|This document is available in two formats: this web page (for browsing content), and PDF (comparable to original document formatting). To view the PDF you will need Acrobat Reader, which may be downloaded from the Adobe site. For an official signed copy, please contact the Antitrust Documents Group.|
Claude F. Scott, Esq.
Pam Cole, Esq. (CA Bar No. 208286)
Renata B. Hesse, Esq.(CA Bar No. 148425)
Phillip R. Malone, Esq. (CA Bar No. 163969)
U.S. DEPARTMENT OF JUSTICE, ANTITRUST DIVISION
450 Golden Gate Avenue, Rm. 10-0101
San Francisco, CA 94102-3478
(415) 436-6683 (Fax)
Attorneys for Plaintiff the United States of America
Also filed on behalf of 10 Plaintiff States (See signature block)
UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA
SAN FRANCISCO DIVISION
TABLE OF CONTENTS
TABLE OF AUTHORITIES
AlliedSignal, Inc. v. B.F. Goodrich Co., 183 F.3d 568 (7th Cir. 1999)
Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209 (1993)
Brown Shoe Co. v. United States, 370 U.S. 294 (1962)
Burch v. Goodyear, 554 F.2d 633 (4th Cir. 1977)
California v. American Stores Co., 872 F.2d 837 (9th Cir. 1989) rev'd on other grounds, 495 U.S. 271 (1990), relevant part reinstated, 930 F.2d 776, 777 (9th Cir. 1991);
Coastal Fuels of P.R., Inc. v. Caribbean Petroleum Corp., 79 F.3d 182 (1st Cir. 1996)
FTC v. Cardinal Health, Inc., 12 F. Supp. 2d 34 (D.D.C. 1998)
FTC v. Consolidated Foods Corp., 380 U.S. 592 (1965)
FTC v. Elders Grain, Inc., 868 F.2d 901 (7th Cir. 1989)
FTC v. H.J. Heinz, 246 F.3d 708 (D.C. Cir. 2001)
FTC v. Ind. Fed'n of Dentists, 476 U.S. 447 (1986)
FTC v. PPG Indus., Inc., 628 F. Supp. 881 (D.D.C. 1986), aff'd in part, 798 F.2d 1500 (D.C. Cir. 1986)
FTC v. Procter & Gamble Co., 386 U.S. 568 (1967)
FTC v. Staples, Inc., 970 F. Supp. 1066 (D.D.C. 1997)
FTC v. Swedish Match, 131 F. Supp. 2d 151 (D.D.C. 2000)
FTC v. University Health, 938 F.2d 1206 (1lth Cir. 1991)
Hawaii v. Standard Oil Co., 405 U.S. 251 (1972)
Hospital Corp. Of America v. FTC, 807 F.2d 1381 (7th Cir. 1986)
Image Tech. Servs., Inc. v. Eastman Kodak Co., 125 F.3d 1195 (9th Cir. 1997)
New York v. Kraft Gen. Foods, Inc., 926 F. Supp. 321 (S.D.N.Y. 1995)
Olin Corp. v. FTC, 986 F.2d 1295 (9th Cir. 1993)
Oltz v. St. Peter's Cmty. Hosp., 861 F.2d 1440, 1446 (9th Cir. 1988)
Palmer v. BRG of Georgia, Inc., 498 U.S. 46 (1990)
R.C. Bigelow, Inc. v. Unilever N. V., 867 F.2d 102 (2nd Cir. 1989)
Rebel Oil Co. v. Atlantic Richfield Co., 51 F.3d 1421 (9th Cir. 1995)
Times-Picayune Publishing Co., v. United States, 345 U.S. 594 (1953)
United States v. Addyston Pipe & Steel Co., 85 Fed. 271 (6th Cir. 1898)
United States v. Archer-Daniels-Midland Co., 866 F.2d 242 (8th Cir. 1988)
United States v. Baker Hughes, Inc., 908 F.2d 981 (D.C. Cir. 1990)
United States v. Brown, 936 F.2d 1042 (9th Cir. 1991)
United States v. Cooperative Theatres, Inc., 845 F.2d 1367 (6th Cir 1988)
United States v. E.I. duPont de Nemours & Co., 351 U.S. 377 (1956)
United States v. Engelhard Corp., 126 F.3d 1302 (1lth Cir. 1997)
United States v. Fischbach and Moore, Inc., 750 F.2d 1183 (3rd Cir. 1984)
United States v. General Dynamics Corp., 415 U.S. 486 (1974)
United States v. Gillette Co., 828 F. Supp. 78 (D.D.C. 1993)
United States v. Koppers Co. Inc., 652 F.2d 290 (2nd Cir. 1981)
United States v. Long Island Jewish Med. Ctr., 983 F. Supp. 121 (E.D.N.Y. 1997)
United States v. Marine Bancorp., 418 U.S. 602 (1974)
United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001)
United States v. Philadelphia Nat'l Bank, 374 U.S. 321 (1963)
United States v. Rockford Memorial Corp., 898 F.2d 1278 (7th Cir. 1990)
United States v. Rockford Memorial Corp., 717 F. Supp. 1251 (N.D. Ill. 1989)
United States v. SunGard Data Systems, Inc., 172 F. Supp. 2d 172 (D.D.C. 2001)
United States v. Syufy Enters., 903 F.3d 659 (9th Cir. 1990)
United States v. United Tote, Inc., 768 F. Supp. 1064 (D. Del. 1991)
United States v. UPM-Kymmene Oyj, 2003-2 Trade Cas. ¶ 74,101, 2003 WL 21781902 (N.D.Ill. 2003)
United States Anchor Mfg. v. Rule Indus., Inc., 7 F.3d 986, 995-96 (1lth Cir. 1993)
15 U.S.C.§ 18
2A Phillip E. Areeda et al., Antitrust Law (2d ed. 2002)
4 Phillip E. Areeda, et al., Antitrust Law (rev'd ed. 1998)
William J. Baer, Deborah L. Feinstein & Randal M. Shaheen, Taking Stock: Recent Trends in U.S. Merger Enforcement, Antitrust, Spring 2004
Serdar Dalkir et al., Mergers in Symmetric and Asymmetric Noncooperative Auction Markets: The Effects on Prices and Efficiency, 18 Int'l J. Indus. Org. 383 (2000)
Luke Froeb & Steven Tschantz, Mergers Among Bidders with Correlated Values, in Measuring Market Power 31 (Daniel J. Slottje ed., 2002)
Herbert Hovenkamp, Antitrust Law (1999)
Herbert Hovenkamp, Post-Chicago Antitrust: a Review and Critique, 2001 Colum. Bus. L. Rev
Horizontal Merger Guidelines § 1, reprinted in 4 Trade. Reg. Rep. (CCH) (1992, rev'd 1997) ("Merger Guidelines ")
Jerry A. Hausman & Gregory K. Leonard, Economic Analysis of Differentiated Products Mergers Using Real World Data, 5 Geo. Mason L. Rev. 321 (1997)
Steven Tschantz et al., Mergers in Sealed versus Oral Auctions, 7 Int'1 J. Econ. Bus. 201 (2000)
Keith Waehrer & Martin K. Perry, The Effects of Mergers in Open Auction Markets, 34 RAND J. Econ.(2003)
Carl Shapiro, Mergers with Differentiated Products, Antitrust, Spring 1996 23
Plaintiffs have proven their case:
In response, Oracle offered expert testimony to the effect that mergers to duopoly, or even mergers to monopoly, are either competitively benign or indeterminate in their impact on prices. Oracle then backtracked, claiming that a few unusual customers opting for marginal alternatives will discipline price increases for large companies for which such options are not even remotely viable. Significantly, this argument was advanced exclusively by witnesses on Oracle's payroll. The objective testimony compels the opposite conclusion. In the end, Oracle conceded what it and everyone else in the industry knows: Oracle, PeopleSoft, and SAP are "clearly the largest and most ubiquitous EAS vendors in the world." Oracle's Proposed Conclusions of Law ("OPCL")¶ 17.
This three-to-two merger will reduce competition and should not be permitted. Plaintiffs respectfully request that the Court enjoin Oracle's proposed takeover of PeopleSoft.
Oracle's proposed acquisition of PeopleSoft would leave customers with just two vendors for high function Human Resource Management ("HRM") and Financial Management System ("FMS") software and their accompanying services by combining the second and third-largest of the three existing vendors. Basic facts about the companies and the transaction are found at Plaintiffs' Corrected Proposed Findings of Fact ("PF") 1.1-1.4.
Oracle seeks to acquire market share and an ongoing revenue stream without competing for it, while concurrently eliminating its chief competitor. See PF 1.4.3-5. Oracle plans effectively to shut down sales of PeopleSoft products by no longer marketing to new customers, and to jettison PeopleSoft's technology by ending development of major upgrades to its products. See PF 1.4.4. If the transaction closes, Oracle will eliminate its closest competitor; if it does not close, Oracle still will have slowed PeopleSoft's growth, to the benefit principally of Oracle.
In today's global economy, the ability to reduce the costs of running an organization is vital to an organization's success. ERP software provides tools necessary to automate key business processes efficiently. ERP software is often organized into four categories: HRM, FMS, supply-chain management ("SCM") and customer relationship management ("CRM") software.2 Organizations manage employees through HRM software, which generally includes at least the core functions of payroll planning, employee tracking (names, addresses, salaries, job levels, etc.), and benefits administration. See PF 2.1.4; Jt. Sub. Defn ¶ 5. Organizations maintain financial records through FMS applications, which generally include at least the core functions of general ledger, accounts payable, accounts receivable, and cash and asset management. See id. at 2.1.3; Jt. Sub. Defn ¶ 4.
Oracle and PeopleSoft have historically sold their HRM and FMS products in modules that could be separately licensed and implemented, while SAP, at least until recently, sold an all-encompassing product and did not license separate modules. See PF 2.4.3 (Keating, 6/10/04, 901:6-19, Keating 946:18-20). This architecture gave Oracle and Peoplesoft advantages over SAP in certain industries in terms of modularity and openness, see PF 2.4.3 (Keating, 6/10/04, 901: 6-19), and these advantages continue to make their products more attractive to some customers. See PF 2.4.6 (Keating, 6/10/04, 994:3-6).
"High function software" is our term for products that industry participants call "enterprise" software, "up-market" software, or "Tier One" software." See PF 2.1.1l(e.g., Bergquist, 6/8/04, 274:24-275:7; Wilmington, 6/16/04, 1771:5-1772:1; Elzinga, 6/18/04, 2180:22-2181:5). The HRM and FMS enterprise software products of Oracle, PeopleSoft, and SAP are fundamentally different from mid-market software products even though both perform some of the same core functions (e.g., general ledger). Extensive and unrebutted testimony from customers, industry consultants, and industry experts shows that high function software has a variety of characteristics that distinguish it from other types of solutions and make high function software high function. High function software is:
1. "Scalable," in that it tracks thousands of transactions and supports thousands of concurrent users.3 See PF 2.2.1 (e.g., Burgum, 6/23/04, 3011:19-3013:3; Bergquist, 6/8/04, 289:9-290:7; Johnsen, 6/16/04, 1755:11-24).4
3. Able to perform multiple related transactions seamlessly (e.g., provide a "very tight integration between" business processes without having to write customized code) and with a high degree of ease and sophistication.7 See PF 2.2.3 (Wesson, 6/14/04, 1135:16-1136:13.
4. Capable of handing international aspects of a business, such as multiple currencies, multiple languages, and multiple legal regimes. See PF 184.108.40.206
5. Able to accommodate rapid growth, acquisitions, and reorganizations. It captures data and can use it to model the organization for forecasting and reorganizations. PF 2.2.5 (Iansiti, 6/17/04, 2039:23-25 and 2040:10-18).9
6. Able to reflect actual units of business, rather than a pre-set business organization, and usefully link the data from those units. PF 220.127.116.11 (Iansiti, 6/17/04, 2048:9-25).10
Large enterprise customers demand that vendors: (1) have a track record of market credibility, including references from similarly-situated customers, business capabilities, fiscal stability, and a number of successful implementations, "so they won't disappear after next quarter's results." PF 18.104.22.168-2 ; see also PF 2.3.1 & subparts (Bradley, 6/9/04, 600:9-602:3; Bergquist, 6/8/04, 296:13-297:15; 297:20-298:5; Iansiti, 6/17/04, 2037:1-14; Gorriz, 06/15/04, 1375:2-8); (2) invest sufficiently in research and development to ensure that the software is constantly upgraded and maintained, see Iansiti, 6/17/04, 2037:17-20;11 and (3) have a local presence, with a dedicated sales force that understands the enterprise's needs and can match the software to the needs of the enterprise, see Iansiti, 6/17/04, 2037:22-2038:6.12
Some organizations need the special capabilities and functionality of high function FMS and HRM software, and some do not. Consequently, some organizations use it, and some do not. But that is true for every product, and it in no way suggests that customer characteristics define the relevant markets for high function FMS and HRM software. They are defined entirely by what the software can do.
Section 7 of the Clayton Act bars acquisitions "where in any line of commerce in any section of the country, the effect of such acquisition may be substantially to lessen competition, or to tend to create a monopoly." 15 U.S.C. § 18. The statute "was intended to arrest the anticompetitive effects of market power in their incipiency." FTC v. Procter & Gamble Co., 386 U.S. 568, 577 (1967).
"Section 7 itself creates a relatively expansive definition of antitrust liability: To show that a merger is unlawful, a plaintiff need only prove that its effect 'may be substantially to lessen competition.'" California v. American Stores Co., 495 U.S. 271, 284 (1990)(emphasis added). "Congress used the words 'may be substantially to lessen competition' ... to indicate that its concern was with probabilities, not certainties." Brown Shoe Co. v. United States, 370 U.S. 294, 323 (1962). A plaintiff need not show "even a high probability" that the proposed transaction will substantially lessen competition. FTC v. Elders Grain, Inc., 868 F.2d 901, 906 (7th Cir. 1989) (Posner, J.). "[T]he statute requires a prediction, and doubts are to be resolved against the transaction." Id.
Courts should enjoin a transaction under Section 7 if it has the "potential for creating, enhancing, or facilitating the exercise of market powerthe ability of one or more firms to raise prices above competitive levels for a significant period of time." United States v. Archer-Daniels-Midland Co., 866 F.2d 242, 246 (8th Cir. 1988) ("ADM"). In addition to price increases, the exercise of market power can lead to other anticompetitive effects, such as a decline in product or service quality. See, e.g., United States v. Philadelphia Nat'I Bank, 374 U.S. 321, 368-69(1963).
Courts usually begin a Section 7 inquiry by defining the relevant market in terms of the product and the geographic area. See United States v. Marine Bancorp., 418 U.S. 602, 618-23 (1974); Olin Corp. v. FTC, 986 F.2d 1295, 1297 (9th Cir. 1993). Market boundaries are defined by the "reasonable interchangeability of use" or the "cross-elasticity of demand between the product itself and substitutes for it." Brown Shoe, 370 U.S. at 325, see also United States v. E.I. duPont de Nemours & Co., 351 U.S. 377, 400 (1956). Markets "must be drawn narrowly to exclude any other product to which, within a reasonable variation in price, only a limited number of buyers will turn...." Times-Picayune Publishing Co. v. United States, 345 U.S. 594, 612 n.31 (1953). "The proper question to be asked ... is not where the parties to the merger do business or even where they compete, but where, within the area of competitive overlap, the effect of the merger on competition will be direct and immediate." Philadelphia Nat'l Bank, 314 U.S. at 357.
