Proving Anticompetitive Impact: Moving Past Merger Guidelines Presumptions PROVING ANTICOMPETITIVE IMPACT: MOVING PAST MERGER GUIDELINES PRESUMPTIONS John Harkrider* I. Introduction ............................................................... 317 II. Agency Action is Frequently Outcome D eterm inative ........................................................... 320 III. Agency Should Have High Standard of Proof .......... 324 IV. The Appropriate Role of Presumptions .................... 327 A . Introduction ........................................................ 327 B. Coordinated Effects Presumption ...................... 328 C. Unilateral Effects Threshold .............................. 331 D. Should Statistical Presumptions Be A bandoned? ......................................................... 332 V. Forms of Appropriate Evidence ................................ 334 A . Introduction ........................................................ 334 B. Types of Evidence Entitled to Great Weight ...... 335 1. Natural Experiments ..................................... 335 2. Strategic Docum ents ..................................... 336 3. M erger Sim ulation ......................................... 338 C. Types of Evidence Not Entitled to Great W eight ................................................................. 340 1. Customer Testimony ..................................... 340 2. Documents Written by Salespeople and Low-Level Management ................................ 343 V I. C onclusion ................................................................. 344 I. INTRODUCTION Because antitrust challenges to mergers almost always occur prior to consummation, there is rarely direct evidence " John D. Harkrider is a founding member of Axinn, Veltrop & Harkrider, LLP. His practice focuses on antitrust analysis of mergers, particularly Section 7 litigation. The views and opinions expressed in this Article are those of the author and do not necessarily represent the views of Axinn, Veltrop & Harkrider, LLP, or its clients. COLUMBIA BUSINESS LAW RE WE W [Vol. 2005 of whether a merger is likely to have a substantial competitive impact. As a result, litigants frequently utilize a dizzying array of indirect evidence including statistical presumptions, customer affidavits, econometric models, quotes from strategic documents, sales call reports, and e-mails. In short, everything but the kitchen sink, and sometimes even that. Despite, or perhaps because of, the volume of evidence, courts have struggled to move beyond market definition and barriers to entry towards a coherent analysis of likely competitive impact. Indeed, two recent cases, Federal Trade Commission v. Arch Coal Inc.1 and United States v. Oracle Corp.,2 are notable because they represent the most detailed judicial examination of the conditions under which coordinated and unilateral effects are likely. And, at least in one of them, Oracle, the Court openly struggled for guidance on the appropriate requirements of theory, going so far as to request counsel to brief the court on whether unilateral effects were even viable. 3 These two cases, and United States v. SunGard Data Systems, Inc.4 before them, suggest that the Government has had great difficulty in identifying transactions that courts will conclude are likely to cause competitive harm. The purpose of this paper is to discuss why the FTC and DOJ struggle to identify transactions that are likely to cause competitive harm, and to discuss specific proposals to remedy the situation. In particular, I will make the following arguments: 329 F. Supp. 2d 109 (D.D.C. 2004). 2 331 F. Supp. 2d 1098 (N.D. Cal. 2004). See Dawn Kawamoto, Judge's Order May Be Good for Oracle, CNET NEWS.COM, (July 12, 2004), available at http://ecoustics-cnet.com.com/ Judges+order+may+be+good+for+Oracle/2100-101 l3-5266345.html ("The court is aware of only a handful of cases considering unilateral-effects theories .... This approach is one of several and appears to have received substantial criticism from antitrust scholars and little, if any, exposure in the crucible of litigation."). 4 172 F. Supp. 2d 172 (D.D.C. 2001). COLUMBIA BUSINESS LA W REVIEW [Vol. 2005 MOVING PAST GUIDELINES PRESUMPTIONS * First, because most Clayton Act Section 75 cases result in consent decrees, the agency challenging the transaction is in the position of being both prosecutor and judge. As such, in seeking a consent decree, staff should be subject to a heightened burden of proof, and should always have the burden of production and persuasion to the internal decision makers on all elements of a Section 7 offense, including, most specifically, the likelihood of competitive harm. * Second, the burden of proving competitive harm should not be satisfied by presumptions, but rather with an exposition of a specific theory of competitive harm and proof of that theory tied to the facts and circumstances of the transaction and industry. " Third, competitive harm must be demonstrated systematically with objective, and if possible, scientific evidence. Specifically, econometrically tested natural experiments, documents written by senior management about the rationale for the transaction, and merger simulations based upon estimated cross-elasticities should be given great weight, and customer testimony, e-mails from middle management and sales people should be given less weight. There is no question that the DOJ and FTC take their public policy role very seriously and take all efforts to reach the correct result in the cases before them. In addition, merger workshops and other initiatives have made clear that the agency intends to improve the accuracy and efficiency of merger review. Further, the release of the merger challenges data makes clear that the DOJ and FTC have taken pains to both increase the transparency of their process, which, of course, invites the bar to suggest improvements to the process. Thus, in making these suggestions, I do not mean to imply that agency practice is always at variance with the ' 15 U.S.C. § 18 (2000). No. 2:317] COLUMBIA BUSINESS LAW REVIEW [Vol. 2005 proposals set forth in this paper. For example, the release of the merger challenges data indicates that the agencies do not strictly follow the Guidelines'6 Herfindahl-Hirschman Index ("HHI") presumptions that are criticized in this paper. Furthermore, I note that the agencies are releasing a Commentary on the Guidelines,' though I would suggest that the better, but perhaps not as simple to execute, method is to simply revise the Guidelines themselves. II. AGENCY ACTION IS FREQUENTLY OUTCOME DETERMINATIVE At the outset, it is important to recognize that more than half of all significant merger investigations result in enforcement action,8 which in the majority of cases leads to consent decrees or "fix-it-first"9 remedies.'0 The FTC in fact 6 U.S. DEP'T OF JUSTICE & FED. TRADE COMM'N, 1992 HORIZONTAL MERGER GUIDELINES (1992) (with 1997 revisions), reprinted in 4 Trade Reg. Rep. (CCH) 13,104 [hereinafter MERGER GUIDELINES]. 7 See Chairman FTC Deborah Platt Majoras, Looking Forward: Merger and Other Policy Initiatives at the FTC 4-7, Remarks at the A.B.A. Antitrust Section Fall Forum (Nov. 18, 2004), available at http://www.ftc.gov/speeches/majoras/041118abafallforum.pdf. 8 See, e.g., FED. TRADE COMM'N, PERFORMANCE AND ACCOUNTABILITY REPORT, FISCAL YEAR 2004, 45, available at http://www.ftc.gov/ opp/gpra/2004parreport.pdf (In fiscal years 2001, 2002, 2003, and 2004 the FTC took enforcement action in respectively 68%, 68%, 70%, and 55% of its second request merger investigations); Deputy Assistant Attorey-General Thomas 0. Barnett, Antitrust Enforcement Priorities: A Year in Review, Remarks at the A.B.A. Antitrust Section Fall Forum (Nov. 18, 2004), available at http://www.usdoj.gov/ atr/public/speeches/206455.htm ("17% of its investigations resulted in the issuance of second requests, and slightly more than half of those cases led to challenges."). 