Federal Trade Commission v. CCC Holdings Inc.

605 F. Supp. 2d 26, 2009 U.S. Dist. LEXIS 21784

Opinion

MEMORANDUM OPINION 1

ROSEMARY M. COLLYER, District Judge.

American drivers make nearly twenty-five million automobile insurance claims each year and insurers, in turn, spend an estimated $100 billion annually to cover those claims. Most insurers and automotive repair shops use specialized computer software to estimate the cost of repair or the value of replacement in the event of a total loss. These software systems play a critical role in the automotive repair industry. CCC Information Services, Inc. (“CCC”) and Mitchell International, Inc. (“Mitchell”) are two of the largest companies in these markets. Audatex North America, Inc. (“Audatex”), formerly ADP Claims Services Group, is the one other significant competitor for sales of partial loss estimating and total loss valuation software. 2

CCC is a wholly owned subsidiary of CCC Holdings Inc. (“CCC Holdings”), a for-profit corporation, existing and doing business under the laws of Delaware, and headquartered in Chicago, Illinois. CCC Holdings is principally owned by Invest-corp, S.A. (“Investcorp”), a private equity firm with more than $15 billion under management, and funded primarily by investors in Saudi Arabia. The majority owner of Mitchell is Aurora Equity Partners III, L.P. (“Aurora”), a private equity firm based in Los Angeles, California, which has approximately $2 billion of assets.

On April 2, 2008, Defendants CCC Holdings and Aurora entered into a Restructuring Agreement (“Restructuring Agreement”) which contemplates a $1.4 billion “merger of equals” between CCC and Mitchell to be effected no later than March 31, 2009. PX 786. 3 The Federal Trade Commission (“FTC” or “Commission”), through its Bureau of Competition, seeks to preliminarily enjoin the pending transaction, positing that a 3-to-2 merger in the partial loss and total loss software markets would obviously and substantially harm competition. The Court finds the evidence more complicated and uncertain. Nonetheless, because the FTC has “raised questions going to the merits so serious, substantial, difficult and doubtful as to make them fair ground for thorough investigation, study, deliberation and determination by the FTC in the first instance and ultimately by the Court of Appeals,” FTC v. H.J. Heinz Co., 246 F.3d 708, 714-15 (D.C.Cir.2001) (internal quotations omitted); see also FTC v. Whole Foods Mkt., Inc., 548 F.3d 1028, 1035 (D.C.Cir.2008) (Brown, J.); id. at 1042 (Tatel, J., concurring), the Court will issue the requested injunction.

I. PROCEDURAL HISTORY

On November 25, 2008, the FTC found that it had “reason to believe” that the proposed merger would violate the antitrust laws, specifically Section 7 of the Clayton Act, 15 U.S.C. § 18, and issued an administrative complaint challenging the proposed merger. At the same time, the Commission, pursuant to Section 13(b) of the Federal Trade Commission Act (“FTC Act”), 15 U.S.C. § 53(b), and Section 16 of the Clayton Act, 15 U.S.C. § 26, authorized the Bureau of Competition to file the instant Complaint to petition the Court for a temporary restraining order and a preliminary injunction to preserve the status quo pending an administrative adjudication before the FTC; that trial is scheduled to commence no later than March 31, 2009, and the FTC has committed to rendering a final opinion within 90 days of an initial decision by the Administrative Law Judge. See FTC Press Release (Nov. 25, 2008). The merging parties suggest they will abandon the merger if an injunction issues, in part because financing would be too difficult to maintain during the administrative process. The Court held a total of nine days of evidentiary hearings and legal argument. The parties submitted approximately 15 boxes of documentary evidence, including documents obtained through discovery and deposition and FTC Investigation Hearing (“IH”) transcripts, as well as nearly 300 pages of findings of fact and two rounds of highly refined and informative legal briefing.

II. FACTS

There are millions of accidents each year on roads across the United States. If the drivers in those accidents are insured, they file an insurance claim. This leads to an assessment of the extent of the damage to the automobile and an estimate of the cost of parts and labor needed to effect repairs, or if too expensive, replace the vehicle.

A. The Products

1. Estimatics

Costs for repair of damaged vehicles, or partial loss claims, account for well over $30 billion in insurance claims payments annually. See PX 256-003(CCC). 4 Performing partial loss estimations was once a manual process. The appraiser or claims adjuster would rely on information from published sources and perform the calculations either by hand or with a desk calculator. PX 1020 ¶ 27 (Hayes Prelim. Report). Today, all major automobile insurers and the vast majority of the approximately 45,-000 repair facilities subscribe to one or more estimating software products (“Estimatics”). PX 514-013, 015 (Mitchell); PX 531-001 (Mitchell); see also PX 6 ¶ 3 (Mel-lo Dec!., Anderson Behel Body Shop); PX 10 ¶ 2 (Dorn Deck, Dorn’s Body & Paint). These products are much faster than manual calculations, permit analysis of more kinds of information, and are considered more reliable, consistent, and accurate. See, e.g., PX 14 ¶ 4 (Brandt Deck, Hartford Fire Ins. Co. (“The Hartford”)); PX 13 ¶ 3 (Hall Deck, GMAC); PX 20 ¶¶ 3-4 (Rollins Deck, Safeco Ins. Cos. (“Safeco”)); PX 28 ¶ 3 (Wilson Deck, The Hanover Ins. Group, Inc. (“Hanover”)); PX 27 ¶3 (Danford Deck, Ohio Mutual Ins. Group (“Ohio Mutual”)); PX 11 ¶ 3 (Brown Deck, Erie Ins. Group (“Erie”)); PX 30 ¶ 4 (Starnes Deck, State Farm Mutual Auto. Ins. Co. (“State Farm”)). Estimatics products consist of three main components: (1) a spreadsheet that tracks the line items that are a part of a vehicle repair estimate, (2) the database from which parts and labor costs are pulled, and (3) the software that calculates the total cost of repair, taking into consideration overlap times (such as the need to remove a bumper only once to perform two repairs behind or to the bumper). PX 1020 ¶ 12 (Hayes Prelim. Report) (citing Ramamurthy IH Tr. (7/22/08) at 198-200).

There are presently five companies in the United States that sell Estimatics: CCC, Mitchell, Audatex, Web-Est LLC (“Web-Est”), and Applied Computer Resources, but the three major competitors in the Estimatics market today are CCC, Audatex and Mitchell, which collectively hold the lion’s share of the market — approximately 99% 5 — thus earning the nickname, the “Big Three.” See PX 1020, Exs. 2-3 (Hayes Prelim. Report); see also PX 1028-004 (Web-Est). 6 There are two types of Estimatics products: communicating and non-communicating products. Communicating Estimatics tools are used by insurance companies and repair facilities that are part of a direct repair program (“DRP”), whereas non-communicating estimating tools are used by low-end repair facilities that are not part of a DRP. A DRP is a network of strategic alliances between an insurance company and the repair facilities with which it frequently does business. Communicating products allow insurance companies and DRP repair facilities to instantaneously relay information about the status of partial loss claims through specialized communication links, while non-communicating products do not have this capacity. See PX 1020 ¶ 12 (Hayes Prelim. Report); DX 644, App. 2 ¶ 7 (Ordover Coordinated Interaction Report). CCC, Mitchell and Audatex primarily sell communicating Estimatics products. Web-Est and Applied Computer Resources currently sell only non-communicating products.

CCC’s core communicating Estimatics offering is called Pathways, but it also offers a low-end, non-communicating tool called Comp-Est which it markets to the independent repair facility segment. PX 86-008, 056(CCC). 7 To procure the parts and labor data for its Estimatics product, CCC obtained an exclusive license to the Hearst Business Publishing, Inc. “Motor” database in 1998. PX 25 ¶ 22 (Carr Deck, Motor Information Systems (“Motor”)). [Text redacted.].

Mitchell’s communicating Estimatics product is UltraMate, although it also sells a low-end product called Repairmate. PX 680-004-005 (Aurora). Audatex’s Estimatics product is called Audatex Estimating. Unlike CCC, both Audatex and Mitchell have their own proprietary databases, which were developed over many years.

The Estimatics providers generally charge their customers on a per user basis. Currently, the predominant medium for Estimatics software installation is a CD or DVD; monthly software updates must be sent to clients and re-installed on their systems. Web-Est, however, offers a Web-based non-communicating Estimatics product. 8 Seidel, Tr. (1/22 p.m.) at 165:18— 22, 167:7-8 (Web-Est). It currently licenses Mitchell’s database under restrictive conditions which prohibit it from selling to the top fifty insurance companies, from selling communicating Estimatics products to insurance companies or repair facilities, and from integrating with other third-party service providers, such as parts locators, salvage, and other providers. See DX 59 (Mitchell-Web-Est License Agreement); Seidel, Tr. (1/22 p.m.) at 173:24-174:13 (Web-Est). This restrictive license effectively precludes Web-Est from selling to most insurance companies and DRP repair facilities.

2. TLV

Some accidents may prove to be so costly that the insurer will declare the automobile a total loss because the estimated cost of repair approaches or exceeds the vehicle’s value. In such situations, the insurance company will calculate the replacement cost of the automobile, as it existed before the accident, based on model, make, upgrades, condition, mileage, etc., and pay the policyholder the vehicle’s replacement cost. 9 As with estimating repair costs, the total loss valuation process was once done only by hand. Today there are several different methodologies and tools for insurance companies to calculate total loss. One option is the “book” providers— NADA Appraisal Guides (“NADA”), the Kelley Blue Book, the Red Book, and the Black Book (collectively “the Books”)— whose reports are based on local or regional values. Brungger IH Tr. at 121-22 (Mitchell); Linder IH Tr. at 43, 46 (Mitchell); Marushka IH Tr. at 18, 29-30(CCC). These products are available in hard copy and many are available electronically; the electronic versions are updated more frequently than the printed Books. A number of insurance companies perform some or even most of their total loss valuations in-house using a combination of the Books and market research conducted by internal staff in order to obtain a more accurate and localized valuation than the Books can provide alone. Most insurers, however, use total loss software systems (“TLV”) which contain comprehensive databases of vehicle sales information that is regularly compiled from numerous sources and hundreds of localities. See PX 680-007-010 (Aurora); PX 86-008(CCC). The TLV software products provided by CCC; Mitchell and Audatex account for more than 90% of all total loss valuations. PX 513-016 (Mitchell); PX 548-007 (Mitchell). Other providers of total loss services include AutoBid and Vehicle Valuation Services (“WS”). WS is a software product that is used mostly for specialty vehicles such as commercial trucks and classic cars, and for a limited number of non-specialty vehicles as an accommodation to existing clients. See PX 401 ¶¶ 3, 5 (Blitstein Deck, WS).

Until recently, CCC and Audatex were the only providers of TLV. However, Mitchell successfully entered the TLV market in 2005 after two previous failed attempts. CCC’s TLV product is called ValueScope Salvage Solutions. PX 256-039, 046(CCC). Mitchell’s TLV product is called WorkCenter Total Loss, and Audatex’s product is called Autosource.

B. Marketplace Dynamics

In addition to Estimatics and TLV, CCC, Mitchell and Audatex sell a number of other automotive damage products, including workflow management products, business intelligence products, and repair facility or shop management software. DX 82(CCC); PX 560-031 (Mitchell). 10 Mitchell also sells an auto injury estimating, or “medical” product; it is the only of the Big Three companies that offers a medical product. See PX560-031-032 (Mitchell). Insurance companies often purchase a suite of products from one vendor. These “bundles” usually include a core Estimatics product and several add-on products, such as digital imaging, shop management, aftermarket parts data, workflow, and total loss valuation. Conway, Tr. (1/9) at 92:1-18 (Audatex); see also PX 680-007-008, 010 (Mitchell); PX 86-008-009(CCC); PX 244-062(CCC); PX 763 (Mitchell). Contracts for DRP repair facilities also often consist of bundles, including the base Estimatics product and a variety of add-ons. Conway, Tr. (1/9) at 111:9-13 (Audatex); DX 47 (Mitchell); DX 98(CCC). Independent repair facilities tend to purchase only the core estimating product.

The buyers of Estimatics are insurance companies and auto repair facilities, while only insurance companies buy TLV products. Repair facilities account for about 60% of Estimatics revenues, while insurers account for the other 40%. See PX 1020, Ex. 2 (Hayes Prelim. Report). Insurers negotiate longer term contracts with Estimatics and TLV systems suppliers — generally two to five years. Sun, Tr. (1/9) at 96:16-20 (Mitchell). In a typical year, approximately 100 insurance company accounts come up for renewal. Brungger IH Tr. at 185-86 (Mitchell). These contracts can cost anywhere from a few thousand dollars per year for a small insurer to several million dollars per year for a large insurer. Conway, Tr. (1/9) at 94:22-95:4 (Audatex); DX 13 ¶ 69 (Hayes Prelim. Report); see, e.g., DX 252-4, 53 (five-year, multi-million dollar contract with CCC). Most of these insurance contracts are obtained through a secret bidding process whereby the insurance company submits a request for proposal (“RFP”) to the bidders in which it seeks information on the bidder’s proposed prices and product bundle options. RFPs are usually accompanied by a non-disclosure agreement, Sun, Tr. (1/9) at 51-52, 46-47 (Mitchell); Brandt, Tr. (1/8 a.m.) at 23 (The Hartford), and therefore are not supposed to be shared with the bidder’s competitors. Repair facility contracts are also intended to be confidential. See, e.g., Cheskis, Tr. (1/23 a.m.) at 16:5-7,17:21-23 (Gerber Collision & Glass (“Gerber”)); Sun, Tr. (1/9) at 53:24-54:10 (Mitchell); Ramamurthy, Tr. (1/22 a.m.) at 108:19-109:25(CCC); Conway, Tr. (1/12 a.m.) at 39:8-13 (Audatex).

Despite a highly concentrated supplier base and low growth rates in the industry, PX 543-005 (Mitchell); PX 574-002 (Mitchell), these markets are highly competitive today. Brandt, Tr. (1/8 a.m.) at 17:23-18:2, 18:6-7 (The Hartford); Cheskis, Tr. (1/23 a.m.) at 10:12-23 (Gerber); Aquila Dep. at 15 (Audatex); FTC’s PreTrial Br. at 3; Ordover, Tr. (1/23 a.m.) at 76:20-77:8. Although insurance company and repair facility contracts are confidential, the Estimatics and TLV vendors expend significant resources in order to obtain “competitive intelligence” about each other’s pricing, and compete vigorously for customers. Sun, Tr. (1/8 p.m.) at 53-58, 92-93 (Mitchell). The FTC fears that a merger of CCC and Mitchell will destroy the delicate balance that has sustained healthy competition despite the small number of competitors in these markets; hence this lawsuit.

III. APPLICABLE LAW

Section 7 of the Clayton Act prohibits mergers or acquisitions “the effect of [which] may be substantially to lessen competition, or to tend to create a monopoly” in “any line of commerce or in any activity affecting commerce in any section of the country.” 15 U.S.C. § 18. The Supreme Court has explained that Section 7 “deals in probabilities, not certainties.” United States v. Gen. Dynamics Corp., 415 U.S. 486, 505, 94 S.Ct. 1186, 39 L.Ed.2d 530 (1974); see also Brown Shoe Co. v. United States, 370 U.S. 294, 323, 82 S.Ct. 1502, 8 L.Ed.2d 510 (1962); United States v. El Paso Natural Gas Co., 376 U.S. 651, 658, 84 S.Ct. 1044, 12 L.Ed.2d 12 (1964). Thus, to establish a violation of Section 7, the FTC need not show that the challenged merger or acquisition will lessen competition, but only that the loss of competition is a “sufficiently probable and imminent” result of the merger or acquisition. United States v. Marine Bancorp., 418 U.S. 602, 623 n. 22, 94 S.Ct. 2856, 41 L.Ed.2d 978 (1974) (quoting United States v. Continental Can Co., 378 U.S. 441, 458, 84 S.Ct. 1738, 12 L.Ed.2d 953 (1964)).

When the FTC has “reason to believe” that the consummation of a merger will violate the antitrust laws, it may seek a preliminary injunction to prevent the merger until the agency has had an opportunity to adjudicate the merger’s legality in an administrative proceeding. 15 U.S.C. § 53(b). Section 13(b) of the FTC Act “provides for the grant of a preliminary injunction where such action would be in the public interest.” Heinz at 714; see 15 U.S.C. § 53(b). In enacting Section 13(b), Congress “ ‘demonstrated its concern that injunctive relief be broadly available to the FTC.’ ” Heinz, 246 F.3d at 714 (quoting FTC v. Exxon Corp., 636 F.2d 1336, 1343 (D.C.Cir.1980)); H.R.Rep. No., 93-624, at 31 (1973), reprinted in 1973 U.S.C.C.A.N. 2417, 2533; see also Whole Foods, 548 F.3d at 1042 (Brown, J.) (“the FTC—an expert agency acting on the public’s behalf—should be able to obtain injunctive relief more readily than private parties.”). Thus, the FTC “is not required to prove, nor is the court required to find, that the proposed merger would in fact violate Section 7 of the Clayton Act” in order for a preliminary injunction to be issued. FTC v. Staples, Inc., 970 F.Supp. 1066, 1070 (D.D.C.1997) (citations omitted). Rather, the FTC “need only show that there is a ‘reasonable probability’ that the acquisition may substantially lessen competition.” Id. at 1072.

Section 13(b) authorizes a court to issue a preliminary injunction “[u]pon a proper showing that, weighing the equities and considering the Commission’s likelihood of ultimate success, such action would be in the public interest.” 15 U.S.C. § 53(b). The Court must balance these considerations on a sliding scale. Heinz, 246 F.3d at 714; FTC v. Elders Grain, Inc., 868 F.2d 901, 903 (7th Cir.1989); see also Whole Foods, 548 F.3d at 1035 (Brown, J.). Thus, “[a] greater likelihood of the FTC’s success will militate for a preliminary injunction unless particularly strong equities favor the merging parties.” Whole Foods, 548 F.3d at 1035 (Brown, J.).

The equities will often weigh in favor of the FTC because “ ‘the public interest in effective enforcement of the antitrust laws’ was Congress’s specific ‘public equity consideration’ in enacting” Section 13(b). Id. (quoting Heinz, 246 F.3d at 726). If the FTC meets its burden of showing that it is likely to succeed on the merits, it “creates a presumption in favor of preliminary in junetive relief,” id., which the merging parties may rebut by showing that, contrary to traditional antitrust theory, the public equities weigh in favor of the merger. See Whole Foods, 548 F.3d at 1035 (Brown, J.); see also FTC v. Arch Coal, Inc., 329 F.Supp.2d 109, 160 (D.D.C.2004); cf. Heinz, 246 F.3d at 727 n. 25 (noting that private equities are afforded little weight in Section 13(b) cases). If the merging parties are able to make such a showing, the FTC is required to show a greater likelihood of success on the merits. Whole Foods, 548 F.3d at 1035 (Brown, J.) (citing FTC v. Weyerhaeuser Co., 665 F.2d 1072, 1087 (D.C.Cir.1981)).