The "Government is not required to delineate § 7 markets by 'metes and bounds,'" United States v. General Dynamics Corp., 415 U.S. 486, 521 (1974) (Douglas, J. dissenting), and it is "always possible to take pot shots," United States v. Rockford Mem'l Corp., 898 F.2d 1278, 1285 (7th Cir. 1990) (Posner, J.). The "purpose of the inquiries into market definition and market power is to determine whether an arrangement has the potential for genuine adverse effects on competition...." FTC v. Ind. Fed'n of Dentists, 476 U.S. 447, 460 (1986).
"A 'market' is any grouping of sales whose sellers, if unified by a monopolist or a hypothetical cartel, would have market power in dealing with any group of buyers." Rebel Oil Co. v. Atlantic Ritchfield Co., 51 F.3d 1421, 1434(9th Cir. 1995). See also Image Tech. Servs., Inc. v. Eastman Kodak Co., 125 F.3d 1195, 1203-04 (9th Cir. 1997); Olin Corp., 986 F.2d at 1299-1300. The "hypothetical monopolist" test is derived from Supreme Court precedent, see E.I. duPont, 351 U.S. at 391-92 and 400-01, and has been refined and made operational by government guidelines. U.S. Department of Justice and FTC, Horizontal Merger Guidelines § 1, reprinted in 4 Trade. Reg. Rep. (CCH) ¶ 13,104 (1992, rev'd 1997) ("Merger Guidelines").13
Professor Elzinga defined markets after carefully analyzing a broad array of evidence. He reviewed Oracle's Discount Approval forms, which revealed that Oracle discounts most aggressively against PeopleSoft and SAP. PF 3.3.5, 22.214.171.124, 3.8.1, 126.96.36.199, 188.8.131.52. He considered the perceptions of market participants, including independent market research organizations and "Big 5" consulting firms, which clearly distinguished between "Tier 1" vendors and "Tier 2" or "Mid-Market" vendors. PF 3.2.6, 3.2.7. He considered Oracle's win-loss customer survey data, which consistently identified PeopleSoft as Oracle's most frequent competitor in larger accounts. PF 3.3.4. He reviewed extensive documentary evidence. Finally, Professor Elzinga carefully considered the importance of price discrimination, high margins, and the relationship of license and maintenance fees to the total cost of ownership (TCO).
In contrast, Professor Hausman merely raised the specter of various alternatives, such as outsourcing, incumbent systems, and bundling, without evaluating whether they were sufficiently good alternatives to merit placement in the same market as high function software. At trial, Professor Hausman testified that he had "not tried to determine" whether Oracle, PeopleSoft, and SAP could jointly raise price 5-10%. PF 184.108.40.206.1; see also PF 220.127.116.11.2 (acknowledging this to be the hypothetical monopolist test). And he testified that he had not thought in detail about whether all packaged software vendors could satisfy the hypothetical monopolist test with respect to financial or HR software, but thought that they could with respect to financial software. Hausman, 6/29/04, 3976:25-3977:25. Professor Hausman also indicated that he had changed his opinions since his deposition because of new information. Hausman, 6/29/04, 3981:3-17.14 This new information apparently caused Professor Hausman to realize that he had not actually analyzed key issues on which he had previously provided opinions.
Professor Hausman also testified that some customers have only a single bidder in the final round, so it follows that they consider competition unnecessary. Whatever the reasons some customers have for inviting a single bidder, such as an existing price hold, Hausman, 6/29/04, 4008:3-4009:17, Professor Hausman's regressions demonstrate that those customers who have more bidders benefit from additional discounts induced by competition.15 PF 18.104.22.168-8.
"The touchstone of market definition is whether a hypothetical monopolist could raise prices." Coastal Fuels of P.R., Inc. v. Caribbean Petroleum Corp., 79 F.3d 182, 198 (1st Cir. 1996). Thus, only products that prevent a hypothetical monopolist from imposing a small but significant increase in price should be included in the relevant market. See, e.g., United States v. Microsoft Corp., 253 F.3d 34, 51-54 (D.C. Cir. 2001) (affirming exclusion of "middleware" and other products from the relevant market for Intel-compatible PC operating systems as, inter alia, either too costly or not sufficiently similar to constrain defendant's prices).
Products that are not functionally similar to those sold by the merging firms generally should be excluded from the relevant market. See id. at 53-54 (excluding middleware products from the relevant market for operating systems because no such "product could now, or [will] soon,. . . serve as a platform for popular applications [comparable to Microsoft's Windows], much less take over all operating system functions.").
Products do not compete in the same relevant market merely because they offer functionality similar to that offered by the products of the merging firms. See U.S. Anchor Mfg. v. Rule Indus., Inc., 7 F.3d 986, 995-96 (11th Cir. 1993); ADM, 866 F.2d at 246; FTC v. Staples, Inc., 970 F. Supp. 1066, 1074-75, 1078 (D.D.C.1997), United States v. Gillette Co., 828 F. Supp. 78, 82-83 (D.D.C. 1993). Products substantially similar to those sold by the merging firms, but which nevertheless would not prevent a hypothetical monopolist from raising prices, should not be included in the relevant market. See ADM, 866 F.2d at 246 (sugar not in the relevant market for high fructose corn syrup, despite being functionally interchangeable, because sugar price sufficiently high that buyers of high fructose corn syrup would not switch in the face of a small but significant increase in price).
The antitrust laws, Merger Guidelines, and sound economic analysis all compel the same conclusion: High function FMS and HRM software are the relevant product markets. Other products or services could not constrain a 5-10% increase by a hypothetical monopolist provider of either type of high function ERP software.16
High function HRM and FMS software have more robust, cost-effective capabilities than any alternative products or services. While it is possible to complete some tasks performed by high function software by other means, this does not answer the market definition question. See U.S. Anchor Mfg., 7 F.3d at 995-96; ADM, 866 F.2d at 246; Staples, 970 F. Supp. at 1074-75, 1078. The relevant question is whether customers view these other means as viable and cost effective alternatives when faced with an increase in price of high function software. The evidence proved conclusively that they do not: These options provide less functionality, slower speed, higher risk, and greater overall expense. See generally PF Section 3 (e.g., PF 3.2.1, 3.2.4, 22.214.171.124, 3.8.3, 3.8.5). Thus, these other solutions, individually and collectively, are not viable and cost effective alternatives and would not constrain small but significant prices increases in high function FMS and HRM software. PF 126.96.36.199
Verizon needs to process several billion HRM and FMS transactions per month, Bradley, 6/9/04, 599:16-600:8, and high function HRM and FMS software is by far the least expensive option for doing so. Bradley, 6/9/04, 606:22-608:8 ( she would not switch to alternatives even if there were a 10% price increase because the "business case" would not justify it).
Prior to becoming the co-President of the defendant in this case, Charles Philips wrote to his Morgan Stanley clients: "the back office applications market for global companies is dominated by an oligopoly comprised of SAP, PeopleSoft, and Oracle. The market is down to three viable suppliers who will help re-automate the back office business processes for global enterprises for years to come." PF 188.8.131.52 (P3068, Phillips dep., 6/3/04, 157:7-22; P2290 at MS 00914 (emphasis added)). The record demonstrates the accuracy of his analysis: Virtually all customers with complex requirements would not switch to other products or services in response to a 10% price increase.18
These customer views are the product of detailed analyses, and many went through a detailed, interactive sales process involving comprehensive requests for proposals, followed by extensive evaluation processes. Nextel identified 615 business requirements for HR software, and 1,238 business requirements for Accounting and Finance software. P4077, NEXTEL-000289-318; P4076, NEXTEL-000223-269. The State of North Dakota developed a 500-page requirements document that it matched against the functionality offered by Microsoft Great Plains, Oracle, and PeopleSoft. Wolfe, 06/16/04, 1554:02-11. Robert Bullock of CH2M Hill explained: "We wanted to make sure we had all the bases covered with regards to what we were looking at. So, we wanted to make sure they didn't leave any stone unturned that we would later find out to be a hole in their capabilities." Bullock, 6/7/04, 212:24-213:16; see also PF 184.108.40.206. Each of these customers, as well as the other witnesses who testified, all arrived at no more than three optionsOracle, PeopleSoft, and SAP.19 See PF 3.2.1 20
Oracle apparently could not find even one of its roughly 4,000 customers to testify that a small but significant price increase by Oracle, PeopleSoft, and SAP would cause it to turn to other options. To the contrary, Oracle's featured customer witness, Mr. Peters of Emerson, testified that the value of consolidating Emerson's multiple divisions on a standard FMS platfonn is so great that Oracle could have increased its license fee by at least 10% without causing Emerson to look elsewhere.
The hard data from Oracle's discount approval forms (withheld for months by Oracle) also validate the markets for high function FMS and HRM software. These data consistently show that Oracle's sales people identified PeopleSoft and SAP as Oracle's most frequent competitors for large accounts. When seeking a discount, Oracle's front-line sales people cited as the primary justification competition from PeopleSoft 122 times and cited competition from SAP 81 times. Notably, the next closest competitor, Lawson, had only 16 mentions. Other vendors, outsourcers and legacy systems appeared even less frequently. Elzinga, 6/18/04, 2177:3-20; P3175 and P4015B. Oracle's quibbles with these data succeeded only in pointing out that Professor Elzinga's tabulation presented a conservative assessment of the competition among Oracle, PeopleSoft, and SAP because the tabulation included entities such as Manugistics, that do not even produce HRM or FMS software. Elzinga, 6/18/04, 2330:5-2331:6.21
* This tabulation reports the number of times that each alternative is listed as a discount justification on the relevant discount approval forms. Relevant forms pertain to U.S. customers for high function FMS or HRM software or E-Business Suite with net license prices of $500,000 or more where the alternative is reported.
**J.D. Edwards is now owned by PeopleSoft.
Source Elzinga Supplemental Report, 5/3/2004, Exhibit S-2.
For cost-effective solutions to complex FMS and HRM requirements, [RED.], IBM Global Services, BearingPoint, [REDACTED MATERIAL] recommend Oracle, PeopleSoft, and/or SAP as the best alternatives for clients seeking to automate complex human resource or financial management processes.22 BearingPoint's Mr. Keating put it succinctly: SAP, Oracle, and PeopleSoft "are the only ones that provide a product that will be acceptable to a large company in terms of product capabilities, vendor reliability, scalability, cost, handling multiple currencies, multiple calendars, and multiple charts of accounts."23
IBM's Senior Vice President of Software Steve Mills concurred:
[T]he larger-size businesses are looking for sophisticated packages that have a lot of function that can match scalability. If you have thousands or even tens of thousands employees, any old accounting package is not going to do. It has to match the nature of vow business, and the three dominant providers of that technology in the market today are SAP, PeopleSoft and Oracle.
PF 220.127.116.11 (P3193, Mills dep., 5/27/04, 128:5-129:25 (emphasis added)).
Moreover, Forrester Research's 2004 survey of a number of HRM vendors, issued two weeks ago, concluded "that the three large ERP vendorsSAP, PeopleSoft, and Oracledominate the high end of the market with sophisticated HRMS offerings." PF 18.104.22.168 (P3324, at 4 (emphasis added)). The survey further noted that multinational organizations are limited to Oracle, PeopleSoft, and SAP because the "global features and localizations for HRMS tend to be far stronger among the big three ERP vendors." PF 22.214.171.124 (P3324, at 8).
Oracle's high-level officials view the structure of the market the same way. Larry Ellison testified in deposition that Oracle's biggest competitors for both FMS and HRM ERP application software are PeopleSoft and SAP. PF 126.96.36.199 (P3171, Ellison dep., 01/20/04, 164:8-165:7; 183:3-11, 184:11-16). Further, until Oracle launched its bid for PeopleSoft, Oracle's annual reports consistently identified PeopleSoft and SAP as Oracle's primary competitors in the ERP applications software market. PF 3.3.3 (P2048 at 9, P2049 at 9, P2050 at 9).
Oracle's expert economist, Professor Hausman, explained why the requirements of some large complex customers give Oracle's, PeopleSoft's, and SAP's products special status.
they [large complex enterprises are] got not getting the same product. They are getting a different product than the small people . . . the 10,000-seat puts much greater strain on the software. Because you are going to have many more concurrent transactions, soand are you going to have much more complicated software. That's what I mean by not getting the same product.
Hausman, 6/29/04, 4057:20-24, 4058:6-15 (emphasis added).
Mid-market vendors, such as Lawson, would not defeat a small but significant price increase by the three providers of high function FMS and HRM software. The products supplied by these vendors lack the breadth and depth of functionality, scalability, and reliability that the high function products supplied by Oracle, PeopleSoft and SAP possess. PF 188.8.131.52 (Bergquist, 6/8/04, 277:22-278:3). For example, Novell, a company Oracle placed on its witness list but did not call live at trial, testified that J.D. Edwards' FMS software was not viable for Novell. PF 184.108.40.206 (D8107, Anderson dep., 6/3/04, 170:11-170:17). The customer testimony cited in Section III.C.2 above confirms this view.
Oracle argues that Microsoft is already in the high function market, but the overwhelming evidence in the record refutes that notion. Oracle cites an internal Microsoft email suggesting ways to explain Axapta's scope to customers, DF at ¶53, but neglects to cite the deposition testimony of Mr. Pollie, Vice President of U.S. Sales, about the email: in "no way is it [Axapta] robust enough to service the complex centralized locations of a large enterprise account." P3254R, Pollie dep., 5/26/04, 199:5-18.24 Oracle also relies on an Axapta brochure, but fails to cite to testimony describing its context, D7173, Burgum dep., 5/13/04, 45:5-49:24, or noting that some of the statements in its Axapta brochures, "do not represent real-life situations" and "could be determined as misleading." P3254R, Pollie dep., 5/26/04, 105:10-25; 235:9-236:14; D5333.
Oracle relies at length on one MBS customer, Esselte, but ignores the fact that Esselte has yet to implement Axapta, PF 220.127.116.11.1 ,CITE, and that the implementation may well not be successful, given that Microsoft views Esselte as having a maximum of 200 concurrent users per location, consistent with Microsoft's conclusion that Axapta is "stretched" at 350 users, while Esselte plans to put 1000 concurrent users at a smaller number of sites. P3254R, Pollie dep., 5/26/04, 87:12-24; Burgum, 6/23/04, 3016:11-3017:2; D7159, Spund dep., 5/25/04, 176:5-16; D7159, Spund dep., 5/25/04, 167:7-168:4 (implementation has been a "rough start").25
Oracle, PeopleSoft, and SAP recognize the fundamental difference between their flagship products and those offered by the mid-market vendors, and each offers separate scaled-down products for mid-market customers. Oracle produced "Special Edition North America" for mid-market customers. The product has significantly fewer components than Oracle's E-business Suite. PF 18.104.22.168 (P3070 Prestipino Dep., 5/18/04, 36:25-37:23). PeopleSoft purchased J.D. Edwards' EnterpriseOne product, which is sold successfully only to mid-market organizations because it lacks functionality and scalability. PF 22.214.171.124. SAP introduced software products with less complex functionality in order to sell effectively to smaller organizations. PF 126.96.36.199 (Knowles, 6/23/04, 2907:3-2907:7).