9 See U.S. DEP'T OF JUSTICE, ANTITRUST DIVISION POLICY GUIDE TO MERGER REMEDIES 26 (2004), available at http://www.usdoj.gov/atr/ public/guidelines/205108.htm ("A fix-it-first remedy is a structural remedy that the parties implement and the Division accepts before a merger is consummated. A fix-it-first remedy eliminates the Division's antitrust concerns and therefore the need to file a case."). 10 See, e.g., Assistant Attorney General R. Hewitt Pate, Antitrust Enforcement Oversight 9, Statement before the Committee on the Judiciary, United States House of Representatives (July 24, 2003) COLUMBIA BUSINESS LA W REVIEW [Vol. 2005 No. 2:3 17] MOWNG PAST GUIDELINES PRESUMPTIONS sets a target that between 60% and 80% of "HSR second request investigations" should result in enforcement action." available at http://www.usdoj.gov/atr/ public/testimony/201190.htm. In his statement, Assistant Attorney General Pate noted: Since June 2001, the Division has challenged thirty-four mergers it deemed anticompetitive, and we have been successful in thirty-one of the thirty-two matters that have thus far reached a conclusion. Nine of these matters were resolved by consent decree, twelve through a "fix-it-first" restructuring, seven were abandoned after the Division indicated that it would file suit, and three-General Dynamics/Newport News, Hughes/Echostar, and SGL Carbon/ Carbide/Graphite Group-were abandoned after the Division filed suit. The Division was unsuccessful in seeking to block the SunGard/Comdisco merger, a transaction the Division asserted was likely substantially to lessen competition in the market for shared "hotsite" disaster recovery services. Two of the merger challenges remain in litigation. Id. Moreover, the DOJ and FTC Annual Reports to Congress for the fiscal years 1997 to 2003 show that on average approximately 70% of the DOJ merger challenges led to consent decrees and fix-it-first remedies, and that on average approximately 60% of the FTC merger challenges led to consent decrees. See U.S. DEP'T OF JUSTICE AND FED. TRADE COMM'N, ANNUAL REPORT TO CONGRESS, FISCAL YEAR 2003 8-9, 13 (fifteen DOJ challenges, of which five resulted in fix-it-first remedies and five in consent decrees; twenty-two FTC challenges, of which eight resulted in consent decrees); FISCAL YEAR 2002 at 8-9, 11 (ten DOJ challenges, five fix- it-first, two consent decrees; twenty-four FTC challenges, twelve consent decrees); FISCAL YEAR 2001 at 14-15, 19 (thirty-two DOJ challenges, twenty fix-it-first, eight consent decrees; twenty-three FTC challenges, eighteen consent decrees); FISCAL YEAR 2000 at 8-9, 20 (forty-eight DOJ challenges, sixteen fix-it-first, eighteen consent decrees; thirty-two FTC challenges, nineteen consent decrees); FISCAL YEAR 1999 at 10, 20 (forty- seven DOJ challenges, sixteen fix-it-first, twenty consent decrees; thirty FTC challenges, eighteen consent decrees); FISCAL YEAR 1998 (fifty-one DOJ challenges, twenty-four fix-it-first, eleven consent decrees; thirty- three FTC challenges, twenty-three consent decrees); FISCAL YEAR 1997 (thirty-one DOJ challenges, eight fix-it-first, thirteen consent decrees; twenty-eight FTC challenges, seventeen consent decrees); available at http://www.ftc.gov/bc/hsr/hsrinfopub.htm. " See FED. TRADE COMM'N, PERFORMANCE AND ACCOUNTABILITY REPORT, supra note 8, at 44. MOVING PAST GUIDELINES PRESUMPTIONSNo. 2:317] COLUMBIA BUSINESS LAW REVIEW [Vol. 2005 The success of the government in obtaining consent decrees is in part driven by the fact that the potential cost of a full stop injunction outweighs by a large margin the relief typically proposed through consent decree. 12 Consequently, where the agency seeks a remedy, the typical equilibrium is that the parties will agree to divestitures within the hell-or- high-water clause. While there is no question that the agency is able to obtain consent decrees in some cases because they have a high chance of prevailing in Court, the asymmetry between the cost to the parties of a consent decree and the cost to the parties of losing in Court is so significant that parties will enter into consent decrees even where the agency has a low probability of prevailing in Court. What this means is that the agency's determination to seek a remedy frequently is outcome determinative. The concern that the parties may enter into consent decrees notwithstanding the merits of the Government's case depends, in part, on one's view of the relative importance of a policy which blocks some competitively neutral or beneficial mergers while blocking most competitively harmful mergers. The question is how many beneficial mergers we are willing to give up in the process. In evaluating this tradeoff, it is important to note that stopping competitively benign mergers is not costless. Not only does it cause significant transaction costs, but it also undermines significant allocative efficiencies. Contracts move assets to a buyer who has a higher utility for the acquired assets than the seller. Not only is that increment in utility lost by stopping or reformulating the sale, but the consent decree involves a sale to a buyer who did not purchase the asset in the pre-merger world. In other words, a consent decree runs the risk that the divested asset is being sold to a buyer who may have a lower utility for the " For a sample of the relief the DOJ typically requires in consent decrees, see, e.g, U.S. DEP'T OF JUSTICE, ANTITRUST DIVISION, ANNUAL REPORT, FISCAL YEAR 1999, 41-86 (Appendix B), available at http://www.usdoj.gov/atr/public/4523.pdf. COLUMBIA BUSINESS LA W REVIEW [Vol. 2005 MOVING PAST GUIDELINES PRESUMPTIONS asset than the seller had in the pre-merger world. Indeed, the FTC divestiture study13 makes quite clear that many divestitures run a significant risk that the divested assets may not be used as efficiently as they were used in the pre- merger world. 14 Fortunately, or unfortunately, we are not free to make the tradeoff between erring by blocking a beneficial merger and erring by allowing a harmful merger as we see fit. This is so because Congress and the Courts have made clear that Section 7 is intended as a prophylactic measure and therefore tolerates some errors.15 Thus the case law makes clear that courts and the agencies are not required to seek certainty in identifying transactions that are likely to harm consumer welfare,'6 but only are required to identify "3 FED. TRADE COMM'N, BUREAU OF COMPETITION, A STUDY OF THE COMMISSION'S DIVESTITURE PROCESS (1999), available at http://www.ftc.gov/os/1999/08/divestiture.pdf. 14 Id. at 16-29. 15 See, e.g., Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc., 429 U.S. 477, 485 (1977) ("Section 7 of the Act proscribes mergers whose effect may be substantially to lessen competition, or to tend to create a monopoly. It is, as we have observed many times, a prophylactic measure, intended primarily to arrest apprehended consequences of intercorporate relationships before those relationships could work their evil") (emphasis in original) (internal quotations and citations omitted). 