IV. ANALYSIS

A. Likelihood of Success on the Merits

The burden of showing likelihood of success on the merits is met if the Commission has “raised questions going to the merits so serious, substantial, difficult and doubtful as to make them fair ground for thorough investigation, study, deliberation and determination by the FTC in the first instance and ultimately by the Court of Appeals.” Heinz, 246 F.3d at 714-15 (internal citations omitted); see also Whole Foods, 548 F.3d at 1035 (Brown, J); id. at 1042 (Tatel, J., concurring). At the same time, the Court may not “simply rubber-stamp an injunction whenever the FTC provides some threshold evidence; it must ‘exercise independent judgment’ about the questions § 53 commits to it.” Whole Foods, 548 F.3d at 1035 (Brown, J.) (citing Weyerhaeuser, 665 F.2d at 1082). 11

In United States v. Baker Hughes Inc., 908 F.2d 981, 982-83 (D.C.Cir.1990), the D.C. Circuit adopted an analytical approach to Section 7 cases which has been followed in subsequent Section 13(b) cases. See, e.g., Heinz, 246 F.3d at 715; Arch Coal, 329 F.Supp.2d at 116. First, to meet its initial burden, the government must show that the proposed merger would lead to “undue concentration in the market for a particular product in a particular geographic area.” Baker Hughes, 908 F.2d at 982. Such a showing creates a “ ‘presumption’ that the merger will substantially lessen competition.” Id. Upon such a showing, the burden shifts to the defendants to rebut the presumption with evidence that “ ‘shows that the market-share statistics [give] an inaccurate account of the [merger’s] probable effects on competition’ in the relevant market.” Heinz, 246 F.3d at 715 (quoting United States v. Citizens & S. Nat’l Bank, 422 U.S. 86, 120, 95 S.Ct. 2099, 45 L.Ed.2d 41 (1975)) (alterations in original). If the defendants succeed in rebutting the presumption that the merger will lessen competition, “ ‘the burden of producing additional evidence of anticompetitive effects shifts to the government, and merges with the ultimate burden of persuasion, which remains with the government at all times.’ ” Heinz, 246 F.3d at 715 (quoting Baker Hughes, 908 F.2d at 983).

The Supreme Court has cautioned, however, that while “statistics reflecting the shares of the market controlled by the industry leaders and the parties to the merger are ... the primary index of market power[,] ... only a further examination of the particular market — its structure, history and probable future — can provide the appropriate setting for judging the probable anticompetitive effect of [a] merger.” Brown Shoe, 370 U.S. at 322 n. 38, 82 S.Ct. 1502. In order to adequately address these factors, “the merging parties are entitled to oppose a [Section 13(b) ] preliminary injunction with their own evidence, and that evidence may force the FTC to respond with a more substantial showing” of the merger’s probable anticompetitive effects. Whole Foods, 548 F.3d at 1035 (Brown, J.). In the end, “antitrust theory and speculation cannot trump facts, and even Section 13(b) cases must be resolved on the basis of the record evidence relating to the market and its probable future.” Arch Coal, 329 F.Supp.2d at 116-17; see also Brown Shoe, 370 U.S. at 322 n. 38, 82 S.Ct. 1502. Thus, an analysis of the likely competitive effects of a merger requires determinations of (1) the relevant product market, (2) the relevant geographic market, and (3) the transaction’s probable effect on competition in those markets. See Marine Ban-corp., 418 U.S. at 618-23, 94 S.Ct. 2856; Gen. Dynamics, 415 U.S. at 510-11, 94 S.Ct. 1186; see also Arch Coal, 329 F.Supp.2d at 117.

1. Prima Facie Case

Aprima facie Section 7 case “rests on defining a market and showing undue concentration in that market.” Whole Foods, 548 F.3d at 1036 (Brown, J.) (citing Baker Hughes, 908 F.2d at 982-83). Courts generally begin their analysis of a Section 7 case by defining the relevant market. See, e.g., Marine Bancorp., 418 U.S. at 618-23, 94 S.Ct. 2856; FTC v. Swedish Match, 131 F.Supp.2d 151, 156 (D.D.C.2000). But see Whole Foods, 548 F.3d at 1036 (Brown, J.) (noting that “this analytical structure does not exhaust the possible ways to prove a § 7 violation” (citing El Paso Natural Gas Co., 376 U.S. at 660, 84 S.Ct. 1044)). A relevant market has two components: (1) the relevant product market and (2) the relevant geographic market. The “relevant product market” identifies the product and services with which the defendants’ products compete. The “relevant geographic market” identifies the geographic area in which the defendants compete in marketing their products or services. The FTC bears the burden of proof and persuasion in defining the relevant market. Arch Coal, 329 F.Supp.2d at 119.

Once the relevant market is defined, the court can determine market concentration. The standard measure for market concentration is the Herfindahl-Hirschmann Index (“HHI”). See Heinz, 246 F.3d at 716. Under the Federal Trade Commission and U.S. Department of Justice Horizontal Merger Guidelines, a market with a post-merger HHI above 1800 is considered “highly concentrated,” and mergers that increase the HHI in such a market by more than 100 points “are presumed ... likely to create or enhance market power or facilitate its exercise.” Fed. Trade Comm’n & U.S. Dep’t of Justice Horizontal Merger Guidelines (1992), as revised (1997) (“Merger Guidelines”) § 1.51. Although the Merger Guidelines are not binding on the Court, they provide a “useful illustration of the application of the HHI.” FTC v. PPG Indus., Inc., 798 F.2d 1500, 1503 n. 4 (D.C.Cir.1986). Moreover, the D.C. Circuit explained in Heinz that a merger to duopoly which increased the premerger HHI of 4,775 by 510 points “create[d], by a wide margin, a presumption that the merger w[ould] lessen competition” in the relevant market. Heinz, 246 F.3d at 716.

a. Relevant Product Markets

In determining relevant product markets, courts have traditionally considered two factors: “[1] the reasonable interchangeability of use and [2] the cross-elasticity of demand between the product itself and substitutes for it.” Brown Shoe, 370 U.S. at 325, 82 S.Ct. 1502. In other words, the question is “whether two products can be used for the same purpose, and if so, whether and to what extent purchasers are willing to substitute one for the other.” Staples, 970 F.Supp. at 1074 (internal quotations omitted). Relevant markets will generally include producers who, given product similarity, have the ability to take significant business from each other. Arch Coal, 329 F.Supp.2d at 119.

If consumers can substitute the use of one product for another, those products will be deemed “functionally interchangeable.” Id.; see also Staples, 970 F.Supp. at 1074 (finding that office supplies sold by an “office superstore” like Staples are functionally interchangeable with office supplies sold at “mass merchandisers” like Wal-Mart). Courts generally will include functionally interchangeable products in the same product market unless factors other than use indicate that they are not actually part of the same market. See, e.g., United States v. Archer-Daniels-Midland Co., 866 F.2d 242, 246 (8th Cir.1988) (even though beet sugars and high-fructose corn sugars were functionally interchangeable, they did not belong to the same product market because government price support for beet sugars meant that prices for corn sugars could be raised substantially without feeling the competitive impact of beet sugar prices). Cross-elasticity of demand refers to the “ ‘responsiveness of the sales of one product to price changes of the other.’” Staples, 970 F.Supp. at 1074 (quoting United States v. E.I. du Pont de Nemours & Co., 351 U.S. 377, 400, 76 S.Ct. 994, 100 L.Ed. 1264 (1956)). However, courts should “exclude any ... product to which, within reasonable variations in price, only a limited number of buyers will turn.” Times-Picayune Publ’g Co. v. United States, 345 U.S. 594, 612, 73 S.Ct. 872, 97 L.Ed. 1277 (1953). 12

Courts have relied on several “practical indicia” as aids in identifying the relevant product market, including: industry or public recognition of a submarket as a separate economic entity, the product’s peculiar characteristics and uses, unique production facilities, distinct customers, distinct prices, sensitivity to price changes, and specialized vendors. Brown Shoe, 370 U.S. at 325, 82 S.Ct. 1502; see also Arch Coal, 329 F.Supp.2d at 120. These indicia can be applied to augment the analyses of interchangeability and cross-elasticity of demand.

1. Estimatics Market

Defendants do not dispute the FTC’s contention that Estimatics is so far superi- or to the paper-based systems that their customers would not return to the old methods even if Estimatics prices increased substantially. See, e.g., PX 6 ¶2 (Mello Decl.); PX 7 ¶4 (Kostakis Deck, Angelo’s Auto Body); PX 20 ¶ 4 (Rollins Deck); PX 27 ¶ 3 (Danford Deck); PX 11 ¶ 3 (Brown Deck). Thus, Defendants concede that the core Estimatics software product is a relevant product market. See Ordover, Tr. (1/23) at 204:20-25. 13

2. TLV Market

Although the Defendants do not dispute the FTC’s market definition for Estimatics, they disagree with its definition of the total loss valuation market. The FTC defines the relevant total loss valuation market to include only TLV sold in the United States by CCC, Mitchell and Audatex. PX 1020 ¶ 48 (Hayes Prelim. Report); see also Hayes, Tr. (1/12 a.m.) at 84:12-22. The Defendants argue that the FTC has not met its initial burden of showing a properly defined product market because it fails to include other sources of total loss valuations that insurance companies currently use or to account for the fact that some insurance companies perform their total loss valuations in-house rather than using TLV products.

Defendants have demonstrated that insurers of auto physical damage currently use a variety of solutions to perform total loss valuations. In addition to the total loss software sold by CCC, Mitchell and Audatex, some insurance companies use the Books to perform some or most of their total loss valuations in-house. Brandt, Tr. (1/8 a.m.) at 47:5-9 (The Hartford); Dibble, Tr. (1/23 a.m.) at 38:23-39:18 (Infinity Ins. Co.). In fact, some of the largest auto insurers in the country rely on the Books for most of their total loss needs. See DX 49 (Mitchell); DX 50 (Mitchell). For instance, CCC’s CEO, Githesh Ramamurthy, testified that, of the 700,000 total loss claims that [text redacted] processes each year, it calculates roughly 500,000 in-house using NADA. Ramamurthy, Tr. (1/22 a.m.) at 97:4-8(CCC); see also Brandt, Tr. (1/8 a.m.) at 47:5-10 (The Hartford) (Progressive Insurance Company (“Progressive”) uses NADA to calculate roughly 75% of its total loss claims). 14 Those insurance companies that perform some of their valuations internally may consult other data sources as well, such as Autotrader, or conduct their own research by contacting salvage yards and auto dealerships. See, e.g., Dibble, Tr. (1/23 a.m.) at 39:8-13 (Infinity Ins. Co.). Defendants point to these shifts between products as evidence of a single market between Estimatics and the Books. See IIB Phillip Areeda et al., Antitrust Law ¶ 562a, at 371 (3d ed. 2007) (“[A]ctual shifts between two products in response to — or even without — changes in their relative prices indicate a single market.”).

This evidence is factually unchallenged by the FTC but the agency contends that it does not have the force Defendants would imply, given that over 90% of all total loss claims are calculated using TLV sold by one of the three major competitors. PX 513-016 (Mitchell); PX 548-007 (Mitchell). 15 According to the FTC, there is sufficient evidence in the record to conclude that the TLV products sold by CCC, Mitchell and Audatex represent a separate product market from other total loss valuation methods.

Dr. John Hayes, an economics expert testifying for the FTC, conducted a critical loss analysis to determine whether a hypothetical monopolist could profitably impose a SSNIP in the TLV market, an approach described in the Merger Guidelines. PX 1020 ¶¶ 26, 42-48 (Hayes Prelim. Report). 16 Through his analysis, Dr. Hayes concluded that TLV software constitutes a relevant product market. Id. ¶ 48. Dr. Hayes was aware that some insurance companies use the Books to conduct valuations, but he admitted that he did not know how pervasively internal valuations are used or how insurance companies produce internal valuations. Hayes, Tr. (1/21 p.m.) at 156:24-157:18. Thus, when calculating total loss market shares, Dr. Hayes did not include valuations that were performed internally. Id. at 149:22-150:8.

Although Defendants’ expert, Dr. Janusz Ordover, did not reach a conclusion as to whether the Books are or are not part of the same market as TLV, Ordover, Tr. (1/23 p.m.) at 205:18-22, a critical failure to support Defendants’ challenge to the FTC’s market definition for TLV, the Defendants discredit Dr. Hayes’s conclusions, arguing that his critical loss analysis is flawed because he incorrectly assumed that insurance companies lack the economies of scale to perform total loss valuations in-house. Hayes, Tr. (1/21 p.m.) at 150:14:151:4, 153:2-154:20, 157:2-158:17. The evidence is clear that CCC lost a large insurance company as a customer when that company decided to do most of its total loss calculations in-house using NADA. DX 86. “[C]ourts have generally recognized that when a customer can replace the services of [an external product] with an internally-created [ ] system, this ‘captive output’ (i.e. the self-production of all or part of the relevant product) should be included in the same market.” United States v. SunGard Data Sys., Inc., 172 F.Supp.2d 172, at 193, 186 (D.D.C.2001) (quoting FTC v. Cardinal Health, Inc., 12 F.Supp.2d 34, 48 (D.D.C.1998)) (alterations in original) (internal quotations omitted); accord Merger Guidelines § 1.31 (vertically integrated firms are included in the market “to the extent that such inclusion accurately reflects their competitive significance in the relevant market prior to the merger”).

Notwithstanding this gap in Dr. Hayes’s analysis, the real-world evidence shows that Books and TLV are not part of the same product market. See Arch Coal, 329 F.Supp.2d at 116 (“[A]nti-trust theory and speculation cannot trump facts.”). The Book vendors do not consider themselves to be in competition with CCC, Mitchell, and Audatex. PX 17 ¶¶ 7-13 (Stanton Deck, NADA); PX 19 ¶ 5 (Fournier Deck, Automobile Red Book); PX 29 ¶ 8 (Cross Deck, Nat’l Auto Research). In fact, CCC and Mitchell offer access to NADA as an add-on to their TLV products. PX 241-023(CCC); PX 643-011 (Mitchell). Most insurance companies do not view the Books as an adequate substitute for TLV products. See PX 664-005 (Mitchell); Brown Dep. at 90-90 (Erie); see also Hall, Tr. (1/8 a.m.) at 79:15-17, 81:2-4 (GMAC); Danford Dep. at 58 (Ohio Mutual only uses Books for commercial and recreational vehicles that are not in the TLV databases); PX 11 ¶ 7 (Brown Deck). In fact, the majority of insurers who were contacted by the FTC stated that they would not switch from TLV products to the Books even if facing a price increase of 10% or more. See, e.g., Adamson Dep. at 31-32, 40-41 (EMC Ins. Co.); PX 20 ¶ 14 (Rollins Deck) (“In the event of a 10% price increase in total loss valuation products, Safeco would not be likely to switch to using NADA, Kell[e]y, or other ‘book’ products for total loss estimates. The book products are only guides, which do not provide a local market evaluation of comparable vehicles.”). Similarly, almost none of the Defendants’ internal documents refer to the Books when describing the competitive landscape. 17 To the extent that the Books are mentioned, they are referred to as “ancillary databases.” PX 643-011 (Mitchell document containing a competitive analysis between CCC, Mitch ell and Audatex, and referring to NADA and other Books as “ancillary databases”). 18

The market participants view TLV as separate from the Books for good reason. The evidence reveals that the Books simply are not as accurate, detailed, or up-to-date as TLV. CCC, Mitchell and Audatex use a consistent methodology across all vehicles that includes recent sales of comparable vehicles. PX 1964-009-010(CCC); PX 1958-034-036 (Mitchell). Their methodology accounts for, among other factors, year, make, model, body style, engine, mileage, upgrades, condition, and includes numerous databases. PX 1964-006, 010 (CCC’s TLV product “equates to a vehicle database of approximately 64,00 [sic] unique vehicles, which is substantially more than Red Book and Kelley Blue Book”). CCC’s TLV database is “compiled by surveying over 4,400 car dealerships in more than 350 markets twice each month and through 1,800 publications.” PX 99-020(CCC). Mitchell’s database contains over [text redacted] million electronic records including the Power Information Network (PIN) data received from JD Power and Associates. PX 643-011 (Mitchell); PX 541-024 (Mitchell). The valuation methodologies used by the Books include subjective data and do not account for all of the vehicle’s options, mileage, or condition as precisely as TLV programs. Hall, Tr. (1/8 a.m.) at 79:18-80:8, 80:19-81:8 (GMAC); PX 1964-006, 010(CCC). For instance, NADA uses wholesale and retail data sets and manipulates that information using analytical modeling and judgment by its editorial staff to estimate vehiele value. PX 17 ¶ 3 (Stanton Deck); PX 260-00KCCC); PX1964-016 (CCC Value-Scope Methodology & Settlement Tips). Thus, TLV has substantially different valuation methodologies than the Books. See Brown Shoe, 370 U.S. at 325, 82 S.Ct. 1502 (peculiar characteristics of the product indicate separate product markets).

In addition, TLV provides valuations in local markets, while the Books provide only regional data that are rarely localized. See PX 20 ¶ 14 (Rollins Deck); PX 11 ¶ 7 (Brown Deck); PX 28 ¶ 11 (Wilson Deck) (stating that “pricing information in guidebooks ... does ... not contain the same level of specificity as local market comparables.”); PX 1964-010(CCC) (“Where NADA will have one region covering seven states, CCC will segment each state into an average of eight local markets.”); PX 17 ¶ 9 (Stanton Deck); PX 256-048(CCC). Furthermore, TLV systems’ calculations are said to comply with applicable state regulations of each state, PX 81-038 (CCC Response to [text redacted] RFP); PX 116-039 (CCC Response to [text redacted] RFP); PX117-044, 051 (CCC Response to [text redacted] RFP), whereas the Books do not.

Nor does CCC, Mitchell, or Audatex price TLV against the Books. TLV costs approximately $ [text redacted] per estimate, PX 548-007, whereas the Books charge about 5-35 cents per use. Stanton Dep. at 101-03 (NADA); PX 19 ¶ 2 (Fournier Deck). Of course, “products competing against one another in a differentiated product market may have widely different prices.” United States v. Oracle Corp., 331 F.Supp.2d 1098, 1121 (N.D.Cal.2004). 19 If part of the same product market, the Books and the Defendants’ TLV products are undoubtedly differentiated products. The variance in price therefore may not be surprising. 20 However, there is no evidence to suggest that the price of TLV is sensitive in any way to changes in pricing by Book vendors, or vice-versa. Cf. PX643-008 (Mitchell) (setting pricing strategy by reference to CCC and Audatex only). In addition, there is scant evidence of any direct competition between TLV and the Books.

In sum, the practical indicia — particularly industry recognition of a separate market; TLV’s peculiar characteristics including especially accurate, up-to-date valuations, speed, reliance and defensibility, and ability to interface with estimating products; and sensitivity to price changes only against other TLV products — support the conclusion that TLV software products represent a relevant product market.

Despite this analysis, the fact remains that some of the largest insurance companies in the country perform a significant percentage of their total loss valuations in-house using the Books and that some others use the Books or other valuation methods for a small percentage of their total loss claims, rather that TLV. What this demonstrates, at best, is that some of the largest insurance companies in the country have the resources and scale to perform most of their total loss valuations in-house profitably in lieu of using TLV, and that a few middle tier insurance companies use the Books to supplement TLV software, but not that the Books constitute a viable substitute for TLV for the vast majority of insurers. 21 When TLV software is used to calculate 90% of the total loss claims made in this country, and the majority of the remaining 10% includes the largest insurers that do most of their calculations in-house, it is clear that the computer-based TLV products provided by these three suppliers constitute a relevant product market. The Books are, in the main, a complementary product and not a competitive product.

