In re Polyurethane Foam Antitrust Litigation

*229

CLASS CERTIFICATION MEMORANDUM OPINION AND ORDER

JACK ZOUHARY, U.S. DISTRICT JUDGE

Introduction

In these consolidated proceedings, two putative classes allege that dominant firms in the flexible polyurethane foam market engaged in a decade-long conspiracy to fix, raise, and maintain the price of foam products. Pending before this Court are Motions for Class Certification by two groups of plaintiffs: Direct Purchasers (Doc. 584) and Indirect Purchasers (Doc. 577). Defendants opposed the Motions (Docs. 680, 682, 683 & 686), and Plaintiffs replied (Docs. 740, 741 & 744). The parties papered the docket with numerous expert reports (Does. 581-83, 584-10 at 22-42, 584-14, 590, 594, 679-1, 680-1, 682-1, 682-2, 683-2, 743, 744-13, 744-48, 744-49, 752, 892-1, 892-2, 892-3 & 893-1). (Other parties in this case, Direct Action Plaintiffs, proceed separately from the putative classes.)

This Court then held oral argument on both Motions, and heard testimony from several experts — Leitzinger, Lamb, Ordover, and Burtis (Docs. 898, 938, 954, 957, 959-60 & 967). There is more. The parties engaged in a parade of filings with post-heating supplemental authorities and other materials covering a range of topics (Docs. 963-64, 985, 990, 1020, 1024, 1031, 1035, 1085-86 & 1098). After an exhaustive review of the record, this Court grants both Motions for the reasons that follow.

Background

The Parties and Products

Polyurethane Foam. Market

The polyurethane foam (“foam”) market consists of several types of products. Foam *230 products are created from a mixture of toluene diisocyanate (“TDI”) and polypropylene glycol (“polyol”) (Doc. 584-14 at 10; Doc. 581 at 13 (explaining “all of the TDI produced in the United States” is used in foam production); Doc. 682-1 at 37; Doc. 680-1 at 40). Pricing for each chemical is influenced by petroleum prices (Doc. 682-1 at 41-42 (quoting a Foamex SEC filing attributing foam input price increases to “oil and natural gas prices and the current geopolitical instability and its impact on oil production and prices”)). The principal chemical manufacturing plants for TDI and polyol, run by industry-dominant firms like Lyondell Chemical Company (now LyondellBasell Industries), Huntsman, Dow, and BASF, are or were located in the Gulf Coast region (Doc. 682-1 at 40-41; Doc. 680-1 at 43-44; Doe. 584-14 at 44 n.226). Together, these two chemicals account for roughly ninety percent (90%) of foam production costs (Doc. 584-14 at 10 (estimating TDI and polyols to be “80 to 90 percent of slab-stock manufacturing costs”); Doc. 581 at 13 (same); Doc. 682-1 at 38 (quoting various Defendants’ internal cost-share estimates from 85 to 90 percent of aggregate cost); Doc. 680-1 at 41)).

Water also figures in the foam blend, serving as a “blowing agent” that produces an exothermic reaction in the foam blend causing it to expand (Doc. 680-1 at 40). Product-specific additives, discussed below, are important in determining foam type. But, like water content, the cost of these additives does not cause large variations in the cost of producing foam products of comparable foam volume — TDIs and polyols are dominant throughout.

From here, production processes and alterations to the general foam blend effectively split the generic foam market into “sub-markets.” Some foam, termed “rigid,” is used for building or automotive insulation (see Doc. 680-1 at 38 (noting Woodbridge’s production of rigid foam products); Doc. 581 at 12 (noting that rigid foam accounts for 46 percent of polyurethane foam demand, with the remainder tied to flexible foam)). “Molded” foam, primarily found in automobile cushions, is formed after the foam blend is “poured into custom tooled molds the shape of the desired product” (Doc. 680-1 at 40-41). Foam is near ubiquitous, appearing in a vast array of product uses. “Packing peanuts,” for example, derive from the same general foam blend (Doc. 682-1 at 51). Flexible foam alone amounts to 1.2 billion pounds per year in domestic consumption (Doe. 581 at 13).

But the present Motions, while broad in scope, do not encompass the entire foam industry. Rather, only two foam sub-markets, distinguished from other sub-markets by their production processes and end uses, are relevant: slabstock and underlay.

Slabstock

As noted above, the foam blend undergoes an exothermic reaction with the addition of a blowing agent, typically water; that reaction generates a release of heat, forming bubbles within the foam blend. “The bubbles in the foam produce ‘cells’ containing air” (Doc. 682-1 at 37; Doc. 680-1 at 40). For slab-stock, the expansion process occurs as foam is continuously “poured” onto a moving conveyor belt (Doc. 584-14 at 9). That conveyor belt is equipped with “sides from [three to four feet] high” (Doc. 581 at 13), with length and width dimensions that vary depending on intended uses (Doc. 584-14 at 9 & n.16). Cured slabstock, which resembles a loaf of bread, is also referred to as a “bun” (id. at 9 & n.15).

But not all slabstock is created equal. Foam manufacturers regularly vary foam blends to produce varying characteristics or “grades” (see id. at 15 (noting combustion modifiers and other additive types)). Slab-stock can vary by density, with more dense foam typically providing better support and comfort (Doc. 682-1 at 47; Doc. 680-1 at 50). Slabstock can also be more or less firm, measured according to an industry-standard Indentation Force Deflection calculation (Doc. 584-14 at 17; Doc. 682-1 at 48; Doc. 680-1 at 51). “Support factor,” a measure of “deep down support,” can also be adjusted in the production process (Doc. 680-1 at 51), as can resilience or springiness, a quality measured by dropping a steel ball onto foam and measuring the ball’s rebound height as a percentage of the drop height (Doc. 584-14 at 19). The paities identify other ways slab-stock can vary (see, e.g., id. 18-19 (describ *231 ing hysteresis, flex fatigue, roller shear, tear strength, air flow, and combustibility qualities of slabstoek and how those qualities are measured)). These characteristics “are determined by the specific formulation of chemicals that are combined to produce [the foam], and are largely independent of each other,” such that high resilience foam can be more or less dense (Doc. 680-1 at 52).