1 See, e.g., United States v. Penn-Olin Chem. Co., 378 U.S. 158, 177 (1964) ("[Tlhe mandate of the Congress is in terms of the probability of a lessening of substantial competition, not in terms of tangible present restraint."); Brown Shoe Co. v. United States, 370 U.S. 294, 323 (1962) ("Congress used the words may be substantially to lessen competition, to indicate that its concern was with probabilities, not certainties.") (internal quotations and citations omitted); Hospital Corp. v. FTC, 807 F.2d 1381, 1389 (7th Cir. 1986) (Judge Posner) (noting that Section 7 of the Clayton Act does not require proof of actual consumer harm, but rather involves "[a] predictive judgment, necessarily probabilistic and judgmental."). Note, however, that Section 7 of the Clayton Act is concerned with a lessening of competition "which is sufficiently probable and imminent," United States v. Continental Can Co., 378 U.S. 441, 458 (1964), not with "ephemeral possibilities," Brown Shoe, 370 U.S. at 323, or "remote possibilities." United States v. Falstaff Brewing Corp., 410 U.S. 526, 555 (1973). No. 2:317] COLUMBIA BUSiNESS LAW REViEW [Vol. 2005 transactions that are reasonably likely to lessen competition. 17 The balance of this paper will consider whether current agency practice is well suited to meet this minimal standard. III. AGENCY SHOULD HAVE HIGH STANDARD OF PROOF The current agency practice is not well suited to the objective determination of truth. As noted before, the agency is often in the position of both prosecutor and judge. Furthermore, the process is not transparent. Specifically, despite some recent improvement, staff does not always clearly articulate a theory of competitive harm, and further, in the context of evidence presented by third parties, may be legally prevented from disclosing the source of its evidence. 8 As a result, staffs theory is not as carefully vetted before the relevant decision maker as would occur in a judicial setting. Given all of this, and the fact that staff has the ability to obtain third-party data, one would think that staff should be subjected to a heightened burden of proof, with the responsibility to prove relevant market, barriers to entry, and likelihood of competitive harm by clear and convincing evidence. The Guidelines, however, are purposefully silent on who has the appropriate burden of production and persuasion. 9 In practice, this means that staff has the burden of proof of market definition, and, perhaps, the barriers to entry, but the parties have the burden of proof on competitive effects, particularly where a presumption is 7 See, e.g., Penn-Olin, 378 U.S. 158, 171 (noting that a violation of Section 7 of the Clayton Act is established when a "reasonable likelihood" of a substantial lessening of competition is shown). 18 See 15 U.S.C. § 1313(c)(3) (2000) ("while in the possession of the custodian, no documentary material, answers to interrogatories, or transcripts of oral testimony, or copies thereof, so produced shall be available for examination, without the consent of the person who produced such material. ..."). '9 MERGER GUIDELINES, supra note 6, § 0.1 ("the Guidelines do not attempt to assign the burden of proof, or the burden of coming forward with evidence, on any particular issue.") COLUMBIA BUSINESS LA W REVIEW [Vol. 2005 MOVING PAST GUIDELINES PRESUMPTIONS triggered.2 0 Furthermore, it seems that for those elements where staff assumes the burden of proof, it satisfies this burden by a preponderance of the evidence, which is consistent with the relevant case law.21 When considering whether this approach is likely to satisfy the requirement that there be a reasonable likelihood that a merger will substantially lessen competition, it is useful to consider that the ultimate probability that an event will occur is a function of the individual probabilities of each element, at least where these elements are independent.22 For example, let's say that one is slightly more than 50% confident 23 that the market is appropriately defined, that entry is unlikely, and that unilateral or coordinated effects are likely. Assuming that each element is independent of the other, the odds that an anticompetitive effect will occur post- merger is given by multiplying the individual probabilities- 50% * 50% * 50%--or 12.5%.24 20 Id. § 1.51(c) ("The presumption may be overcome by a showing that factors set forth in Section 2-5 of the Guidelines make it unlikely that the merger will create or enhance market power or facilitate its exercise"). 21 See, e.g., United States v. Oracle Corp., 331 F. Supp. 2d 1098, 1132 (N.D. Cal. 2004) ("[in order to sustain plaintiffs' product market definition the court must find, by a preponderance of the evidence, that plaintiffs' have shown an articulable and distinct product market .... ); United States v. SunGard Data Systems, Inc., 172 F. Supp. 2d 172, 180 (D.D.C. 2001) ("The United States has the ultimate burden of proving a Section 7 violation by a preponderance of the evidence."). ' For a discussion of the so-called "product rule" and the role of probabilistic judgments in the legal process, see Neil B. Cohen, Confidence in Probability: Burdens of Persuasion in a World of Imperfect Knowledge, 60 N.Y.U. L. REV. 385 (1985). 2" The evidentiary standard "by a preponderance of the evidence" is commonly understood to require a probability exceeding 50% that alleged facts are true. See, e.g., United States v. Shonubi, 895 F. Supp. 460, 470 (E.D.N.Y. 1995) ("A survey of judges in the Eastern District of New York found general agreement that a preponderance of the evidence translates into 50+ percent probability.") (internal quotations omitted). 24 Cf. Alan A. Fischer and Robert H. Lande, Efficiency Consideration in Merger Enforcement, 71 CAL. L. REV. 1582, 1585 n.354 (1983). ("[Diecisionmaker A might decide that he is 70 percent certain he has defined the product market correctly. Suppose additionally that A is 80 No. 2:317] COLUMBIA BUSINESS LAW REVIEW Of course, these elements will not always be independent. For example, a narrow market definition may make entry less likely. This exercise, however, is useful to illustrate the point that anticompetitive harm is unlikely if the agency is able to narrowly prevail on market definition, barriers to entry, and the likelihood of coordinated or unilateral interaction. An alternative approach is either to impose a requirement upon staff that each element be proved by clear and 25convincing evidence, or alternatively, that the relevant decision-maker be required to step back after all the elements are presented, and assess whether anticompetitive effects are likely in light of the relative strength and weakness of the proof on each element. In considering the impact of the proposal above, it is useful to apply the clear and convincing evidence approach to recent cases brought and lost by the agencies. 26In Oracle, the critical issues included market definition, likelihood of unilateral effects,27 and the ability of SAP to 281 reposition. While the government could fairly view that the evidence on market definition and unilateral effects was clear and convincing given the weight and volume of percent sure he has defined the geographic market properly and 90 percent sure that significant entry barriers exist. Assuming that these determinations are statistically independent, A will conclude there is a .70 x .80 x .90 = .504... chance that there is market within which significant market power could arise."). Cf. FTC v. Consolidated Foods Corp., 380 U.S. 592, 605 (1965) (Stewart, J., concurring) (disagreeing with the Court's evaluation of the evidence: "Ithe law does not require proof that competition certainly will be lessened by the merger. But the record should be clear and convincing that the requisite probability is present."); United States v. M.P.M., Inc., 397 F. Supp. 78, 91 (D. Col. 1975) (citing J. Stewart's concurring opinion in Consolidated Foods) (applying clear and convincing evidence standard in Section 7 case). 