The difference in the two product classes shows up in the pricing. PF 188.8.131.52 (P3037, Knowles dep. 5/3/04, 78:9-78:15). Oracle's discount approval forms demonstrate that mid-market software vendors do not impact Oracle's bottom-line offers on sales of high function software. PF 184.108.40.206 (Elzinga, 6/18/04, 2179:5-14; P3175; P4015A; P4015B). PeopleSoft prices its mid-market products differently than it does the products it targets at the up-market. PF 220.127.116.11 (Wilmington, 6/17/04, 1972:5-19; see P4965, at PS-TE2251790, PS-TE2251794).
The structures of Oracle, PeopleSoft, and SAP's sales forces similarly reflect segmentation into high function and mid-market solutions. Oracle dedicates salespeople to "mid-market" organizations. PF 18.104.22.168 (P3033, Henley dep., 5/4/04, 116:12-117:4). PeopleSoft established a separate sales force for customers with under $1 billion in annual revenues. PF 22.214.171.124 (Wilmington, 6/16/04, 1765:23-1766:16). SAP has dedicated salespeople to sell to organizations with less than $1.5 billion in revenues. PF 126.96.36.199 (Knowles, 6/23/04, 2814:13-2814:21). In addition, system integrator BearingPoint has a dedicated mid-market business group focused on consulting opportunities at smaller organizations that are not as geographically dispersed. PF 188.8.131.52 (Keating, 6/10/04, 864:12-17, 863:11-18, 864:22-865:2).26
The record amply demonstrates that customers' cobbling together multiple best-of-breed (or "point") solutions will not defeat a small but significant increase in the price of high function FMS or HRM software. Oracle's Chief Executive Officer, Mr. Ellison, testified that best-of-breed is a dying solution, characterizing it as the "gift that keeps on giving" because organizations incur significant ongoing costs associated with upgrades and maintaining interfaces. PF 184.108.40.206 (P3172, Ellison dep., 01/20/04, 133:17-135:20; see also PF 3.6.2 (Oracle's Answer, Response to Allegation 16; Ellison, 6/30/04, 4317:12-4317:14) (integrated suites of high function HRM and FMS software provide superior functionality and better performance than a package of best-of-breed solutions). Organizations using best-of-breed solutions can find it "staggeringly" difficult to perform critical functions, such as access and store data, because the information is fragmented across databases. PF 220.127.116.11 (P3171, Ellison dep., 01/20/04, 139:9-140:4).27 These strongly held views of Oracle's most important executive demonstrate that best-of-breed solutions cannot constrain prices of high function HRM and FMS software.
The unrebutted customer testimony tracks Mr. Ellison's assessment. Multiple customers testified that they would not adopt a best-of-breed approach even if the license cost of the high function software solution sold by Oracle, PeopleSoft, and SAP was 10% higher. PF 18.104.22.168 (Cichanowicz, 6/14/04, 1077:10-24; Hatfield, 6/7/04, 137:20-138:2). Given the benefits of integration, most customers simply do not perceive a best-of-breed package of different software solutions as a substitute for the high function HRM or FMS software sold by SAP, Oracle, and PeopleSoft. PF 3.6.1 (P3041, Patel dep., 6/03/04, 86:21-87:4, 92:2-16; Gorriz, 06/15/04, 1430:13-15).28 Rather, customers receive the most value with pre-integrated high function FMS and HRM software because it is purchased in one instance, is easier to pre-test, runs on one technology and one hardware, and, most significantly, costs less. PF 22.214.171.124 (Hatfield, 6/7/04, 93:16-94:8; Wesson, 6/14/04, 1150:16-1151:20; P3189-R at DOJ-DC-NOV-000025).29
Outsourcers also would not defeat a small but significant price increase by vendors of high function FMS and HRM software. Outsourcers do not provide independent competition for high function software because, as [REDACTED] explained, the outsourcers in HRM or FMS only use Oracle, PeopleSoft, or SAP software for large, complex enterprises. PF 126.96.36.199 (P3198, [REDACTED MATERIAL]).30 They resell the software, or in some cases, facilitate its direct licensing by the customer from one of the three dominant vendors. In addition, outsourcing is primarily (if not exclusively) an HR phenomenon. PF 3.9.1 (Elzinga, 6/18/04, 2197:12-2198:11). It is irrelevant for customers seeking a solution for their financial management needs.
Moreover, outsourcers rarely focus on very large companies (because these customers want control over their own systems), favoring customers with "non-complex HR functions" and "limited international requirements." PF 39.5 (D5192 at 19, 26). Fidelity, for example, has only ten customers with more than 10,000 employees. PF 188.8.131.52 (Sternklar, 6/23/04, 3169:11-21). When outsourcers use their own software, they are virtually never in competition with Oracle, PeopleSoft, and SAP. PF 3.10.1 (Iansiti, 6/17/04, 2070:9-25). Mr. Knowles testified that he has "never seen ... a loss" to an outsourcer against SAP's HR suite. PF 184.108.40.206 (P3037 Knowles dep. 5/3/04, 272:9-272:19; 272:20-273:11 (emphasis added)).
Professor Elzinga's analysis of the Discount Approval Forms demonstrates that outsourcing vendors are not a significant pricing constraint on Oracle. PF 220.127.116.11 (Elzinga, 6/18/04, 2197:12-2199:11; P3175). The customers explained why, consistently testifying that they would not switch from one of the three available vendors to an outsourcing solution in the event that high function software prices rose 10%-15%. PF 18.104.22.168 (Maxwell, 6/9/04, 687:10-16); PF 22.214.171.124.4 (Cichanowicz, 6/14/04, 1079:4-18); PF 126.96.36.199 (Mearns, 6/24/04, 3302:20-3303:6).
Oracle's last fallback is incumbent solutions, i.e., "doing nothing." Enterprise class HRM and FMS software has been available for about a decade. Prior to its development, companies used a variety of less integrated and less robust software to perform HRM and FMS functions. Oracle's logic, based heavily on the unsupported testimony of Dean Campbell, is that these out-of-date systems substantially constrain pricing for the state-of-the-art high function software solutions offered by Oracle, PeopleSoft, and SAP.
The record evidence is strongly to the contrary. Professor Elzinga's tabulation of Oracle's data demonstrates that the possibility of remaining with an incumbent solution does not constrain prices for high function FMS and HRM software vendors. PF 3.8.1 (Elzinga, 6/18/04, 2393:9-18; 2192:21-2197:11). He concluded, based on his review of the discount approval forms, that incumbent systems do not "disciplin[e] the Oracle pricing the way the two manufacturers of high-function FMS software and HRM software do, and that's SAP and PeopleSoft." PF 188.8.131.52 (Elzinga, 6/18/04, 2179:5-14; 2193:19-2194:22). The systems integrators agree.31
The testifying customers offered the same assessment. Mr. Cichanowicz of Nextel, [RED.], and Oracle 's witness, Mr. Peters of Emerson, each testified that they would not have remained on incumbent systems in response to a 10% price increase from the high function vendors. See, e.g., PF 184.108.40.206 (Cichanowicz, 6/14/04, 1078:9-17); [REDACTED] [Under Seal]; Hatfield, 6/17/04, 138:3-19; Peters (Emerson), 6/14/04, 1269:24-1274:23). In the rare instances that a customer begins the procurement process but decides to "do nothing," it is typically due to a problem in funding, not a lack of intent to procure. See, e.g., PF 220.127.116.11 (Glover, 6/15/04, 1490:6-22) (explaining that Greyhound fully intended to purchase software, but held off only because a new CEO cut the project's budget)). In fact, it is risky and costly for customers to maintain incumbent systems for long periods of time because of the need to comply with changing regulations and business practices. AIMCO's Wesson explained: "maintain[ing] the product ourselves ... there's no support for it outside the walls of our company, and for a strategic application for a company of our size, that's pretty risky." PF 18.104.22.168 (Wesson, 6/14/04, 1142:10-22).32
Side-stepping the facts, the law, and basic economics, Oracle wrongly argues that the existence of distinct high function FMS and HRM product markets is somehow untenable because customers often purchase other software at the same time. Oracle's premise is that competition cannot be meaningfully analyzed in just high function FMS and HRM software. This baseless and counterintuitive logic would prevent defining FMS and HRM software markets, even if Oracle tried to acquire PeopleSoft and SAP. 33
In fact, Oracle frequently sets prices and discounts that are targeted exclusively at competition to sell FMS and HRM software. First, Oracle often discounts stand-alone sales of HRM or FMS, usually based on the presence of a competing bid from PeopleSoft. An Oracle sales representative justified a 73.2% discount on a potential sale of only HRM software to Jacobs Engineering Group because PeopleSoft offered an 80% discount.34 Likewise, a 70% discount request was justified on a sale of just HRM software to PepsiCo because PepsiCo "favored" PeopleSoft's HRM and it "was offering tremendous discounts .... We require the discount to remain competitive on price." The sales representative added that "PEOPLESOFT WILL WIN THIS DEAL IF WE CANNOT OFFER THIS DISCOUNT."35
Oracle's argument also ignores that it gives blended discounts on multi-product sales specifically targeted to prevent PeopleSoft or SAP from selling FMS or HRM software. Competing with PeopleSoft for a sale to Gap Stores, Oracle offered an "additional larger discount on HR Applications to seed them with the modules to encourage the Gap to move to Oracle HR rather than Upgrade to PeopleSoft 8.0."36 The sales representative requested an 88% blended discount comprised of 93% off HRM, but only 80% off the other applications.37 A sales representative recently requested a 50% "bundled" discount covering multiple products, including Oracle's database and eBusiness Suite of applications, when Macromedia wanted to replace their PeopleSoft HR because PeopleSoft (that does not market a database product) offered a 60 to 70% discount.38
Oracle's approach also runs headlong into a half-century of case law on market definition. "The product market includes the pool of goods or services that enjoy reasonable interchangeability of use and cross-elasticity of demand." Oltz v. St. Peter's Cmty. Hosp., 861 F.2d 1440, 1446 (9th Cir. 1988); ( United States v. E.I. du Pont de Nemours & Co., 351 U.S. 377 (1956). The "salad bar" of additional software that Oracle seeks to lump into the competitive mixincluding CRM, middleware, and even databases, see OPCOL ¶ 10are not interchangeable with FMS and HRM software. As the leading treatise explains: "grouping complementary goods into the same market is not only economic nonsense, it also undermines the rationale for the policy against monopolization or collusion in the first place." 2A Phillip E. Areeda et al., Antitrust Law ¶ 565a, at 331 (2d ed. 2002). Bundling its dominant PC operating system with, for example, a computer mouse, would not somehow prevent Microsoft from exercising monopoly power in the PC operating system market. And Oracle's practice of also selling other products and services (for which there is additional competition) will not protect customers of FMS and HRM software in the two-firm market that Oracle seeks to create.
"A 'market' is any grouping of sales whose sellers, if unified by a monopolist or a hypothetical cartel, would have market power in dealing with any group of buyers." Rebel Oil Co. v. Atlantic Richfield Co, 51 F.3d 1421, 1434 (9th Cir. 1995). Thus, geographic markets are defined so that the analysis of "the effect of [the] acquisition," 15 U.S.C. § 18, does not ignore any "producers [that] substantially constrain the price-increasing ability of the monopolist or hypothetical cartel." Id. at 1434; Merger Guidelines §1.0 (hypothetical monopolist test applied to both product and geographic markets). In this case, the geographic scope of the relevant market is the United States because "there was no evidence that [U.S. customers] could effectively turn outside of [the United States] for alternate sources of [the product]."39 Oltz, 861 F.2d at 1447. Further, the relevant customers are business units operating in the United States, regardless of where the corporate headquarters may reside.
While vendors of high function HRM and FMS software do business and compete with each other internationally, a hypothetical monopolist of U.S. sales (or a cartel of Oracle, PeopleSoft, and SAP) could impose a small but significant price increase. Customers in the United States market would not turn to any vendors not now selling in the United States in the event of a price increase, both because there are no such vendors, Iansiti, 6/17/04, 2025:2-13, and because they are purchasing not just a product, but also a relationship with the vendor that requires that the vendor have a significant U.S. presence. PF 4.1.2 (Elzinga, 6/18/04, 2152:21-2156:10, 2203:22-2206:19; Cichanowicz, 6/14/04, 1080:16-25; Gorriz, 6/15/04, 1378:3-9; Knowles, 6/23/04, 2901:20-2902:22).
Licensing high function HRM or FMS software customers entails entering into an on-going relationship with a vendor that involves lengthy on-site evaluation and demonstration processes, initial product support, and critical continuing maintenance and upgrades. As Professor Elzinga explained:
What you buy when you buy this product is, you buy a relationship; that is, you don't just buy code, you buy [a] relationship that begins before you even make the - before you even cut the deal, before you even actually choose the vendor, you meet for many months in the  purchase process of trying to learn which vendors will meet your needs if you're a prospective customer. And then after you make the purchase, once, again, it's very different from just taking on code. That relationship continues as the FMS software or HRM software is installed, implemented, bugs are taken out. It's maintained. It's upgraded. And that's really where the competition takes place in that local relational sense.
Elzinga, 6/18/04, 2152:21-2156:10, at 2155:13-25; 2203:22-2206:19, at 2205:12-13.40 Consequently, U.S. customers can turn only to a vendor with extensive domestic operations that makes them capable of providing the necessary services.
The customer testimony consistently stressed the importance of a vendor having domestic operations. Nextel explained that it requires a "relationship" with an HRM or FMS vendor that has "a presence in the United States physically" so that it can "see and talk to [the vendor] on a consistent basis for such a strategically important function." Cichanowicz, 6/14/04, 1080:16-25. Mr. Hatfield of Cox Communications stated that he could not, either by traveling outside of the United States or otherwise, license software from abroad to counter a domestic price increase. Hatfield, 6/7/04, 134:9-135:1; 141:4-21. When DaimlerChrysler purchased PeopleSoft HRM software, it purchased the software for its German operations in Europe and for its U.S. operations in the United States because of the importance of consultation in the presale phase and technical support for software implementation and operation. Gorriz, 6/15/04, 1378:3-25.41
SAP's business model further confirms the existence of a separate United States market. SAP America has over 4,000 employees in the United States who sell, market, support, and maintain SAP ERP software in the United States. P3036 Knowles dep., 12/3/03, 15:23-16:11. The company has entirely separate marketing, finance, and sales organizations in the United States, which are headquartered in Philadelphia and operate in multiple U.S.-based regional offices. Knowles, Tr., 6/23/04, 2902:23-2903:4; P3036, Knowles dep., 12/3/03, 15:23-16:11. This expensive U.S. infrastructure is a requirement of business reality.42 Competing for U.S. customers of high function HRM and FMS software requires providing a host of services within the geographic boundaries of the United States. If SAP could sell U.S. customers high function HRM and FMS software simply by e-mailing its software from Germany, it would. SAP does not because it cannot.