In any event, the FTC is not “required to settle on a market definition at this preliminary stage,” Whole Foods, 548 F.3d at 1036 (Brown, J.), and inclusion of the Books in the market would have an insig nificant effect on the market shares because over 90% of total loss claims are calculated using TLV. PX 513-016 (Mitchell) (chart showing that CCC, Audatex and Mitchell were used to evaluate a combined 90% of all total loss claims made in 2007, while NADA and “other” products were used for approximately 10% of valuations); PX 548-007 (Mitchell) (table showing that book valuations are used for approximately 10% of all total loss claims while TLV software is used for 90% of claims). Even including the Books and other valuation methods, the combined CCC/Mitchell would still have over a 50% share of the broader total loss valuation product market. See PX 513-016 (Mitchell). The corresponding HHIs likewise would still establish a strong prima facie case for the FTC. Cf. Arch Coal, 329 F.Supp.2d at 158 (concluding that the HHI increase of 49 to 2,103 was “far from compelling,” and thus the prima facie case was “fairly weak.”). 22

b. The Geographic Market

Unlike some other Section 7 cases, the geographic market is not contested here. The relevant geographic market for both products in this case is the United States. Audatex competes in many foreign countries; CCC and Mitchell currently compete only in the United States. More significantly, although software products can be designed, manufactured and sold almost anywhere, the databases for Estimatics and TLV sold in the United States must contain data that apply to vehicles driven in the U.S., wherever manufactured. Hayes, Tr. (1/12 p.m.) at 131:2-10. Similar auto physical damage products sold outside the United States that rely on a database of foreign vehicles cannot provide partial loss and total loss valuations for most vehicles driven in this country. PX 1020 ¶ 50 (Hayes Prelim. Report).

c. Market Concentration

Market share and concentration statistics can establish a presumption of harm and shift the burden of proof to Defendants to demonstrate that the presumption does not accurately reflect a merger’s likely effects on competition in the relevant market(s). United States v. Baker Hughes, 731 F.Supp. 3, 11-12 (D.D.C.1990).

1. Concentration in the Estimatics Market

There are five competitors in the Estimatics market: CCC, Audatex, Mitchell, Web-Est, which sells the software product WebEst, and Applied Computer Resources, which sells the software product Crash-writeR. Of these, CCC, Mitchell and Audatex dominate the market. Ramamurthy IH Tr. at 157-158(CCC); PX 1028-004, 017 (Web-Est); Ramamurthy, Tr. (1/22 a.m.) at 40:20-41:7, 74:1-10(CCC). Defendants’ expert, Dr. Ordover, agreed that CCC, Mitchell, and Audatex are the only three competitors for insurance customers in the Estimatics segment and that neither Web-Est nor Applied Computer Resources provides any services to insurers. Ordover, Tr. (1/23 p.m.) at 205:1-7. The Big Three are currently the only competitors for DRP repair facilities as well. PX 4 ¶5 (Daly Deck); PX 27 ¶¶4, 7 (Danford Deck); PX 11 ¶ 4 (Brown Deck); PX 25 ¶ 15 (Carr Deck); PX 20 ¶ 5 (Rollins Deck); Kostakis Dep. at 18 (Angelo’s Auto Body Shop). At this point in time, Web-Est and Applied Computer Resources offer only non-communicating Estimatics products to low-end repair facilities that are not part of a DRP. Web-Est and Applied Computer Resources together have a tiny fraction of the Estimatics market— approximately 1% combined 23 — and they currently do not compete to any meaningful extent with CCC, Mitchell and Audatex. Removing Web-Est and Applied Computer Resources from the market share data would have “almost no bearing” on the Estimatics market concentration calculations. Hayes, Tr. (1/12 p.m.) at 124:24-125:5. This is no surprise considering that the contract between Web-Est and Mitchell precludes Web-Est from offering communicating Estimatics to insurance companies or DRP facilities and from selling any Estimatics to the top 50 insurance companies. See DX 59 (Mitchell-Web-Est License Agreement); Seidel, Tr. (1/22 p.m.) at 173:24-174:13.

In 2007, CCC had approximately a 48% share of the revenue in the Estimatics market, Audatex had a" 30% share of the market, and Mitchell had a 21% share. PX 1020-039, Ex. 2 (Hayes Prelim. Report); see also Hayes, Tr. (1/12 p.m.) at 120.T5-122.T0. Thus, the combined CCC/Mitchell would hold almost a 70% market share, and would be more than twice the size of Audatex North America. PX 1020 ¶49, Ex. 2 (Hayes Prelim. Report). The merger would increase the HHI in Estimatics from approximately 3,650 to 5,685 — an increase of 2,035 points. Id. ¶ 60. The post-merger HHI would therefore be higher than it was in Heinz, which “create[d], by a wide margin, a presumption that the merger w[ould] lessen competition.” Heinz, 246 F.3d at 716; see also FTC v. H.J. Heinz, Co., 116 F.Supp.2d 190, 195-96 (D.D.C.2000) (merger would have given the new company approximately 32% of the market and increased HHI by 510 to 5,285).

Because insurance companies and repair facilities operate in different, albeit intertwined sub-markets, it also is useful to break down the numbers between insurance companies and repair facilities. In the repair facility segment, CCC’s Estimatics market share as of 2007 was approximately 55.7%, Audatex had a 25.2% share, and Mitchell came in third with an 18.1% share. PX 1020-040, Ex. 3 (Hayes Prelim. Report); see also Tr., Hayes (1/12 p.m.) at 136:3-137:9; Ramamurthy, Tr. (1/22 a.m.) at 63:19-25. Web-Est and Applied Computer Resources each had an approximate 0.5% share. If the merger is consummated, the combined CCC/Mitchell would hold approximately a 73.8% share of the Estimatics market for repair facilities. See PX1020, Ex. 3 (Hayes Prelim. Report). With respect to the insurance company segment, Audatex led with a 38.5% market share, CCC held a 35.0% market share, and Mitchell came in third at 26.5%. Id. Post-merger, CCC/Mitchell would hold a 61.5% share of the insurance segment of the Estimatics market. See id.

2. Concentration in the TLV Market

The only providers of total loss software systems are CCC, Audatex and Mitchell. But see PX 534-005 (Mitchell) (listing three main competitors in total loss valuation segment as CCC, Audatex and Auto-Bid). In 2007, CCC held approximately 60.7% of the TLV market, followed by Audatex with an approximate 34.8% share, and Mitchell at third with only a 4.5% share. PX 1020, Ex. 4 (Hayes Prelim. Report); Hayes, Tr. (1/12 p.m.) at 128:8-23, 129:8-13; see also PX 514-019-020 (Mitchell) (pie graph projecting CCC with 51%, Audatex with 42% and Mitchell with 7% of the TLV market in 2008). A combined CCC/Mitchell would hold about a 65% market share. Hayes, Tr. (1/12 p.m.) at 128:13-23 (Dr. Hayes); PX 1020, Ex. 4 (Hayes Prelim. Report); see also PX 514-019 (Mitchell). The pre-merger HHI in this market currently exceeds 4,900. The merger would result in an HHI of 5,460, with a 545 delta, Hayes, Tr. (1/12 p.m.) at 128:24-129:7; PX 1020 ¶ 64 (Hayes Prelim. Report), creating the presumption that the merger would lessen competition. Heinz, 246 F.3d at 716.

3. Merger to Duopoly

The FTC repeatedly proclaimed that this transaction represents a “merger to duopoly,” that is, a 3-to-2 merger, as if that settles the question. It asserts that the Estimatics and TLV markets really consist only of the Big Three — CCC, Mitchell, and Audatex. Defendants reject the notion that the descriptive term “duopoly” accurately depicts the post-merger competitive landscape because “Web-Est and a host of other new entrants offer innovative products that can transform the Estimatics market,” and [text redacted]. Defs.’ Posh-Trial Brief at 25. The Court concludes that the accuracy of this descriptor has minimal significance to the analysis here.

The FTC relies heavily on Heinz for the proposition that a “merger to duopoly” is destined for a preliminary injunction because “no court has ever approved a merger to duopoly under similar circumstances.” 246 F.3d at 717. The FTC overlooks the significance of the phrase “under similar circumstances” in Heinz, id., and thus over-reads the case. Instead of making a generalized holding relevant to all 3-to-2 company mergers, the Court of Appeals focused on the nature of the baby food market at issue in Heinz, in which high barriers to entry and total transparency in pricing underscored the risk of coordination. While pricing is not completely secret in the instant markets, the characteristics of Estimatics and TLV are worlds apart from baby food. For starters, the software products are complex, the price quotes to insurers are confidential, and the products are usually sold in complex bundles that may include both Estimatics and TLV or just one of these, as well as various other software products. What is clear from this preliminary record is that this situation is not Heinz. The question, therefore, is not simply whether this merger would constitute a “merger-to-duopoly,” but rather, whether the presumption of anticompetitive effects holds up, for preliminary relief, given the way these markets operate in fact.

As defense counsel admitted at the inception of this case, the HHIs in these markets are “very, very, high.” Scheduling Conf. at 10-11. Because of the high market concentrations and HHIs in the pre- and post-merger Estimatics and TLV markets, the FTC has established a strong prima facie case that a merger between CCC and Mitchell would violate Section 7 of the Clayton Act. See Heinz, 246 F.3d at 716. But that is just the beginning of the inquiry.

2. Rebuttal Arguments

Upon the showing of a prima facie case, the burden shifts to Defendants to show that traditional economic theories of the competitive effects of market concentration are not an accurate indicator of the merger’s probable effect on competition in these markets or that the procompetitive effects of the merger are likely to outweigh any potential anticompetitive effects. The courts have not established a clear standard that the merging parties must meet in order to rebut a prima facie case, other than to advise that “[t]he more compelling the prima facie case, the more evidence the defendant must present to rebut [the presumption] successfully,” Baker Hughes, 908 F.2d at 991. 24 Even in cases where the FTC has made a strong prima facie showing:

[i]mposing a heavy burden of production on a defendant would be particularly anomalous where, as here, it is easy to establish a prima facie case. The government, after all, can carry its initial burden of production simply by presenting market concentration statistics. To allow the government virtually to rest its case at that point, leaving the defendant to prove the core of the dispute, would grossly inflate the role of statistics in actions brought under section 7. The Herflndahl-Hirschman Index cannot guarantee litigation victories.

Id. at 992. With these words of caution in mind, the Court turns to the Defendants’ arguments.

a. Barriers to Entry

Defendants’ first point of rebuttal is that both the Estimatics and TLV markets will lack significant barriers to entry after the proposed merger, and that existing competitors are “poised for future expansion.” Id. at 988-89. A variety of factors, including the absence of significant entry barriers in the relevant market, can rebut a prima facie case. Id. at 984, 987 (“The existence and significance of barriers to entry are frequently, of course, crucial considerations in a rebuttal analysis.”); see also Heinz, 246 F.3d at 717 n. 13 (“Barriers to entry are important in evaluating whether market concentration statistics accurately reflect the pre- and likely post-merger competitive picture.”). “Ease of entry is the ability of other firms to respond to collusive pricing practices by entering to compete in the market.” Cardinal Health, 12 F.Supp.2d at 54-55. Even in highly concentrated markets, if there is sufficient ease of entry, others might enter to compete and undercut the likely anti-competitive effects of a merger. Id. In other words, entry is one way in which post-merger pricing practices can be forced back down to competitive levels. Id.-, see also Baker Hughes, 908 F.2d at 987 (“In the absence of significant barriers, a company probably cannot maintain a supracompetitive pricing for any length of time.”); Merger Guidelines, § 3.0 (“A merger is not likely to create or enhance market power or to facilitate its exercise, if entry into the market is so easy that market participants, after the merger, either collectively or unilaterally could not profitably maintain a price increase above premerger levels.”). “Determining whether there is ease of entry hinges upon an analysis of barriers to new firms entering the market or existing firms expanding into new regions of the market.” Cardinal Health, 12 F.Supp.2d at 55.

According to the Merger Guidelines, entry or expansion must be “timely, likely, and sufficient in its magnitude, character and scope to deter or counteract the competitive effects” that otherwise may be likely to result from a merger that significantly enhances market concentration. Merger Guidelines § 3.0; see also Chi. Bridge & Iron Co. v. FTC, 534 F.3d 410, 427-29 (5th Cir.2008); United States v. Visa USA, Inc., 163 F.Supp.2d 322, 342 (S.D.N.Y.2001) (entry must be “timely, likely, and [of a] sufficient scale to deter or counteract any anticompetitive restraints”).

1. History of Entry in the Estimatics and TLV Markets

“The history of entry into the relevant market is a central factor in assessing the likelihood of entry in the future.” Cardinal Health, 12 F.Supp.2d at 56 (citing Merger Guidelines § 3.1); see also Baker Hughes, 908 F.2d at 988; United States v. Waste Mgmt., Inc., 743 F.2d 976, 982 (2d Cir.1984); United States v. Tote, 768 F.Supp. 1064, 1076 (D.Del.1991). The FTC argues that history has proven entry into the Estimatics and TLV markets to be very difficult. The Defendants disagree.

With respect to Estimatics, the FTC contends that Audatex was the last successful entrant with its own database, and that was over thirty years ago. Defendants counter that there have been several recent examples of successful new Estimatics vendors:

• Comp-Est — Founded in 1990 and grew to more than 5,000 customers by the time it was purchased by CCC in 2003 (which still offers the product today to low-end repair shops);

• Focus Write, LLC (“Focus Write”)— Started in 2005 by the founder of Comp-Est and quickly grew to 1,500 customers but then floundered due to management issues;

• Web-Est — Founded in March 2008; has already doubled its customer base; and

•Applied Computer Resources — Began offering Crash-writeR in 1993; currently has 600 customers and sells four products to the automotive repair industry.

This evidence of supposed past successes fails to carry much water. Whatever brief success Comp-Est enjoyed, it has disappeared as a competitor through its purchase by CCC in 2003. PX 86-013(CCC). 25 In 2008, Eric Seidel, a former investor in Focus Write, became the founder and CEO of Web-Est, and purchased the assets of Focus Write in an effort to recoup his investment. Web-Est and Applied Computer Resources are the only two recent entrants who are still in the market, and they held collectively less than a 1% share of the Estimatics market in 2007. For contrast, Crash-writeR is used by fewer than 600 repair facilities whereas CCC’s communicating Estimatics product, Pathways, is used by more than 16,000. PX 406; PX 99-018(CCC). Thus, even a “successful” commercial entry may not be sufficiently successful to affect the analysis here.

The history of successful entrants is additionally clouded by the fact that the companies identified by the Defendants compete® only in the “low end market,” that is, the small, independent repair facilities that are not part of any DRP and that have traditionally been underserved or ignored by CCC, Mitchell, and Audatex. See PX 759 (Applied Computer Resources & Mitchell e-mail correspondence). These independent repair facilities do not require connectivity to insurers or the other add-on features offered by the large vendors and demanded by larger customers. See id. This distinction is reflected in the relative prices: Crash-writeR retails for $135 per month, PX 760, while Pathways lists at about [text redacted] per month, PX 1407(CCC).

Moreover, the last decade has seen a decrease in the number of Estimatics providers. In the 1990s, there were almost a dozen companies offering some form of Estimatics, including Dupont and Sherwin Williams. Carr Dep. at 20-21 (Motor). By the end of the 1990s, the only remaining Estimatics providers were CCC, Mitchell, Audatex, and Comp-Est. Id. at 20. The former competitors have either exited the market or been acquired by CCC, Mitchell, or Audatex. PX 25 ¶ 18 (Carr Decl.).

In the TLV market, Mitchell successfully entered the market in 2005, obtaining over a 4% share of the market by 2008, but that success followed two failed attempts, ten years of effort, and millions of dollars of investment. PX 1020 ¶ 99 (Hayes Prelim. Report). Additionally, Mitchell already had a large share of the Estimatics market. It could capitalize on its strong reputation and relationship with insurance companies, as well as its ability to offer connectivity between its Estimatics product and its new TLV product, in order to gain significant shares of the TLV market. There is no evidence that a firm without an Estimatics product has ever successfully entered the TLV market. This history suggests that barriers to entry in these markets are significant.

2. Defendants’ Reliance on Historic Barriers to Entry

Defendants have touted the historic barriers to entry into the Estimatics and TLV markets. Mitchell advertised the barriers to entry in countless financing and internal documents over the past few years. For instance, in a 2006 memorandum analyzing whether to purchase Mitchell, Aurora’s analysts stated that:

[Text redacted.]

PX 629-001, 004 (Aurora) (emphasis added). 26 Aurora identified a “variety of significant barriers to entry”: a proprietary auto physical damage database; a unique auto injury (medical) database; a large installed customer base; a small, low-growth industry with few primary competitors; and specialized regulatory knowledge of an editorial staff comprising [text redacted] professionals. PX 629-001-002. Separately, Aurora praised the “medium to high” switching costs from one of the Big Three to another and the “stable” competition between Mitchell, Audatex and CCC who have competed against each other for “a number of years.” PX 629-004. Mitchell itself has repeatedly cited the “significant barriers to entry” in its financing documents. See, e.g., PX 607-014 (Rating Agency Presentation) (“[significant barriers to entry through unique combination of data, software, communications and relationships”); Sun, Tr. (1/8 p.m.) at 42:12-17 (Mitchell); PX 583-027-028 (Mar. 2007 Confidential Info. Mem. for Senior Secured Credit Facilities) (citing customer network and experience, the Mitchell database, medical database, and high switching costs as significant barriers to entry); PX 560-027 (July 2004 Confidential Info. Memo, for Senior Secured Credit Facilities). Nor was this recitation of barriers saved only for external financiers. In Aurora’s 2007 “Annual Meeting Presentation” to its investors after the acquisition of Mitchell, Aurora listed several “Significant Barriers to Entry” in Mitchell’s businesses, including its “Customer Network and Experience,” “The Mitchell Database,” and customers’ “High Switching Costs.” PX 691-049 (Aurora).

Likewise, CCC and its private equity owner, Investcorp, have repeatedly noted high barriers to entry in these markets. Upon completing the acquisition of CCC, Investcorp issued a press release in which its Chief Operating Officer stated that “[t]his market ... has high barriers to entry.” PX 1401-001. CCC and Invest-corp also told rating agencies that there are high barriers to entry in these markets. See PX 161-011 (Draft Rating Agency Presentation) (“Key rating considerations [include] [v]ery high barriers to entry with two primary competitors”). 27 According to CCC/Investcorp, the “likelihood” of a new entrant into Estimatics is “very low” because of:

• high overall satisfaction with CCC customers, no customer demand for change

• Fragmented market: CCC penetrated in over 20,000 body shops

• Dual network effect with both insurers and body shops

• CCC software involves over a decade of trials and is built on complex algorithms

• Source data on labor for autobody shops difficult to compile.

PX 161-023. CCC/Investcorp described the likelihood of a new entrant into TLV as “low” because:

• CCC Total Loss product is based on bi-monthly surveys of approximately 4,000 dealerships and 2,000 publications

• Differentiated product that is a critical component to insurer decision making

• Generating accurate, up-to-date valuations requires significant resources

• Database built on inspected “steel”: unique in the industry

• Some states allow usage of book rates, which eliminates arguments with policyholders/claimants; however, most insurance customers will pay for the most accurate valuation

• Mitchell reportedly building product to compete with CCC and [Audatex], although no evidence that product would be a robust alternative. 28

PX 161-023. See also PX 1420-024(CCC) ([text redacted]). Audatex documents concur. PX 585-047 (Audatex).