26 331 F. Supp. 2d at 1123-65, 1175. Id. at 1166-73, 1175. ' The court did not reach this issue because it concluded that plaintiffs "failed to show an area of localized competition between Oracle and PeopleSoft." Id. at 1172-73. [Vol. 2005 MOVING PAST GUIDELINES PRESUMPTIONS customer testimony, the evidence on SAP's ability to reposition was not. SAP was, after all, the largest firm in the relevant market 29 and clearly had the resources to reposition.30 The only question is whether it would. In Arch Coal, the critical issue was likelihood of post- merger coordination. While there appeared to be some evidence that firms wished to be able to reduce market output, 31 there was, at least according to the court, little evidence that this had occurred in the past.32 To the contrary, evidence seemed to suggest that spikes in demand and supply made coordination extremely unlikely.33 IV. THE APPROPRIATE ROLE OF PRESUMPTIONS A. Introduction Whatever the appropriate standard of proof, it is clear that the element of likely competitive effect cannot be satisfied by presumption. In general, the purpose of a presumption is to assist a court or agency in reaching conclusions where direct evidence is unavailable. Presumptions are appropriate where we have sufficient experience with an event to predict its likely consequence without a detailed examination in a particular case. Yet it is clear both from the empirical literature and the FTC's and DOJ's own enforcement Id. at 1167 ("SAP ranked highest.., with a 39 percent market share"). o In its post-trial brief, Oracle pointed out that plaintiffs did not proffer any evidence that SAP was unable to reposition itself. According to Oracle, plaintiffs' expert Elzinga admitted to seeing no reason why SAP could not reposition into the parts of the market in which it was not dominant. See Oracle Corporation's Corrected Post-Trial Brief at 34, United States v. Oracle Corp., 331 F. Supp. 2d 1098 (N.D. Cal. 2004) (No. C 04-0807 VRW), available at http://www.oracle.com/peoplesoft/ OraclePostTrialBrief.pdf. "I 329 F. Supp. 2d at 137, 139-40 & 158. 32 Id. at 140 & 158. Id. at 140, 142-43, 146 & 158. No. 2:317] COLUMBIA BUSINESS LAW REVIEW practice that this does not describe the presumptions found in the Merger Guidelines. B. Coordinated Effects Presumption Section 1.51(c) of the Guidelines provides that mergers that result in a post-merger HHI of at least 1800 and a delta of at least 100 are presumed to be likely to create or enhance market power or facilitate its exercise. At the outset it is important to note that the coordinated effects presumption could have some utility in that it gives information to the business community as to the antitrust risk presented by the deal. This information in turn gives the contracting parties the opportunity to purchase contractual insurance in the form of hell-or-high-water clauses and termination fees. This efficiency, though, quickly turns into an inefficiency to the extent that the agencies in practice use statistical thresholds much higher than the ones set forth in the Guidelines (see infra). In evaluating the validity of the coordinated effects presumption, we should start by noting that there is little empirical support for the presumption that mergers above an 1800/100 threshold are likely to create or enhance market power or facilitate its exercise. While many studies have found that there is a general relationship between price (orcost)and . 34 cost) and concentration, a significant number of studies have found that there is not.35 Moreover, studies of the " See generally Richard Schmalensee, Inter-Industry Studies of Structure and Performance, HANDBOOK OF INDUSTRIAL ORGANIZATION Ch. 6 (Richard Schmalensee & Robert Willig, eds., 1989); see also Ronald W. Cotterill, Market Power in the Retail Food Industry: Evidence from Vermont, 68 REV. ECON. & STAT. 379, 379-86 (1986) (positive and statistically significant relationship between HHI and price); Ronald W. Cotterill & C. David Harper, Market Power and the Demsetz Quality Critique: An Evaluation for Food Retailing, FOOD MARKETING POLY CENTER, RES. REP. No. 29, U. CONN. (1994) (same). "' See, e.g., Phil Kaufman & Charles R. Handy, Supermarket Prices and Price Differences: City, Firm, and Store-Level Determinants, USDA, ERS, TECH. BULL. 1776 (1990) (firm market share and HHI index were negatively but insignificantly correlated with price of goods at 616 [Vol. 2005 MOVING PAST GUIDELINES PRESUMPTIONS relationship between concentration and price (or cost) have been subject to considerable empirical criticism, including the failure to appropriately define relevant product and geographic markets,36 the use of accounting data to measure economic profits, 37 and the failure to control for non-price competition (such as quality).8 supermarkets chosen from twenty-eight cities selected at random); Craig M. Newmark, A New Test of the Price-Concentration Relationship in Grocery Retailing, 33 ECON. LETTERS 369, 369-73 (1990) (negative and insignificant relationship between concentration and basket of goods in twenty-seven cities). ' See, e.g., Richard J. Sexton & Mingxia Zhang, An Assessment of Market Power in the U.S. Food Industry and its Impact on Consumers 21, Paper prepared for the Conference on "The American Consumer and the Changing Structure in the Food System," Arlington, Virginia, (May 4-5, 2000) available at http://www.ers.usda.gov/briefing/foodmarketstructures/ conferencepapers/sexton.pdf. Price-concentration studies frequently use SIC codes as a proxy for market definition. For example, one study examined the link between concentration and price within a market consisting of SIC Code 2015, which would include turkeys, other poultry and liquid, dried, and frozen eggs, all of which belong in separate markets. See Sanjib Bhuyan & Rigoberto A. Lopez, Oligopoly Power in the Food and Tobacco Industries, 79 AM. J. AGRIC. ECON. 1035, 1035-43 (1997). Other studies have defined relevant markets as "fresh fruits" and "fresh vegetables." Albert J. Reed & J. Stephen Clark, Structural Change and Competition in Seven U.S. Food Markets, USDA, ERS, TECH. BULL. No. 1881 (2000), available at http://usda.mannlib.cornell.edu/reports/general/tb/tb1881.pdf. In a critique of a study finding no link between price and concentration, Werden found that the ratio of the annual dollar volume of commerce in the alleged relevant market to the value of shipments for the corresponding four-digit SIC code was less than 25% for 77.8% of the alleged relevant markets, and less than one percent for 32.5% of the markets, levels that are "far too aggregated to permit the measurement of any effects." See Gregory J. Werden, U.S. Dep't of Justice, The Effect of Antitrust Policy on Consumer Welfare: What Crandall and Winston Overlook 5, EAG Discussion Paper, (Jan. 2003), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=384100. "7 See, e.g., Franklin M. Fisher & John J. McGowan, On the Misuse of Accounting Rates of Return to Infer Monopoly Profits, 73 AM. ECON. REV. 82, 82-87 (1983). " See, e.g., Cotterill & Harper, supra note 34; Sexton & Zhang, supra note 36, at 18. No. 2:317] COLUMBIA BUSINESS LAW REVIEW But leaving aside these empirical questions, even those studies that find a relationship between price and concentration do not support the proposition that a post- merger HHI of 1800 and a delta of 100 are, in fact, the appropriate thresholds. Moreover, the agencies, as they freely admit, do not rigorously follow the HHI presumptions set forth in the Guidelines. The agencies issued second requests and obtained relief in cases where the post-merger HHI