Also confirming the existence of a separate United States market is the fact that the three vendors have different levels of competitive product strengths in the U.S. compared to other regions of the world. Knowles, Tr. 6/23/04, 2940:10-2941:24. SAP has had relatively little success penetrating the U.S. banking industry, even though it is successful in this sector in Europe. Id. SAP's Knowles testified as to one reason: SAP's "software is not designed to compute average daily balances for whatever reason, because that's, you know, in Europe that's not the way they run their banking operations. We do need to remember that we did grow out of a European centric need development organization." P3036 Knowles dep., 12/3/03, 126:8-127:4, at 126:25-127:4.
Further, because arbitrage is not possible, a hypothetical monopolist of United States sales (or a cartel of Oracle, PeopleSoft, and SAP) could raise prices in just the United States. Unlike the situation for many commodities, a price increase in just the United States would not be defeated through the importation of the same product sold at a lower price abroad. License restrictions preclude resale. P2060, ORCL-EDOC-00046835 to ORCL-EDOC-00046836 & ORCL-EDOC-00046839. Moreover, the software is configured for specific customer needs and national requirements; it simply will not work if installed on a different customer's systems (or vice versa). Elzinga 6/18/04, 2156:11-2158:21, at 2158:15-21. Confirming this is the fact that prices for high function HRM and FMS software in the United States are not impacted by prices abroad and vice versa. Elzinga, 6/18/04, 2156:11-2157:22.43
The evidence thus demonstrates that the relevant product markets are high function HRM and FMS software and the relevant geographic market is the United States.
Section 7 of the Clayton Act sets forth the standard for determining whether a merger should be prohibited: The Court should enjoin any transaction where "the effect of such acquisition may be substantially to lessen competition, or to tend to create a monopoly." 15 U.S.C. § 18. Courts and antitrust enforcement agencies use various analytical tools to guide their assessment of the potential anticompetitive effects of a transaction. For example, they examine shares and concentration within a defined relevant market. Courts and agencies also examine the competitive process. A transaction harms competition if it creates a monopolist or enables the combined firm individually to raise pricesso-called "unilateral effects"by, for example, merging closest geographic competitors or the closest substitute products in a differentiated product market. Similarly, harm to competition may arise if a transaction could make it easier for the remaining firms to collude, either expressly or tacitlyso-called "coordinated effects" in setting prices or allocating customers. All of these tools point to the conclusion that Oracle's proposed acquisition of PeopleSoft would substantially lessen competition. Ultimately, these tools are only guides in the application of the statutory mandate. Setting such labels aside, the evidence is overwhelming that customers benefit from head-to-head competition between Oracle and PeopleSoft for the sale of high function HRM and FMS software and that this transaction would deprive these customers of this benefit. Under these circumstances, the Court should find that the acquisition is likely "to lessen competition" within the meaning of Section 7.
A transaction is presumed illegal under Section 7 of the Clayton Act if the combined entity would have a significant market share in a sufficiently concentrated market. Philadelphia Nat'l Bank, 374 U.S. at 363; California v. Am. Stores Co., 872 F.2d 837, 842 (9th Cir. 1989) rev'd on other grounds, 495 U.S. 271 (1990), relevant part reinstated, 930 F.2d 776, 777 (9th Cir. 1991). Plaintiffs establish a prima facie case of a § 7 violation by demonstrating "that the merger would produce 'a firm controlling an undue percentage share of the relevant market, and [would] result [ ] in a significant increase in the concentration of firms in that market.'" FTC v. H.J. Heinz Co., 246 F.3d at 708, 715 (D.C. Cir. 2001) (quoting Philadelphia Nat'l Bank, 374 U.S. at 363) (modifications in original).
Evidence that a merger will result in the combined firm having a large market share, particularly when fewer than three significant rivals would remain in the market post-merger, is sufficient to demonstrate a likelihood of anticompetitive effects. R.C. Bigelow, Inc. v. Unilever N. V., 867 F.2d 102, 110(2d Cir. 1989). Indeed, according to the D.C. Circuit "no court has ever approved a merger to duopoly under similar circumstances." H.J. Heinz, 246 F.3d at 717.44
Courts often apply the Merger Guidelines' approach for assessing pre- and post-merger concentration with the Herfindahl-Hirschman Index ("HHI"). See, e.g., H.J. Heinz, 246 F.3d at 716; AlliedSignal, Inc. v. B.F. Goodrich Co., 183 F.3d 568, 574 (7th Cir. 1999); American Stores,872 F.2d at 842. The HHI is calculated by summing the squares of the market shares of all firms in the market. Merger Guidelines ¶ 1.5. "Sufficiently large HHI figures establish [a] ... prima facie case that a merger is anti-competitive." H.J. Heinz, 246 F.3d at 716.
Oracle's acquisition of PeopleSoft would significantly increase concentration in the already highly concentrated high function FMS and HRM software markets. In high function FMS, the pre-merger HHI is 2,813, and it would rise by 1,020 points to 3,833 post merger; the combined Oracle-PeopleSoft share would be 47.4%. In high function HRM, the pre-merger HHI is 3,835, and it would rise by 1,872 points to 5,707 post merger; the combined Oracle-PeopleSoft share would be 69.7%. These figures are significantly above the Merger Guidelines' thresholds (§ 1.51) for presuming that the transaction would substantially reduce competition.
Even if the relevant geographic market were worldwide, the pre-merger HHI would exceed 3,000 (a mathematical certainty with only three sellers in the market), and the post-merger HHI would increase by a substantial amount. Accordingly, the acquisition presumptively harms competition regardless of whether the geographic market is the United States or worldwide.
The government's primary theory of competitive harm involves "unilateral effects": By eliminating head-to-head competition between Oracle and PeopleSoft, the proposed acquisition would cause Oracle to offer lower discounts to U.S. customers on its high function HRM and FMS software. Oracle would find it in its interest to raise prices, without any coordination with its remaining rival. Oracle also would have less incentive to improve and support its products.
"Unilateral effects" describes a class of theories, and different unilateral effects theories apply depending on the nature of the competitive process. What all have in common is that the merged firm finds it in its unilateral self-interest to raise price, reduce output, or otherwise act anticompetitively. The leading antitrust treatise endorses four distinct unilateral effects theories:
(a) creating a monopoly or dominant firms; (b) perpetuating a monopoly or dominant firm by eliminating a nascent rival; (c)) giving one firm stronger control of its "niche" in a product-differentiated market; or (d) strengthening a firm's power to make noncompetitive bids that buyers will be unable to refuse.
4 Philip E. Areeda, et al., Antitrust Law ¶ 910, at 54 (rev'd ed. 1998). Points (c)) and (d) are ways of articulating Plaintiffs' unilateral theories in this case. The Merger Guidelines (§ 2.2) explain: "Unilateral competitive effects can arise in a variety of different settings. In each setting, particular other factors describing the relevant market affect the likelihood of unilateral competitive effects. The settings differ by the primary characteristics that distinguish firms and shape the nature of their competition." The Merger Guidelines (§ 2.21) distinguish two broad categories of firms, and the one relevant in this case is that in which "products are differentiated" and "individual sellers compete more directly with those rivals selling closer substitutes." In a footnote, the Merger Guidelines (§ 2.21 n.21) mention the subcategory of markets in which firms "are primarily distinguished by their relative advantages in serving different buyers or groups of buyers, and buyers negotiate individually with sellers."
Merger challenges by the federal enforcement agencies based on unilateral effects theories have been common and generally accepted for more than a decade. See FTC v. Swedish Match, 131 F. Supp. 2d 151 (D.D.C. 2000); FTC v. Staples, Inc., 970 F. Supp. 1066 (D.D.C. 1997); see also William J. Baer, Deborah L. Feinstein & Randal M. Shaheen, Taking Stock: Recent Trends in U.S. Merger Enforcement, Antitrust, Spring 2004, at 15, 18.45
For horizontal mergers, the Supreme Court adopted a presumption of illegality based on market shares "with respect to mergers whose size makes them inherently suspect." Philadelphia Nat'l Bank, 374 U.S. at 363. The Court's presumption applies regardless of the particular competitive effects theories used as an analytical tool for the case. This case presents with market shares high enough to be inherently suspect, and the presumption of illegality applies.
Courts have employed the Philadelphia Nat'l Bank presumption, based on market shares and market concentration in cases premised on unilateral effects theories.46 See FTC v. Swedish Match, 131 F. Supp. 2d 151, 166-67 (D.D.C. 2000) ("Because of the market share and concentration levels, the Court finds that the [FTC] established a presumption . . . [that the acquisition] is likely to substantially lessen competition . . . ."); FTC v. Staples, Inc., 970 F. Supp. 1066, 1083 (D.D.C. 1997) ("By showing that the proposed transaction . . . will lead to undue concentration, . . . the [FTC] establishes a presumption that the transaction will substantially lessen competition.").47
With differentiated products, the Merger Guidelines (§§ 2.211-2.212) emphasize a factual inquiry into the closeness of the products of the merging firms, in particular how often each is the second choice when the other is the first choice, and the ability of rivals to replace lost competition. The Merger Guidelines also discuss circumstances in which market shares are likely to be indicative of not only the frequency with which the merging products are first choices, but also the critical frequency with which they are second choices. In that context, the Guidelines create a presumption of competitive harm when the combined market share exceeds 35% for the merging firms, but the Guidelines do not indicate that a lesser combined share precludes a merger challenge or a finding of significant anticompetitive effects. Oracle's expert, Professor Hausman, agreed that significant unilateral effects were certainly possible with a lesser combined share. Hausman, 6/29/04, 4042:17-21, see also Jerry A. Hausman & Gregory K. Leonard, Economic Analysis of Differentiated Products Mergers Using Real World Data, 5 Geo. Mason L. Rev. 321, 338 (1997) (arguing against a 35% safe harbor).
Judicial experience in analyzing mergers under unilateral effects theories generally has tracked the Guidelines' approach. See Swedish Match, 131 F. Supp. 2d at 169-70 (focusing on the closeness of the merging firms' products and the ability of rivals to replace the lost competition); Long Island Jewish, 983 F. Supp. at 142-144 (finding that a price increase by the merging hospitals would be defeated by substitution to nearby non-merging hospitals); Staples, 970 F. Supp. at 1086-87 (focusing on the ability of rivals to replace lost competition); Kraft Gen. Foods, 926 F. Supp. at 352-53 (focusing on whether the merging products were "the first and second choices of a significant number of consumers"); Gillette, 828 F. Supp. at 84-85 (focusing in part on the ability of rivals to replace lost competition).
Swedish Match and Staples, two cases enjoining mergers on the basis of unilateral effects theories, involved differentiated consumer products.48 Like the products in those cases, high function HRM and FMS software are highly differentiated. The products of Oracle, PeopleSoft, and SAP have different heritages based on the historic strengths of their respective products and their software architectures. PF 22.214.171.124.1; Keating, 6/10/04, 897:23-899:3; Hausman, 6/29/04, 4018:2-16. [REDACTED MATERIAL] explained that there are recognizable differences among Oracle, PeopleSoft, and SAP applications, with the functionality of some products fitting better in some organizations than for others. P3198, [REDACTED MATERIAL] ; P3200, [REDACTED MATERIAL] (UNDER SEAL).
Oracle and PeopleSoft also are each others' closest competitor. See generally PF 7.1.1, 7.1.2. An Oracle report, from the quarter preceding the announcement of Oracle's tender offer, explicitly states: "PeopleSoft is our #1 Competitor: and "SAP is our #2 competitor: in application sales. PF 126.96.36.199 (P2093 at ORCL-EDOC-00042674; Block dep., 12/16/03, 233:17-234:18). As noted above, Oracle's sales people seeking approval for discounts cite competition from PeopleSoft far more often than competition from other vendors, and Oracle documents cite PeopleSoft as its most significant competitor. P2093 at ORCL-EDOC-00042674; Block Dep., 12/16/03, 233:17-234:18. As noted above, Oracle's sales people seeking approval for discounts cite competition from PeopleSoft far more than competition from other vendors, and Oracle documents cite PeopleSoft as its most significant competitior. P2093 at ORCL-EDOC-00042674; Block Dep., 12/16/03, 233:17-234:18. One important reason that Oracle and PeopleSoft are the first and second choices for many customers is that many currently use both PeopleSoft and Oracle software, and these "dual footprint" customers provide excellent sales opportunities for the two companies because of the benefits of consolidating ERP applications on a single vendor's platform. Keating, 6/10/04, 930:20-931:11, 997:19-999:6. Bearing Point identified more than 1,200 companies that run Oracle Financials and PeopleSoft HR. Keating, 6/10/04, 931:12-932:3.
SAP's Knowles also explained that SAP faces serious obstacles with respect to U.S. customers. Many perceive SAP to have a higher total cost of ownership than either Oracle or PeopleSoft. PF 188.8.131.52.8 (Knowles, 6/23/04, 2950:24-2951:2); see Wilmington, 6/16/094, 1816:9-1817:7. Knowles added that implementing SAP's software can be a lengthy and costly process for an organization because the software is higher priced and more complex than other solutions. PF 184.108.40.206.9 (P3036, Knowles dep., 12/3/03, 152:11-152:19). Knowles also acknowledged that SAP has struggled to penetrate organizations in the services sector. PF 220.127.116.11.2 (P3037, Knowles dep., 5/3/04, 67:21-68:7). Along the same lines, [REDACTED MATERIAL] testified that SAP currently has "zero" penetration in the top twenty banks in the United States, and that most of them use Oracle for FMS. PF 18.104.22.168.9 [REDACTED]. Surveys confirm SAP's substantial disadvantage. P0017, at ORCL-EDOC-00140867, ORCL-EDOC-00140869. And Oracle and PeopleSoft both position themselves as quicker and cheaper to implement than SAP. Knowles, 6/23/04, 2950:8-2950:16; P3036 Knowles dep., 12/3/03, 153:19-154:18.
Customer testimony provided further confirmation that Oracle and PeopleSoft often both present better alternatives in the United States than SAP. North Dakota eliminated SAP during the first round of its procurement process because SAP's price was too high and it did not provide the necessary functionality. Wolfe, 06/16/04, 1546:13-24. Neiman Marcus eliminated SAP from consideration as a vendor of high function software because SAP was not strong in retail and had a high cost of implementation. PF 22.214.171.124, Maxwell, 6/9/04, 690:21-691:14. Greyhound determined that Oracle and PeopleSoft were its two best options for human resources management software. Glover, 6/15/04, 1495:22-23. Cox's first two choices were Oracle and PeopleSoft. PF 126.96.36.199.1. AIMCO excluded SAP from consideration because configuration of SAP's software is more complex from that of Oracle and PeopleSoft and thus more expensive. Wesson, 6/14/04, 1182:15-1183:05. Similarly, SAP's implementation at the engineering and construction firm Fluor was very difficult and costly. Fluor continues to have trouble with SAP's software and recently suspended making maintenance payments. PF 188.8.131.52.7 (P3150).
The Big Five system integrators also believe SAP is often the third choice of many U.S. customers. According to Bearing Point's Keating, SAP has long been the least flexible of the three vendors in the way it has sold its HRM and FMS products. PF 184.108.40.206.11 (Keating, 6/10/04, 901:6-20, 946:18-20; P3200, [REDACTED MATERIAL](UNDER SEAL)). Keating also testified that even though SAP has tried to be more flexible and modular, it has not closed the gap with Oracle and PeopleSoft in terms of modularity and openness. PF 220.127.116.11.16 Keating, 6/10/04, 993:16-994:2. [REDACTED] testified that SAP is less likely to discount than Oracle and PeopleSoft. PF 18.104.22.168.12, P3198, [REDACTED MATERIAL].