3. Technical Barriers

In addition to this historical evidence of Defendants capitalizing on barriers to entry, the very nature of the products in question illustrate why this has been the case. Estimatics products consist of a parts and labor database, and software that interacts with the database to calculate the total cost of repair. PX 1020 ¶ 12 (Hayes Prelim. Report) (citing Ramamurthy IH Tr. (7/22/08) at 198-200). The difficulty and cost of developing and maintaining an entirely new parts and labor database that is accepted by the market would be significant barriers to new entrants. See, e.g., PX 691-049 (Aurora); PX 161-023(CCC). Such a database would need to cover at least 95% of all automobiles in the U.S. market, PX 28 ¶ 5 (Wilson Deck), and include parts data on all major makes and models of vehicles sold in North America, including “historical information for many years.” PX 25 ¶ 12 (Carr Decl.). 29 The database also would need to have comprehensive information on labor times and costs, which would involve an enormous number of expert-supervised time studies. PX 25 ¶ 13 (Carr Decl.). Furthermore, customer feedback on the parts and labor information is developed over a period of years. Linder IH Tr. at 62 (Mitchell); PX 25 ¶¶ 9, 12 (Carr Decl.) (“there likely would be a high error rate ... because [the entrant’s database] would not benefit from corrections ... accumulated over time based on customer feedback”). Thus, to create a credible database, a new entrant would need to regularly perform “highly resource-intensive tasks” such as “data revisions and time studies,” PX 7 ¶ 6 (Kostakis Decl.), by using parts data from the original equipment manufacturers, and continually update and refíne its data. PX ¶ 5 (Brown Decl.). 30

Mitchell spends over $ [text redacted] million annually, and Audatex $ [text redacted] million annually, just to maintain, update, and refine their databases. Brungger IH Tr. at 93 (Mitchell); PX 516-003 (Mitchell). CCC pays $ [text redacted] million annually for exclusive use of the Motor database. PX 86-234(CCC). It would take a number of years, untold thousands of man-hours, and millions of dollars of investment to create and maintain a competitive parts and labor database. See, e.g., PX 573-009 (Mitchell’s database “is unique and would require significant capital and development time to replicate”). But see PX 25 ¶¶ 14-15, 20 (Carr Decl.) (total sunk costs for building the Motor database was only about $ [text redacted] million). No firm has done so successfully since the mid-1990s.

Defendants argue, however, that this is a phantom barrier because new entrants will not be required to develop a new database after the merger. CCC has agreed that, upon consummation of the merger, it will relinquish its exclusive rights to license Hearst’s Motor database, allowing any competitor or entrant the opportunity to obtain immediate access to a comprehensive, fully updated database of parts and services. Ramamurthy, Tr. (1/22 a.m.) at 90-91, 91-92:8; DX 218 (CCC Letter to Motor). Defendants argue that the renewed open access to the Motor database is likely to generate additional new entry. The use of “existing” or “idle facilities” can facilitate entry into a market. FTC v. Occidental Petroleum Corp., No. 86-900, 1986 WL 952, at *8, 1986 U.S. Dist. LEXIS 26138, at *21 (D.D.C. Apr. 29, 1986) (reasoning that the presence of idle factories for the production of polyvinyl chloride resin — a thermoplastic resin derived from the chemicals ethylene and chlorine' — could reduce barriers to entry because “[fjirms can enter the market through means other than de novo plant construction in considerably less time than it would take to build an entirely new facility”). Here, access to the Motor database certainly could make the Estimatics market more appealing to potential entrants.

However, economic realities may prevent a new entrant from pursuing a license to use the Motor database. CCC current ly pays $ [text redacted] million per year for its exclusive license of the Motor database. PX 86-2S4(CCC). While a new entrant might be expected to pay significantly less for a non-exclusive license, that negotiated price remains unknown and unknowable. A price that approached the $ [text redacted] million range that Mitchell and Audatex spend annually to maintain their databases may be prohibitively expensive for new entrants who also must offer lower prices to attract customers. Because this is a low-growth industry with few new customers, see PX 574-002 (Mitchell), new entrants would have to win business from incumbents to generate sufficient revenues to remain a going concern for any length of time. It is unclear whether they could do so profitably, or if the projected profits would be sufficient to entice new entrants, given the market power that would be held by CCC/Mitchell and Audatex. See Merger Guidelines § 3.3 (entry considered “likely” if it would be profitable at pre-merger prices, and if such prices could be secured by the entrant). Notably, no potential new entrant has contacted Motor about the availability of its database since the merger was announced in April 2008. PX 25 ¶ 20 (Carr Deck). [Text redacted.]. Id. ¶22 (Carr Deck). 31

Despite these hurdles, it cannot be gainsaid that the release of the Motor database would significantly reduce the most critical barrier to entry in Estimatics. However, access to a complete and respected Estimatics database appears to be only one of many related barriers to entry.

As with Estimatics, a database for TLV also requires the compilation and integration of millions of data points collected from thousands of sources. For example, Audatex’s database draws on “approximately 110 million vehicle records” per year, and contains data from over 11,000 dealerships and over 3,500 other local market resources. PX 534-019; see also PX 1943-007, 009; PX 688-075. CCC has a “vehicle database of approximately 64,00 [sic] unique vehicles.” PX 1964-010(CCC). Both CCC’s and Mitchell’s TLV products include vehicle history reports by Vehicle Identification Number (VIN) to verify the condition of the vehicle. PX 81-032-036 (CCC Response to [text redacted] RFP); PX 116-035-038 (CCC Response to [text redacted] RFP); PX 643-011 (Mitchell). Thus, development of a TLV database poses similar challenges to those relating to creation of an Estimatics database.

Besides the database, the second main component of Estimatics and TLV products is the software that computes repair estimates or, in the case of TLV, vehicle valuations. With respect to Estimatics, development of complex, customizable, and integrated software is another significant barrier to entry. Firms must design custom interfaces that support specific communication protocols. See PX 25 ¶ 18 (Carr Deck); see also Carr Dep. at 23 (Motor) (features and functionalities of Estimatics software are important to customers). CCC’s Estimatics software “involves over a decade of trials and is built on complex algorithms.” PX 204-006. It can communicate via the Internet or a private network, PX 81-006 ([text redacted] RFP), and offers a patented “Compare” feature that automatically alerts users when alternative parts are available. PX 81-019(CCC). CCC began developing its Pathways product in 1993-94, but did not roll it out until 1996 even though it did not have to create its own database. Dickens IH Tr. at 115-117(CCC). Despite advances in computer technology, it is projected that developing new Estimatics soft ware would take one to two years. Linder IH Tr. at 147:12-14 (Mitchell); accord PX 680-093-094 (Aurora); Lukens Dep. at 95-97 (APU Solutions, Inc. (“APU”)) (Mitchell executive told an APU executive that the combined Mitchell-CCC platform would take 3-5 years to complete). 32 Focus Write developed a software interface over a period of about 18 months, PX 610-001, but its software “logic” was considerably improved once Focus Write was given access to Mitchell’s software. PX 545-003 (2006 Focus Write Letter).

Similarly, TLV software is also powered by complex algorithms that interact with the database and drive the TLV system. Sun IH Tr. at 59-60 (Mitchell); Sun, Tr. (1/8 p.m.) at 47:199 — 48:19 (Mitchell) (to penetrate the TLV market, Mitchell partnered with J.D. Power, its source for the core database and development of algorithms for the software). In a combined effort, WS and CCC spent eighteen to nineteen months developing a TLV software product for commercial trucks alone. Blitstein Dep. at 18-19(WS). CCC financed and provided the computer programs for the project. Id. at 17-18.

Although small repair facilities can use and perhaps may even prefer a basic Estimatics tool without all of the add-on programs, insurance companies and DRP repair facilities, which constitute the overwhelming majority of the market, require communicating products that offer connectivity between the insurance company and the repair facility. For these customers, the Estimatics product must be able to communicate the status of partial loss claims through the Internet or other specialized communication links. See PX 1020 ¶ 12 (Hayes Prelim. Report); see also PX 81-006 (CCC’s Estimatics product can communicate via the Internet or a private network). In addition, because many customers, particularly insurance companies, prefer to purchase their auto physical damage and TLV products in bundles from one vendor, 33 successful vendors must be able to offer a full portfolio of products. See PX 25 ¶ 18 (Carr Decl.). 34 This point is demonstrated by the business practices of the three major competitors: In addition to the core estimating products, the bundles offered by CCC, Mitchell and Audatex may include digital imaging, shop management, aftermarket parts data, workflow, audit tools, and total loss valuation, among others; they also may offer free or cut-rate training on the use of the products as well as variable pricing based on frequency of use or contract duration. Conway, Tr. (1/9) at 92:1-18 (Audatex); PX 680-007-008, 010 (Mitchell); PX 86-008-009(CCC); PX 244-062 (CCC’s Response to [text re dacted] RFP). Without a full suite of products to offer, a new firm would face enormous difficulties in challenging the established companies for business from insurance companies and DRPs.

Indeed, partially as a result of product integration and connectivity between repair facilities and insurance companies, CCC has achieved “[d]eep integration with customers [that] supports a 95% retention rate.” PX 99-011(CCC). This “stickiness” affects both insurance companies and repair facilities because if an insurance company is unwilling to switch, the repair shops that are part of its DRP typically will not switch either. This is because insurance companies either “mandate” that members of its DRP use the same Estimatics product used in-house or recommend (issue a “soft mandate”) that they use the same product. PX 531-002 (“Insurance DRP Mandates continue to drive market.”); PX554-036 (Mitchell). Thus, relationships with large insurance carriers are an important dynamic in the industry because such relationships are:

the leading driver of revenue as carriers are not only the estimatics vendors’ largest customers in terms of direct revenue, but also generate secondary revenue streams through repair shops that are affiliated with the insurance carriers’ DRP programs. These insurance carrier relationships are generally longstanding and difficult to break once established as the switching costs associated with retooling internal systems and forcing system turnover at the repair shop level are steep. In addition, the service provided by estimatics vendors are critical to the operation of insurance carriers’ claims resolution process and any disruption would have negative ramifications for the carrier’s operations.

PX 585-047 (Audatex). 35

Additionally, “integrated services, technological infrastructure in place, and long-term customer contracts all contribute to prohibitively high switching costs,” PX 583-028 (Mitchell), thus making it difficult for a new entrant to gain market share from an incumbent vendor. See PX 543-016 (Mitchell). Furthermore, “[s]witching vendors is a long, difficult process, often taking as much as 12 to 18 months to fully transition to a new provider.” PX 585-029 (Audatex); PX 574-003, 016 (Mitchell) (A customer switching vendors “would be forced to retrain its employees, which can take as long as 3^4 days per employee for an insurance carrier, and make a significant capital expenditure to switch out [the] embedded technological infrastructure”). These disincentives to switching have resulted in an incumbent winning percentage of close to 90%. PX 574-003 (Mitchell); PX 629-003 (Mitchell); Conway, Tr. (1/9) at 121:4-5 (Audatex); Ramamurthy Dep. at 47(CCC); Hayes, Tr. (1/12 p.m.) at 161:25-162:11. See also PX 668-021 (Mitchell) (“Particularly among Top 25, Core Vendors are entrenched — switching rare”).

Furthermore, on this preliminary record, it appears that reputation, scale, and relationships also serve as lesser barriers to entry. The significance of reputational barriers to entry in antitrust analysis is a somewhat unsettled question. See Tote, 768 F.Supp. at 1075. Reputation can be a considerable barrier to entry where customers and suppliers emphasize the im portance of reputation and expertise. See Chi. Bridge, 534 F.3d at 437-38 & n. 17 (while “general reputation” alone is not an effective barrier to entry, reputation for “industry-specific traits” of expertise and experience was considered significant in the storage tank-building industry); Cardinal Health, 12 F.Supp.2d at 57 (reasoning that the strength of reputation of the merging companies over smaller companies in the pharmaceutical wholesale market served as barriers to growth of smaller competitors); Tote, 768 F.Supp. at 1075 (stating that the lack of a record of demonstrated performance served as a significant barrier to entry in the totalisator market, an industry that creates systems that support a common form of wagering at racetracks). But see Waste Mgmt, 743 F.2d at 984 (“[W]e fail to see how the existence of good will achieved through effective service is an impediment to, rather than the natural result of, competition.”); United States v. Consol. Foods Corp., 455 F.Supp. 108, 119 (E.D.Pa.1978) (“Particularly in selling to institutional customers brand names are not significant.”).

The record illustrates that reputation for experience and scale are legitimate barriers to entry in the Estimatics and TLV markets. As a precondition to bidding, most insurance companies require multiple years of audited financial statements showing financial stability, and multiple references from other Estimatics customers. PX 94-016 (CCC Response to [text redacted] RFP); PX 95 at 5 (CCC Response to [text redacted] RFP); PX 687-07, 026, 027 ([text redacted] RFP); PX 118-065-066, 067 (CCC Response to [text redacted] RFP); PX 116-008 (CCC Response to [text redacted] RFP); PX 117-004, 005, 028 (CCC Response to [text redacted] RFP); PX 686-006 ([text redacted] RFP). Insurance customers demand a “strong [s]hop footprint ... [,] a proven management team that is dedicated to customer service ... [, and] a proven Estimatics product with a clear roadmap for future enhancement and options.” PX204-007 (Audatex). Scale is important in this industry, in part, because a vendor has to be large enough to afford the continuous research and development necessary to be competitive in Estimatics and TLV. Sun Dep. at 101 (Mitchell); see also Daly Dep. at 76-77 (Allstate) (“I would suspect that some of these smaller companies simply didn’t have the track record or resources to the extent that we felt comfortable partnering with them on such a ... significant initiative and product.”). In addition, insurance customers place great importance on extensive customer support capabilities and may require Estimatics suppliers to affirm that they have a single point of contact available 24 hours a day, seven days a week. See, e.g., PX 118-009 (CCC Response to [text redacted] RFP).

On these factors, the Big Three have an enormous advantage. CCC, Mitchell, and Audatex are enormous companies in comparison to their remaining competitors, much less a new entrant. Mitchell has over 1,400 employees, Sun Dep. at 17 (Mitchell), of which [text redacted] to [text redacted] are dedicated to insurance customer support, Sun, Tr. (1/9) at 75:4-8 (Mitchell). It has [text redacted] to [text redacted] people across the country in its collision sales department, Sun Dep. at 39, and an editorial staff of [text redacted] to maintain its database, PX 572-026 (Mitchell). Mitchell has a technical assistance center staffed from 5 a.m. to 5 p.m. and offers the option of emergency technical assistance 24 hours a day, seven days a week. PX 1958-043 (Mitchell’s Response to [text redacted] RFP); see also PX 681-003 ([text redacted]). CCC has over 834 full-time employees, including 225 product development personnel, and 36 client consultants who are scattered throughout the country and are responsible for training, customer service, and technical support. See PX 94-012, 019 (CCC Response to [text redacted] RFP). CCC also has over 120 field employees able to service insurance customers, PX 95-022 (CCC Response to [text redacted] RFP), 65 account managers and field support personnel covering all 50 states in its Insurance Services Division, and 60 representatives in its Automotive Services Group. PX 116-004 (CCC Response to [text redacted] RFP).

In terms of experience, Mitchell’s Executive Vice President for Auto Physical Damage, Marc Brungger, has been with the Company for 20 years. PX 691-046 (Aurora). “Mitchell also has an editorial staff with an average of over 17 years of industry experience.” PX 573-009 (Mitchell). CCC’s management team has an “[a]verage of 10+ years of industry and company experience.” PX 99-011 (CCC Lenders’ Presentation). In addition, given the number of individuals who have defected to a competitor and the relative dearth of novices in this industry, see, e.g., Conway, Tr. (1/9 a.m.) at 91:21-92:24 (Conway worked at Mitchell before moving to Audatex); PX562-33 (Tony Aquila was employed by Mitchell before leaving for Audatex); DX 30-8 (Eric Seidel was a former investor in Focus Write), the evidence suggests that there are a limited number of people who have significant experience on the seller side of this industry.

Relationships also play a significant role in insurance companies’ and repair facilities’ choice of vendor. CCC, Audatex and Mitchell have a large head start over the recent and potential new entrants in establishing customer relationships. Through its 60 year history of providing estimating solutions, Mitchell has established an extensive network of thousands of insurance staff and collision repair facility users that would be difficult to replicate. See PX 573-009 (“Mitchell has a long-standing and deep relationship with over [text redacted] insurance companies and over [text redacted] collision body shops through its 60 year history.”). Likewise, “CCC has a highly diversified, stable customer base” and “strong, long-term customer relationships with a 95% customer retention rate.” PX 99-011, 007(CCC).

Finally, the return on investment in the Estimatics and TLV markets may make entry of new firms unlikely. The cost of creating the software (and potentially a database) are up-front sunk costs. The ability of new entrants to recoup those costs may be limited by factors that define these markets: (1) Estimatics and TLV markets are “very mature with not a lot of room for growth,” PX 543-005 (Mitchell) (emphasis in original); (2) customers do not frequently switch firms, PX 618-027 (Mitchell); PX 161-007, Oll(CCC); PX 543-016 (Mitchell); PX 574-003 (Mitchell); PX 583-026 (Mitchell); and (3) insurance companies enter long-term contracts for Estimatics and TLV, so only about a third of insurers are even available for bid in a given year, PX 583-026 (Mitchell). These factors led Mitchell to conclude that the “small revenue opportunity” in Estimatics “should deter any potential new competitor from making the considerable capital investment it would require to develop the database and communications infrastructure necessary to compete in this space.” PX 629-002 (Mitchell). Further consolidation of these markets is likely to make potential new entrants even more wary.

4. Web-Est: The “Fix” To Barriers to Entry ?

Defendants rely heavily on the potential future of Web-Est to counter the FTC’s case on barriers to entry. Web-Est was formed in March 2008 with Mitchell’s financial backing around the same time CCC and Mitchell announced the merger; Mitchell holds a [text redacted] % interest in Web-Est. Seidel, Tr. (1/22 p.m.) at 173:11-15 (Web-Est). Web-Est now offers a Web-based non-communicating estimating product to low-end repair facilities. Seidel, Tr. (1/22 p.m.) at 165:18-22, 167:7-8 (Web-Est). 36 It does not currently participate in the TLV market. Web-Est’s Estimatics system is based, in part, on a license permitting it to use Mitchell’s parts database, but the license agreement contains a restrictive covenant that prohibits Web-Est from selling to the top fifty insurance companies, from selling communicating Estimatics products to insurance companies or repair facilities, and from integrating with other third-party service providers, such as parts locators, salvage, and other providers. See DX 59; Seidel, Tr. (1/22 p.m.) at 173:24-174:13 (Web-Est).

Facing FTC resistence to the merger, Mitchell followed suit to CCC’s release of its exclusive hold on the Motor database. Mitchell and Web-Est have entered into a new licensing agreement which will become effective as soon as the merger closes. Seidel, Tr. (1/22 p.m.) at 175:20-23, 177:12-19 (Web-Est); DX 423 (Mitchell). Pursuant to the new agreement, Mitchell will remove virtually all of the restrictions on Web-Est’s rights to sell to insurance companies and repair facilities. Seidel, Tr. (1/22 p.m.) at 175:20-25 (Web-Est); DX 423 (Mitchell).