Slabstoek with a given set of characteristics may vary by form (see, e.g., Doc. 682-1 at 49). A bun can be sold in the same form in which it leaves the conveyor belt (Doc. 584-14 at 20). Or, extending the industry’s apparent penchant for baking metaphors, a bun can become a “roll” by slicing the bun lengthwise and then winding the resulting foam sheets into a roll. Or a bun can be otherwise fabricated, either by the bun manufacturer itself or by another firm, through various post-pouring technologies “involving] combining foam with another material, such as a nonwoven substrate or fiber,” or joining foam types together (id. at 21).

Underlay

Underlay, or carpet cushion, is produced in two primary forms. Most commonly, underlay is composed of “rebond” (id. at 23). Re-bond joins together slabstoek “scrap” or “trim” generated during the fabrication process — for example, the waste material created in the course of cutting a bun to some desired shape. The scrap is shredded further, and then combined in a mold with a binding agent. Heat and pressure are applied, forming a rebond bun or log (id.). A rebond bun can then be sliced lengthwise to produce underlay (Doc. 581 at 15-16). A rebond log is “rotated against a knife blade to peel the log into a long sheet of rebond foam padding” (Doc. 584-14 at 23). In both cases, a backing material of some type is applied to the rebond sheets (id.). Two-thirds of rebond production costs are attributable to the price of scrap (Doc. 680-1 at 48 & n,89). Another 20 percent of costs can be traced to the binding agent (Doc. 967 at 73). Not all the rebond-producing Defendants internally source the scrap used to produce rebond. For example, Mohawk “has always fulfilled all of its scrap foam needs on the open market” (Doc. 679-1 at 9).

In recent years, domestic scrap’s place in the rebond “mix” has declined, due in part to the rising price of domestic slabstoek (Doc. 682-1 at 46 (noting a shortage in domestic scrap following Hurricanes Katrina and Rita)). Between 2000 and 2008, underlay producers substantially increased the use of “take up” in producing rebond; “take up” is underlay removed from a prior use in residential or commercial buildings (id. at 43-44; Doe. 680-1 at 46-47). Similarly, over the same period, scrap imports increased (Doe. 682-1 at 43-44; Doc. 680-1 at 46-47). Re-bond demand has declined since 2004, owing to shifts in consumer preferences for wood or tile flooring products (Doc. 581 at 16).

Carpet cushion also can contain “prime underlay.” Prime underlay is not formed from the waste of fabrication operations or post-consumer recycled materials; instead, prime underlay is poured in the same manner as slabstoek, for the immediate purpose of serving as foam for underlay (Doc. 584-14 at 24). Prime underlay is a very small portion of the underlay market, amounting to “less than one percent” of underlay sales in 2009 (id.). See also Doc. 682-1 at 43 n.83; Doc. 679-1 at 27 n.58.

Finally, like slabstoek, underlay of a particular type can vary according to product attributes. For instance, underlay can include noise dampening qualities and antimicrobial coatings. And, like slabstoek, its density can vary. These varying attributes affect underlay prices (Doc. 679-1 at 12).

Plaintiffs and the Proposed Classes

The pending Motions seek certification of classes that include most of Defendants’ immediate customers, as well as all end-use purchasers, for certain foam products. Seven Plaintiffs (Doc. 52 at ¶¶ 6-12) seek certification of a Direct Purchaser class, defined to include (Doc. 584 at 1):

All persons or entities that purchased flexible polyurethane foam (but excluding molded foam) directly from Defendants and/or their co-conspirators from January 1, 1999 to July 31, 2010 for purchase from or delivery into the United States. Excluded from the Class are governmental *232 entities, Defendants, their co-conspirators, and their officers, employees, agents, representatives, parents, subsidiaries and affiliates.

Indirect Purchasers, 1 individual consumers and “authorized managing agents” for hotels and other entities operating in various states, seek certification of a class consisting of (Doc. 577 at 1-2):

All persons or entities in Alabama, Arizona, California, Colorado, [the] District of Columbia, Florida, Hawaii, Illinois, Iowa, Kansas, Maine, Massachusetts, Michigan, Minnesota, Mississippi, Missouri, Nebraska, Nevada, New Hampshire, New Mexico, New York, North Carolina, North Dakota, Oregon, Rhode Island, South Dakota, Tennessee, Vermont, West Virginia, and Wisconsin who purchased products containing flexible polyurethane foam [“product” here defined to include only carpet underlay, bedding, and upholstered furniture products], not for resale, which were manufactured, produced or supplied by Defendants or their unnamed co-conspirators from January 1, 1999 to the present. Excluded from the Class are governmental entities, Defendants, their co-conspirators and their representatives, parents, subsidiaries and affiliates.

Defendants

As a group, Defendants dominate the domestic market for the manufacture and sale of flexible polyurethane foam (“flexible foam”). Defendants’ internal records trace a gradual increase in Defendants’ combined control of the “North American flexible slab production” market over the course of the Class Period — from 80 percent in 1999 to 96 percent of “total U.S. production” in 2010 (Doc. 584-14 at 26-27). See also Doc. 581 at 24 (calculating a Herfindahl-Hirschman Index (“HHI”) 2 as “moderately concentrated” for the 2004 foam market); id. at 25-26 (HHI scores of 2,868 and 2,200 for Defendants’ foam used in the 2009 bedding and furniture markets). A similar upward trend is reflected in Defendants’ share of the underlay market, from 91 percent in 2001 to 94 percent in 2009 (Doc. 584-14 at 27). See also Doc. 581 at 25 (calculating a 2009 HHI score of 2,073). Individual Defendants during the Class Period have engaged in these two general markets, among others, to greater or lesser degrees (see, e.g., Doc. 680-1 at 53 fig. 4 (showing Defendants’ 2005-10 relevant sales, and the percentage of that total sales figure attributable to bun, roll, fabricated foam and carpet cushion sales)).