SAP cannot reposition its products to replace the lost competition. See PF 22.214.171.124.1-5. This is especially true for those many customers who have already established Oracle and PeopleSoft within their HRM and FMS footprints. Elzinga, 6/18/04, 2372:5-21. SAP has spent enormous sums to improve its product, and still continues to be perceived as the most rigid, least flexible of the three products. Elzinga, 6/18/04, 2370:15-2373:4 at 2372:22-2373:4. Only after years of focus on solutions for manufacturers did SAP manage to extend its functionality to create broader FMS and HRM offerings. Keating, 6/10/04, 898:11-899:3, 901:6-20, 946:18-20; P3200, [REDACTED MATERIAL] (UNDER SEAL). SAP has unsuccessfully attempted for years to sell its software to the North American banking industry. Knowles, 6/23/04, 2877:24-2878:9; Keating, 6/10/04, 993:3-10. Even SAP's partnership with Accenture has not led to any new sales to financial services organizations, Knowles, 6/23/04, 2947:10-2947:18, and Accenture has questioned SAP's ability to sell into such organizations. P3152. Indeed, if SAP could easily reposition it already would have done so.
The Staples court held: "Since prices are significantly lower in markets where Staples and Office Depot compete, eliminating this competition with one another would free the parties to charge higher prices in those markets, especially those in which the combined entity would be the sole office superstore." 970 F.Supp. at 1082. Just as in Staples, the loss of the head-to-head competition between the merging firms in this case can be expected to result in higher prices and substantial consumer harm. Elzinga, 6/18/04, 2208:11-18, 2236:9-17.
This case differs from Staples in the precise mechanism leading to price increases. Unlike consumer products with uniform pricing, competition in this case involves a bidding process that is separate for each customer. P4014B; McAfee, 6/21/04, 2598:23-2599:8. Copious evidence documents the fact that discounts vary considerably across customers, depending on the particular circumstances of each customer and the competition to supply each customer. See PF 7.2.2. Systematic analysis of Oracle's data on E-Business Suite sales and from its discount approval forms and sales representative survey reports indicates that Oracle discounts significantly more than otherwise when in competition with PeopleSoft. See PF 7.3.3.
Because the price competition to sell high function HRM and FMS software is specific to the particular customer, the effects of the merger differ across customers according to the significance of the head-to-head competition between Oracle and PeopleSoft. See, e.g., Jonathan B. Baker, Unilateral Competitive Effects Theories in Merger Analysis, Antitrust, Spring 1997, at 21.49
Professor McAfee's conclusion that customers of high function HRM and FMS software would face higher prices as a result of the proposed merger is based on three independent analyses. McAfee, 6/21/04, 2448:24-2449:2 First, he examined in detail twenty-five specific instances of competition in which PeopleSoft caused Oracle to offer significantly higher discounts. McAfee, 6/21/04, 2449:3-7. Second, using regression analyses, Professor McAfee established that when PeopleSoft is a competitor, Oracle offers higher discounts. McAfee, 6/21/04, 2449:13-19;2532:23-2533:3. Third, Professor McAfee simulated the proposed merger using a standard model from auction theory, which predicted substantial price increases following the proposed acquisition. McAfee, 6/21/04, 2533:4-9. Each of the three different approaches that Professor McAfee used to examine head-to-head competition between Oracle and PeopleSoft led to the same conclusion: Customers of high function software would face higher prices as a result of the proposed merger. McAfee, 6/21/04, 2533:15-2534:2.
Professor Hausman acknowledged the applicability of unilateral effects theory, specifically that a merged firm can significantly raise its prices if the competitive constraint imposed by the other firm has been removed. PF 126.96.36.199. And his regressions demonstrated that competition caused both PeopleSoft and Oracle to increase discounts. Professor Hausman testified that a high degree of accuracy is required in setting different prices to different customers. Hausman, 6/28/04, 3874:5-20. On cross, he opined that negotiation offers many ways to reduce the risk of denying a discount; yet if ultimately Oracle insists on not discounting an additional 5 percent, Oracle would have to be right 94 percent of the time. Hausman, 6/29/04, 4026:14-22. Today, however, Oracle does not offer all its customers its highest discount. Instead, it charges different prices to different customers depending on such factors as who it is competing against and the greater functionality of software required by larger customers. Indeed, Professor Hausman acknowledged that Oracle charges its largest customers more for their software through the use of per seat licensing. The largest customers pay more because they need a higher function product; indeed, in Professor Hausman's opinion, the product they need is a different product than what smaller customers need. Hausman, 6/29/04 4057:3-4063:11 (questioning by the Court). Thus, Oracle today has sufficient knowledge about its customers' situations to identify customers to charge more, and it does so regularly.50
Large customer benefits have been produced by the head-to-head battle between Oracle and PeopleSoft to make sales of high function HRM and FMS software. Oracle's documents show that the companies that received large discounts last year as a result of the Oracle-PeopleSoft competition include: Neiman Marcus, Ameritrade, Qualcomm, TAC Worldwide, Hannover Compression, Hudson Highland Group, and the State of Ohio. See PF 188.8.131.52. At trial, the customer witnesses repeatedly testified to the effectiveness of Oracle-PeopleSoft competition in producing lower prices. PF 7.2.2.
Oracle's "2003 Win/Loss Analysis" for its North American Sales Organization indicates it registered 25 losses in Ql FY03, 16 of which were to PeopleSoft, the most of any company. Oracle lost 36.8% of the time it was up against PeopleSoft. PF 184.108.40.206.3 (P2090, ORCL-EDOC-00038647). The same year, PeopleSoft was identified as a competitor against Oracle 51 times in deals in which Oracle discounted by over 50 percent, whereas SAP was mentioned 30 times. See P2095, ORCL-EDOC-00055322-23; Block Dep., 12/16/03, 248:5-250:6.
Still another "win/loss" analysis for that year again contained similar information, again indicating that PeopleSoft is Oracle's strongest competitor. In a 2003 Oracle e-mail captioned "Win/Loss Survey & OSO," Oracle highlighted its win/loss statistics against PeopleSoft, SAP, J.D. Edwards, and Siebel for special attention. Oracle notes that it won $5.7 million and lost $16 million against PeopleSoft in Q3 FY2003. By comparison, it won $1.8 million and lost $8.8 million against SAP, and won $3.3 million and lost $5.6 million against J.D. Edwards. The report notes that Oracle wins about 50% of the time against PeopleSoft, but PeopleSoft wins 3:1 based on revenue. PF 220.127.116.11.10, P2132, ORCL-EDOC-00054305-06.
In Oracle's discount approval forms, PeopleSoft is the competitor cited most frequently as the basis for a discount request. SAP, while a significant competitor, appears much less frequently. Elzinga 6/18/04, 2176:9-2179:17, at 2177:3-11; P3175; P4015B. PeopleSoft's Approval Matrix Request Forms leads to the same conclusion. In a number of opportunities in which the sales representative indicated that Oracle was the competitor (or, in one instance, Oracle/SAP), the company authorized discounts up to 85% off the already discounted price.51
The Oracle/PeopleSoft rivalry is responsible for substantial innovations in high function FMS and HRM software. As Mr. Ellison explained at trial, "if there's no competition, innovation would be wasted effort." Ellison, 6/30/04, 4313:25-4314:11. Oracle and PeopleSoft have engaged in feature-by-feature "leap-frogging" innovation competition. Oracle's head of product development, Mr. Wohl, testified at length how PeopleSoft for years was seen as setting the standard for HRM software, but that Oracle had matched and then surpassed some aspects of PeopleSoft's functionality through vigorous research and development. Mr. Wohl added that, for both their HRM and FMS products, "there are a number of features we've added because PeopleSoft introduced a feature, some of PeopleSoft's customers and prospective customers found it attractive, and we raced to add the feature. Wohl, 6/25/04, 3378:10-19. PeopleSoft approaches innovation competition with Oracle with the same intensity. For example, according to Wohl, PeopleSoft made substantial progress closing certain gaps in the functionality of its FMS products versus those of Oracle. Wohl, 6/25/04, 3391:18-3392:1.
In this industry, having three firms vigorously engaging in the innovation process is more likely to result in a stream of innovation than having two firms. PF 18.104.22.168 (Elzinga, 6/18/04, 2380:4-10). The competitive alternative eliminated by the merger is particularly attractive for enterprises with a "split footprint" of PeopleSoft HRM and Oracle FMS software. For such enterprises, the alternative of consolidating onto either Oracle or PeopleSoft has advantages over switching one of the installed applications to SAP. Consolidation offers better integration across the applications and it avoids the cost and risk of retraining personnel. PF 22.214.171.124. l(Wohl, 6/25/04, 3412:1-17; P3200, [REDACTED MATERIAL] (UNDER SEAL); Elzinga, 6/18/04, 2371:18-2372:4). There are at least 1,200 of these "split footprint" customers. PF 126.96.36.199.2 (Wohl, 6/25/04, 3411:22-24; Keating, 6/10/04, 931:12-932:7). And Professor Elzinga has explained that it is very unlikely SAP could manage to "reposition" itself to become an appealing option for these consumers. PF 188.8.131.52.3( Elzinga, 6/18/04, 2372:5-21).
While effects of the proposed merger on price competition are unilateral, the merger may also have coordinated effects arising from implicit customer or industry allocations that manifest themselves in the ways Oracle and SAP market or develop their products. Coordinated effects are distinguished from unilateral effects because they arise not from the merged firm's pursuit of its unilateral self-interest, but rather from the likelihood of coordinated interaction, explicit or tacit, among the firms that remain post merger. "[I]f conditions are ripe, sellers may not have to communicate or otherwise collude overtly in order to coordinate their price and output decisions; at least they may not have to collude in a readily detectable manner." Elders Grain, Inc., 868 F.2d at 905.
As the Court noted in its July 10 Order, various market factors may be relevant in assessing the likelihood of coordinated interaction as well as the types of coordination that could work profitably. The Guidelines (§2.1, emphasis added) mention
the availability of key information concerning market conditions, transactions and individual competitors; the extent of firm and product heterogeneity; pricing or marketing practices typically employed by firms in the market; the characteristics of buyers and sellers; and the characteristics of typical transactions."52
The fact that high function software is a differentiated product and that pricing is not transparent make price coordination between Oracle and SAP unlikely. "But alternative forms of collusion may prove satisfactory. For example, a product, territorial, or customer division agreement may divide up the market in such a fashion that each firm can set its own profit-maximizing price in its own assigned market niche." 12 Herbert Hovenkamp, Antitrust Law ¶ 2002f2, at 23 (1999).
The proposed merger of Oracle and PeopleSoft would enhance the likelihood of tacit coordinated interaction between Oracle and SAP through a form of market or customer allocation. Recognizing that "terms of coordination may be imperfect and incomplete inasmuch as they omit some market participants, omit some dimensions of competition, [or] omit some customers," Guidelines §2.11, Oracle and SAP could reach an understanding based on "mutual trust and forbearance," Hospital Corp. of Am. v FTC, 807 F.2d 1381, 1391 (7th Cir. 1986) that each will "stop . . . calling on each other's accounts."53 Or they could coordinate to avoid competing in particular verticals. As Oracle and SAP make decisions about what verticals and product specializations to invest in, coordinated interaction is likely to affect their product design choices.54 Post merger, each would have a stronger incentive to avoid significant development efforts in the other's strong verticals. In a two-firm market, it is reasonable to expect that a strong attack on one's firm's dominant areas would be met by a similar response.55
Two factors in these markets indicate that the acquisition would significantly increase the likelihood of coordination of this sort. First, Oracle and SAP have different areas of relative strength and expertise across industries and customers. PF 184.108.40.206.17-.21. As Mr. Ellison testified "that's the way the application business works, it's divided by industry." P3171, Ellison Dep. 01/20/04 143:1-3. For example, in banking, Oracle is strong in the U.S., while SAP is strong in Europe, P3036, Knowles Dep.,12/3/03, 126:8 - 127:14; SAP dominates the oil and gas industry, while Oracle is strong in high tech manufacturing and telecommunications, PF2.4.4; P3171, Ellison Dep. 01/20/04. 174:12-18. With three competitors there are more overlaps in areas of strength (especially between Oracle and PeopleSoft), P3171, Ellison Dep., 01/20/04, 175:9 -176:2; P4482 at SAP62528-SAP62529 ("Several [Oracle and PeopleSoft] product lines are heavily overlapped and target customers are almost identical") than there would be with two, so it is less likely now that the firms will coordinate to avoid attacking each other's areas of relative strength or to allocate the areas that are in doubt. Second, in a two-firm market, detection of cheating is virtually assured. And as the Merger Guidelines (§ 2.12) indicate, "[c]redible punishment. . . may not need to be any more complex than temporary abandonment of the terms of coordination."56
Plaintiffs cannot and do not say, of course, that particular anticompetitive coordinated interaction will occur; rather the law asks whether the merger "will make it easier" for such coordinated interaction to occur, Elders Grain, 868 F.2d at 905, and thus has the "potential for creating, enhancing, or facilitating the exercise of market power," ADM, at 246. In this case, the acquisition would enhance the risk of coordinated anticompetitive effects in the markets for high function softwarea risk based on the very product differentiation that is also a foundation of the risk of unilateral anticompetitive effects. Thus, coordinated effects analysis provides relevant and appropriate guidance to the Court in assessing the harmful effects of this acquisition.
Several customer witnesses currently using PeopleSoft HRM or FMS software explainec that the loss of competition would likely cause them considerable harm, despite the fact that they were not planning to purchase completely new systems in the near term. They anticipate receiving regular upgrades, including not only bug fixes and regulatory updates, but also product enhancements. PF 220.127.116.11.57 But as Mr. Ellison testified, customers' maintenance contracts do not guarantee that innovation will be delivered; it is "guaranteed by market dynamics." PF 18.104.22.168. Competition guarantees enhancements in two ways. Vendors must keep their existing customers satisfiedby providing upgrades and high quality servicein order to compete effectively in the near term for sales from new customers. PF 22.214.171.124-2, 126.96.36.199-10, 188.8.131.52.1, 184.108.40.206. Moreover, vendors compete for each others' installed base of customers, PF 220.127.116.11.4-5, 18.104.22.168.1-2, and this competition places at risk future revenue from a vendor's installed base. Without the spur of Oracle/PeopleSoft competition, customers reasonably fear that these innovation benefits of competition will wither or disappear. PF 22.214.171.124-9.
Consistent with this fear, Oracle's plan not to upgrade PeopleSoft products would harm existing PeopleSoft customers by reducing the quality of their stream of enhancements and services. In doing so, Oracle faces little risk of losing to SAP the many existing PeopleSoft customers with an Oracle-PeopleSoft footprint, for whom migration to Oracle would be the next-best alternative. PF 126.96.36.199.1 -.2; PF 3200 (UNDER SEAL); Elzinga, 6/18/04, 2371:18 - 2372:4.