is in the 1800/100 range.39 However, the markets involved in these cases represented at most just forty of the 1263 markets in which the DOJ or FTC sought enforcement between 1999 and 2003.40 If the overwhelming majority of cases in the 1800/100 range do not, upon a fuller examination, raise anticompetitive issues, then it is difficult to justify a presumption that they do. Indeed, this very point was made by the court in Arch Coal, citing to that very same enforcement data for the proposition that mergers that barely exceed the 1800/100 range are entitled to only a weak presumption.4' " Specifically, according to the FTC's study of merger investigations between 1996 and 2003, the FTC opened investigations of forty-four "markets" where the post-merger HHI was between 1800 and 2000 and the delta was higher than 100, of which thirty-four resulted in some sort of enforcement action. FED. TRADE COMM'N, HORIZONTAL MERGER INVESTIGATION DATA, FISCAL YEARS 1996-2003, table 3.1 (Revised Aug. 31, 2004), available at http://www.ftc.gov/os/2004/08/040831horizmergersdata 96-03.pdf. 40 See U.S. DEP'T OF JUSTICE & FED. TRADE COMM'N, MERGER CHALLENGES DATA, FISCAL YEARS 1999-2003, table 1 (Dec. 18, 2003), available at http://www.ftc.gov/os/2003/12/mdp.pdf. There were nineteen cases where the delta was between 100-199 and the HHI was over 1800; seventeen where the post-merger HHI was under 1800 and the delta over 100; and two where the HHI was over 1800 and the delta between 0-99. Id. 41 329 F. Supp. 2d at 129. [Vol. 2005 No. 2:3 17] MOWNG PAST GUiDELINES PRESUMPTIONS C. Unilateral Effects Threshold Section 2.211 of the Guidelines provides that where the merger has a post-merger HHI and delta that fall outside of the safe harbors, and the products are differentiated, and the firms have a combined share of 35%, the agency will presume that a significant number of customers will regard the products of the merging parties as their first and second choice. It is conceded that the 35% presumption is relatively innocuous. It is, after all, not a presumption that unilateral effects are likely, but simply a presumption with respect to consumer preferences. Moreover, one would think that it would not be difficult to overcome this presumption with objective evidence concerning the attributes of the products of the merging parties. The fact that a presumption is relatively innocuous should not be a sufficient reason to retain it. At the outset, there is no a priori reason to believe that market share gives an accurate view as to the relative preferences of consumers. Taking an example from breakfast cereal, assume that one leading product is crunchy without sugar, and another is a sugar flake cereal. Assume further that 50% of consumers have a strong preference for crunchy non-sugared cereal, and another 50% have a strong preference for sugar flake cereal. Does this mean that a significant number of customers regard Frosted Flakes and Grape Nuts as their first and second choice if their combined share were higher than 35%? Against this drawback, the 35% threshold has little functional utility. Unlike the HHI thresholds, the 35% presumption does not serve a gate-keeping role. Specifically, the Guidelines leave open the possibility that where there is direct evidence that a significant number of customers regard the products of the merging firms as close substitutes, unilateral effects may be likely, even where the products offered by the merging firms have a combined share well under 35%.42 Thus, it is unclear whether the 35% 42 MERGER GUIDELINES, supra note 6, § 2.221. No. 2:317) MOVING PAST GUIDELINES PRESUMPTIONS COLUMBIA BUSINESS L4WREVIEW [Vol. 2005 presumption either sheds transparency into agency practice, or allows parties to allocate risk. More importantly, the use of the 35% threshold, coupled with the requirement that the market be highly concentrated for the presumption to be triggered, is so conceptually incoherent43 that it has brought considerable confusion into the economics of unilateral effects. It is therefore .perhaps not surprising that both the bar, as well as the courts, make frequent errors in applying unilateral effects theory. Indeed, in Oracle, Judge Walker openly struggled to understand the requirements of a unilateral effects case, suggesting at times that Oracle and PeopleSoft would need to have near monopoly power over an identifiable group of customers,44 something that is neither required by the Guidelines nor by economic theory. D. Should Statistical Presumptions Be Abandoned? Given the foregoing, the question should be fairly asked, should the Guidelines presumptions be abandoned? The cost of the presumptions is considerable. Because they are not supported by sufficient empirical evidence they increase the risk that where they are relied upon, courts and agencies will enjoin mergers that do not present competitive harm. Because they are not followed in practice, they increase the risk that parties purchase more contractual insurance than is required. Because of their importance, they force parties and the DOJ and FTC to treat market definition as outcome determinative, leaving little time for ' Cf Jonathan B. Baker & Steven C. Salop, Should Concentration be Dropped from the Merger Guidelines?, 33 LUWLA L. REV. 3, 11-12 (2001) (asserting that concentration is rather one of the least than one of the most relevant factors in assessing unilateral effects in differentiated markets); Armando E. Rodriguez & Malcolm B. Coate, Merger Pitfalls in Practice: Three Case Studies, 20 U. PA. J. INT'L ECON. L. 793, 802 (1999) (criticizing the 35% presumption on the ground that product positioning, not market share, is relevant to the likelihood of unilateral effects in differentiated markets). 1 331 F. Supp. 2d at 1118, 1123. COLUMBIA BUSINESS LA W REVIEW (Vol. 2005 No. 2:3 17] MOVING PAST GUIDELINES PRESUMPTIONS analysis of competitive effects. Indeed, it is notable that Arch Coal, along with the FTC's analysis of the Cruise Line45 mergers, represents one of the few express applications of coordinated interaction theory in a specific merger. Against this, it is also important to recognize the importance of statistical thresholds in guiding the business community and preventing a return to Von's Grocery, a merger with a post-merger HHI of less than 300 and a delta of less than forty.46 Given all of this, I recommend the following. First, the HHI thresholds should be raised to reflect agency practice. This will allow parties to appropriately allocate risk. Second, the presumption in Section 1.51(c) should be eliminated and the relevant language should be changed to make clear that mergers above the appropriate threshold "may" potentially raise significant competitive concerns, but are not presumed to do so. Third, because the 35% threshold does not offer a safe-harbor, and because there is no empirical or theoretical support for such a screen, it should be eliminated from the Guidelines.47 Alternatively, the 35% threshold, or perhaps a lower number, should be used as a safe-harbor under which the agencies presume that the costs of enforcement outweigh the likelihood of anticompetitive harm. " See Statement, Fed. Trade Comm'n, Royal Caribbean Cruises, Ltd./P&O Princess Cruises plc and Carnival Corporation/P&O Princess Cruises plc (FTC File No. 021 0041), available at http://www.ftc.gov/os/2002/10/ cruisestatement.htm. 46 This HHI and delta range is calculated based on the market shares mentioned in Justice White's concurring opinion. United States v. Von's Grocery Co., 384 U.S. 270, 281 (1966) (White, J., concurring); cf. Robert H. Lande, Resurrecting Incipiency: from Von's Grocery to Consumer Choice, 68 ANTITRUST L.J. 875 (2001) (calculating the post-merger HHI and delta in Von's Grocery at less than 300 and 20). 