Carpenter manufactures polyurethane foam products, polyols, polyester fibers, and expanded polystyrene products (Doc. 682-1 at 33). See also Doc. 744-49 at 36 n.151 (noting Carpenter’s self-sourced polyols). Carpenter had $269 million in 2009 foam sales (Doc. 682-1 at 33).

FXI produces foam for the home, healthcare, electronics, industrial, personal care, and transportation markets. FXI’s products include finished goods, sub-assemblies, ser *233 vices, and raw material for OEMs, fabricators, and retailers. FXI had $247 million in 2009 foam sales. In June 2009, FXI purchased assets of the former Foamex International Inc. (“Foamex”) at a bankruptcy sale. Defendants represent FXI as “a distinct company [from Foamex] with different ownership that did not exist before 2009” (id. at 34). (Defendants claim certain Plaintiffs’ experts omit the FXI/Foamex distinction in their references to FXI.)

Future Foam manufactures polyurethane foam for the furniture industry. Future Foam also operates carpet cushion plants and fabricates foam. Future Foam had $221 million in 2009 sales (id.).

Flexible Foam manufactures carpet cushion and flexible foam for bedding, furniture, and specialty applications. Flexible Foam had sales of $269 million in 2009 (id. at 34-35).

Woodbridge primarily manufactures molded foam products for the automotive and transportation industries, and also manufactures rigid foam insulation products, technical foams, and unfabricated and fabricated products (see Doc. 680-1 at 38). Woodbridge had foam sales of $52 million in 2009 (Doc. 682-1 at 35).

Leggett & Platt manufactures products for the residential, automotive, and transportation markets (Doc. 680-1. at 39). Rebond underlay accounts for the overwhelming majority of Leggett & Platt’s sales to direct purchasers (Doc. 679-1 at 5). Leggett & Platt sold its slabstock production and sales operations in March 2007, and now only produces underlay using purchased scrap (id. at 7). Leggett & Platt had $185 million in 2009 underlay sales (Doc. 682-1 at 36).

Hickory Springs similarly manufactures flexible foam products for the furniture and bedding industries, totaling $180 million in 2009 sales (Doc. 682-1 at 36).

Mohawk produces carpet cushion products made from recycled flexible polyurethane foam, latex rubber, and other fibers (id.). Mohawk’s relevant sales were limited to carpet underlay during the Class Period (Doc. 679-1 at 5), with only a “small fraction,” $92 million out of $5 billion, in 2009 total revenues from such underlay sales (Doc. 680-1 at 40).

Vitafoam sold its American polyurethane foam manufacturing operations in 2005-06 (Doc. 683-2 at 8), and differs from other Defendants in two respects: Vitafoam has settled with Direct Purchasers; and Vita-foam is a Department of Justice (“DOJ”) colórate leniency applicant — a decision on that application in still pending (Doc. 682 at 23). Vitafoam also cooperates with Plaintiffs in hopes of securing damages limitations under the Antitrust Criminal Penalties Enhancement and Reform Act of 2004 (“ACP-ERA”), in the event judgment is entered against Vitafoam (Doc. 686 at 5).

Distribution Chains

Earlier discussion identified links in the foam distribution chain. Chemical manufacturers, like BASF, sell TDIs and polyols to slabstock manufacturers, or “foamers,” like Flexible Foam. Flexible Foam might then sell unfabricated foam buns or partially fabricated foam rolls to foam fabricators. Further foam fabrication may be followed by sale of the resulting fabricated foam product to an original equipment manufacturer (“OEM”), like a mattress producer. Or the fabricated foam can be sold directly to a retailer or distributor, without further finishing. In either case, after purchasing flexible foam from a Defendant, an OEM, retailer, or distributor eventually sells a product incorporating flexible foam to an end-use consumer (see Doc. 581 at 17-18, 18 fig. 2; Doe. 680-1 at 129 fig. 25).

The underlay distribution chain is similar to the slabstock and fabricated foam distribution chains. Again, chemical manufacturers sell inputs to foamers who then pour slab-stock. Foamers sell unfabricated slabstock to foam fabricators, who in turn sell trim generated during fabrication of slabstock to rebond producers. Alternatively, rebond manufacturers can purchase take-up from vendors of that input, or imported scrap from brokers, or rely on internal scrap. Rebond is then produced from these primary inputs. The same firm, or a different firm, then fashions underlay from the rebond bun or log. Underlay is then sold to retailers, dis *234 tributors, and buying co-ops. Builders, or homeowners directly, then purchase the underlay for end-use in commercial and residential buildings (Doc. 581 at 19 fig. 3; Doe. 680-1 at 130 fig. 26).

Defendants fill a number of these roles in the supply chain. Carpenter, a producer of polyols, is both a chemical manufacturer and a foamer (Doc. 680-1 at 36), while Mohawk is not a foamer, and instead manufactures only underlay using purchased scrap (id. at 40). The fact that not all Defendants engage in the slabstock and underlay markets to the same extent means that situations like the following hypothetical can arise: Flexible Foam might sell slabstock to Fabricator A. Fabricator A could then sell scrap to Mohawk. Or Flexible Foam could sell its internal scrap to Mohawk, directly or through a scrap broker. It may even be that a Defendant firm could sell an unfabricated bun to a fabricator, and then purchase, from the same fabricator, scrap cast off from the earlier-sold bun for use in rebond production.