Oracle suggested that harm to existing PeopleSoft customers is irrelevant under Section 7 because it might happen even if a new owner other than Oracle took over PeopleSoft. But as the CIO of Pepsi Americas explained, it is just "common sense" that a new owner other than Oracle would not purchase the company to let the product wither away. PF 188.8.131.52. Customers estimated the possible cost to them of making a switch to avoid this quality deterioration as a result of lost competition; some anticipated costs as high as $30, $50, or $100 million. PF 184.108.40.206; 220.127.116.11-16; 18.104.22.168. One customer had even already incurred the cost of additional software, purchased now to avoid the anticipated post-merger price increase. PF 22.214.171.124.2.1, 126.96.36.199.12.
Oracle's proposed acquisition of PeopleSoft will significantly increase concentration in the already highly concentrated relevant markets, and is therefore presumptively anticompetitive. The record on the competitive process confirms this presumption. Customers enjoy significant benefits as a result of intense head-to-head competition between Oracle and PeopleSoft, such as larger discounts, lower prices, and better products from innovations. By eliminating this intense competition with its closest rival, the transaction will enable Oracle unilaterally to scale back the discounts it currently must offer to win business from PeopleSoft, which will result in higher prices for customers of high-function FMS and HRM software involving half a billion dollars in commerce. It also will reduce the need for Oracle to continue to improve its products in order to keep and gain customers. Finally, the transaction will make it easier for the two remaining market participants - Oracle and SAP - to coordinate their activities, such as by implicitly ceding certain complementary industry strengths, to the detriment of customers. Thus, regardless of the label applied to the analysis, the transaction will substantially lessen competition within the meaning of Section 7.
Oracle has failed to meet its burden of demonstrating that new entry into the relevant markets will occur, much less that it will relieve the transaction's anticompetitive effects. See Am. Stores Co., 872 F.2d at 842-43: United States v. Baker Hughes, Inc., 908 F. 2d. 981, 987 (D.C. Cir. 1990). Entry must be timely, likely, and of sufficient magnitude to prevent a significant exercise of market power. See Rebel Oil Co., 51 F.3d at 1440-41 (9th Cir. 1995) (citing Merger Guidelines' "timely, likely, and sufficient" test). Oracle must not only show entry could happenalthough the technical possibility alone is a huge challenge herebut also that it likely would happen, meaning that it is both technically possible and economically sensible in response to a post-merger price increase. See FTC v. Cardinal Health, Inc., 12 F. Supp. 2d 34, 56 (D.D.C. 1998) (quoting Merger Guidelines § 3.3).
Any firm attempting to enter these markets faces enormous challenges. Potential entrants face tough technical obstacles that are likely to take years to overcome.58 See United States v. United Tote, Inc., 768 F. Supp. 1064, 1075-77 (D. Del. 1991) (technology a barrier when likely software entrants needed 18 to 24 months to re-tool their software product, customers demanded 100% reliability, and past attempted entrants demonstrated repeated technological failures); 5 Areeda & Hovenkamp, Antitrust Law ¶ 1128c4, at 103 (2d ed. 2003) ("A firm is not likely to enter a market for which it lacks the capital, technical resources, marketing channels, or skills."). High function software is enormously complex, with order-of-magnitude differences from mid-market products in scope and scalability, sophistication, depth, and breadth of functionality, and extensive built-in configuration options, among other characteristics. See generally PF 2.2 & subparts. Moreover, a potential entrant would need to gain a deep understanding of the business processes of potential customers before even beginning the process of creating an application to automate those business processes.59 See PF 188.8.131.52, 10.1.9.3; Bergquist, 6/8/04, 333:8-15. Because the sale of high function applications software involves a lengthy and detailed sales process, see PF 2.7 et seq., an entrant also would have to build a substantial, dedicated direct sales force, PF 184.108.40.206-220.127.116.11, committed to understanding the requirements of large and complex organizations in different industries, and able to work to match the software as much as possible to the needs of the customer. Iansiti, 6/17/04, 2037:l-2038:6.60
In addition, customers of high-function applications look not only at functionality but also at credibility, reputation, and ability to sustain a long-term relationship. P3171, Ellison dep., 01/20/04, 45:2-46:8; PF 18.104.22.168. A new entrant would have to demonstrate not only its fiscal stability, which is critical, (Iansiti, 6/17/04, 2037:1-2038:6), but also its capability for sustained, future R&D, Iansiti, 6/17/04, 2037:1-2038:6. And it would have to establish a track record of successful implementations. Bergquist, 6/8/04, 296:7-297:15; 297:20-298:10.61 Even after spending the years needed to develop its initial high-function product, it would need to spend one to three years more attracting and developing the "reference customers" necessary to compete effectively. Bergquist, 6/8/04, 321:23-322:1, 322:10-19, 322:21-325:6.62
The difficulties and low likelihood of successful entry are vividly illustrated by the failed attempt by J.D. Edwards in the 1990s to sell to the up-market. The company, which was the fourth largest enterprise application provider, Allen, 6/10/04, 746:18-747:14, embarked on an ambitious project to develop a new product architecture that would allow its applications to run on multiple hardware platforms, with multiple databases and multiple operating systems, Allen, 6/10/04, 771:11-21. It hired additional programmers, 1000 additional sales and marketing personnel, and spent approximately $1 billion over six to seven years. Allen, 6/10/04, 770:22-771:10; 774:5-776:2. J.D. Edwards abandoned the effort in 2001, after determining that it "didn't have the products, services, and ultimately reputation necessary to satisfy the requirements of upmarket customers." Allen, 6/10/04, 777:1-13. Before abandoning the effort, Edwards calculated that to bring a competitive product to the up-market would have required an additional three to five years of work and a "boatload" of money. Allen, 6/10/04, 786:4-787:2 (emphasis added).
The time required for entry into the high function FMS and HRM markets far exceeds judicial and the Merger Guidelines timeliness thresholds, which generally are two years. See United States v. Syufy Enters., 903 F.2d 659, 666 n.l1 (9th Cir. 1990) (citing Merger Guidelines § 3.3 (2 year time frame)); FTC v. PPG Indus., Inc., 628 F. Supp. 881, 885 (D.D.C. 1986) (finding high entry barriers when acquiring technical expertise delayed entry for two- to six-years), aff'd in part, 798 F.2d 1500 (D.C. Cir. 1986); Olin Corp., 113 F.T.C. 400 (1990) (two year standard in the Merger Guidelines), petition for reh. denied, 986 F.2d 1295 (9th Cir. 1993); United Tote, 768 F. Supp. at 1075-77 (two-year test for entering large-scale gambling software market when starting from middle-scale market).63
The evidence at trial, including testimony from four Microsoft witnesses and numerous company documents, overwhelmingly demonstrates that Microsoft does not intend to enter the high-function HRM or FMS markets. See PF 8.3.2, 8.3.3, 8.3.6. The head of Microsoft's business applications division, Doug Burgum, testified that to enter the large enterprise market would be a "formidable task that would take more money than I would be willing to recommend that Microsoft spend." PF 22.214.171.124 (Burgum, 6/23/04, 3024:3-22). Even if Microsoft were willing to make such a financial investment, it would take many years and be a "very long . . . and costly road" to enter. P3255R, Ayala dep., 5/18/04, 145:14-146:20; 147:4-150:10. Microsoft has never considered building a product or changing its product to enter the large enterprise space, due to the gap in functionality, gap in a direct sales force, and gap in consulting services. PF 126.96.36.199 (Burgum, 6/23/04, 3023:17-3024:2; 3001:17-3002:l).64
Microsoft has no direct sales force that could sell high function applications, and has no plans to develop one. In fact, Microsoft's sales model, company-wide, is to sell indirectlyabout 98% of Microsoft's sales are through partners. PF 188.8.131.52 (P3255R, Ayala dep., 5/18/04, 157:7-16; 240:4-17). Developing the extensive and highly skilled direct sales forces of incumbents would be a costly and time consuming undertaking. PF 8.1.2 (JDE hired over 1,000 sales and marketing personnel and it still was not enough). Nor does Microsoft have any of the implementation and consulting capability needed to sell to and serve customers of high function applications, and it has no intention of building such capability. Burgum, 6/23/04, 2995:3-18; 2992:19-2993:21.
The difficulty that Microsoft, like any potential entrant, would face in entering the high function markets is demonstrated by Microsoft's Project Green, an ongoing effort to develop a next-generation, mid-market set of applications. The Project Green applications have already been under development at Microsoft for roughly 2 years, yet version 1 will not be ready for market until 2008, at the earliest. Even then, the product will have less functionality than Microsoft's current mid-market products. PF 184.108.40.206.1 (Burgum, 6/23/04, 3056:9- 3059:17); P2533R at MS-OPCID 19698 (p. 6 of 54). Not until version 2 does Microsoft expect Project Green to achieve functionality comparable to Microsoft's current products, but that version is not expected to be released until 2010 or 2011 at the earliest. PF 220.127.116.11.2 (Burgum, 6/23/04, 3061:7-3062:6). And even at that point the new Microsoft applications will not be able to serve the needs of complex, high function customers that Oracle, PeopleSoft, and SAP's products serve today. Iansiti, 6/17/04, 2061:11-24; PF 18.104.22.168.1-22.214.171.124.5.65
Lawson's focus on three verticalshealth care, retail, and public sector, would not change should Oracle acquire PeopleSoft. Lawson considers Oracle and PeopleSoft to be "horizontal players" because their focus, unlike Lawson's, is not restricted to a limited number of verticals. PF 126.96.36.199 (Coughlan, 6/28/04, 3661:25-3662:10). In a November 2003 presentation to the Lawson board of directors, CEO Jay Coughlan recommended that Lawson continue its vertical focus rather than become a "horizontal player" like Oracle and PeopleSoft. PF 188.8.131.52-12 (Coughlan, 6/28/04, 3679:9-15; 3679:16-22). Coughlan explained that Lawson did not have the functional capabilities 27 to compete successfully against Oracle and PeopleSoft outside its delineated verticals and beyond the mid-market. PF 5.4.14 (Coughlan, 6/28/04, 3671:5-17); [REDACTED] [Under Seal]. This recommendation was affirmed in a June 24, 2004 earnings call, where Coughlan stated that Lawson's vertical focus would not change, regardless of the outcome of this trial. PF 184.108.40.206 (Coughlan, 6/28/04, 3666:17-25-3667:1).
Lawson's inability to compete against Oracle and PeopleSoft in high function software is borne out by past experiences. Despite a four year effort to position itself in the financial services vertical, this vertical has yet to generate even 5% of Lawson's overall revenues. PF 220.127.116.11 (Coughlan, 6/28/04, 3668:14-20). And, even though public sector is one of Lawson's delineated verticals, Lawson does not have a single statewide FMS implementation in any state government. PF 5.5.45 (Coughlan, 6/28/04, 3727:5-7). Lawson also believes it would be difficult to develop an international payroll product. Coughlan, 6/28/04, 3737:24-3738:10.
Lawson acknowledges that it does not possess the industry knowledge or product functionality to compete against Oracle and PeopleSoft on a horizontal level, PF 18.104.22.168 (Coughlan, 6/28/04, 3673:10-20); [REDACTED] [Under Seal]. According to Mr. Ellison, Lawson lacks the size and research and development dollars to compete across the board in a lot of different industries. PF 22.214.171.124-2 (P3171, Ellison dep., 01/20/04, 236:8-238:13).66
Oracle also has not met its burden of showing that efficiencies will outweigh the substantial competitive harm that this transaction will produce. Only "extraordinary efficiencies" can rebut a showing that a transaction is likely to substantially reduce competition in a highly concentrated market, such as the high function HRM and FMS software markets. Heinz, 246 F.3d at 720. Moreover, efficiencies claims are subject to "rigorous analysis" to ensure that they "represent more than mere speculation and promises about post-merger behavior," Id. at 721. Oracle must show that its alleged efficiencies are: (1) verifiable, i.e., supported by "credible evidence," Staples, 970 F. Supp at 1089; see FTC v. University Health, 938 F.2d 1206,1223 (1lth cir. 1991): (2) not attributable to reduced output or quality, see Rockford Mem'l, 717 F. Supp. at 1290; 4A Areeda et al., Antitrust Law, ¶ 974b3, at 64( personnel cuts resulting from reduction in ouput are not efficiencies); (3) merger specific, see Cardinal Health, 12 F. Supp.2d at 62-63; and (4) greater than the transaction's substantial anticompetitive effects, see id. at 63. Oracle fails these tests.
Verification of cost savings requires the defendant to provide "backup [and], source[s]" and "explain the methods used to calculate" the numbers. Staples, 970 F. Supp. at 1089. In Staples, the court rejected the defendants' alleged huge cost savings in their entirety, in part because they were in "large part unverified, or at least the defendants failed to produce the necessary documentation for verification." Id. at 1089; see also University Health, 938 F.2d. at 1223-24.67
The totality of Oracle's cost savings evidence is a July 2003 spreadsheet (most of which is unrelated to the efficiencies claim) and several pages of testimony from Safra Catz.68 Oracle extracted the $1.18 billion number from the spreadsheet by subtracting highly speculative estimates of projected reductions in PeopleSoft and J.D. Edwards' costs for sales, research and development, and administration in fiscal year 2005 ("FY05") from the companies' costs in those areas in fiscal year 2003 ("FY03") PF9.2.4. The entire FY05 "calculation" (which drives the cost savings amount) is contained in only two columns of the stale spreadsheet.69
Equally important, the numerical "inputs" (e.g., headcount reductions) for the FY05 cost calculation are all "hard-coded"i.e., they are not the product of any modeling but were simply typed into the spreadsheet. As Professor Zmijewski explained, these hard-coded numbers lack any verifiable backup in the form of documents, data, citations, analyses or calculations.70 Ms. Catz confirmed this at trial, testifying that "no document that [she] know[s] of explains how the inputs were developed.71 Indeed, Oracle's investment banker, who worked on the model with Ms. Catz, said repeatedly that he was not aware of the "facts" or "assumptions" used to determine most of the hard-coded inputs for the FY05 cost estimate.72 This barren record falls far short of that required for the Court to credit Oracle's cost savings claims. See Staples, 970 F. Supp. at 1089. The transaction's hostile nature does not excuse the minimalist nature of Oracle's record. As Professor Zmijewski explained, there were multiple types of materials that Oracle could have presented to the Court to support the purported cost savings claims. For example, Oracle certainly has numerous business documents and data that relate to the relationship between output and costs in the ERP software industry. PF9.2.12.
Oracle asks the Court to rely on the unverified personal judgment of Mr. Ellison and Ms. Catz. If an anticompetitive merger could be justified simply on the unverified "say-so" of the acquirer's executives, every anticompetitive merger would be approved. The law is not so naive, and thus there is a verification burden, which Oracle has not met.
In addition, Oracle improperly ignored the $1.385 billion in integration-related expenses it will incur in order to acquire PeopleSoft.73 D7132; P3357, at slide 6; P3004R, at 34 and 41. Oracle's casual dismissal of this huge sum as a mere "one time" cost directly conflicts with settled law and the Merger Guidelines. See Rockford Mem'l , 717 F. Supp. at 1289; Merger Guidelines § 4 ("Cognizable efficiencies are assessed net of costs produced by the merger or incurred in achieving those efficiencies.").