[Text redacted.]

The new license agreement also resets the clock on Web-Est’s initial five year license to the Mitchell database, and gives Web-Est an option to renew for an additional five years. Seidel, Tr. (1/22 p.m.) at 175:9-14 (Web-Est); DX 423 (Mitchell). Web-Est also will be permitted to use other databases if it wishes, and can integrate with other vendors’ products. Id. Furthermore, if the merger closes, Mitchell will relinquish its equity in Web-Est as well as its purchase option for Web-Est. Id.-, Sun Tr. (1/9 p.m.) at 65:11-25 (Mitchell).

[Text redacted.]

The ability and willingness of current competitors to expand their foothold in the market and/or reposition greatly reduces the anticompetitive effects of a merger, and is essentially equivalent to new entry. See Arch Coal, 329 F.Supp.2d at 148 (“Defendants have shown that the post-merger fringe capacity in the [market] would be more than sufficient to absorb any increase in demand caused by any production lag coordinated by the ‘big three’ producers .... ”). Defendants argue that the release of these restrictions will shatter the only major barrier to entry, will enable Web-Est to expand its business to new heights, and will usher in a new era of dynamic competition in these markets. They offer numerous points of evidence in support of their conclusion.

Defendants contend that Web-Est’s web-based product will make it a “game changer” in the industry. Seidel, Tr. (1/22 p.m.) at 193:11-21 (Web-Est). They argue that Audatex’s experiences in the U.S. market confirm that the concept of web-based products is gaining ground in the auto physical damages industry. Cheskis, Tr. (1/23 a.m.) at 11:3-23 (Gerber uses Audatex’s web-based product in sixteen facilities in the United States); DX 225-33-35 (Audatex [text redacted]). Additionally, Web-Est already has a communicating Estimatics product embedded in its non-communicating product which it can turn on almost immediately, Seidel, Tr. (1/22 p.m.) at 176:1-3, 186:19-187:3 (Web-Est); is in the process of developing [text redacted], id. at 188:10-189:12, thus eliminating the barrier of product portfolio in Estimatics and TLV.

Defendants suggest that these plans have already begun to reap significant rewards. Since its formation last spring, Web-Est has more than doubled its initial customer base; it now sells Estimatics to about 11% of its addressable market. Id. at 165:23-25, 174:14-21. 37 Looking forward, Eric Seidel plans an aggressive price-cutting strategy and forecasts sales to DRPs and insurers within the first year and annual revenues of $14.6 million within five years. DX 30 at 6-7, 50 (Web-Est Oct. 2008 Business Plan). Mr. Seidel expects to add 5,000 customers within five years, including three insurance carriers within the next year. Id. at 5, 28.

[Text redacted.]

Although Defendants acknowledge that Web-Est is not nearly as large as Audatex, CCC or Mitchell, they assert that Web-Est does not plan “to go out and try to replicate a CCC, a Mitchell, Audatex,” because they are “brick and mortar” Estimatics firms. Id. at 193:11-21. Defendants argue that Web-Est believes it will become the Estimatics industry’s version of Netflix which upended Blockbuster and revolutionized the movie rental business. Id. at 193:22-194:19.

Mr. Seidel, who testified at the preliminary injunction hearing, is an impressive man and the Court has no doubt that he has the entrepreneurship, experience, and drive that bode well for Web-Est’s future. Then again, of course, Mr. Seidel is wholeheartedly behind the proposed merger because it would free his hands to compete as he wishes. Nonetheless, the Court cannot ignore that Web-Est is a ten to twelve person company with current projected annual revenues of $ [text redacted] million (assuming the merger is effected). Id. at 218:14-19; PX 1003-006-007, 033 (Web-Est). It will be an ant to an elephant, compared to a post-merger CCC/Mitchell, at $450 million in revenues and over 2,000 employees. See PX 529-006 (transaction valued at approximately $1.4 billion and CCC/Mitchell expects to earn $460 million in annual sales at inception); see also Seidel, Tr. (1/22 p.m.) at 205:3-6 ([text redacted] ). Assuming that web-based products will succeed (society is increasingly using the web, including the courts for electronic case filing), insurance companies will still demand customer support services and suppliers still will need large sales forces to woo them. The growth curve to take Web-Est from where it is now to where it would need to be to compete with CCC/Mitchell seriously enough to cause price competition is extraordinarily steep and inevitably long. That journey is made more difficult by the fact that Mitchell will no longer lend its expertise to Web-Est, nor will it hire out its staff to Web-Est. Sun, Tr. (1/8 p.m.) at 13:5-9 (Mitchell).

Moreover, CCC, Mitchell and Audatex have been refining their Estimatics software for over a decade. See PX 204-006(CCC). In discussing whether entry is “sufficient,” the Merger Guidelines suggest that:

where the concern is unilateral price elevation as a result of a merger between producers of differentiated products, entry, in order to be sufficient, must involve a product so close to the products of the merging firms that the merged firm will be unable to internalize enough of the sales loss due to the price rise, rendering the price increase unprofitable.

Merger Guidelines § 3.4. From this record, it is simply unknown how many insurance companies and DRP repair facilities will find Web-Est’s communicating Estimatics product to be up to par, or at what pace.

And, Web-Est cannot be considered a truly independent actor because Mitchell will continue to be so involved in its business. In order to be accepted, “curative divestitures” must be made to a new competitor that is “in fact ... a willing, independent competitor capable of effective production in the ... market.” White Consol. Indus. v. Whirlpool Corp., 781 F.2d 1224, 1228 (6th Cir.1986) (emphasis added). Mitchell’s counsel has similarly observed that it is a “problem” to allow “continuing relationships between the seller and buyer of divested assets after divestiture, such as a supply arrangement or technical assistance requirement, which may increase the buyer’s vulnerability to the seller’s behavior.” Richard G. Parker & David A. Balto, Evolving Approach to Merger Remedies, Antitrust Report (May 2000), available at http://www.ftc.gov/ speeches/other/remedies.shtm (citing FTC, A Study of the Comm’n’s Divestiture Process (1999), available at http://www.ftc.gov/ os/1999/08/divestiture.pdf). Although Mitchell will relinquish its financial interests in Web-Est upon consummation of the merger, Web-Est will continue to license Mitchell’s database for at least five years, with a continuing option for five more.

[Text redacted.]

At its rosiest, Web-Est’s most optimistic financial projection predicts annual revenues of approximately $ [text redacted] million by 2013 (of which $ [text redacted] million would come from insurance companies). PX 1003-033 (Web-Est); Seidel, Tr. (1/22 p.m.) at 198:19-23, 202:25-205:2 (Web-Est). This is admirable growth. By contrast, however, in 2007, Mitchell had $ [text redacted] million in total revenue and an EBITDA 38 of $ [text redacted] million. PX 530-004 (Mitchell). CCC had $ [text redacted] million in total revenue and an EBITDA of $ [text redacted] million. Id.; PX 86-034(CCC) (also estimating Solera’s global revenues in 2007 at $ [text redacted] million). A merged CCC/Mitchell would be even bigger. According to the Merger Guidelines, entry is “timely” if it can be achieved “within two years from initial planning to significant market impact.” Merger Guidelines § 3.2. (“significant market impact” means “significant impact on price in the relevant market”). While two years may be a short time frame by which to judge successful entry in this industry, even within five years, Web-Est hopes to make hardly a splash compared to Mitchell’s current market share. Whatever Web-Est’s plans and aspirations for the future may be, it is very unlikely to be able to compete effectively, i.e., affect pricing, within five years or even soon thereafter. See In re Chi. Bridge & Iron Co., 138 F.T.C. 1024, 1071 (2005) (“the mere fact that new entrants and fringe firms have an intent to compete does not necessarily mean that those firms are significant competitors capable of replacing lost competition”).

5. Predictive Analytics

Predictive analytics, an internal method of calculating future estimates based on an insurance company’s own empirical data, Marushka Dep. at 14-15(CCC), may supplant the Estimatics software sold by CCC, Mitchell and Audatex one day, and the Defendants hint that it may be sooner rather than later when they suggest that the merger of two outdated platforms cannot hurt anyone. It is acknowledged by CCC, however, that there is “nothing in production ... that we’ve actually seen” that might replace today’s Estimatics. Ramamurthy IH Tr. at 203:13-19(CCC). CCC’s General Manager, Insurance Services Group, stated that no insurance company currently uses predictive analytics to calculate partial loss estimates and he does not know whether predictive analytics are allowed under insurance regulations. Marushka Dep. at 15(CCC). In other words, while “there has been talk about ... it,” Ramamurthy IH Tr. at 203(CCC), no insurance companies have implemented it. See Marushka Dep. at 15. The Court finds this evidence of new entry too speculative to rely upon. 39

b. Coordinated Effects

Merger law “rests upon the theory that, where rivals are few, firms will be able to coordinate their behavior, either by overt collusion or implicit understanding in order to restrict output and achieve profits above competitive levels.” Heinz, 246 F.3d at 715 (citing PPG Indus., 798 F.2d at 1503). The theory follows that, absent extraordinary circumstances, a merger that results in an increase in concentration above certain levels “raise[s] a likelihood of ‘interdependent anticompetitive conduct.’ ” PPG Indus., 798 F.2d at 1503 (quoting Gen. Dynamics, 415 U.S. at 497, 94 S.Ct. 1186); see also FTC v. Univ. Health, Inc., 938 F.2d 1206, 1218 n. 24 (11th Cir.1991) (high concentration makes it “easier for firms in the market to collude, expressly or tacitly, and thereby force price above or farther above the competitive level”); Elders Grain, 868 F.2d at 905. Successful coordinated interaction entails “[1] reaching terms of coordination [2] that are profitable to the firms involved and [3] an ability to detect and punish deviations that would undermine the coordinated interaction.” Merger Guidelines § 2.1. Whether a merger will make coordinated interaction more likely depends “on whether market conditions, on the whole, are conducive to reaching terms of coordination and detecting and punishing deviations from those terms.” Id. Because the FTC has established a prima facie case, the burden is on the Defendants to demonstrate “structural barriers,” unique to this industry, that are sufficient to defeat the “ordinary presumption of collusion” that attaches to a merger in a highly concentrated market. Heinz, 246 F.3d at 725.

“The combination of a concentrated market and barriers to entry is a recipe for price coordination.” Id. at 724. Yet, despite the high HHI numbers currently displayed in these markets, by all accounts, Estimatics and TLV are highly competitive markets today and there is no evidence of past coordination. Danford Dep. at 120 (Ohio Mutual); Brandt, Tr. (1/8 a.m.) at 17:23-18:7 (The Hartford); Cheskis, Tr. (1/23 a.m.) at 10:12-23 (Gerber); Aquila Dep. at 15 (Audatex); Conway, Tr. (1/9) at 169:10-21 (Audatex). Compare with Elders Grain, 868 F.2d at 905 (“there is a history of efforts to fix prices in the industry”); Hosp. Corp. of Am. v. FTC, 807 F.2d 1381, 1388 (7th Cir.1986) (“there is a tradition ... of cooperation between competing hospitals in Chattanooga”). “While proof of prior cooperative behavior is relevant, it is not a necessary element of likely future coordination in violation of Section 7.” Arch Coal, 329 F.Supp.2d at 116. But Defendants do not rest there. Rather, they argue that these markets provide a perfect example of the HHI’s failure to measure anticompetitive effects because, despite their high concentration, “[t]he un disputed market realities here present a perfect storm of factors that impede coordination.” Defs.’ Posh-Trial Brief at 16. According to Defendants, those factors include: (1) product heterogeneity; (2) lack of price transparency; (3) complexity and lack of standardization with respect to pricing and products; (4) firm heterogeneity; (5) large, infrequent contracts; (6) high fixed costs relative to variable costs; and (7) the presence of sophisticated buyers. See Ordover, Tr. (1/23 a.m.) at 90:1— 92:5.

1. Characteristics of the Products and Competitors

“[Rjeaching terms of coordination may be limited or impeded by product heterogeneity.” Merger Guidelines § 2.11; see also, e.g., Hosp. Corp. of Am., 807 F.2d at 1390. “Estimatics software is differentiated along a number of important dimensions,” including “accuracy and coverage of the underlying data for parts and estimated labor times, perceived differences in the ease of use of the competing products, and the quality and timeliness of technical support services.” PX 1020 ¶ 71 (Hayes Prelim. Report). Dr. Ordover opined that on a scale of zero to ten, these products are closer to a ten in terms of heterogeneity. Ordover, Tr. (1/23 a.m.) at 85:2-11. Insurance companies demand customized Estimatics and TLV products tailored to fit their individual needs. Ramamurthy, Tr. (1/22 a.m.) at 99:8-100:2(CCC). Some insurers even require that bidders commit to spending $500,000 to $600,000 per year to develop software unique to that customer. Id.

In addition to product heterogeneity, coordination may be impeded by a lack of “standardization of pricing or product variables on which firms could compete.” Merger Guidelines § 2.11. The base Estimatics product is often sold in highly customized bundles that include other auto physical repair products and various add-on products to suit the particular requirements of the customer, thus resulting in significant product variability. Ramamurthy, Tr. (1/22 a.m.) at 88:3-7(CCC); Conway, Tr. (1/9) at 93:24-95:5 (Audatex) (“With the number of products that [Audatex] offer[s] [to insurers], you could come up with thousands of different combinations of those products.”); id. at 111:9-13 (“On the [repair] shop side with the number of different add on products,” Audatex could offer “hundreds” of different bundles). Along with the base estimating product, the bundles can be comprised of any number or combination of other products and add-ons, as well as varying levels of installation support, customer service, training and integration. Rollins Dep. at 17 (Safeco); Kostakis Dep. 36, 37-39 (Angelo’s Auto Body Shop); see also DX 98 (CCC Price Sheet); DX 47 (Mitchell’s August 2007 price list for repair facilities is thirty-one pages long, consisting of almost 400 product and pricing options). These customized bundles have highly complex and varying pricing metrics including flat fees for the entire bundle, individual product pricing, and almost everything in between. See, e.g., DX 243-10-21 (CCC Pathways Services Agreement).

Defendants also assert that the likelihood of coordination in these markets is further diminished by the heterogeneity exhibited among the firms in these markets. “[R]eaching terms of coordination may be limited or impeded by firm heterogeneity, for example, ... the production of another product that tends to be used together with the relevant product.” Merger Guidelines § 2.11. Defendants argue that because the sale of add-ons and other products in conjunction with Estimatics and TLV is an increasingly important area of focus and source of revenue growth, “the merged firm’s unique interest in enhancing sales of such related offerings further impedes any interest in coordinated interaction.” Defs.’ Post-Trial Brief at 20. They point to Mitchell’s unique medical bill-review services as an example. See PX 583-027-028 (Mitchell); PX560-031032 (Mitchell).

While Defendants’ arguments do have force, particularly with regard to the insurance industry, their points are weakened by their own admissions that there is a “minimum differentiation in offerings,” PX 1420-007(CCC); see also PX 253-010(CCC). “Generally speaking there are ... three or four [bundles] that are purchased consistently from one insurance company to the next in most all of the bids.” Conway, Tr. (1/9) at 94:9-11 (Audatex). The “standard package” includes the basic estimating system, the imaging systems, an aftermarket products tool and communication. Id. at 94:12-18. While it is true that Mitchell is the only one of the Big Three firms that offers a medical bill-review product, with limited other exceptions, CCC, Mitchell and Audatex have essentially the same suite of product offerings and add-ons, and are therefore unable to separate themselves from the pack on a consistent basis through unique sets of products. See PX 1420-007 (CCC document showing the different add-ons in Estimatics) ([text redacted]); PX 253-010(CCC) (noting that [text redacted]). But see PX 550-022 (Mitchell) (“Continue to look for points of differentiation”); see also Carr Dep. at 23 (Motor).

2. Dynamics of the Marketplace

In addition to product and firm heterogeneity and variability, Defendants argue that the unique marketplace dynamics make coordination unlikely. First, Defendants assert that the lack of pricing transparency for Estimatics and TLV is a significant impediment to coordination. “Reaching terms of coordination may be limited or impeded ... by firms having substantially incomplete information about the conditions and prospects of their rivals’ businesses,” Merger Guidelines § 2.11, and the ability to detect deviations from the terms of coordination is limited “if key information about specific transactions or individual price or output levels is [not] available routinely to competitors.” Id. § 2.12. See also Oracle, 331 F.Supp.2d at 1166 (“Without homogeneity [of product offerings] or transparency [in pricing], the market conditions are not conducive to coordinated effects, either tacit or express.”). The bidding process for insurance companies is conducted on a confidential basis, and bids are usually accompanied by a non-disclosure agreement. Sun, Tr. (1/9) at 46:22-47:4 (Mitchell); Brandt, Tr. (1/8 a.m.) at 23:19-22 (The Hartford). Though these competitors go to great lengths to obtain “intelligence” about each other’s pricing, Sun, Tr. (1/8 p.m.) at 55:13-58:21 (Mitchell); Sun, Tr. (1/9) at 75:20-77:5 (Mitchell), and the record contains instances where one vendor has obtained partial bids, price lists, or other pricing information about their competitors, e.g., PX 85-059(CCC); PX 506 (Mitchell on Audatex pricing); PX 623 (Mitchell), PX 696 (Mitchell); PX 713 (Mitchell on Audatex pricing); PX 714 (Mitchell on Audatex pricing); PX 715 (Mitchell on CCC pricing); PX 716 (Mitchell on CCC pricing), pricing for insurance company contracts is not “routinely available.” 40 Furthermore, many times when CCC, Audatex or Mitchell ob tain so-called competitive information on pricing, the information is either misleading or simply false (presumably to force a lower bid). See, e.g., DX 425-1 ([text redacted] ); Conway, Tr. (1/12 a.m.) at 28:14-19 (Audatex “expect[s]” that customers have provided inaccurate information on occasion with the hopes of driving down Audatex’s prices).

The FTC contends that CCC, Mitchell and Audatex obtain a substantial amount of information after a bidding process, which they can use to set future prices. See Sun, Tr. (1/8 p.m.) at 55:6-9 (Mitchell); see also PX 172-002 (CCC e-mail); PX 598-001 (Mitchell [text redacted]); PX 179-003(CCC). Given the degree of product variability in the industry, past pricing information may be suggestive but not necessarily helpful for the next bid, particularly a bid for the same insurer because contracts with insurance companies can last from two to five years, during which time prices have been shown to alter significantly. See Arch Coal, 329 F.Supp.2d at 141 (“It is true that industry publications make some market information available among producers. However, the information published in those sources is limited, imperfect, and largely unreliable and untimely.”).