Eventually a product reaches its “end-use.” Indirect Purchasers describe three relevant end-use categories: (1) carpet underlay; (2) “bedding (i.e., mattresses, pillows, and toppers)” and; (3) “upholstered furniture products, including upholstered sofas and chairs.” These three categories account for three-fourths of all domestic foam consumer uses (see Doc. 581 at 15 & tbl. 1). In consumption terms, underlay is the largest of these end-use markets, followed by bedding and then furniture (id.). Defendants’ flexible foam products appeal’ in each of the end-use mai’kets to varying degrees (see Doc. 680-1 at 54 fig. 5).

The Alleged Conspiracy

Plaintiffs allege Defendants joined in a conspiracy to fix and raise the price of flexible polyurethane foam, and to allocate customers in that market (Doc. 46 at ¶¶ 1, 3; Doc. 371 at ¶¶ 1, 3). The Direct Purchaser Class Period spans more than eleven years, from January 1, 1999 through July 31, 2010 (Doc. 584 at 1). The Indirect Purchaser Class Period shares the same start date, but runs through the present (Doc. 577 at 1).

Plaintiffs allege that law enforcement first learned of the conspiracy when, in February 2010, Vitafoam sought admission to the DOJ corporate leniency program (Doc. 46 at ¶ 60; Doc. 371 at ¶ 72). DOJ issued Vitafoam a “conditional leniency letter ..., which means that Vitafoam has admitted to its participation in a conspiracy to violate the U.S. antitrust laws” (Doc. 46 at ¶ 61; Doc. 371 at ¶ 73). Further, in connection with Vita-foam’s participation in that program, and in cooperation with Canadian antitrust authorities, Vitafoam employees “revealed the mechanisms, participants, duration, and impact of the conspiracy” (Doc. 46 at ¶ 62; Doc. 371 at ¶ 74).

According to these Vitafoam employees, the conspiracy worked as follows: chemical manufacturers would announce a price increase for TDIs or polyols, providing Defendants the “pretext” for a collusive price increase. Defendants would then resort to an “established practice” to “reach an agreement or understanding” on the amount and timing of price increases for their foam products, by phone or in person. Face-to-face talks were typically held two to three times per year and often coincided with foam industry trade shows. Price increase letters also played an important role in reaching and enforcing the conspiratorial agreement. Pri- or to sending customers formal notice, the exchange of draft price increase letters provided a basis for discussing the level and timing of increases. After publishing price increase letters, exchange of the letters helped Defendants police co-conspirator’s compliance with the agreed-on pricing policy (Doc. 46 at ¶¶ 63-66, 96; Doc. 371 ¶¶ 75-78, 107). See also id. at ¶¶ 95-112 (quoting excerpts of telephone and e-mail conversations that purport to show Defendants “implementing and enforc[ing] the conspiracy”).

Plaintiffs then detail the participation of four Vitafoam employees in the conspiracy, discussing their use of these same coordination practices to agree on anticompetitive pricing and naming other Defendants’ employees who joined in these discussions. These Vitafoam employees include: a former Vitafoam President; the then-current Vita-foam President; a former Vitafoam Vice President of Sales; and a succeeding Vice President of Sales (see Doe. 46 at ¶¶ 67-94; *235 Doc, 371 at ¶¶ 79-105). This last Vitafoam employee also appears as “Witness A” in a Canadian sworn information in support of a search warrant, identifying foam industry employees involved in the conspiracy (Doc. 46 at ¶¶ 83-84 & tbls.; Doc. 371 at ¶¶ 95-96 & tbls.).

Plaintiffs allege that every flexible foam price increase letter issued during the Class Period was “the result of conspiratorial discussions among Defendants on pricing” (Doc. 46 at ¶ 94; Doc. 371 at ¶ 105)

Class Certification Standard

Direct Purchasers seek certification of a nationwide class, comprised of individuals who directly purchased products containing slabstock and underlay from Defendants or their coconspirators (Doc. 584 at 1). Indirect Purchasers seek certification of a class that includes residents of twenty-nine (29) states and the District of Columbia who purchased items falling within three end-use products— underlay, bedding, and furniture products— and containing slabstock or underlay manufactured by Defendants or their co-conspirators(Doc. 577 at 1 n.l). Both Motions seek certification of classes under Federal Civil Rule 23(a) and (b)(3) (Doc. 577 at 1; Doc. 584-1 at 34-38).

To be certified under Rule 23, a putative class action first must satisfy Rule 23(a), which provides:

One or more members of a class may sue or be sued as representative parties on behalf of all members only if:

(1) the class is so numerous that joinder of all members is impracticable;

(2) there are questions of law or fact common to the class;

(3) the claims or defenses of the representative parties are typical of the claims or defenses of the class; and

(4) the representative parties will fairly and adequately protect the interests of the class.

Second, the putative class action must satisfy one of three sets of conditions under subsection (b) of the same Rule. Relevant here, Rule 23(b)(3) requires this Court to determine if:

[Questions of law or fact common to class members predominate over any questions affecting only individual members, and that a class action is superior to other available methods for fairly and efficiently adjudicating the controversy. The matters pertinent to these findings include:

(A) the class members’ interests in individually controlling the prosecution or defense of separate actions;

(B) the extent and nature of any litigation concerning the controversy already begun by or against class members;

(C) the desirability or undesirability of concentrating the litigation of the claims in the particular forum; and

(D) the likely difficulties in managing a class action.