Finally, and fundamentally, even if the cost savings occur, they do not constitute efficiencies because they apparently flow from plans for a dramatic reduction in the output of the combined firm. Cost savings from anticompetitive reductions in output or services do not count as efficiencies. See Rockford Mem'l, 717 F. Supp. at 1290 (rejecting some headcount reductions as efficiencies because "savings in these areas would occur . . . from a drop in production"); Merger Guidelines at § 4 ("cognizable efficiencies ... do not arise from anticompetitive reductions in output or service"); 4A Areeda, et al., Antitrust Law ¶ 974b3, at 64 (rev'd ed. 1998) (" layoffs are not necessarily evidence of efficiencies at all;" reductions in personnel may be because the firm "is reducing its output in order to reflect newly acquired market power"). As Professor Elzingza explained:
[S]omething's economically efficient if you can produce the same rate of output using fewer inputs .... If somebody cuts out a product line and reduces output, [economists] don't think of that as a social efficiency.
Elzinga, 06/18/04, 2161:1-14-2162:1; PF126.96.36.199 and 4
Professor Zmijewski showed clearly at trial that Oracle's model projects substantial declines in output.74 Oracle did not provide the Court with any documentation or analyses to support its claim that its planned cost savings are somehow unrelated to the projected plummets in volume.75 Specifically, Oracle failed to show, as it must, that the projected 20% declines in PeopleSoft and J.D. Edwards customers, and the 61% drop in licenses purchased by these customers, are independent of Oracle's plans to eliminate PeopleSoft's entire sales force and 50% of its research and development staff.76 Absent such a showing, Oracle's cost savings claims amount to no more than the unsurprising observation that if one produces less, one has lower costs.
Innovation claims require verification and merger specificity. See Heinz, 246 F.3d at 723 ("In the absence of reliable and significant evidence that the merger will permit innovation that otherwise could not be accomplished, the district court had no basis to conclude that the FTC's showing was rebutted by an innovation defense."). Oracle's innovation claims do not meet these standards. They are little more than vague promises of a "superset" product without any back-up documents or analyses relating to how much it will cost, when it will be ready for market or why acquiring PeopleSoft is important to its development. Ms. Catz conceded at trial that: (1) she did not know when Oracle would provide the alleged "superset" of products, Catz, 06/25/05, 3533:8-16; and (2) there are no contemporaneous business documents discussing the purported "superset." Catz, 06/25/05, 3534:20-3535:7. Ms. Catz also testified in her May deposition that Oracle has not prepared a cost budget for any particular purported modules or features of the product.77
Oracle's innovation claims also are not merger specific. See Cardinal Health, 12 F. Supp.2d at 62-63. Ms. Catz testified that Oracle could build on its own all of the components of the superset, regardless of whether Oracle acquires PeopleSoft. Catz, 06/25/05, 3536:19-22, and Mr. Ellison and Mr. Kutnick both testified that Oracle was not trying to purchase PeopleSoft for its technology.78 As for the alleged benefits of scale from the transaction, Mr. Ellison testified that Oracle can achieve the scale it wants without acquiring PeopleSoft, either by acquiring other companies or competing aggressively for new customers.79 The complete lack of verification and evidence of merger specificity is fatal to Oracle's innovation claims. See Heinz, 246 F.3d at 723 (requiring "reliable and significant evidence" when efficiencies based on "innovation defense").
The threatened loss or damage from the proposed acquisition establishes the Plaintiff States' standing under two different capacities: (1) as parens patriae on behalf of their citizens and each state's general welfare and economy, and (2) as representatives of government agencies that are past purchasers or likely future purchasers of high function HRM and FMS software. See Hawaii v. Standard Oil Co., 405 U.S. 251, 259-61 (1972). Plaintiff States have standing as parens patriae because the substantial lessening of competition proven at trial also establishes the threatened loss or damage to the general welfare and economy of each plaintiff state in which the relevant competition takes place. See Burch v. Goodyear, 554 F.2d 633, 634-35 (4th Cir. 1977). Oracle admitted that it licenses, sells, and markets its products throughout the United States and in each of the Plaintiff States. See Oracle Corporation's Answer at Allegation 4; P3056, Response to Requests for Admissions. Oracle also admitted that its sales "represent a regular, continuous and substantial flow of interstate commerce, and have had a substantial effect upon interstate commerce" with and in each of the Plaintiff States. Id. In addition, Plaintiff States have proprietary standing as representatives of current or likely future public sector purchasers of high function HRM and FMS software. See, e.g., P1944, State of Ohio Project Update Fall 2003; D8103, Johnson Dep., 5/28/04, 93:14-94:1; P1938; P1939; P1940; P1942; P1943. For these reasons, Plaintiff States have standing to seek injunctive relief to prevent injury to their general,welfare and economies as well as their proprietary interests.
CERTIFICATE OF SERVICE
I hereby certify that on July 13, 2004,1 caused a true and correct copy of the foregoing PLAINTIFFS' POST-TRIAL BRIEF REDACTED PUBLIC VERSION to be served on the individuals listed below.
Jay L. Himes, Esq.
3See PF 188.8.131.52 (P3036, Knowles dep., 12/3/02, 103:20-104:2) (SAP software can handle 47,000 concurrent users); Burgum, 6/23/04, 3016:11-3017:2 (Microsoft's Axapta, is "stretched" at 350 concurrent users).
4 High function software can process several billion transactions per month, see PF 184.108.40.206 (Bradley, 6/9/04, 599:16-600:8), and can generate reports from hundreds of distinct entities, see PF 220.127.116.11 (Iansiti, 6/17/04, 2039:2-2040:7). In contrast, Lawson's mid-market products permit only five levels of organization in FMS and only three in HRM and have no extensive data sharing across entities. PF 18.104.22.168 (Iansiti, 6/17/04, 2046:19-2047:5).
5 Up-market customers look for a high degree of configurability in the software, with options pre-built, Bergquist, 6/8/04, 285:23-287:17, to avoid the substantial disadvantages of customization. See P2896, at ORLITF0091414, P3038, Godwin dep., 01/20/04, 131:22-132:4.
6 Configurability built into software allows for a user to select among thousands of options to fit its specific business practices. See PF 22.214.171.124 (Iansiti, 6/17/04, 2121:16-2123:4); see also Bergquist, 6/8/04, 285:18-22; PF 126.96.36.199 (Wolfe, 06/16/04, 1565:21-1568:12; 1600:02-14).
7 Examples include tracking employees by skills, job classes, job codes, union affiliations, and any number of other criteria. See Glover, 6/15/04, 1484:8-1485:23; see also PF 188.8.131.52 (P3050, Kender dep., 5/11/04, 79:11-81:17); PF 184.108.40.206 (P2208, at ORLIT-EDOC-00244083) (benefits and cost savings of integrated suites).
8 Burgum, 6/23/04, 3011:25-3012:4; PF 220.127.116.11 (P3038, Godwin dep., 1/20/04, 182:9-12) (multiple language support); PF 18.104.22.168 (Iansiti, 6/17/04, 2057:17-2058:3) (able to deal with currency fluctuations on a real-time basis). Mid-market vendors cannot not provide this. See Bergquist, 6/8/04, 291:7-24.
10See Iansiti, 6/17/04, 2039:6-11 and 20-23 (consolidated data and the ability to "drill down"); PF 22.214.171.124 (Iansiti, 6/17/04, 2042:2-12) (allows users in different locations, different jurisdictions, and perhaps different countries speaking different languages to have access to the same information at the same time); PF 126.96.36.199 (Iansiti, 6/17/04, 2049:16-2050:5).
11 PF 188.8.131.52 (P3171, Ellison dep., 1/20/04, 48:8-49:8) (customers do not just want better functionality, they want it from a vendor that will continue to improve the product over a 5-10 year period). Oracle has 6,000 R&D personnel worldwide working on enhancements to the 1li E business suite applications. See PF 184.108.40.206 (P3038, Godwin dep., 1/20/04, 88:6-92:8).
13 The Merger Guidelines (§§1.0-1.2) define relevant markets by asking whether a hypothetical monopolist over a candidate group of products and geographic area would maximize its profits by imposing at least small but significant price increase. This test is referred to as the "hypothetical monopolist test" or the "SSNIP test." If in response to a Small but Significant and Nontransitory Increase in Price (a "SSNIP"), enough buyers would substitute away from the candidate market, the hypothetical monopolist would not impose the SSNIP and so the candidate market must be expanded until the hypothetical monopolist would impose at least a SSNIP.
14 At his deposition, Professor Hausman offered a different opinion that Oracle, PeopleSoft, and SAP could not jointly raise price, and that all packaged software firms jointly would not satisfy the hypothetical monopolist test. PF 220.127.116.11.3, 18.104.22.168.2. As the First Cirsuit held in a case involving Professor Hausman, such inconsistencies between an expert's opinions are grounds for exclusion. Hausman 6/29/04, 4039:8-4041:22; see Coastal Fuels of P.R., Inc., v. Caribbean Petroleum Corp., 175 F.3d 18, (1st Cir. 1999) ("the more likely basis for the exclusion [of Professor Hausman's testimony] was the district court's belief that there was considerable and unjustified variance between the expert's Rule 26 report and his testimony").
15 At his deposition, Professor Hausman also claimed that his empirical work showed that adding one or two vendors made no significant difference in the discount. PF 22.214.171.124. But he had only his "expert judgment" for concluding that an effect as large as a 16% price increase was not significant. Hausman, 6/29/04, 3987:8-3994:6; 3998: 9-11 ("Q. Now, in terms of economic significance in this case, would you consider a 16 percent price increase to be significant? A. It could be, yes.") When questioned on specific regressions, he conceded some effects were significant, contrary to his claim. PF 126.96.36.199. Moreover, his regressions show substantial effects on Oracle and PeopleSoft discounts from having one or two competitors generally, PF 188.8.131.52, and on Oracle from having SAP or PeopleSoft as a competitor, PF 184.108.40.206.
16 Oracle makes much of the fact that plaintiffs' experts used different criteria in analyzing data. But real-world data are never perfect, and each expert used professional judgment in focusing as best as was possible on sales to new customers in the relevant markets. Oracle's discussion of this question reflects its willful misunderstanding of a basic point: The market is defined by the characteristics of the product, not the identity of the customer.
17 Oracle created confusion by limiting the relevant market to basic "core" functionality that appears in both mid-market and high function products. The ability simply to record a general ledger entry, however, does not make a product high function. The ability to record and analyze general ledger entries in dozens of languages across dozens of jurisdictions and across hundreds of legal entities through a single integrated product is a "high function" capability.
18See PF 220.127.116.11.2 (Gorriz, 6/15/04, 1381:11-19 (Daimler would not consider vendors beyond Oracle, PeopleSoft, and SAP in response to a 10% increase in HR software prices)); PF 18.104.22.168.4 (Cichanowicz, 6/14/04, 1077:10-24, 1078:9-17, 1079:4-18 (Nextel would not consider a best-of-breed solution, outsourcing, or remaining on its legacy system in response to 10% license fee increase)); PF 22.214.171.124.3 (Wolfe, 6/16/04, 1569:15-1570:16 (North Dakota would not consider Microsoft, Lawson, outsourcing, SCT, developing its own software, or remaining on its current system in response to 10% price increase)); PF 126.96.36.199.5 (Bradley, 6/9/04, 607:8-15 (Verizon would not develop an in-house software solution in response to a small but significant price increase from Oracle, PeopleSoft, and SAP)); PF 188.8.131.52.5 (Maxwell 6/9/04, 670:9-18, 686:25-687:9 (Neiman Marcussame)); PF 184.108.40.206.6 (P3020, [REDACTED TEXT] [Under Seal]; PF 220.127.116.11.1 (Hatfield, 6/7/04, 136:10-18, 137:1-139:8 (Cox would not have considered mid-market software, writing its own software, a best-of-breed approach, outsourcing, or upgrading its existing system if Oracle and PeopleSoft increased the license and maintenance price of their FMS software 10%)).
20 Bradley, 6/9/04, 598:4-599:11; Gorriz, 6/15/04, 1428:14-18; Maxwell, 6/9/04, 664:25-665:3, 685:21-686:1; Cichanowicz, 6/14/04, 1066:11-17; Hatfield, 6/7/04, 96:7-12, 114:22-115:10; P3020, [REDACTED MATERIAL] [Under Seal]; P3041, Patel dep., 6/3/04, 81:24-82:4.
21 Q Doesn't that make you worry a little bit about the methodology in coming up with this form? A No, if anything, we erred on the side of listing everything that was there, even if we didn't actually think it was a competitor.") (Emphasis added.)
23 PF 18.104.22.168 (P203 at 11; Keating, 6/10/04, 912:15-916:7 (emphasis added)); see PF 22.214.171.124 (Keating (Bearing Point), 6/10/04, 867:15 - 868:6; P3183 [REDACTED MATERIAL][Under Seal]; see Thomas (IBM), 6/9/04 474:16-475:5).
24 Mr. Pollie testified that Microsoft solutions are not appropriate for large enterprises as they lack the required "experience, infrastructure, partner training, system integrator support, years in providing successful solutions to very large corporations, and functionality." P3254R, Pollie dep., 5/26/04, 241:1-243:2. Further, the description of Microsoft's target markets that was actually distributed to the sales team, unlike the rejected wording Oracle cites, reaffirmed that Microsoft does not target large enterprises. P0838; P3254R, Pollie dep., 5/26/04, 236:15-240:12; P3255R, Ayala dep., 5/18/04, 236:24-238:5.
25 Similarly, Oracle refers to [RED.], while failing to mention that it is a holding company with limited requirements, and only 10 or 15 Great Plains users. P3254R, Pollie dep., 5/26/04, 69:7-14. Another example of where Oracle cites to the document, but not the testimony explaining its meaning can be found by comparing the documents cited in DF ¶ 55 (D5379R, D5387R) with the testimony at D7174, Pollie dep., 5/26/04, 126:14-127:5, 132:18-133:2, 134:24-135:11, 136:17-137:7, 139:23-153:6.
26 Oracle wrongly claims that United States v. Engelhard Corp., 126 F.3d 1302 (11th Cir. 1997), requires any alternative ever considered by customers, no matter how soundly rejected, to be included in the relevant market. Englehard actually held that because the government failed to offer any evidence on an important form of substitution, "determining whether a hypothetical monopolist could profitably raise [its] prices [would be] pure guesswork." 126 F.3d at 1307. No form of substitution identified by Oracle has been ignored in this case. All forms of substitution in the aggregate are insufficient to prevent a small but significant price increase in high function HRM and FMS software.
28 Additional specific trial testimony underscored the relative shortcomings of best-of-breed solutions. Fleet Bank replaced its best-of-breed system with an integrated Oracle-based outsourcing system because the fragmented and complex architecture was costly to maintain and created user confusion. PF 126.96.36.199 (Mearns, 6/24/04, 3305:23-3308:4, 3313:24-3314:8.). AIMCO's representative testified that the company did not seriously consider a "best-of-breed" solution to automate its complex HR processes because the total cost is relatively high. PF 188.8.131.52 (Wesson, 6/14/04,1150:16-1151:20). And Mr. Gorriz of Daimler concluded that a best-of-breed route simply was not worth the risk: "[I]f best of breed would mean that we have to buy a couple of software pieces and glue them together and hope that it runs, I would say, no, that's not a true option for us." PF 184.108.40.206 (Gorriz, 06/15/04, 1429:22-1430:2); see also PF 3.6.2.l(Cichanowicz, 6/14/04, 1110:15-1111:02 (point solutions not a viable alternative).