Pricing is not as obscure in the repair facility segment. Like insurance company bids, repair facility contracts are confidential. See, e.g., Sun, Tr. (1/9) at 53:19-54:10 (Mitchell); Ramamurthy, Tr. (1/22 a.m.) at 108:23-25(CCC); Conway, Tr. (1/12 a.m.) at 39:8-17 (Audatex). Repair facility contracts also often consist of bundled products, including the base Estimatics product and a variety of add-ons. Conway, Tr. (1/9) at 111:9-113:2 (Audatex). However, despite confidential repair facility contracts and product bundling, Estimatics pricing for the repair shop segment is “transparent.” Seidel, Tr. (1/22 p.m.) at 227:9-14, 226:2-24 (Web-Est) (stating that it is possible to get the price points of the Big Three’s Estimatics products by calling up shops and asking what they are paying); see also, e.g., PX 179-003(CCC); PX 173 (CCC [text redacted]); PX 742-008 (Mitchell began a matching program in December of 2008 in which it offered to match CCC and Audatex on the condition that Mitchell had the “competitive quote faxed in”).

Likewise, TLV prices are more transparent than Estimatics prices for insurance companies. Because TLV is-sold on a per valuation basis, there is only one price to track. See PX 172-002 (CCC email) (“Mitchell is already in [the] door with estimating and is offering total loss @ $ [text redacted]”); PX 722 (Mitchell email) (Mitchell agreed on $ [text redacted] per valuation for a certain insurer “in part because CCC came in at $ [text redacted].”). However, TLV prices are becoming less transparent as vendors turn increasingly toward the “one throat to choke concept,” i.e., bundling, whereby they purchase Estimatics, TLV and add-ons from a single source. See PX 543-019-020 (Mitchell); PX 550-025 (Mitchell); PX 1432-003(CCC); PX 1402 (CCC web-page promoting the CCC One total repair platform). Even when TLV is sold separately, these three companies still have differing price metrics based on a variety of factors including, inter alia, the relationship with the customer and the size of the order. See DX 186-26 (CCC contract with [text redacted]) (per valuation price, plus surcharge for certain methods of ordering, as well as additional fees for different vehicles); DX 187-10 (CCC contract with [text redacted]) (per valuation price for cars, additional for trucks, with separate minimum purchase requirements for both).

On balance, as Dr. Ordover suggests, the pricing information in these markets is best described as “shrouded,” Ordover, Tr. (1/23 a.m.) at 88:8-22, ie., neither as transparent as the FTC would wish nor as secret as the Defendants would now prefer.

Defendants als'o contend that long-term, high-value Estimatics and TLV contracts in the insurance industry lower the incentives to coordinate and increase the incentives for cheating. See id. at 85:15-23. Insurance contracts typically range from two to five years, and can be priced anywhere from a few thousand dollars a year for a small insurer to several million dollars a year for a large insurer. Conway, Tr. (1/9) at 96:16-20, 94:22-95:4 (Audatex); see, e.g., DX 252-4, 53 (five-year contract with CCC). Defendants and their expert assert that the insurance contracts, particularly for Tier One insurers, are just too valuable for any one of the three competitors to agree to forego, whether implicitly or explicitly. The Merger Guidelines provide that “[wjhere large buyers likely would engage in long-term contracting, so that the sale covered by such contracts can be large relative to the total output of a firm in the market, firms may have the incentive to deviate” from the terms of coordination. Merger Guidelines § 2.12; see also United States v. Archer-Daniels-Midland Co., 781 F.Supp. 1400, 1416, 1423 (S.D.Iowa) (finding coordination unlikely where transactions in the market were “relatively large” and “infrequent”). And, although individual repair facility contracts are not particularly large, Defendants argue that because CCC, Audatex and Mitchell earn large portions of their revenues from the sale of non-Estimatics products, they have an incentive to deviate from any potential terms of coordination in order to sell additional products.

In a similar vein, Defendants argue that the bargaining power and sophistication of their insurance company customers further impede coordination. A sophisticated customer base makes price coordination more difficult, see Baker Hughes, 908 F.2d at 986 (sophisticated “buyers closely examine available options and typically insist on receiving multiple, confidential bids for each order”), but buyer power is greater when there are few buyers in the market. See Elders Grain, 868 F.2d at 908 (“A concentrated and knowledgeable buying side makes collusion by sellers more difficult.”). Although the top fifty insurance companies account for approximately 80% of the total revenues in physical damage claims market, PX 543-019, the insurance market as a whole is not concentrated. The HHI for the nineteen largest insurers (ie., the only insurers with a greater than 1% market share), who together represent 76% of the entire insurance market, is only 631 — not very concentrated. See PX 515— 028; see also PFF ¶ 197. However, it is true that the larger automobile insurance companies have enough buying power to demand customized products and to use their leverage regularly to keep Estimatics prices low for repair facilities in their DRP networks. See Cheskis, Tr. (1/23 a.m.) at 13:16-14:13 (Gerber); see also Hall, Tr. (1/8 a.m.) at 94:15-24 (GMAC); Ramamurthy, Tr. (1/22 a.m.) at 110:15-111:6(CCC). Collectively, the heterogeneity of the base products and customized bundling, the largely confidential pricing, and the high-value insurance contracts tend to make tacit coordination less likely than the huge HHIs might predict.

But this conclusion is not the end of the analysis. Defendants ignore a number of other factors present in these markets that would tend to confirm the HHI’s predictions regarding the likelihood of coordination. Estimatics and TLV are “stable” markets in which the same three companies have been competing against each other for over a decade, making the market participants very familiar with each other. PX 583-024 (Mitchell); Sun, Tr. (1/8 p.m.) at 20:11-22 (Mitchell); see also PX 574-002 (Aurora); PX 554-054, 055 (Mitchell) (“the industry will not dramatically change over the next 5-10 years”); PX 1970-027 (CCC-Mitchell) (“The $40 billion auto damage claims industry is large and stable____”). Defendants point to a lack of price transparency as the “[k]ey information” that is not “available routinely.” See Merger Guidelines §§ 2.11 & 2.12. But pricing is not the only “key information” that is contemplated by the Guidelines. Indeed, the Guidelines state that if “key information about specific transactions or individual price or output levels is available routinely to competitors, it may be difficult for a firm to deviate secretly.” Id. § 2.12 (emphasis added). These three firms know what products each offers, what insurance accounts each has won, and, often, the identities of the final two bidders for an insurance company contract. See, e.g., Balbirer IH Tr. at 73(CCC) (“Obviously we know what their products are, they know what our products are, and we’re in the marketplace. They know who we sell to and we know who they sell to.”). In particular, each of these three companies knows which one of them sells to which of the top twenty-five insurance companies. PX 543-011-012 (Mitchell) (identifying top twenty-five insurers, who won the contract, and claims volume). CCC, Audatex and Mitchell have sought to “gather as much competitive information [about each other] as possible” for a number of years, Sun, Tr. (1/8 p.m.) at 57:17-58:1 (Mitchell), and have largely succeeded in their efforts.

These competitive markets are stable in part because they are very mature and have little room for growth. PX 543-005 (Mitchell); see also PX 1970-033 (CCC-Mitchell). 41 Estimatics is “90%+ saturated,” making it a “game [of] market share take-away.” PX 630-017. But market shares have also remained stable over the last several years. See PX 583-024 (Mitchell); PX 543-020 (Mitchell) (“not a lot of switching going on”). “Taking [m]arket [s]hare is [d]ifficult” in this industry, largely because of the high switching costs and the time associated with switching products. PX 543-016 (Mitchell); see also PX 634-002 (Mitchell e-mail) (cost of switching generally is $ [text redacted] per user); PX 585-029 (Audatex); PX 574-003, 016 (Mitchell). Because of the difficulty of gaining market share, CCC, Audatex and Mitchell have all shifted their focus toward selling more products to their existing customers rather than engaging in price wars over each other’s customers. See PX 543-019, 020; PX 550-024-025, 030; see also [text redacted]; PX 632-034. This can lead to even greater stabilization of market share and greater segmentation of the market, thus increasing the incentives and lowering the impediments to tacit coordination. See Merger Guidelines § 2.11 (“Firms coordinating their interactions need not reach complex terms concerning the allocation of the market output across firms or the level of the market prices but may, instead, follow simple terms such as ... stable market shares, or customer or territorial restrictions.”).

Nor are high switching costs the only characteristics that make these markets conducive to tacit coordination in the form of market stabilization or customer allocation. Product heterogeneity, which in some ways can reduce the likelihood of coordination, can also lead to greater segmentation of the market and more entrenched market shares. 42 Decreasing the number of firms who can service the larger contracts from three to two is likely to accelerate the trend of market stability, as gaming market share will be marginalized, and increased customization and integration of products will become the primary goal.

Finally, the argument that high fixed costs relative to variable costs creates an incentive to seek volume, while broadly appealing, is not as applicable to the Estimatics and TLV markets as it might be to other software markets. When fixed costs are high relative to variable costs, there is a “strong economic incentive[ ] to produce at close to full capacity” that “works against the likelihood of any collusive price raising scheme which would require output restrictions.” Archer-Daniels, 781 F.Supp. at 1423. As Dr. Ordover noted, industries with high fixed costs relative to variable costs are likely to have no barriers to expansion, and no capacity constraints on production. Ordover, Tr. (1/23 a.m.) at 91:16-92:1. Here, while the Big Three firms have no capacity constraints, the Estimatics and TLV markets are already “very mature” and offer little room for growth. PX 543-005 (Mitchell). Capacity is limited by customer demand, not vendor cost, and the volume of claims has been growing at a slow rate over the last several years. See id. Moreover, because of the time and high costs associated with switching products, it is very difficult to gain market share. Accordingly, Dr. Hayes calculated that “it’s not in [the merged firm’s] interests to reduce prices; in fact, it’s in their interest to raise prices.” See Hayes, Tr. (1/21 a.m.) at 69:11-72:1.

The Defendants point to a Solera financial document to support their contention that Audatex sees the merger as a potential opportunity to increase its competition for CCC’s and Mitchell’s existing customers. In that document, Solera stated:

We view potential industry consolidation as a positive for Solera [because]: [1] It provides an opportunity for Solera to gain share in North America as its two competitors will be distracted; [and][2] it likely creates a more rational North American market, as the newly combined competitors would be highly levered and capital constrained.

DX 41-AUD0000229 (Audatex) (emphasis added). As indicated by an internal company e-mail from [text redacted], which the FTC championed as a “smoking gun” of sorts, to a profit maximizing business operating under these market conditions, “rational” means avoiding price wars with competitors. PX 552. 43

In a highly concentrated market, with stable market shares, low growth rates and significant barriers to entry, there are few incentives to engage in healthy competition. Although the FTC has exaggerated the legal significance of the “merger-to-duopoly” inquiry, it is clear that CCC/Mitchell and Audatex will likely be the only major players in these markets for the foreseeable future. “[I]t is easier for two firms to collude without being detected than for three to do so,” Am. Hosp. Supply Corp. v. Hosp. Prods. Ltd., 780 F.2d 589, 602 (7th Cir.1986), but price fixing is only one concern of the antitrust laws. A more common concern is “the creation or reinforcement by merger of ... oligopolistic market structures in which tacit coordination can occur.” Heinz, 246 F.3d at 725. With only two dominant firms left in the market, the incentives to preserve market shares would be even greater, and the costs of price cutting riskier, as an attempt by either firm to undercut the other may result in a debilitating race to the bottom.

Nevertheless, Defendants have made a strong argument that despite these characteristics, the market dynamics create a number of incentives to compete, and indeed, have maintained a competitive marketplace to this day. See Mitchell, Tr. (2/17 p.m.) at 43:14-44:15 (“Concentrated though they are everybody agrees that [these markets] are competitive.... We know that this market does not operate the way the presumption indicates. Why? Because it’s highly concentrated, [yet] it’s [still] competitive, that’s the first clue.”). Although Defendants present several arguments why coordination is not likely to occur despite their merger, the FTC has responded with substantial evidence of significant barriers to entry as well as credible evidence that coordination is possible, and even likely, in these markets. Whether the Defendants’ argument that the unique combination of factors in these markets negates the probability that the merger may tend to lessen competition substantially, or whether the FTC is correct that the market dynamics confirm the presumptions that follow its prima facie case, is ultimately not for this Court to decide. As Judge Tatel confirmed in Whole Foods, “[c]ritically, the district court’s task is not ‘to determine whether the antitrust laws have been or are about to be violated. That adjudicatory function is vested in the FTC in the first instance.’ ” 548 F.3d at 1042 (Tatel, J., concurring) (quoting Heinz, 246 F.3d at 714-15). The Defendants’ arguments may ultimately win the day when a more robust collection of economic data is lain before the FTC. On this preliminary record, however, the Court must conclude that the FTC has raised questions that are so “serious, substantial, difficult and doubtful” that they are “fair ground for thorough investigation, study, deliberation and determination by the FTC.” Heinz, 246 F.3d at 714-15.

c. Unilateral Effects

The FTC’s sufficient response to the Defendants’ rebuttal arguments on coordinated effects alone could suffice, but because the parties focused on unilateral effects throughout the evidentiary hearing, a brief analysis on this issue is in order. There are two basic analytical frameworks for analyzing the competitive effects of a merger: coordinated and unilateral effects. Though the distinction between these two frameworks has more significance in law than it does in economics, see Hayes, Tr. (1/12 p.m.) at 138:5-8, 10-11 (describing the distinction between coordinated and unilateral effects as “artificial”), the basic premise underlying the distinction is that the unilateral effects theory surmises that firms do not recognize their shared interest in elevating price, whereas the coordinated effects theory assumes that they do. Id. at 138:15-19. Thus, under unilateral effects theory, economists assume that firms behave independently. Id. at 138:18-19.

There are two main theories that predict unilateral effects: (1) the “dominant firm” theory and (2) the differentiated products theory. Both experts agreed that the “dominant firm” theory, also known as the “network effects” or “tipping point” theory, does not apply here, so the Court turns to the differentiated products theory. See Hayes, Tr. (1/21 p.m.) at 209:18-210:7 (“I do not credit the strength of the network effects being sufficient to undermine Audatex. I [do not give] that argument a lot of weight.”); Ordover, Tr. (1/23 p.m.) at 148:5-18.

The differentiated products theory applies to markets where the products sold by different suppliers are not perfect substitutes for one another. Unilateral effects in a differentiated product market are likely to be profitable under the following conditions: (1) the products must be differentiated; (2) the products controlled by the merging firms must be close substitutes, ie., “a substantial number of the customers of one firm would turn to the other in response to a price increase”; (3) other products must be sufficiently different from the products offered by the merging firms that a merger would make a small but significant and non-transitory price increase profitable for the merging firm; and (4) repositioning must be unlikely. Oracle, 331 F.Supp.2d at 1117-18. As Dr. Ordover observed, the Estimatics and TLV products sold by CCC, Mitchell and Audatex are significantly heterogeneous. See Ordover, Tr. (1/23 a.m.) at 85:2-11. As to the second, third and fourth conditions, the Merger Guidelines advise that:

[s]ubstantial unilateral price elevation in a market for differentiated products requires that there be a significant share of sales in the market accounted for by consumers who regard the products of the merging firms as their first and second choices, and that repositioning of the non-parties’ product lines to replace the localized competition lost through the merger be unlikely.

Merger Guidelines § 2.21. Thus, the key point in contention is whether a significant percentage of consumers view CCC and Mitchell as their first and second choice, and Audatex as a “more distant third.” See PX 1020 ¶ 79 (Hayes Prelim. Report).

Economists employ a separate category of models to predict unilateral behavior, that is, to predict whether a hypothetical profit maximizing firm would find it profitable to raise prices on its own. Hayes, Tr. (1/12 p.m.) at 138:11-15. Dr. Hayes prepared three different models to predict the likelihood of unilateral effects in these markets: (1) a Bertrand simulation of price effects for repair facilities, (2) a bidding model for the sale of Estimatics to insurance companies and (3) a separate bidding model for the sale of TLV to insurance companies.

Dr. Hayes’s Bertrand model, which incorporates data on market shares, diversion ratios, and production costs, predicts a post-merger average price increase of about 30% for Estimatics for repair facilities. Hayes, Tr. (1/12 p.m.) at 153:17-55:15; see also PX 1059 ¶¶ 6-10, 17, Ex. S-2 (Hayes Suppl. Report). Accounting for the cost savings and cross-selling opportunities that would be facilitated by the merger, the model still predicts an average price increase of approximately 19%. PX 1059, Ex. S-2. 44 Dr. Ordover challenges the accuracy of the diversion ratios that Dr. Hayes employs in his model, but even using Dr. Ordover’s suggested diversion ratios, the model still predicts substantial price increases for repair facilities. Hayes, Tr. (1/21 p.m.) at 213:24-214:14.

Because CCC, Mitchell and Audatex bid on Estimatics contracts with insurance companies, Dr. Hayes employed a derivation of an auction model to predict the merger’s effect on prices for the insurance segment of the Estimatics market. PX 1020 ¶ 78 (Hayes Prelim. Report). 45 This model assumes that in a procurement auction, which these bids resemble, though the winner is the supplier that is able to make the most attractive offer to the customer, the winning price is actually determined by the supplier with the second-most attractive offer. Id. Thus, the most attractive offer determines the winner and the second-most attractive offer determines the price. Id. Critically, the model predicts that the merger will cause prices to increase “if the merging suppliers would have been the most attractive and second-most attractive suppliers had they remained separate, and the third-most attractive supplier is viewed as a more distant choice.” Id. ¶ 79 (emphasis added). 46 Therefore, the model attempts to mirror the economic theory. Through a series of calculations, which include data on market shares, profit margin earned on incremental sales, and incumbent winning percentage, see PX 1059 ¶¶24, 26, 27 (Hayes Suppl. Report), Dr. Hayes’s model predicts that insurance companies are likely to receive a price increase of approximately 7% for Estimatics products. Id., Ex. S-7. 47 Thus, “[w]hat the model is telling us is that there [is a] reasonably large fraction of customers that think that ... Audatex is a more distant third.” Hayes, Tr. (1/21 a.m.) at 42:16-18. If the Defendants’ asserted cost-savings are included in the calculations, the average price increase is a more modest 1%. PX 1059, Ex. S-7 (Hayes Suppl. Report).

Finally, Dr. Hayes prepared a similar auction simulation to determine whether there would be price effects for the TLV market. Using similar inputs as the auc tion model for Estimatics, the TLV simulation predicted, without accounting for cost savings, an increase in price to insurance companies of 15.8%. Id., S — 12; see also Hayes, Tr. (1/12 p.m.) at 166:15— 167:17. With cost savings, the model still finds an average price increase of 6% in TLV. PX 1059, S-12 (Hayes Suppl. Report); see also Hayes, Tr. (1/12 p.m.) at 167:18-168:9.

The main problem with Dr. Hayes’s models is that the data and predictions cannot reasonably be confirmed by the evidence on this record. The PTC argues that there is a presumption that a “significant share” of customers prefer the merging parties’ products as their first and second choices if the merged firm has a market share of at least 35%. FTC’s Post-Trial Brief at 47 (citing Merger Guidelines § 2.211). However, this presumption is warranted only if: “each product’s market share is reflective of not only its relative appeal as a first choice to consumers of the merging firms[’] products but also its relative appeal as a second choice, and hence as a competitive constraint to the first choice.” Merger Guidelines § 2.211.