Plaintiffs bear the burden of showing the relevant components of the rule are met. In re Am. Med. Sys., Inc., 75 F.3d 1069, 1079 (6th Cir.1996), Rule 23(a) and (b)(3) require Plaintiffs to establish six requisites before a class may be certified: numerosity, commonality, typicality, adequacy, predominance, and superiority.

Some courts note a seventh “implied” requirement of “ascertainability.” See, e.g., In re High-Tech Employee Antitrust Litig., 289 F.R.D. 555, 563 (N.D.Cal.2013). Both proposed class definitions meet that requirement. For a class to be ascertainable, the proposed definition “must be precise, objective, and presently ascertainable.” It must not rely on “subjective” factors like a class member’s state of mind. “For example, the class may consist of those persons and companies that purchased specified products ... from the defendants during a specified period____” Ann. Manual On Complex Lit. § 21.222 (4th ed.).

At the class certification stage, Plaintiffs must show that there are “in fact sufficiently numerous parties, common questions of law or fact, typicality of claims or defenses, and adequacy of representation” under Rule 23(a), and must further “satisfy through evidentiary proof at least one of the provisions of Rule 23(b).” Comcast Corp. v. Behrend, *236 — U.S. -, 133 S.Ct. 1426, 1432, 1436 , 185 L.Ed.2d 515 (2013) (internal quotation mark omitted) (emphasis in original). The Sixth Circuit recently declined to adopt a standard, embraced in other circuits, that would require “a district court, when deciding whether to certify a class [to] resolve factual disputes by a preponderance of the evidence.” Gooch v. Life Investors Ins. Co. of Am., 672 F.3d 402 , 418 n. 8 (6th Cir.2012) (internal quotation marks omitted). The court concluded there was “no reason to superimpose [on the Circuit’s longstanding “rigorous analysis” standard] a more specific standard”— specifically, a preponderance standard— “than the Supreme Court” did in Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338 , 131 S.Ct. 2541 , 180 L.Ed.2d 374 (2011). Id. But the court also concluded that “the result [in Gooch ] would not differ [if a preponderance standard had been adopted] ... because factual issues [were] not in play.” Id.

The “rigorous analysis” in which this Court must engage may “overlap with the merits of the plaintiffs underlying claim.” Comcast Corp., 133 S.Ct. at 1436 . See Newton v. Merrill Lynch, Pierce, Fenner & Smith, Inc., 259 F.3d 154 , 168 (3d Cir.2001) (“In reviewing a motion for class certification, a preliminary inquiry into the merits is sometimes necessary to determine whether the alleged claims can be properly resolved as a class action.”). But at the same time, this Court has no “license to engage in free-ranging merits inquiries” at this stage of the proceedings. It may consider “[m]erits questions ... to the extent — but only to the extent — that they ai*e relevant to determining whether Rule 23 prerequisites for class certification are satisfied.” Amgen Inc. v. Con. Retirement Plans and Trust Funds, 568 U.S. -, 133 S.Ct. 1184, 1195 , 185 L.Ed.2d 308 (2013). See also Messner v. Northshore Univ. HealthSystem, 669 F.3d 802, 811 (7th Cir.2012) (noting a district court’s class certification inquiry is not “a dress rehearsal for the trial on the merits”); Szabo v. Bridgeport Machines, Inc., 249 F.3d 672, 677 (7th Cir.2001) (“A court may not say something like ... ‘I’m not going to certify a class unless I think that the plaintiffs will prevail’ ”). Where this Court must probe merits questions, it must resolve any material evidentiary disputes. In re Hydrogen Peroxide Antitrust Litig., 552 F.3d 305, 324 (3d Cir.2008) (the task of “[r]esolving expert disputes in order to determine whether a class certification requirement has been met is always a task for” a district court). See also Ellis v. Costco Wholesale Corp., 657 F.3d 970, 982 (9th Cir.2011) (noting a district court erred when “[i]nstead of judging the persuasiveness of the evidence presented, the district court seemed to end its analysis of the plaintiffs’ evidence after determining such evidence was merely admissible”).

This Court’s Rule 23 inquiry requires it to “sift[ ] the [parties’] evidence through the sieve of the legal claims.” In re Whirlpool Corp. Front-Loading Washer Prods. Liab. Litig., 722 F.3d 838, 852 (6th Cir.2013) (“In re Whirlpool Corp.”), cert. denied, Whirlpool Corp. v. Glazer, — U.S. -, 134 S.Ct. 1277 , 188 L.Ed.2d 298 (2014). Plaintiffs must prove: “(1) Defendants violated [the Sherman Act or, in Indirect Purchasers’ case, state-law analogues]; (2) Defendants’ violation caused Plaintiffs to suffer some injury to their business or property (injury-in-fact or impact); and (3) the extent of this injury can be quantified with requisite precision.” In re Cardizem CD Antitrust Litig., 200 F.R.D. 326, 331 (E.D.Mich.2001). Thus, Rule 23 proof is framed in terms of liability, impact, and damages. In addition, because each putative class seeks damages for purchases made during the entire Class period, they must show fraudulent concealment is susceptible of proof on a classwide basis. See Carrier Corp. v. Outokumpu Oyj, 673 F.3d 430, 446 (6th Cir.2012) (listing the three elements of fraudulent concealment, proof of which is necessary to toll applicable statutes of limitations). “The nature of the evidence that will suffice to resolve a question determines whether the question is common or individual.” Blades v. Monsanto Co., 400 F.3d 562, 566 (8th Cir.2005).

Should this Court conclude that either Motion should be granted, this Court must then appoint class counsel, considering under Rule 23(g)(1)(A):

(i) the work counsel has done in identifying or investigating potential claims in th[is] action;

*237 (ii) counsel’s experience in handling class actions, other complex litigation, and the types of claims asserted in th[is] action;

(iii) counsel’s knowledge of the applicable law; and

(iv) the resources that counsel will commit to representing the class.