29 Organizations achieve significant costs savings by implementing integrated HRM or FMS software suites, up to five times as much over the life of the product, as compared to pursuing a best-of-breed approach. PF 220.127.116.11 (Ellison, 6/30/04, 4320:2-4320:17); see also PF 18.104.22.168 P3172, Ellison dep., 01/20/04, 137:15-139:8; Hatfield, 6/7/04, 135:3-10; D8107, Anderson dep, 6/3/04, 93:1-94:5; 102:20-103:9. By contrast, a best-of-breed strategy requires a customer to balance different software vendors, different release schedules and different interfacing strategies. PF 22.214.171.124 P3061, DeSimone dep., 5/19/04,150:12-151:5.
30 Most of the HR outsourcing vendors (Hewitt, Exult, Mellon, IBM, Fidelity, and Accenture) have some type of licensing arrangement with PeopleSoft, Oracle, and/or SAP. PF 126.96.36.199 (Sternklar, 6/23/04, 3158:8-3159:20, 3162:4-16).
31 [REDATED] testified that customers never use the threat of doing nothing to get a better price. PF 188.8.131.52 (P3198, [REDACTED MATERIAL] [Under Seal]). Mr. Keating of Bearing Point similarly testified that less than 10% his clients decide to do nothing at the end of a sales cycle. PF 184.108.40.206 (Keating, 6/10/04, 930:2-19). He continued:
there's reporting requirements they have to meet, there's economies that they want to realize in their business, there's management reporting that they want to be able to use to run their business. They're not about toonce they've gotten the capital approval to go buy that, they very rarely stop in the middle of that and go, no, I'm not going to do it.PF 220.127.116.11.1 (Keating, 6/10/04, 930:2-19).
32 For these reasons, Oracle's reliance on United States v. SunGard Data Sys., Inc., 172 F. Supp. 2d 172 (D.D.C. 2001) is misplaced. Applying the hypothetical monopolist test, the SunGard court found, inter alia, that both merging firms had lost more customers to internal solutions than to any competitor, the court found internal solutions were in the relevant market. Id. at 187-89. In contrast, the record here makes clear that vendors do not lose significant sales to home-grown systems.
33 Oracle also makes a general argument that the Plaintiffs failed to properly consider "dynamic competition" in the form of "stack" or "integration layer" developments. Def. Proposed Conclusions of Law at ¶ 103. But the efforts of Oracle's experts to conjure up the specter of "tectonic shifts" or "paradigm shifts" were wholly unpersuasive - both internally inconsistent and at odds with how Oracle, in its real world business, perceives the competitive environment. See generally PF 10 & subparts. In any event, Oracle ultimately offered no coherent evidence or explanation of how the direct and immediate loss of competition in high function HRM and FMS applications software from the acquisition would be eliminated by these speculative "shifts." Oracle's stack arguments are nothing more than an elaborate distraction from the real issues in this case, a distraction that does not offer any legally or economically supportable basis for casting aside the usual, sound antitrust analysis required by the case law and the Merger Guidelines.
35 P1098, at ORLITE0085428 (emphasis in original); PF at 18.104.22.168.8. See also, e.g., PF 22.214.171.124.; P1176, at ORCL-EDOC-00403221 (describing bidding with PeopleSoft over HRM as reason to discount); PF 126.96.36.199.7; P4964, PS-TE2251423 Wilmington, 6/16/04, 1796:2-1798:2 (Oracle and PeopleSoft "[b]oth discounted the fees," on HRM and "those discounts, . . . were communicated back to us through the Target project team, and that's what drove our decisions to continue to discount our product until we were successful licensing Target."); P1015, at ORLITE0418242-0418244 (discounting HRM justified by extending customer deep discount because customer dissatisfied with past Oracle HR and "this will keep PeopleSoft out."); PF188.8.131.52, P4307, PS-TS2-4166-4167; Wilmington, 6/16/04, 1801:8-l802:23 (describing competition between Oracle and PeopleSoft forcing PeopleSoft to grant 65% discount on HRM to win account); PF 184.108.40.206. (listing similar PeopleSoft documents).
37See 220.127.116.11.18; P1126, at ORLITE0125525. Indeed, there are numerous examples where Oracle sales representatives sought specific discounts on HRM or FMS different from the other items in the bundle. See, e.g., PF 18.104.22.168.1; P1188, at ORCL-EDOC-00320923 (requesting 80% discount on HRM, but 70% on other modules during Oracle's PeopleSoft Replacement campaign); P1153, ORCL-EDOC-00029030 (offering 90% discount on HRM products but 80.23% on all others because customer prefers PeopleSoft HRM and its offer was $100K less.
38 P1015, ORLITE0418242-0418244 (blended discount offered on suite justified in part by competition with PeopleSoft in HRM and FMS). See also, e.g., P1189, ORCL-EDOC-00351542-351544 (offering 64% discount on bundle of HRM and technology tools justified as attempt to replace PeopleSoft as customer's HRM)
39 Because "Congress' foremost concern in passing the antitrust laws was the protection of Americans," Pfizer Inc. v. Gov't of India, 434 U.S. 308, 314 (1978), the proper focus is on the alternative sources of supply for U.S. customers. A market defined as the United States, i.e., U.S. consumers, does not imply that foreign consumers would not be harmed by the merger, but rather recognizes that harm to foreign consumers is not a basis for a prohibition of a merger under U.S. antitrust law.
40 Professor Elzinga also explained that the Elzinga-Hogarty geographic market test is not applicable here because it was designed for product markets that are literally shipped from factory to customer, that have substantial transportation costs relative to the value of the product, and for which there are no legal impediments to shipment across geographic boundaries. Elzinga, 6/18/04, 2154:1-21 & 2157:2-22. The sales process, demonstration, implementation, and upgrading elements make high function HRM and FMS software very different from the products envisioned by that test. Also the Elzinga-Hogarty test does not work in the presence of geographic price discrimination, as exists in this case. See United States v. Rockford Mem'l Corp., 717 F. Supp. 1251, 1267 n.12 (N.D. Ill. 1989).
41 Oracle is incorrect in its contention that the Global 2000 corporations headquartered outside the United States are excluded from the relevant market. Any such corporation with operations in the United States that purchases here, such as Daimler-Chrysler, is included.
43 Relying on a last-minute calculation, Dr. Hausman argued for a world market on the basis that average discounts in the United States were the same as elsewhere. But this argument does not prove that prices are linked, and it ignores the rampant price discrimination across and within those areas.
45See also Herbert Hovenkamp, Post-Chicago Antitrust: a Review and Critique, 2001 Colum. Bus. L. Rev. 257, 333 ("Unilateral effects theories have proven to be among the most useable and robust contributions of the post-Chicago revolution in antitrust economics." "Unilateral effects methodologies for analyzing mergers must be regarded as, if anything, more reliable than the methodologies used for evaluating mergers under the traditional concerns about increased concentration.")
46 The Supreme Court accepted a unilateral effects theory in a conglomerate merger case. In FTC v. Consolidated Foods Corp., 380 U.S. 592 (1965), it rejected a challenge to an FTC administrative decision premised on the theory that the merged firm could unilaterally require suppliers to make unwanted purchases. The Court further held that commanding "a substantial share of a market" supported the finding of anticompetitive harm. Id. at 600.
47 In other unilateral effects cases, courts held the presumption was not established because market shares were not sufficiently high. United States v. Long Island Jewish Med. Ctr., 983 F. Supp. 121, 145 (E.D.N.Y. 1997); New York v. Kraft Gen. Foods, Inc., 926 F. Supp. 321, 363 (S.D.N.Y. 1995); United States v. Gillette Co., 828 F. Supp. 78, 83-84 (D.D.C. 1993).
48 The context in which the unilateral effects of mergers have principally been analyzed, in the economic literature and in actual merger cases, is that of differentiated consumer products which are not sold through a bidding process and for which each product had a single price available to all consumers within any given geographic market. Economists analyze competition in that setting using the Bertrand oligopoly model, which posits that competitors compete strictly on the basis of price. The Bertrand model, and common sense, predict larger post-merger price increases the greater the extent to which buyers of either merging product consider the other merging product to be their second-best alternative. This analysis is presented by Carl Shapiro, Mergers with Differentiated Products, 10 Antitrust, Spring 1996 at 23, and it is discussed by many others, including many cited in the Court's July 10 order.
49See also Serdar Dalkir et al., Mergers in Symmetric and Asymmetric Noncooperative Auction Markets: The Effects on Prices and Efficiency, 18 Int'l J. Indus. Org. 383 (2000); Luke Froeb & Steven Tschantz, Mergers Among Bidders with Correlated Values, in Measuring Market Power 31 (Daniel J. Slottje ed., 2002); Steven Tschantz et al., Mergers in Sealed versus Oral Auctions, 7 Int'l J. Econ. Bus. 201 (2000); Keith Waehrer & Martin K. Perry, The Effects of Mergers in Open Auction Markets, 34 RAND J. Econ. 287(2003).
51See P4092 (PS-TS2000040-41); P4094 (PS-TS2000076-77); P4089 (PS-TS2000236); P4090 (PS-TS2000381-82); P4091 (PS-TS2000385-86); P4098 (PS-TS2000803-04); P4099 (PS-TS2000805-06); P4300 (PS-TS2001707-08); P4301 (PS-TS2001904-05); P4302 (PS-TS2002325-27); P4303 (PS-TS2002397-99); P4307 (PS-TS2004165-67); P4308 (PS-TS2004318-21); P4311 (PS-TS2004924-26); P4313(PS-TS2005076-80).
52 Oracle and SAP would have substantial opportunity, due to the "characteristics of typical transactions," to learn "key information" about each prospective transaction. See generally PF 2.7. Elders Grain and Brooke Group list factors that make a market susceptible to coordination. Some of these factors are present here: there would be just two firms postmerger (compared to five in Elders Grain), there are "no close substitutes for [the product]"; and "entry into the industry is slow." Id. But, just as in Elders Grain (where a preliminary injunction was upheld), "the factors that make a market more or less amenable to being cartelized are not all on one side in this case." Id.
53United States v. Cooperative Theatres, Inc., 845 F.2d 1367, 1368 (6th Cir. 1988) (booking agent services). Even explicit customer or market allocations are not novel or uncommon. United States v. Addyston Pipe & Steel Co., 85 Fed. 271 (6th Cir. 1898), involved allocations, as did Palmer v. BRG of Georgia, Inc., 498 U.S. 46, 49 ("Each agreed not to compete in the other's territories."); United States v. Fischbach and Moore, Inc., 750 F.2d 1183, 1188 (3d Cir. 1984) ("conspiring to allocate electrical construction projects"); United States v. Koppers Co., Inc., 652 F.2d 290, 292 (2d Cir. 1981) ("the two firms divide[d] the state between them"); United States v. Brown, 936 F.2d 1042, 1044 (9th Cir. 1991) (billboard locations) ("explicit written agreement was terminated . . . , [but] was honored by both companies for fifteen more years.").
54 Even in the current three-firm environment, the firms shy away from costly investments in areas where another firm dominates. P3171, Ellison Dep. 01/20/04, 172:3 - 174:11 ("you're not getting any return out of your investment or a very, very small return"); P3036, Knowles Dep.,12/3/03, 126:8 -128:19 ("it requires more investment on our part to really be a contender in banking. ... but we can get our growth in other areas, so I think we'll focus our attention on other areas.").
55 Of course, any such effect that led to maintaining or increasing the differentiation between Oracle and SAP products would exacerbate the unilateral effects discussed above; knowing that SAP does not have a strong presence in a particular industry, Oracle would have less incentive to discount its pricing or to press for innovative new functionality.
56 In addition, the two firms have a profitable related business arrangement because SAP is the largest reseller of Oracle databases. This arrangement gives the two an additional reason to continue to maintain good relations, as well as a possible vehicle for threatened punishment for deviation from terms of coordination. PF 22.214.171.124. Cf. U.S. v. UPM-Kymmene Oyj, 2003-2 Trade Cas. ¶ 74,101, 2003 WL 21781902 at *8 (N.D.Ill. 2003) (upstream supply relationship between acquiring firm and major competitor meant that "[v]igorous competition is unlikely.")
58 It took Oracle three or four years to develop HRM functionality that was "starting to be reasonably competitive" with PeopleSoft's product, P3033, Henley dep., 5/4/04, 223:3-228:3; 228:11-229:23, even though Oracle already had long experience and a proven track record in developing and selling high function FMS products. It took PeopleSoft three years to develop the initial version of its FMS product, and five to seven years to have a "sufficient," truly competitive product that served the global requirements of high function FMS customers, Bergquist, 6/8/04, 264:4-12, even though PeopleSoft too was an established, experienced vendor of high function (HRM) applications. PF 8.2.1-8.2.2.
60 Peoplesoft, Oracle, and SAP each have very large, targeted sales forces that are specialized by industry verticals and which sell directly to large customers. Iansiti, 6/17/04, 2054:22-2055:7; Knowles, 6/23/04, 2903:5-9.
62 Complex companies will only purchase high function software from vendors that can provide a reference from a company in the same industry. E.g., PF126.96.36.199; Gorriz, 6/15/04, 1375:9-15 (DaimlerChrysler); Bradley, 6/9/04, 600:9-602:3. See PF 188.8.131.52.
63 Oracle asserts that the standard two-year test for timeliness should be lengthened because the software products at issue are "durable goods." Oracle CoL ¶ 94. Leaving aside that the products are not so durable, no case has accepted a period for entry as long as the five or more years Oracle suggests here.
64 Non-established vendors such as Microsoft, have "zero opportunity" for licenses in the large enterprise market for a number of reasons, including: (1) cost; (2) necessary footprint to support the customers; (3) product capability; and (4) lack of established reputation. PF 184.108.40.206 (P3255R, Ayala dep., 5/18/04, 70:9-71:23).
66 Finally, Oracle makes the claim that IBM's management "considered reentering the applications market," when in fact the opposite occurred. DF at ¶143. As it does elsewhere in the Findings, Oracle cites to certain language in a document but does not mention specific testimony explaining the document or its meaning, see PF 220.127.116.11-.5 (IBM has "refused to look at any and all application company evaluations of purchase."); similarly, compare DF at ¶ 226 with testimony cited in PF 10.1.8.
67 Professor Zmiejewski explained that verification consists of determining whether efficiencies claims: (1) were correctly calculated, (2) had a reasonable factual basis for inputs and assumptions, and (3) used standard, accepted methodologies and analyses. Zmiejewski, 7/1/04, 4515:10 - 4516:12; P3357.
68 Notably, Oracle did not request that any of its four experts even attempt to verify the accuracy of the purported cost savings, or perform any independent analyses. Rather, Professors Hausman and Teece, in the waning moments of their testimony, conceded that they had taken Oracle's numbers without question, and made a few cursory unsubstantiated comments about the purported probabilities that Oracle would achieve such savings.
72 PF18.104.22.168; P3346, CSFB 30(b)(6) (Scarborough) dep., 06/03/04, 91:10-92:13, 102:1-12, 104:16-19, 105:16-108:2, 111:7-112:15, 116:9-117:6, 136:11-138:16, 150:11-151:22, 154:20-155:8, 156:15-158:10, 193:21-195:13, 206:6-207:7, 208:22-211:6.