41 Cf Baker and Salop, supra note 43, at 12 (recommending "that the Merger Guidelines remove any suggestion of a 35% market share safe harbor in the evaluation of the possibility of unilateral competitive effects among firms selling differentiated products.") (internal quotations omitted). MOVING PAST GUIDELINES PRESUMPTIONSNo. 2:317] COLUMBIA BUSINESS LAW REVIEW V. FORMS OF APPROPRIATE EVIDENCE A. Introduction Having argued that the staff should have a high internal burden in seeking relief, and that presumptions cannot be relied upon to satisfy that burden, the question should fairly be raised, what sort of evidence should the agencies and courts rely upon to demonstrate the likely competitive effect of the merger? Specifically, I suggest that great weight should be given to the following evidence: * natural experiments that allow one to test directly the merger's effect in the pre-merger world; " the views of senior management that motivated the transaction, including credible evidence that management did not believe that it would allow the merged entity to increase price; and " merger simulations, especially where the simulations use estimated cross-elasticities rather than assumptions based upon market share. In contrast, the following evidence should be given less weight: * customer affidavits that are not drawn from a representative sample and ask the customers to opine on issues for which they have not yet been qualified; and * documents written by middle or low level management and salesmen. [Vol. 2005 MOVING PAST GUIDELINES PRESUMPTIONS B. Types of Evidence Entitled to Great Weight 1. Natural Experiments The first type of credible evidence of likely anticompetitive harm is natural experiments that allow one to perform an econometric analysis of prices, profits, or output, under circumstances that resemble the post-merger world. Perhaps the most cited example of a natural experiment occurred in Federal Trade Commission v. Staples, Inc. where prices on a basket of goods were lower in markets where Office Depot and Staples were both present than where only one was present.48 Natural experiments are most frequently used where there is significant geographic or temporal variability in competitive conditions.49 While the required conditions are not always present, natural experiments frequently are the best evidence of the likely anticompetitive effect of a transaction. Unlike merger simulations, there are few assumptions in the model, other than the assumption that the future will be similar to the past. Given the value of natural experiments, where conditions exist in the pre-merger world to directly test the government's theory of competitive harm, the failure to put forward that evidence should be noted. Or put more specifically, if the market is highly concentrated and the government's theory is coordinated effects, the agency should have an affirmative obligation to show that coordination is currently occurring. The court found that the government 48 970 F. Supp. 1066, 1075-77 (D.D.C. 1997). 4' In the Cruise Line mergers, for example, the FTC was able to conduct various natural experiments because it had a few years worth of data at its disposal which showed a temporal variation in the competitive conditions of the industry. For a discussion of these natural experiments, see Mary T. Coleman, David W. Meyer & David T. Scheffman, Empirical Analyses of Potential Competitive Effects of a Horizontal Merger: The FTC's Cruise Ships Merger Investigation, 23 REV. IND. ORG. 121 (2003), available at http://www.ftc.gov/be/riocruise0703.pdf. No. 2:317] did this in Federal Trade Commission v. Cardinal Health5° and Hospital Corporation of America v. Federal Trade Commission51 where it prevailed, and the court found that it failed to do this in Arch Coal52 and United States v. Archer- Daniels-Midland Co. 53 where it did not. Where the government's theory is price discrimination, the government should have the obligation to show price discrimination is occurring in the pre-merger world. The government failed to do this in Oracle54 and SunGard55 and lost both cases. If the government's theory is that the reduction in the number of bidders from three to two increases price, then the government should have the affirmative burden to show that prices are higher where there are only two qualified bidders, something it again failed to do in both Oracle56 (e.g., in verticals where it was alleged that SAP was not an actual or credible bidder) and SunGard.5 ' 2. Strategic Documents The second form of credible evidence is high-level strategic documents predicting the merger's impact on 50 12 F. Supp. 2d 34, 65 (D.D.C. 1998). " 807 F.2d 1381, 1388-89 (7th Cir. 1986). 12 329 F. Supp. 2d at 140. '3 781 F. Supp. 1400, 1421 (S.D. Iowa 1991). 331 F. Supp. 2d at 1172-73. " 172 F. Supp. 2d at 193 n.25. The court did not reach the price discrimination claim in SunGard because it concluded plaintiffs failed to establish the relevant product market. Id. 56 Plaintiffs showed only that Oracle offered substantial discounts where it competed with PeopleSoft. 331 F. Supp. 2d at 1169. They failed to show that Oracle's discounts were less substantial where it only competed with SAP as opposed to where it competed with both PeopleSoft and SAP. " As in Oracle, the government's theory in SunGard was clearly that a three to two reduction would increase price. 172 F. Supp. 2d at 181 ("According to the government ... the proposed acquisition would create a duopoly."). However, the court did not reach an analysis of this claim as the government failed to establish the relevant product market it proposed. 172 F. Supp. 2d at 193 n.25. COLUMBIA BUSINESS LA W REVIEW [Vol. 2005 competition. An Item 4(c) document"' stating that the purpose of the transaction is to eliminate a competitor and increase price is highly probative. Yet, while the agencies clearly place great weight on documents and testimony that the acquiring party intends to increase price or reduce output post-merger, the agencies seem to place no weight on the absence of that evidence, especially where the motivating purpose behind the transaction is simply to realize financial synergies, as is often the case. Indeed, the FTC reports that between 1996 and 2003 it took enforcement action in 70% of the cases where there were no documents clearly predicting post-merger competitive harm. 9 This dichotomy is curious. The merging parties typically have the most knowledge about the underlying market and the likely impact of the transaction of the combined firm's profitability. Furthermore, there can be little question that the likely impact of the transaction on price or profitability is a key factor in setting the purchase price. The idea that the government would know more about the likely competitive impact of a merger than a firm who just invested $100 million to acquire a direct competitor seems to commit the cardinal sin of substituting the views of government for those of the marketplace. Of course, the government does not consider the absence of such evidence because it believes that the parties can hide their true intent. It is true that sophisticated companies will carefully avoid making such express statements when drafting Item 4(c) documents, though the ability of corporate counsel to "scrub" 4(c)s has been substantially reduced by the Sarbanes-Oxley Act.6° It is difficult to believe, however, that management would commit hundreds of millions of dollars See FED. TRADE COMM'N, ANTITRUST IMPROVEMENTS ACT NOTIFICATION AND REPORT FoRM, Item 4(c) (Sept. 10, 2002), Appendix to 16 C.F.R. § 803 (2003). '9 See FED. TRADE COMM'N, HORIZONTAL MERGER INVESTIGATION DATA, supra note 39, tables 5.2 and 6.2. 