This Court may also consider other relevant factors outlined in Rule 23(g)(l)(B)-(D). Appointed class counsel must “fairly and adequately represent the interests of the class.” Rule 23(g)(4).

Plaintiffs Satisfy Rule 23(a)

Each putative class satisfies the numerosity requirement of Rule 23(a)(1), a point Defendants do not dispute. “[W]hile there is no strict numerical test, ‘substantial’ numbers usually satisfy the numerosity requirement.” Daf fin v. Ford Motor Co., 458 F.3d 549, 552 (6th Cir.2006). As noted below in this Court’s predominance discussion, expert analysis shows each proposed class contains (at minimum) thousands of members.

Nor do Defendants seriously dispute that each putative class has identified “questions of law or fact common to the class.” Rule 23(a)(2). Plaintiffs need identify only one common question of law or fact for purposes of “commonality.” See Wal-Mart Stores, Inc., 131 S.Ct. at 2556 . See also Sprague v. General Motors Corp., 133 F.3d 388, 397 (6th Cir.1998). The resolution of a common question must “drive the resolution of the litigation.” Wal-Mart Stores, Inc., 131 S.Ct. at 2551 . See also In re Deepwater Horizon, 739 F.3d 790, 811 (5th Cir.2014) (characterizing a plaintiffs commonality burden as requiring “evidence to demonstrate that a particular contention is common, but not that it is correct” or true). Plaintiffs’ identified common questions do just that — they are common and critical, in that resolution of all putative class members’ claims depend on a single answer provided by a factfinder (see Doc. 584-1 at 35 (noting common questions to include the existence of the alleged price-fixing and customer allocation conspiracy, antitrust injury, and aggregate damages calculations); Doc. 578 at 23-24 (same)).

Rule 23(a)(3) demands that “claims or defenses of the representative parties [be] typical of the claims or defenses of the class.” This element of the Rule serves to “limit the class claims to those fairly encompassed by the named plaintiffs claims.” Gen. Tel. Co. of the Nw., Inc. v. EEOC, 446 U.S. 318, 330 , 100 S.Ct. 1698 , 64 L.Ed.2d 319 (1980). “[A] plaintiffs claim is typical if it arises from the same event or practice or course of conduct that gives rise to the claims of other class members, and if his or her claims are based on the same legal theory.” In re Am. Med. Sys., Inc., 75 F.3d at 1082 .

Defendants argue in passing that the claims of certain Indirect Purchasers are atypical of the class claims (Doc. 680 at 22). Defendants note that two Indirect Purchasers — Kathleen Nolan and Kirsten Luenz— received rebates or other discounts when purchasing the products on which their claims are predicated — in Nolan’s case, a pillow, and in Luenz’s case, mattresses (id.). But Defendants do not, and cannot, dispute that Nolan and Luenz assert claims based on a legal theory, which, if proven, would also entitle those less thrifty class members to recover. See Sprague, 133 F.3d at 399 (noting typicality does not exist if “a named plaintiff who proved his own claim [does] not necessarily prove[] anybody else’s claim.”). See also In re Cardizem CB Antitrust Litig., 200 F.R.D. 297, 304 (E.D.Mich.2001) (“[C]laims in antitrust price-fixing cases generally satisfy Rule 23(a)(3)’s typicality requirement, even if members purchase different quantities and pay different prices”). If Nolan and Luenz secure damages in this ease, they would do so only because a factfinder has accepted Indirect Purchasers’ proof establishing the alleged conspiracy and found credible and persuasive expert analysis of impact and damages. So too for the other class members’ claims.

Defendants’ standing-based challenges to Nolan and the Parker entities fare no better. (See Doe. 742 at 79-80 (Nolan testimony establishing Nolan and her spouse jointly purchased the pillow on which Nolan’s claim depends); id. at 82, 84, 86, 88 (assigning to Parker claims of the Parker-related enti *238 ties)). Similarly, Defendants argue that “the entities on whose behalf Driftwood is suing specifically disclaim any legal agency relationship between Driftwood and them” (Doc. 680 at 23). That assertion is based on the deposition testimony of a Driftwood corporate deponent who does not offer any legal conclusion as to whether an agency relationship exists between Driftwood and the various hotels on whose behalf Driftwood sues (Doc. 680-5 at 5). Nor do the management agreements preclude Driftwood from suing on a managed hotel’s behalf — in fact, each of the management agreements obligates Driftwood to act in that capacity. For five hotels, the management agreements specifically identify Driftwood as an “agent” of the hotel it manages and, further, obliges Driftwood to sue on the hotels’ behalf (see Doc. 742 at 95, 105, 118-19, 130, 143-44, 157,172, 182). For three other hotels, Driftwood is not described as an “agent,” but is nonetheless required to sue on the managed hotel’s behalf (id. at 198, 232, 270). Thus, typicality exists with respect to the claims of the putative classes’ representatives.

Finally, Rule 23(a)(4) requires a finding that “the representative parties will fairly and adequately protect the interests of the class.” This component of Rule 23 focuses this Court’s attention on the qualities of both the class representatives and proposed class counsel. “The representative must have common interests with unnamed members of the class, and [ ] it must appear that the representatives will vigorously prosecute the interests of the class through qualified counsel.” Vassalle v. Midland Funding LLC, 708 F.3d 747, 757 (6th Cir.2013) (alterations omitted) (quoting In re Am. Med. Sys., Inc., 75 F.3d at 1088 ). But “[b]eeause few people are ever identically situated, it is easy to paint an image of the class representative’s interests as peripherally antagonistic to the class. That depiction does not make [a] plaintiff an inadequate representative.” Gooch, 672 F.3d at 429 . See also Matamoros v. Starbucks Corp., 699 F.3d 129, 138 (1st Cir.2012) (“Put another way, to forestall class certification the intra-class conflict must be so substantial as to overbalance the common interests of the class members as a whole.”). A supposed conflict should doom class certification only if that conflict is “fundamental.” Ward v. Dixie Nat. Life Ins. Co., 595 F.3d 164, 180 (4th Cir.2010).