As part of his investigation, Dr. Hayes reviewed a sample of eighteen bidding contests for Estimatics between 2003 and 2007. PX 1020 ¶80 (Hayes Prelim. Report). In seventeen of those contests, CCC was the victor, while Mitchell was one of the finalists in nine and Audatex was the runner-up in eight. Id. Mitchell and Audatex were the finalists in the one remaining instance. Id. ¶ 80 n. 126. But, approximately one hundred insurance contracts are up for renewal each year, Brungger IH Tr. at 186:1-22 (Mitchell), meaning that Dr. Hayes reviewed a total of eighteen bidding contests out of approximately 400 that occurred during that time span. This fraction of auctions is not large enough to rely on as a representative sample of the entire insurance market. See Oracle, 331 F.Supp.2d at 1167 (“Drawing generalized conclusions about an extremely heterogeneous customer market based upon testimony from a small sample is not only unreliable, it is nearly impossible.”); see also SunGard, 172 F.Supp.2d at 182-83. On the TLV side, Dr. Hayes reviewed data from a larger set of recent bid events, but only four of them identified who the two finalists were. See PX 1020 ¶ 88 (Hayes Prelim. Report). CCC and Mitchell were one of the finalists in all of these instances; CCC and Audatex were the two finalists in no instances. Id. Again, this sample is too small to rely on to make broad conclusions about a highly fragmented market.

With respect to repair facilities, the diversion ratios are the best indicators of whether a significant share of the market views CCC and Mitchell as their first and second choices, and Audatex a more distant third. Dr. Hayes’s diversion ratios are derived from a two-year old survey of thirty-one former CCC customers which notes that the results “cannot be projected to the population as a whole due to the limited number of completes.” PX 1423-005(CCC); Hayes, Tr. (1/21 p.m.) at 175:6-13. The warning that the diversion ratios cannot be projected to CCC’s entire customer portfolio is evidence of its unreliability. Moreover, the twenty-nine responses that Dr. Hayes used to calculate his diversion ratios account for less than 0.2% of all of CCC’s repair facility customers. Ramamurthy, Tr. (1/22 a.m.) at 62:7-8 (CCC has approximately 22,000 repair facility customers). This data is even more unreliable than the data relating to insurance company choices, especially considering that Bertrand models typically rely on large volumes of data to estimate diversion ratios. Ordover, Tr. (1/23 p.m.) at 153:20-154:5; see also id. at 154:12-155:7 (the Bertrand model is most reliable in markets where there are tens of thousands of data points, such as a consumer products market); see also Hayes, Tr. (1/21 p.m.) at 176:6-8 (“Certainly 31 customers is not a terribly large number. More customer information would be useful.”). Moreover, Dr. Hayes had no data from Mitchell or Audatex regarding their diversion ratios. Instead, he used CCC’s diversion ratios to predict Mitchell and Audatex’s diversion ratios based on their respective market shares. 48

Alternatively, the Merger Guidelines state that a combined market share of at least 35% may be sufficient to presume that the merger will result in unilateral effects if “data on product attributes and relative product appeal show that a significant share of purchasers of one merging firm’s product regard the other as their second choice.” Merger Guidelines § 2.211. Evidence of this sort is sorely lacking in each segment of these markets. Dr. Hayes could not identify any characteristics of Estimatics or TLV products that might make Audatex a more distant third choice for certain insurers. Nor could he identify any characteristics of a particular class of insurers that might cause them to view Audatex as a more distant third — with the possible exception of lack of familiarity. Furthermore, Dr. Hayes relied on statements of only six out of approximately 300 insurance companies for his conclusion that Audatex’s Estimatics product is a “more distant third” choice for a significant share of insurance companies. Of those six insurance companies, none of them actually stated that they viewed Audatex as a more distant third choice, nor could they identify any characteristics of Audatex’s product that would make it a more distant third choice for them. See Hall, Tr. (1/8 a.m.) at 89:8-21 (GMAC did not “know anything about Audatex”); Brandt, Tr. (1/8 a.m.) at 20:3-6, 41:8-18 (The Hartford was “impressed” with all three of the firms’ offerings and a director within its auto and liability practice department opined that CCC, Audatex and Mitchell all had a relatively similar look, touch and feel); id. at 41:4-5 (“Q. You’re not suggesting [Audatex offers] an inferior product are you? A. Oh, no.”).

The evidence is similarly lacking for TLV. Dr. Hayes could not identify any insurers — let alone any group or class of insurers — that view Audatex’s TLV product as a “more distant third” choice. The only evidence that Dr. Hayes found that might suggest that Audatex’s product is less desirable in some respect was one reference about higher valuations. Hayes, Tr. (1/21 a.m.) at 98:15-21. All the same, he continued the faulty assumption that some insurers view Audatex as a more distant third for Estimatics to assume that those same insurers view it as a more distant third in TLV. See Hayes, Tr. (1/21) at 96:12-23.

Evidence relating to repair facilities is also notably absent. The vast majority of repair facilities who testified or were contacted by the parties ranked Audatex as first or second in Estimatics. See, e.g., Kostakis Dep. at 40 (Angelo’s Auto Body Shop); DX 25 ¶ 8 (Benjamin Deck, ABRA Auto Body & Glass). In fact, a 2006 Collision Week survey ranked Audatex as the “Best Overall Estimating System Provider in 2006,” finding Audatex superior to CCC and Mitchell in 8 of 8 categories for Estimatics. DX 43; see also DX 33 at 00064042, 0064044 (survey showing that shops using two estimating systems most frequently use Audatex and CCC). The evidence showing that Audatex is at least comparable to CCC and Mitchell does not necessarily mean that Audatex is not viewed as a more distant third choice by some share of the market. Even so, the absence of any evidence of identifiable characteristics of these firms’ products or their customers that might make Audatex a more distant third option makes it impossible to reach the conclusion that Audatex is a more distant third choice. 49

Without credible evidence that Audatex is a more distant third choice for a significant share of the market to support the predictions of Dr. Hayes’s models, the Court cannot conclude that the merger is likely to result in unilateral price elevations.

d. Efficiencies

Finally, Defendants argue that the efficiencies that will result from the merger will offset any potential anticompetitive effects of the merger. The Merger Guidelines recognize that “mergers have the potential to generate significant efficiencies by permitting a better utilization of existing assets, enabling the combined firm to achieve lower costs in producing a given quantity and quality than either firm could have achieved without the proposed transaction.” Merger Guidelines § 4. Although the Supreme Court has not sanctioned the efficiencies defense in Section 7 cases, “the trend among lower courts is to recognize the defense.” Heinz, 246 F.3d at 720. “The courts have recognized that ‘in certain circumstances, a defendant may rebut the government’s prima facie case with evidence showing that the intended merger would create significant efficiencies in the relevant market.’ ” United States v. Long Island Jewish Med. Ctr., 983 F.Supp. 121, 146-47 (E.D.N.Y.1997) (quoting Univ. Health, 938 F.2d at 1222); see also Arch Coal, Inc., 329 F.Supp.2d at 150. However, courts have rarely, if ever, denied a preliminary injunction solely based on the likely efficiencies. See Heinz, 246 F.3d at 720; FTC v. Swedish Match N. Am., 131 F.Supp.2d 151, 171-72 (D.D.C.2000); Tote, 768 F.Supp. at 1084-85. And in a highly concentrated market characterized by high barriers to entry, the parties opposing a preliminary injunction must provide “proof of extraordinary efficiencies” in order to rebut the presumption of anticompetitive effects. Heinz, 246 F.3d at 720; see also Phillip E. Areeda, et al, IVA Antitrust Law ¶ 971f, at 47 (2d ed. 2006) (requiring “a showing of ‘extraordinary’ ” efficiencies where the “post-merger market’s HHI is well above 1800 and the HHI increase is well above 100”). Moreover, in such circumstances, “the court must undertake a rigorous analysis of the kinds of efficiencies being urged by the parties in order to ensure that those ‘efficiencies’ represent more than mere speculation and promises about post-merger be havior.” Heinz, 246 F.3d at 721. The Defendants have not demonstrated here that their efficiencies are verifiable, see Staples, 970 F.Supp. at 1089, or that the cost savings achieved through efficiencies are likely to be greater than the transaction’s likely anticompetitive effects, see Cardinal Health, 12 F.Supp.2d at 62-63.

1. Cost Savings

Efficiencies enabled by a merger “can create incentives to reduce prices, and if these efficiencies are sufficiently large, they may fully offset any price elevation that might otherwise flow from the merger.” PX 1020 ¶ 75 (Hayes Prelim. Report). Defendants assert that the efficiencies achieved through the merger will produce cost savings of at least $48 to $55 million per year resulting from the elimination of redundant or overlapping functions and the consolidation of product lines. DX 27 at 6-21 (PriceWaterhouseCoopers (“PWC”) Synergy Analysis); DX 28 at 4-13 (CCC/Mitchell Confidential Synergy Review); Balbirer IH Tr. at 26-27, 50-5HCCC); PX 746-006, 007 (Mitchell). These projected potential cost savings exceed 20% of the companies’ combined cost base. DX 28 at 6 (CCC/Mitchell Confidential Synergy Review).

The projected cost savings would indeed be substantial and such cost savings were fundamental to the parties’ decision to merge. See, e.g., Sun, Tr. (1/9) at 9:8-15 (Mitchell). However, the vast majority of these cost savings will be realized only if, and when, Defendants consolidate to a single software platform. DX 27 at 6-21 (PWC Synergy Analysis); Sun, Tr. (1/9) at 19:20-22:1 (Mitchell); Balbirer IH Tr. at 68(CCC); DX 28 at 4-13 (CCC/Mitchell Confidential Synergy Review). CCC and Mitchell estimate that it will take two to three years to combine their operations and eliminate redundant software systems, DX 115-48 (CCC/Mitchell Confidential Offering Mem.), but they acknowledge that it might take as long as ten years to complete the integration, and the nature of the ultimate software product is uncertain, see PX 755 ([text redacted]); see also Sun, Tr. (1/8 p.m.) at 77:9-12 (Mitchell); PX 756 (Mitchell [text redacted]).

The Merger Guidelines advise that “[delayed benefits from efficiencies (due to delay in the achievement of, or the realization of customer benefits from, the efficiencies) will be given less weight because they are less proximate and more difficult to predict.” Merger Guidelines § 4 n. 37. Thus, while the Court does not doubt that Defendants intend to consolidate the two firms’ software into one platform, it cannot place great weight on the predicted cost savings resulting from that consolidation because there is no telling when those savings might begin to accrue or whether they will actually materialize and not be absorbed in the consolidation effort.

Defendants admit that the merger will result in one-time costs of approximately $27.5 million over the first three years, thus lessening their projected cost savings to approximately $20-27 million per year during that time period. DX 27 at 2, 7 (PWC Synergy Analysis). Even these numbers are speculative and are supported by little more than lawyer argument. Mitchell hired two financial consultants to analyze and confirm that the merger makes financial sense. See Sun, Tr. (1/9) at 18:14-18 (Mitchell). In their final estimates, both Bain & Co. (“Bain”) and PWC predicted that the merger would result in a net loss (in terms of costs versus synergies) at the end of three years. PX 1823 (Bain); see also Sun, Tr. (1/9) at 37:9-19 (Mitchell). But see DX 115-48 (CCC/Mitchell Confidential Offering Memorandum predicting net synergies of $11.2 million in year one); DX 27 at 7 (PWC Synergy Analysis). Defendants argue, however, that Bain’s final projections do not accurately portray the likely synergies because its analysis (a) treats as “costs” certain expenditures, such as investment in research and development, without allowing such expenditures to show efficiency gains, Sun, Tr. (1/9) at 24:22-25:4 (Mitchell), and (b) lists integration expenses without attempting to quantify the benefits that will arise from non-overlapping business units, id. at 25:10-26:2 (Mitchell). But Bain and PWC are Defendants’ financial consultants, not the FTC’s, and Mr. Sun admitted that he did not “have any reason to believe that these guys ... did a bad job.” Id. at 37:17-20. Thus, without overanalyzing the accuracy of their predictions based on incomplete information, it is enough to conclude that the record is far from clear what the net savings would be; hence, Defendants have not demonstrated that the cost savings are likely to counteract the potential anticompetitive effects.

Even assuming arguendo that the Defendants will achieve significant cost savings in a timely manner, there is no evidence to suggest that a sufficient percentage of those savings will accrue to the benefit of the consumers to offset the potential for increased prices. See Univ. Health, 938 F.2d at 1223 (defendant asserting efficiency defense “must demonstrate that the intended acquisition would result in significant economies and that these economies ultimately would benefit competition and, hence, consumers”). First, platform consolidation would create significant one-time switching costs for a large number of repair facilities. At least a quarter of all repair facilities- — 11,000 or more — use only one software platform. PX 635-018 (Mitchell). Those repair facilities currently using the platform that ultimately is not chosen for consolidation will incur one-time costs related to switching to the surviving platform. Rollins Dep. at 84 (Safeco); see also PX 1020 ¶ 11 (Hayes Prelim. Report). Meanwhile, as of the end of 2007, only 1,500 use both CCC and Mitchell. See Ramamurthy Dep. at 72-73(CCC); Balbirer IH Tr. at 189-190(CCC); PX 632-033 (Mitchell); PX 516-006 (Mitchell). Thus, fewer than one in thirty repair facilities — those that currently use both CCC and Mitchell — will realize any costs savings by licensing one fewer product. See PX 516-006 (Mitchell). Furthermore, a substantial number of insurance companies would likewise incur switching costs.

Second, while reducing the costs of doing business provides several advantages for the merged firm, these advantages could show up in higher profits instead of benefitting customers or competition. See Hayes, Tr. (1/21 a.m.) at 87:7-88:14. Mr. Ramamurthy admits that CCC will give its shareholders much of any savings. Ramamurthy IH Tr. at 74-75(CCC). CCC’s CFO, Andrew Balbirer, similarly stated that the synergies from the deal would either be invested in new products or go to company profits. Balbirer IH Tr. at 50-51(CCC). Likewise, Mr. Sun of Mitchell stated that the cost savings are likely to go to “building value added products” rather than lowering consumer costs. Sun IH Tr. at 49-50 (Mitchell).

2. Innovations

Defendants also assert that the merger will lead to innovations. Innovation claims are “often a speculative proposition,” see Heinz, 246 F.3d at 722, and such claims are often looked upon with skepticism for good reason:

[Wjhen the two firms are already among the largest in the market, there is no empirical basis for thinking that even larger firms would produce more R & D. We would therefore limit the defense to instances in which the two merging firms can show that their size forces them to accept higher per-unit costs for research and development than larger firms in their market must pay, and that the merger will enable them to achieve some figure closer to parity. .

IVA Areeda, et al., supra, ¶ 975g, at 94. Defendants assert that as a result of the cost savings generated by the merger, the merged entity will spend considerably more on new product research and development than the two companies spend individually. See DX 28 at 2-3 (CCC/Mitchell Confidential Synergy Review); Ramamurthy IH Tr. at 50-51 (CCC), 82-83; see also Ramamurthy, Tr. (1/22 a.m.) at 111:16-112:11 (CCC) (stating that the merger would “allow us to throw more resources” at development of [text redacted]). The CEOs for both Mitchell and CCC testified that the merged firm will spend approximately 50% more on new product research and development than the combined amount the two companies currently spend as separate companies. Sun, Tr. (1/9 a.m.) at 39:24-40:23 (Mitchell); Ramamurthy, Tr. (1/22 p.m.) at 148:20-25 (CCC) (stating that the combined firm will deploy $10-15 million more towards innovation). But there is no telling whether these aspirations of greatly enhanced investment in R & D will become a reality as the combined firm is saddled with the burden of paying off its combined debt. Furthermore, there is little evidence that these promises of increased R & D spending are merger-specific. Mr. Balbirer testified, for instance, that CCC could already afford to increase R & D funding on its own if its “shareholders were willing to trade off their return.” Balbirer IH Tr. at 198(CCC). This is a standard conflict between investors and management. There is no reason to believe that the merger will eliminate this pervasive problem.

In sum, although the merger may produce substantial efficiencies in the future, such efficiencies are too far afield and too speculative to overcome the strong presumption of anticompetitive effects created by the large HHIs and the high barriers to entry in the Estimatics and TLV markets.

B. Equities

When the FTC shows a likelihood of success on the merits, it will usually be able to obtain a preliminary injunction because the public equities often converge with the “success on the merits” analysis because “ ‘the public interest in effective enforcement of the antitrust laws’ was Congress’s specific ‘public equity consideration’ in enacting” Section 13(b). Whole Foods, 548 F.3d at 1035 (Brown, J.) (quoting Heinz, 246 F.3d at 726). However, the “likelihood of success” analysis and the “public equities” analysis are legally different points and the latter should be analyzed separately, no matter how strong the agency’s case on the former. 50 But since, as discussed above, the FTC has demonstrated a likelihood of success on the merits as a preliminary matter, the Defendants must now show that, despite the likely anticompetitive effects of their proposed merger, the merger would nonetheless benefit their customers. Only “public equities” that benefit consumers can override the FTC’s showing of serious questions on the merits. Whole Foods, 548 F.3d at 1041 (Brown, J.). For instance, if potential merger partners can present credible evidence that the merged company will lower consumer prices because of extraordinary efficiencies (even when the same efficiencies might not suffice to overcome the presumption in favor of the FTC’s prima facie case on the merits), and those efficiencies and lowered prices will be lost forever if the merger is preliminarily enjoined, the public equities in favor of the merger might outweigh the FTC’s likelihood of success on the merits. However, it is clear that a “ ‘risk that the transaction will not occur at all,’ by itself, is a private consideration that cannot alone defeat the preliminary injunction.” Id. (citing Heinz, 246 F.3d at 726; Weyerhaeuser, 665 F.2d at 1082-83).

The Defendants predict that the merger will produce cost savings of 20%. This would be an extraordinary degree of efficiencies. However, the evidence fails to induce confidence that such cost savings will in fact occur, much less when they might be realized. In addition, the evidence leaves uncertainty as to whether cost savings will be enjoyed by customers or by shareholders.

Contrary to this conclusion, Defendants urge the Court to recognize that “[t]he combined company will be able to offer an integrated product that incorporates the best features of each company’s standalone product portfolio.” DFF ¶ 374. The Court accepts and agrees that the evidence supports this statement. “The merging parties envision using the combined resources of the two companies and spending more on new product research and development than they spend individually to improve existing and create new products and reduce costs” for customers. Id. ¶ 375. The Court accepts and agrees that the Defendants envision such a future. “[CJustomers and consumers [will] benefit from more innovative products.” Id. ¶ 380. The Court agrees and accepts this statement as well.

Both timing and free choice undercut Defendants’ predictions of the public value of the potential merger. That is, the parties predict that it will be at least three to five years before their separate software platforms might be merged, and perhaps as much as ten years before it makes sense to do that. During that lengthy waiting period, the serious questions raised by the FTC might ripen into fact and produce a coordinated marketplace through regional or customer allocation or other coordinated terms. During that lengthy waiting period, the evidentiary base for Dr. Hayes’s predictions of unilateral effects might mature.

While the merging parties “envision” using their combined resources to increase research and development of new products, once a merger is effected, their choices would be wide open and they might reasonably decide to use profits to reduce debt instead of increase research, given the debt structure of the newly-merged company and its share of the markets for Estimatics and TLV. The Court does not doubt their intentions, but cannot bank on them for purposes of this preliminary analysis of the equities here.

The only “hard” equity point made by Defendants is that their customers will benefit from more innovative products. Even though such a result is uncertain, this benefit must be weighed against the public benefit. In that weighing, the Court uses a sliding scale. The uncertainty of the public benefit of innovative products is too long in coming and too uncertain in result to hold much weight against the FTC’s interest in enforcing the anti trust laws. The equities favor granting a preliminary injunction.

y. CONCLUSION

For the foregoing reasons, the FTC’s Motion for Preliminary Injunction [Dkt. # 4] will be granted. A memorializing order accompanies this Memorandum Opinion.