'o See Sarbanes-Oxley Act of 2002, Pub. L. No. 107-204 § 802, to be codified at 18 U.S.C. § 1519 (prescribing criminal penalties for altering documents). MOVING PAST GUIDELINES PRESUMPTIONSNo. 2:317] COLUMBIA BUSINESS LAW REVIEW based, in part, on the ability of the merged entity to increase price or reduce output and not leave a paper trail, especially in the synergy model. Furthermore, it would seem that the true views of management would be exposed in depositions or courtroom testimony, and, in fact, discovering this intent is a role that courts are uniquely qualified to play. 3. Merger Simulation Merger simulation is the third form of credible evidence of likely anticompetitive harm. It should be noted that merger simulations are a mathematical representation of the price effect that would result given the observed facts and the model's assumption(s). Where either observed facts or assumptions are incorrect, the model's predictions are not reliable.61 For example, if a firm can reposition itself to replace competition lost as a result of a merger, the price effect predicted by a simulation may be eliminated. In considering the inferences that can be made from a properly performed simulation, it is important to recognize that there are generally two types of merger simulation: 62those that use estimated cross-elasticities and those thatassume cross-elasticities based on market shares.! 61 Cf Gregory J. Werden, Luke M. Froeb, & David T. Scheffman, A Daubert Discipline for Merger Simulation, 18 ANTITRUST 89 (2004) ("[TIhe reliability of any particular application of merger simulation should be gauged by examining the modeling process, which is at least as much art as science."); Daniel L. Rubinfeld, Market Definition with Differentiated Products: The Post/Nabisco Cereal Merger, 68 ANTITRUST L.J. 163, 180 (2000) ("These difficulties may mean that in some cases complex simdlations will contribute little more than can be learned about the anticompetitive incentive of the merging firms to raise price from the demand elasticities alone."). 62 A merger simulation model that uses estimated cross-elasticities is the so-called Almost Ideal Demand System. See, e.g., Angus Deaton & John Muellbauer, An Almost Ideal Demand System, 70 AM. ECON. REV. No. 3, at 312 (1980). 6 A merger simulation model that uses assumed cross-elasticities is the so-called Antitrust Logit Model. See, e.g., Gregory J. Werden & Luke M. Froeb, The Antitrust Logit Model for Predicting Unilateral Competitive Effects, 70 ANTITRUST L.J. 257 (2002). Another example is the so-called [Vol. 2005 MOVING PAST GUIDELINES PRESUMPTIONS In general, models that assume cross-elasticities based on market shares, while useful, are not independent evidence of likely anticompetitive impact. Rather, they simply follow and are dependent upon the market share calculations and other assumptions. A similar point was made by Judge Walker in Oracle when he specifically criticized the DOJ's model for using assumed market share-based elasticities as an input instead of estimated cross-elasticities.6 In contrast, provided that the assumptions underlying the model are correct, simulations that use estimated cross- elasticities can provide strong independent evidence of the likely anticompetitive impact of the transaction, even where the market definition is incorrect. Of course, it should be noted that such evidence is not always available. 5 While retail scanner data is most frequently used, other types of data may also be suitable; cross-elasticities can also be estimated using panel data, transaction level data obtained through civil investigative demand, or from conjoint survey data.66 PCAIDS model. See, e.g., Roy J. Epstein & Daniel L. Rubinfeld, Merger Simulation: A Simplified Approach With New Applications, 69 ANTITRUST L.J. 883 (2002). 331 F. Supp. 2d 1098, 1170 (N.D. Cal. 2004) (finding that the results of the merger simulation were unreliable because the market shares used in the simulation were calculated based on unreliable data). ' Cf. Werden and Froeb, supra note 63, at 259 n.7 ("[Tihe available data tend not to be rich enough to identify accurately the large number of separate substitution relationships being estimated."). For support for survey based competitive analysis, see e.g., Jonathan B. Baker & Daniel L. Rubinfeld, Empirical Methods in Antitrust Litigation: Review and Critique, 1 AM. L. & ECON. REV. No. 1/2, at 386-435 (1999); Gregory J. Werden, Luke M. Froeb, & Timothy J. Tardiff, The Use of the Logit Model in Applied Industrial Organization, 3 INT'L J. ECON. Bus. No. 1, at 85-107 (Feb. 1996); Gregory J. Werden, Luke M. Froeb, & Timothy J. Tardiff, The Demsetz Postulate and the Effect of Mergers in Differentiated Product Industries, in ECONOMIC INPUTS, LEGAL OUTPUTS: THE ROLE OF ECONOMISTS IN MODERN ANTITRUST (Fred McChesney ed. 1998); Shari S. Diamond, Reference Guide on Survey Research, in REFERENCE MANUAL ON SCIENTIFIC EVIDENCE 229-276 (FJC 1994), available at http://www.tjc.gov/public/pdf.nsf/lookup/8.sur-res.pdf/$File/ 8.sur_res.pdf. No. 2:317] COLUMBIA BUSINESS LAW REVIEW EVol. 2005 C. Types of Evidence Not Entitled to Great Weight Unfortunately, the vast majority of evidence presented in an antitrust case, as well as the vast majority of the evidence produced in a second request, has comparably little utility in helping us predict post-merger effects. 1. Customer Testimony Customer testimony-typically in the form of affidavits- has long been the somewhat weak backbone of antitrust cases. The FTC reports that between 1996 and 2003, it took enforcement action in 98% of the cases where there were• 67 strong customer complaints. This is so because customers are thought to be uniquely qualified to answer the central question in most Section 7 cases, namely, what is the likely 68impact of the merger on the prices that customers pay. Curiously, however, the FTC also reports that between 1996 and 2003 it also obtained consent decrees in roughly half of• 69 the cases where there were no strong customer complaints, a somewhat striking number given the agency's views on the reliability of customer opinion. Customer affidavits have several significant flaws that generally undermine their validity. First, customer affidavits are written for the purpose of litigation, and 67 See FED. TRADE COMM'N, HORIZONTAL MERGER INVESTIGATION DATA, supra note 39, tables 7.1 and 8.1. 6 One of the two main types of evidence the FTC mentions in its merger investigation data, are "strong customer complaints," which it defines as a customer's "credible concern that a significant anticompetitive effect would result were the transaction allowed to proceed." Id. See also David Scheffman, Malcolm Coate, & Louis Silvia, Twenty Years of Merger Guidelines Enforcement at the FTC: An Economic Perspective, 71 ANTITRUST L.J. 277, 304 (2003) ("In the early years of merger review under the 1982 Guidelines, the importance of customer complaints and hot documents was fully appreciated. FTC staff and outside parties have become ever more thorough and sophisticated in attempting to solicit and assess customer opinions.") (internal quotations omitted). 