Plaintiffs identify the proposed class representatives’ “common interests” to establish the existence of the alleged price-fixing conspiracy and to recover damages (Doe. 584-1 at 37; Doc. 578 at 25). Each touts the skills of proposed class counsel, and argue the time and resources invested by proposed class counsel to date show proposed class counsel will vigorously litigate the class claims (Doe. 584-1 at 37-38; Doc. 578 at 26).

Defendants do not dispute either group of proposed class counsel are “qualified counsel” for the purposes of Rule 23. This Court agrees. Instead, Defendants point to characteristics of specific class representatives, and argue these characteristics create conflicts between class representatives, as well as with the class. With respect to Direct Purchasers, Defendants note the breadth of the proposed class — encompassing slabstock, underlay, and fabricated foam— and then emphasize two facts: five of seven class representatives produced and sold fabricated foam “in direct horizontal competition with Defendants” and so made sales of fabricated foam that “benefitted” from the inflated pricing environment created by the alleged conspiracy (Doc. 682 at 65); and two of seven class representatives sourced large portions of their flexible foam purchases from firms who are not defendants, and so are “hardly representative of purchases by the class Plaintiffs seek to create.” (Id. at 66). Direct Purchasers counter that in earlier downstream discovery orders (see Docs. 420 & 458), this Court rejected the premise of Defendants’ conflicts argument (see Doc. 744 at 6).

This Court concludes Direct Purchasers satisfy the adequacy requirement. That two of the seven class representatives purchased large amounts of flexible foam products from non-defendant firms does not change the fact that the same class representatives also seek recovery of alleged antitrust overcharges on purchases from Defendants. See In re Warfarin Sodium Antitrust Litig., 391 F.3d 516 , 532 (3d Cir.2004). This is hardly a “funda *239 mental conflict” (if a conflict at all). See Schlaud v. Snyder, 717 F.3d 451, 458 (6th Cir.2013) (noting a “clear conflict” between named representatives, all of who opposed union representation “in any form,” and members of the class, many who approved the collective bargaining agreement and its challenged mandatory dues provision). Nor is Defendants’ conflicts argument, touching on the competitor foam fabricator class representatives, a basis for denying certification of a Direct Purchaser class. This Court’s prior downstream discoveiy orders do not resolve this issue, but they come close.

Specifically, Defendants’ conflicts argument necessarily would require the competitor foam fabricators who wish to serve as class representatives to not only show they have individual standing to bring an antitrust claim (Doe. 584-14 at 130), but to also show lost profits (or lower profit margins) during the Class Period in the relevant areas of competition. After all, how else could these competitors avoid Defendants’ painting them as “benefitting” from the alleged conspiratorial pricing? But the Hanover Shoe line of eases, embraced by this Court in its downstream discovery orders, deem an antitrust plaintiffs lost profits irrelevant for purposes of impact and damages:

It is also clear that if the buyer, responding to the illegal price, maintains his own price but takes steps to increase his volume or to decrease other costs, his right to damages is not destroyed. Though he may manage to maintain his profit level, he would have made more if his purchases from the defendant had cost him less. We hold that the buyer is equally entitled to damages if he raises the price for his own product. As long as the seller continues to charge the illegal price, he takes from the buyer more than the law allows. At whatever price the buyer sells, the price he pays the seller remains illegally high, and his profits would be greater were his costs lower.

Hanover Shoe, Inc. v. United Shoe Mach. Corp., 392 U.S. 481, 489 , 88 S.Ct. 2224 , 20 L.Ed.2d 1231 (1968). See also Hawaii v. Standard Oil Co. of Cal., 405 U.S. 251 , 262 n. 14, 92 S.Ct. 885 , 31 L.Ed.2d 184 (1972).

This Court joins those courts that have found the logic of Hanover Shoe extends to also reject conflicts arguments based on the fact that a representative plaintiff may have “benefitted” from the anticompetitive conduct they challenge. Teva Pharm. USA, Inc. v. Abbott Labs., 252 F.R.D. 213, 227 (D.Del.2008); Meijer, Inc. v. Warner Chilcott Holdings Co. III, Ltd., 246 F.R.D. 293, 304 (D.D.C.2007) (holding “Defendants’ arguments that the [representative parties] actually benefitted from the delayed entry of [a generic drug] into the market due to the generic bypass phenomenon are irrelevant as a matter of law, and cannot serve to demonstrate that a conflict exists between Plaintiffs’ interests and those of the [“benefitting” parties] with respect to this litigation”). Defendants’ conflicts arguments, based on the fact of foam fabricator competition, are speculative, and are leveled in the face of a proposed damages methodology that would first calculate classwide damages, and then award each representative party and class member the full amount of damages to which the Sherman Act entitles them, based on transaction-level data. See Newberg on Class Actions § 3:64 (5th ed.) (noting “this form of conflict should not preclude a finding of adequacy on merely speculative terms as, for example, it will almost always be the case that some member in a large class prefers the status quo for some reasons”). And in any event, were these hypothetical conflicts to become conflicts in fact (e.g., in the context of a class settlement), mechanisms exist for addressing this issue. See Kohen v. Pacific Inv. Mgmt. Co. LLC, 571 F.3d 672, 680 (7th Cir.2009).