1 .This Memorandum Opinion was provided to the parties in final form on March 9, 2009, but public release was delayed to ensure that no confidential business information that had been submitted under seal was being released. Based on input from Defendants and certain interested third parties, some competitively sensitive information has been redacted and other minor adjustments have been made to the opinion. The phrase "[Text redacted]” has been inserted to signify where text has been redacted from the opinion.

2 . Audatex is a subsidiary of Solera Holdings, Inc. ("Solera”), a publicly-traded holding company based in San Diego, California.

3 . Technically, Mitchell’s holding company would be merged into CCCM Parent (DE), Inc., a Delaware corporation, which, through a series of pass-through entities, would be held primarily by Investcorp and Aurora, as well as by the managing shareholders of CCC Holdings and Mitchell, and a mezzanine fund established by Goldman Sachs. PX786-018019.

4 . Citations to the trial record are in the following form: references to Plaintiff's exhibits — "PX ([Company])”; references to Defendants’ exhibits — "DX ([Company] )”; references to the preliminary injunction hearing transcript' — "[Name], Tr. ( [Date]) at [page]: [line] ([Company])”; references to the FTC IH transcripts — "[Name] IH Tr. at [page] ([Company])”; references to deposition designations — "[Name] Dep. at [page] ([Company])"; references to Plaintiff's Findings of Fact — "PFF ¶ [_]"; references to Defendants’ Findings of Fact — "DFF ¶ [_]”. Citations may deviate from these formats as necessary to protect commercially sensitive information.

5 . Unless otherwise indicated, market shares approximate 2007 data, and are not current.

6 . Audatex also has a substantial global footprint, operating in approximately fifty-one countries across six continents. Including its international operations, Audatex is the largest Estimatics company in the world, and is twice the size of its nearest competitor. See PX 89-013 (Audatex).

7 .CCC acquired Comp-Est in 2003. PX 86-013 (CCC).

8 . Audatex also has begun offering a web-based product. See DX 225-33-35 (Audatex).

9 . Insurers will usually replace the damaged vehicle if the cost of repair approaches around 65 to 75% of the vehicles’s value. Brungger IH Tr. at 118 (Mitchell).

10 . These "add-on” products are sold by a large number of companies in addition to CCC, Mitchell and Audatex, and the merger is not challenged by the FTC based on these products.

11 . Defendants take issue with the FTC’s interpretation of the "serious, substantial” question standard set forth in Heinz and Whole Foods, asserting: "[Y]ou can talk about substantial questions, doubtful questions, whatever.... [W]hat those cases say [is that] it simply means nothing other than likelihood of success on the merits.” Parker, Tr. (2/17 p.m.) at 41:24 — 42:3 (Mitchell). While Defendants' statement is literally true, precedents irrefutably teach that in this context “likelihood of success on the merits” has a less substantial meaning than in other preliminary injunction cases. Heinz not only emphasized this point but Whole Foods makes clear that Heinz remains good law. The analysis of likelihood of success "measure[s] the probability that, after an administrative hearing on the merits, the Commission will succeed” in proving that the effect of a merger "may be to substantially lessen competition or tend to create a monopoly.” Heinz, 246 F.3d at 714 (emphasis added).

12 . The Merger Guidelines provide an analytical tool for determining interchangeability and cross-elasticity of demand. The Merger Guidelines define a product market as "a product or group of products such that a hypothetical profit-maximizing firm that was the only present and future seller of those products ('monopolist') likely would impose at least a ‘small but significant and nontransitory' increase in price (SSNIP).” Merger Guidelines § 1.11. In most contexts, the Merger Guidelines consider a price increase of 5% to constitute a SSNIP. Id.

13 . An argument could be made that the Estimatics market should be further subdivided into communicating and non-communicating Estimatics platforms. Cf. Ordover, Tr. (1/23 p.m.) at 205:11-14 ("I did not reach an ultimate conclusion of whether these two are in the same market [for repair facilities] although obviously, people can move back and forth between communicating and non-communicating Estimatics systems”). Communication is required to work with the major insurance companies as part of a repair shop network, so all insurance companies and virtually all DRP repair facilities must use communicating products. See PX 1020 ¶ 12 (Hayes Prelim. Report); DX 644, App. 2 ¶ 7 (Ordover Coordinate Interaction Report). There also is a considerable cost differential between communicating and non-communicating Estimatics products. See PX 760 (Applied Computer Resources has two non-communicating Estimatics products that retail for $99 per month and $135 per month); PX 173 (CCC’s “Competitive Comparison Chart”); PX 1407(CCC). Nonetheless, while communicating Estimatics products may make up a relevant sub-market, that analysis is unnecessary here as it would only further limit the product market definition and minimally enhance the market concentration in the relevant market to a legally insignificant degree. The Court therefore will define the relevant market to include all Estimatics products.

14 . State Farm, Allstate Insurance ("Allstate”) and Progressive, three of the largest auto insurers in the country, as well as American Family, Hanover, Cincinnati Insurance, Country Insurance, Kentucky Farm Bureau, Mississippi Farm Bureau, Assurant Specialty Property ("Assurant”), and Infinity Insurance Company use the Books to perform some or most of their total loss valuations in-house. Stanton Dep. at 18-19, 23, 26-28, 33, 38-39, 45, 50-51 (NADA); Dibble, Tr. (1/23 a.m.) at 39:8-18 (Infinity Ins. Co.); Licause Dep. at 73 (Assurant); Marushka Dep. at 29, 33-34 (CCC); DX 52 at FTC-PA-000004 (listing approved guide sources in Pennsylvania as Red Book, CCC, NADA, Audatex, Autobid, and Mitchell); Conway, Tr. (1/12 a.m.) at 64:1-9 (Audatex) (American Family uses Books for some of its total loss); Brandt, Tr. (1/8 a.m.) at 47:12-22 (The Hartford uses Book data for some of its total loss needs); Hall, Tr. (1/8 a.m.) at 87:11-14, 99:1-6 (GMAC used Books in North Carolina until it was able to negotiate a flat rate with Mitchell that was less expensive).

15 . Mitchell argued in closing argument that it thinks the nine insurers who do some in-house total loss valuations account for about 20% of the total loss claims market, and that the FTC failed to include those in-house valuations in its HHI calculations. Prior to closing argument, however, there seemed to be a consensus that the Books consumed about 10% of the total loss valuation market. See PX 548-007 (Mitchell) (chart showing that “book values” account for 10% of all total loss claims). This figure must include internal calculations performed by insurance companies because only insurance companies (and independent appraisers to some extent) handle total loss claims, with or without TLV software. See Hayes, Tr. (1/12 p.m.) at 130:25-131:1.

16 . Critical loss analysis is a standard tool used by economists to study potentially relevant markets. The analysis is based on the assumption that if the price of a product were increased, some customers would not purchase the product, and if enough customers would not purchase the product, the price increase would not be profitable. PX 1020 ¶ 42 (Hayes Prelim. Report). For the purpose of market delineation, Dr. Hayes was interested in whether or not a hypothetical monopolist over all TLV software could profitably impose a SSNIP on one or more of the products included in the candidate market. Id. ¶ 44.

17 . See PX 1412-020 (pie graph showing 2007 market share data for total loss which refers only to Audatex, Mitchell and CCC); PX 99-022 (CCC); PX 107-018 (CCC); PX 204-002 (CCC); PX 214-002 (CCC); PX 515-036 (Mitchell) (noting the holders of the key insurance accounts for estimating, TLV and DRP shops); PX 583-041 (Mitchell); PX 513-012 (Mitchell); PX513-016 (Mitchell document referring only to CCC and Audatex as competitors); PX 643-008 (Mitchell pricing strategy); PX 629-002 (Aurora) ("Mitchell competes primarily against only two other companies, Audatex and CCC Information Services.”). But see PX 677 (Mitchell) (Why Use Total Logic Versus NADA); DX 86 (CCC’s competitive loss of [text redacted] to NADA); PX 1964-009 (CCC) (“CCC Versus Guidebook Value”).

18 . The Court notes the apt warning that "separate markets are not indicated by documents within A firms that are preoccupied with other A firms. After all, a given producer of A cannot charge more than other A firms and thus may focus entirely on them even though a hypothetical monopolist of product A firms would focus entirely on the price of a close substitute B.” IIB Areeda, supra, ¶ 562a, at 372. If the FTC were relying solely on Defendants’ documents, that evidence might be insufficient. However, various other industry participants from all sides share the view that TLV and the Books are not interchangeable.

19 . A "differentiated product ‘market’ is a market in which sellers compete along more dimensions than price." Oracle, 331 F.Supp.2d at 1121.

20 . The Defendants note that Book values are not the final word on total loss valuations; they are simply an input. Insurance companies incur additional costs by employing their own personnel and software to refine the Book valuations with more data points. Stanton Dep. at 26-27 (NADA); Dibble, Tr. (1/23 a.m.) at 38:23-40:4 (Infinity Ins. Co.). Thus, the true "cost” of using the Books is greater than would appear; how much this real cost approaches the pricing of a TLV product is not admitted in the record.

21 . Cf. United States v. Grinnell Corp., 384 U.S. 563, 574, 86 S.Ct. 1698, 16 L.Ed.2d 778 (1966) ("Defendants earnestly urge that despite these differences [between the central station service and substitute alarm systems], they face competition from these other modes of protection. They seem to us seriously to overstate the degree of competition, but we recognize that ... they 'do not have unfettered power to control the price of their services ... due to the fringe competition of other alarm or watchmen services.’ What defendants overlook is that the high degree of differentiation between central station protection and the other forms means that for many customers, only central station protection will do.”) (quoting United States v. Grinnell Corp., 236 F.Supp. 244, 254 (D.R.I.1964)); accord Whole Foods, 548 F.3d at 1038 (Brown, J.). But see United States v. Engelhard Corp., 126 F.3d 1302, 1306 (11th Cir.1997) ("[I]t is possible for only a few customers who switch to alternatives to make the price increase unprofitable.”).

22 . If each of the Book vendors were included in the TLV market as separate suppliers, the merger would still produce an HHI delta well above 100. See Merger Guidelines § 1.51(c).

23 . Hayes, Tr. (1/12 Tr. p.m.) at 122:4-10; PX 1020, Ex. 2 (Hayes Prelim. Report); see also PX 543-011 (Mitchell).

24 . Baker Hughes expressly rejected the FTC's theory that "defendants can rebut a prima facie case only by a clear showing that entry into the market by competitors would be quick and effective,” and instead accepted the district court’s conclusion that "the acquisiti on was not likely to substantially lessen competition.” 908 F.2d at 983 (emphasis added).

25 . [Text redacted.]

26 . Defendants argue that statements such as these are of little value because the industry representatives who made them are not antitrust lawyers and did not mean that there were entry barriers in an “antitrust sense.” The Court recognizes the caution but, even discounting the statements as puffery, finds that they are supported by the preliminary record.

27 . CCC's CEO, Mr. Ramamurthy, expressed concern that this draft presentation may not accurately reflect the final version of its rating agency presentation because the document was initially drafted by a number of young financial analysts who are not familiar with the auto physical damage industry. To the extent that CCC failed to provide a final version of the presentation to the FTC in discovery, it was appropriate for the agency to rely on statements in the draft. Furthermore, considering the overwhelming number of references to barriers in other documents, it seems highly unlikely that the final version omitted any reference to "high barriers to entry.”

28 . CCC obviously underestimated Mitchell’s ability to penetrate the TLV market, although its unique situation would not readily transfer to another new entrant.

29 . For instance, the Motor database contains over 3 million unique parts. PX 25 ¶ 6 (Carr Decl.). Mitchell's database has "over 7.6 million part descriptions, 2.2 million labor times, 128,000 vehicle configurations, 139,000 graphics and detailed parts information for vehicles up to 20 years old.” PX 573-009 (Mitchell); see also PX 574-001.

30 . Mitchell, Motor, and Audatex each use labor times developed from their own propietary time studies. PX 578-025 (Mitchell); PX 25 ¶ 8 (Carr Decl.); PX 1000-006 (Audatex) ("Forty years of experience, backed by time and labor studies”); PX 585-017 (Audatex).

31 . [Text redacted.]

32 . APU is a technology provider for the alternative parts-supply industry that recently entered into a strategic relationship with Audatex.

33 . See PX 11 ¶ 4 (Brown Decl.) ([text redacted] ); PX 27 ¶ 5 (Danforth Decl.) ([text redacted] ); see also PX 686-013 (Ohio Casualty Ins. Co. ("Ohio Casualty”) RFP); Brandt, Tr. (1/8 a.m.) at 20:18-21:3 ([text redacted]).

34 . See, e.g., Burklin Dep. at 80-81 ([text redacted]); Brungger IH Tr. at 161-162, 158— 159 ([text redacted]); PX 535-022 (Mitchell's TLV product was regarded as a "Killer Product [that] Closes [a] Big Portfolio Gap”); Hayes, Tr. (1/12 p.m.) at 134:14-135:10; PX 401 ¶ 12 (Blitstein Decl.) ("[T]he inclusion of an automobile partial loss estimating system is widely regarded among customers as a necessary feature when considering the purchase of an automobile total loss valuation product.”); PX 534-005 (Mitchell) (Kelley Blue Book and NADA cannot enter TLV market segment because they "do not have the base estimatics business to leverage (as does CCC, [Audatex] and Mitchell)”); PX 71-004 (CCC's total loss functionality will not be optimized unless it is paired with CCC's Estimatics).

35 . The current state of the marketplace for Estimatics products is as described. However, there is evidence that the insurance industry is moving towards open standards, by which the insurer would no longer mandate what software its DRP repair facilities use (potentially opening up the repair shop market considerably). See Ramamurthy, Tr. (1/22 a.m.) at 94:19-24(CCC).

36 . [Text redacted.]

37 . Due to the current restrictions of the Mitchell license, Web-Est currently is limited to pursuing a market of approximately [text redacted] small shops. Seidel, Tr. (1/22 p.m.) at 174:14-17 (Web-Est).

38 . Earnings Before Interest, Taxes, Depreciation and Amortization.

39 . With the movement towards web-based platforms well underway, see DX 225-33-35 (Audatex), and the potential emergence of predictive analytics in the not too distant future, the FTC may be chasing yesterday’s technology, when all is said and done.

40 . One source of competitive intelligence is former employees of one of the Big Three competitors who migrate to another one. Sun Dep. at 10-12 (Mitchell); PX 624-001. [Text redacted.]

41 . [Text redacted.]

42 . “Although homogeneity of products may make the creation and enforcement of a tradi tional cartel easier, tacit collusion may be easier when products are differentiated. When products 'are highly differentiated, direct competition between them is limited, in effect creating a smaller number of critical points for tacit interaction.... The result may be that highly segmented markets ... can sustain high profits more effectively than homogeneous markets.” Lawrence A. Sullivan & Warren S. Grimes, The Law of Antitrust: An Integrated Handbook § 11.2el, at 635 (2d ed. 2006).

43 . [Text redacted.]

44 . Defendants argue that Dr. Hayes's opinions are not based in reality because the combined CCC/Mitchell could not “conceivably get away with the sort of price increase that” his Bertrand model predicts. CCC, Tr. (1/21 p.m.) at 179:22-25. But Dr. Hayes admits that he puts "less weight on the precise numbers than the general directional indications that come out of the analyses.... [W]hether it's going to be 40 percent or 20 percent at each particular supplier, I wouldn’t stake my reputation on that.” Hayes, Tr. (1/21 p.m.) at 161:24, 177:10-12; see also Hayes, Tr. (1/21 p.m.) at 164:11-12 ("The world is complex[;] we can't capture everything in a model that we might like to.”). Thus, it is not particularly troublesome that the extraordinary price increases predicted in the model may not accord with "economic realities.” See Ordover, Tr. (1/23 p.m.) at 151:20-25.

45 . Specifically, Dr. Hayes analyzed the price effects on sales of Estimatics to insurers using "an independent private values English auction with no reserve price. The key role of the 'independent private values’ assumption is to acknowledge that while bidders may know something about their rivals’ costs, they do not know more about those costs than their rivals know. The absence of a reserve price (i.e., a maximum acceptable bid set by the insurer) is justified by the fact that there are no good substitutes for estimatics software ....” PX 1020 ¶ 78 n. 123 (Hayes Prelim. Report).

46 . In addition to inferring from the price increase that the non-merging firm is a more distant third choice, the model also assumes a more distant third. This is because the auction model always predicts some unilateral price increase, which, under economic theory, can occur only where the non-merging firm is a more distant third choice; and because the model was calibrated to generate the 80% price-cost margin exhibited in these markets, which, in combination with the high incumbent winning percentage and market share data, produces a significant price increase. See DX 644 ¶ 24 (Ordover Report) (lowering the economic margin in the model makes "the gap between the challeng[ing bidders] ... small[er],” which, in Dr. Ordover’s view, is an “assumption [that] more accurately reflects the business realities”).

47 .Notably, Dr. Hayes's auction model predicts that Audatex will have to reduce prices for Estimatics by approximately 15% when it is the incumbent bidder, i.e., when it has a nearly 90% chance of winning, even though the combined CCC/Mitchell will be able to raise prices. PX 1059, Ex. S-7 (Hayes Suppl. Report). Defendants make much of the fact that Audatex has aggressively attempted to impede this merger, arguing that Audatex’s opposition to the merger is evidence in itself of the merger’s likely competitive effects. See, e.g., Defs.’ Post-Trial Brief at 1.

48 . The Court does not conclude that the predictions of Dr. Hayes’s Bertrand model are necessarily wrong or that the diversion ratios he used are necessarily incorrect. The Court merely concludes that it cannot rely upon such a limited amount of data. Dr. Ordover concluded that "the diversion ratios that [Dr. Hayes] is putting into his model are not robust, I don’t think they're sound, I don't understand them, and I don’t believe that they necessarily match up with the economic realities of the repair facilities in the marketplace.” Ordover, Tr. (1/23) at 151:20-25 (emphasis added).

49 . Even if Audatex were a distant third for some percentage of the market, a unilateral price increase would be profitable for the merged firm only if it were unlikely that Audatex could reposition itself to replace CCC or Mitchell should they raise prices to supracompetitive levels. See Merger Guidelines § 2.21. Because the only evidence in the record regarding customer preference is that a few insurance companies lack familiarity with Audatex’s products, there is no reason to believe that Audatex would not be a suitable replacement for CCC or Mitchell. See Hall, Tr. (1/8 a.m.) at 89:11-21 (GMAC does not "know anything about Audatex,” but it "plan[s] to learn something about [it]” if the merger is consummated).

50 . If the two factors always merged, then there would be no need for a public equities analysis when the FTC shows a likelihood of success on the merits. The D.C. Circuit’s plurality decision in Whole Foods indicated that a separate equities analysis is necessary. Whole Foods, 548 F.3d at 1035 (Brown, J.) ("The district court did not apply the sliding scale, instead declining to consider the equities .... If, and only if, the district court’s certainty [that the FTC would not succeed on the merits] was justified, it was appropriate for the court not to balance the likelihood of the FTC’s success against the equities.”).