69 See FED. TRADE COMM'N, HORIZONTAL MERGER INVESTIGATION DATA, supra note 39, at tables 7.2 and 8.2. COLUMBIA BUSINESS LAW REVIEW [Vol. 2005 MOVING PAST GUIDELINES PRESUMPTIONS therefore lack the normal indicia of reliability found in many documents written in the ordinary course of business. Second, customer affidavits are almost never written by the affiants, and instead represent efforts by lawyers to put words into the mouths of businessmen. Third, customer affidavits are often procured by the government under implied threat of subpoena, or by the parties either because of prior personal relationship, or because of the customer's desire to curry favor with the merged entity. ° Fourth, customer affidavits almost always result from some selection bias, suggesting that their views do not represent the general customer population, but rather represent those customers who are uniquely suited to be harmed, or benefited by the transaction. As a result, there is rarely little assurance that a judge can extrapolate from the affidavits to the general population, a point that was made in SunGard,1 Cardinal Health,7 2 and United States v. Engelhardt. 7 3 70 Frequently, the merging parties and the government will get contradictory affidavits from the same customers. In this regard, the government's subpoena power can be used to test the conviction of a defendant's affiant. Defendants, thus, may not want to get affidavits unless the customer's view is strongly held and the defendant has an understanding of the antitrust basis for the customer's position. From a government perspective, customers may simply be complaining about changes or disruptions in the market that will cause the customer inconvenience. Such concerns do not necessarily speak to an overall lessening of competition. It is not fatal to defendants, however, if the government flips a pro-merger customer. If the customer's equivocation obscures the issue, the government may fail to carry its burden of proof. Cf United States v. SunGard Data Systems, Inc., 172 F. Supp. 2d 172, 189 (D.D.C. 2001). 71 Id. at 191-92 ("The sampling of customer statements before the Court is minuscule when compared with the entire universe of defendants' shared hotsite customers."). 7 12 F. Supp. 2d 34, 48 (D.D.C. 1998) (finding that the ability of certain customers to switch to in-house solutions in response to a price increase by defendants was not representative for the position of the majority of defendants' customers). '3 126 F.3d 1302, 1306 (11th Cir. 1997) ("No matter how many customers in each end-use industry the Government may have No. 2:317] COLUMBIA BUSINESS LAW REVIEW [Vol. 2005 Fifth, customer affidavits frequently contain opinion testimony of the sort one would expect from an expert. For example, an affiant may testify that it is unlikely that the merged entity will or will not increase price, or that other firms will or will not enter the market, or reposition their products. While there may be circumstances in which a customer would be qualified to give such opinions, the witness still must be qualified as an expert.F4 What's more, the conclusions expressed with respect to product market, anticompetitive effects, or entry, must be substantiated by the sort of market analysis that would be found in an expert's affidavit. It was this very problem that caused the court to reject the customer affidavits in both Oracle75 and Arch Coal.76 In sum, while customer opinion may perform a useful litmus test for the government as to whether it should issue a second request, it is difficult to argue with the view of Areeda and Hovenkamp that customer affidavits are the 77"least reliable" form of evidence in antitrust cases. interviewed, those results cannot be predictive of the entire market if those customers are not representative of the market."). 14 See FED. R. EVID. 702. 76 331 F. Supp. 2d 1098, 1131 (N.D. Cal. 2004) ("If backed by credible and convincing testimony of this kind or testimony presented by economic experts, customer testimony of the kind plaintiffs offered can put a human perspective or face on the injury to competition that plaintiffs allege. But unsubstantiated customer apprehensions do not substitute for hard evidence."). " 329 F. Supp. 2d 109, 145-46 (D.D.C. 2004) ("[W]hile the Court does not doubt the sincerity of the anxiety expressed by SPRB customers, customers do not, of course, have the expertise to state what will happen in the SPRB market."). 77 PHILIP E. AREEDA & HERBERT HOVENKAMP, ANTITRUST LAw J[ 538b at 239 (2002) ("Though not irrelevant, such statements are often unreliable."). While these flaws are inherent in customer affidavits, most of these flaws can at least be addressed at cross-examination either at trial, or, at the very least, in depositions. COLUMBIA BUSINESS LA W REVIEW [Vol. 2005 MOVING PAST GUIDELINES PRESUMPTIONS 2. Documents Written by Salespeople and Low- Level Management Finally, we turn to the vast majority of documents produced in a second request-documents written by individuals who have virtually no ability to make strategic decisions at the company. While there is no question that such documents frequently set forth facts and opinions that help form the texture of the industry, given the tens of millions of dollars spent annually by companies responding to second requests, it is a fair question whether the utility of such documents is outweighed by the costs of production. In making this argument, I wish to exclude transactional documents such as sales call reports. These documents frequently include valuable data concerning how firms respond to competition and can provide valuable input to bidding models and natural experiments. In addition, documents that describe what salespeople think specific customers are doing with respect to dual-sourcing are equally valuable. In contrast, non-transactional documents written by low- level management, and, in particular, sales people, serve little probative value, especially when weighed against the sizeable cost of their collection. While such emails frequently give some evidence as to market realities, as Judge Easterbrook noted in A.A. Poultry Farms, Inc. v. Rose Acre Farms, Inc. those introduced into evidence are just as frequently going to reflect bravado or misleading information. 8 78 881 F.2d 1396, 1402 (7th Cir. 1989) ("Lawyers rummage through business records seeking to discover tidbits that will sound impressive (or aggressive) when read to a jury. Traipsing through the warehouses of business in search of misleading evidence both increases the costs of litigation and reduces the accuracy of decisions. Stripping intent away brings the real economic questions to the fore at the same time as it streamlines antitrust litigation."). No. 2:317] COLUMBIA BUSINESS LAW REVIEW VI. CONCLUSION Over twenty years ago, the Guidelines formally introduced Chicago-style economics into antitrust. By advocating a rigorous approach and HHI safe-harbors, the Guidelines ensured that 1960s and 1970s style cases like Vons Grocery would not be brought in the future. However, as Oracle and Arch Coal make clear, the Guidelines as presently interpreted do not eliminate the risk that the government will at times sue to prevent mergers that courts ultimately determine do not raise a significant risk of anticompetitive effect. Because there is no reason to believe that the cases that the government litigates are any stronger than the cases that it settles, there is a substantial risk that agencies are obtaining consent decrees in hundreds of cases that they would otherwise lose in court. To reduce this risk, the Guidelines should be revised to make clear that the government always has the burden of persuasion on whether the merger is likely to have anticompetitive effect and, further, that the government should not short-circuit this requirement by relying on statistical presumptions but instead must prove this element through the use of credible, objective evidence concerning market performance. [Vol. 2005