lack of competition."(15) The D.C. Circuit ultimately held that Microsoft took actions to discourage the development and deployment of rival Web browsers and Java technologies, in an effort to prevent them from becoming middleware threats to Microsoft's operating system monopoly.
Microsoft's anticompetitive campaign to exclude rival Web browsers went far beyond offering its own Internet Explorer browser to original equipment manufacturers ("OEMs") in a bundle with Windows at no extra charge; it discouraged and threatened to penalize individual OEMs that wanted to pre-install and promote rival browsers. The Division argued that such restrictions "stifled innovation by OEMs that might have made Windows PC systems easier to use and more attractive to consumers," and "harmed consumers who preferred a different browser or none at all."(16) And, as to some of the restrictions, of course, the Court agreed.
Most recently, in Trinko, the Department and the FTC jointly filed (i) an amicus brief urging the Court to grant certiorari on Verizon's appeal from the Second Circuit decision, and (ii) a brief on the merits after the Court granted cert. The Second Circuit held that a customer of AT&T's local phone service had stated an antitrust claim for monopolization under Section 2 by alleging that Verizon had not fulfilled its duties to competitors, as set forth in the 1996 Telecomm Act. The 1996 Act "require[s] incumbent local-exchange carriers to share their own facilities and services on terms agreed upon with new entrants . . . ."(17) Notably, the 1996 Act also includes an antitrust savings clause, providing that "nothing in this Act . . . shall be construed to modify, impair, or supersede the applicability of any of the antitrust laws." As we stated in our brief on the merits: "The 1996 Act's imposition of new duties to assist rivalscoupled with the increasing number of antitrust lawsuits predicated on the alleged noncompliance with the 1996 Acthave given new urgency to careful examination of the circumstances under which antitrust law requires a dominant firm to provide such assistance."
In upholding the plaintiff's claim that Verizon violated Section 2 by failing to meet the standards for assisting competitors imposed by the 1996 Act, the Second Circuit relied on the "essential facilities" doctrine, holding that a monopolist has a duty to provide rivals with reasonable access to facilities that the monopolist controls without which one cannot compete. The court also held that the plaintiff had stated a claim under the questionable theory of monopoly leveraging, which prevents a firm with a monopoly in one market from using its power in that market to gain an advantage in a second market.
In our briefs, we argue that the Second Circuit's decision dramatically expands antitrust liability for failure to assist rivals and, in doing so, that it conflicts with the law of other circuits.(18) The "essential facilities" decision is erroneous because it imposes duties on rivals without requiring that the firm's challenged conduct be exclusionary or predatory i.e., that the refusal to deal does not make economic sense except as an effort to diminish competition.
As we argued in our brief: "Unlike the 1996 Act, the Sherman Act does not impose a duty to sell to rivals in an evenhanded fashion unless the refusal is predatory or exclusionary, i.e., unless the refusal represents a sacrifice of profit or goodwill that makes sense only because it has the effect of injuring competition."
The monopoly leveraging decision is also erroneous because it means that mere use of monopoly power is enough to state a claim under Section 2, which is inconsistent with Supreme Court precedent. The Court made clear in Spectrum Sports, Inc. v. McQuillan, 506 U.S. 447, 459 (1993), that Section 2 "makes the conduct of a single firm unlawful only when it actually monopolizes or dangerously threatens to do so." The monopoly leveraging claim also suffers from the same problem as the essential facilities rationale in that it does not require exclusionary or predatory conduct.
1. The Supreme Court has held that conscious parallelism simply reflects rational behavior by independent firms that correctly perceive and act upon competitive interdependencies without reaching any sort of agreement or mutual understanding with their rivals. Brooke Group v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 227 (1993). Thus, unlike the other forms of coordinated interaction mentioned, conscious parallelism is not reachable under Section 1 of the Sherman Act because such behavior does not entail any act of agreement. Nonetheless, conscious parallelism remains a prime consideration under coordinated effects analysis because the enforcement agencies' reviews of mergers under Section 7 of the Clayton Act look to prospective effects and do not require a prediction of illegal conduct to justify a challenge. Instead, with regard to coordinated effects, a merger need only "create an appreciable danger" of coordination to justify a challenge. United States v. Phila. Nat'l Bank, 374 U.S. 321, 362 (1963).
2. In many of these challenges, the Division also alleged that the merger likely would substantially lessen competition due to unilateral effects.
3. Charles A. James, "Rediscovering Coordinated Effects," August 13, 2002, available at http://www.usdoj.gov/atr/public/speeches/200124.htm.
4. Early economic analysis and case law used a "checklist" approach that often devolved into counting plus and minus sides of the coordination ledger and then applying a subjective (but unstated) weighting scheme to deduce whether conditions were sufficiently conducive to collusion to raise antitrust concerns. The checklist, however, offered neither necessary nor sufficient conditions for coordinated interaction.
5. The Division articulated this two-part inquiry when it explained the basis for challenging the proposed acquisition of Masonite (a leading manufacturer of molded doorskins) by Premdor (a leading manufacturer of interior molded doors). While there were only two major competitors in both the doorskin and door markets prior to the merger, the evidence showed that at least 4 significant factors in the premerger market structure made premerger coordination less likely. The Division then found that the merger would relax each of those constraints. The Competitive Impact Statement provides a detailed analysis.
6. See supra n.1.
7. I was recused from this matter and therefore had no participation in its consideration.
8. Frank H. Easterbrook, When Is it Worthwhile to Use Courts to Search for Exclusionary Conduct?, 2003 Columbia Bus. L. Rev. 101 (2003) (forthcoming).
9. United States v. Aluminum Co. of Am., 148 F.2d 416, 430 (2d Cir. 1945).
10. Olympia Equip. Leasing Co. v. W. Union Tel. Co., 797 F.2d 370, 375 (7th Cir. 1986) (internal quotations omitted).
11. Aspen Skiing Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585 (1985), citing Robert Bork, The Antitrust Paradox 138 (1978).
12. Brief for Appellant, United States of America v. AMR Corp., American Airlines, Inc., and American Eagle Holding Corp., No. 01-3202 (10th Cir. Jan. 11, 2002).
13. Robert H. Bork, The Antitrust Paradox 145 (1978).
14. Brief for Appellees, United States v. Microsoft Corp., Nos. 00-5212, 00-5213 (D.C. Cir. Feb. 9, 2001).
17. Verizon Communications, Inc. v. FCC, 122 S. Ct. 1646, 1654 (2002).
18. The Second Circuit declined to follow the Seventh Circuit's decision in Goldwasser v. Ameritech Corp., 222 F.3d 390 (7th Cir. 2000), upon which the district court in Trinko had relied in dismissing the plaintiff's claim. See Goldwasser, 222 F.3d at 399 (holding that "the duties the 1996 Act imposes" are not "coterminous with the duty of a monopolist to refrain from exclusionary practices").