Plaintiffs Satisfy Rule 23(b)(3)

In recognition of the fact that Rule 23’s predominance requirement is “more stringent” than other elements of the Rule, Amchem Prods. Inc. v. Windsor, 521 U.S. 591, 609 , 117 S.Ct. 2231 , 138 L.Ed.2d 689 (1997), the overwhelming focus of the briefing discusses whether Direct or Indirect Purchasers can show that common questions of fact or law predominate over those questions “affecting only individual members.” Rule 23(b)(3). To carry their Rule 23(b)(3) burden, Direct and Indirect Purchasers offer *240 documentary evidence, deposition testimony, and the expert reports of Leitzinger, Krieger, Gordon (for Direct Purchasers), and Lamb (for Indirect Purchasers). Defendants reply in kind, offering contrary expert reports of Ordover, Burtis (for Mohawk and Leggett & Platt), the Sentinel Group, and Maness (for Vitafoam). 3

At this stage of the litigation, Plaintiffs’ burden as it relates to predominance is “not to prove [ (for example) ] the element of antitrust impact.” In re Hydrogen Peroxide Antitrust Litig., 552 F.3d at 311. Plaintiffs must instead show that the essential elements of their claims are “capable of proof at trial through evidence that is common to the class rather than individual to its members.” Id. at 311-12 (emphasis added). This inquiry necessarily requires this Court to form “some prediction as to how specific issues will play out” in terms of trial proof, particularly when a class certification decision is made on the basis of an incomplete discovery record. In re New Motor Vehicles Canadian Exp. Antitrust Litig., 522 F.3d 6, 20, 27 (1st Cir.2008). The predominance inquiry gauges whether a proposed class is cohesive enough to “warrant adjudication by representation.” Beattie v. CenturyTel, Inc., 511 F.3d 554, 564 (6th Cir.2007) (internal quotation marks omitted) (quoting Amchem Prod., Inc., 521 U.S. at 632 , 117 S.Ct. 2231 ).

The Sixth Circuit has expanded on this predominance burden at some length, in the course of examining recent Supreme Court class action jurisprudence. See In re Whirlpool Corp., 722 F.3d at 858-61 . In this Circuit’s view, Amgen instructs that a plaintiff need not “prove that each element of a claim can be established by classwide proof: What the rule does require is that common questions predominate over any questions affecting only individual [class] members.’” Id. at 858 (emphasis and alterations original) (internal quotation mark omitted) (quoting Amgen Inc., 133 S.Ct. at 1196 ). Comcast did not alter that burden: it merely “reaffirms the settled rule that liability issues relating to injury must be susceptible of proof on a classwide basis to meet the predominance standard.” Id. at 860. See also id. (concluding Amgen and Comcast “are premised on existing class-action jurisprudence”). But see In re Rail Freight Fuel Surcharge Antitrust Litig., 725 F.3d 244, 255 (D.C.Cir.2013). All that is needed is “common evidence and methodology,” not “also common results for members of the class.” Butler v. Sears, Roebuck & Co., 727 F.3d 796, 801 (7th Cir.2013), cert. denied, — U.S. -, 134 S.Ct. 1277 , 188 L.Ed.2d 298 (2014). See also Amgen Inc., 133 S.Ct. at 1196 (noting putative class representatives “need not, at th[e clas certification] threshold, prove that the predominating question will be answered in their favor”).

Defendants attempt to distinguish In re Whirlpool Corp., noting the case involved products liability claims with impact occurring at the point of sale (see Doc. 682 at 21 n.3). But while that ease’s discussion of the predominance standard occurred in a specific factual context — as all Rule 23 cases do— there is no reason to conclude In re Whirlpool Corp.’s instructive analysis is limited to products liability class action litigation. After all, In re Whirlpool Corp. expounds Com-cast and Amgen, which undoubtedly control this Court’s analysis. See Merenda v. VHS of Mich., Inc., 296 F.R.D. 528, 548 *241 (E.D.Mich.2013) (citing In re Whirlpool Corp. for its predominance analysis in the course of certifying a plaintiff class asserting federal antitrust claims). See also Cason-Merenda v. VHS of Mich., Inc., 2014 WL 905828 (E.D.Mich.2014) (“reinstating in full” the district court’s prior opinion and order, following an order from the Sixth Circuit directing the district court to reconsider its prior decision in light of Comcast ). 4

Analysis of whether Plaintiffs satisfy Rule 23’s predominance requirement “begins, of course, with the elements of the underlying cause of action.” Erica P. John Fund, Inc. v. Halliburton Co., 563 U.S. 804 , 131 S.Ct. 2179, 2184 , 180 L.Ed.2d 24 (2011). Therefore, this Court wades through the parties’ proof by examining liability, impact, damages, and, finally, fraudulent concealment. This Court conducts a “rigorous” analysis of Direct Purchasers’ predominance showing, followed by a similar analysis of Indirect Purchasers’ attempt to satisfy the same requirement.

* * *

Direct Purchasers’ Liability Proof

Direct Purchasers allege that the conspiracy centered on the issuance of price increase letters “close in time and for the same or similar amounts and effective dates” throughout the Class Period (Doc. 584-1 at 16). Direct Purchasers submit examples of these price increase letters (584-7 at 2-188). Similarly, Leitzinger prepared a chart, listing for each quarter the Defendants who issued price increase letters; the percentage price increase reflected in the letter; and the foam products to which the letter applied (Doe. 584-14 at 121-22). The price increase letters typically attribute increased foam prices to increases in raw materials prices. Or more specifically, the price increase letters claim that “[suppliers of polyol and TDI have substantially increased their prices to the urethane industry” or that the cost of trim had risen (Doc. 584-7 at 43-44,149). These price increase letters typically applied to broad product segments (see, e.g., id. at 11) (Foamex letter announcing price increase for “all grades of Polyurethane foam”), and provided the starting point for price negotiations with customers (Doc. 584-9 at 163).

Direct Purchasers offer witness declarations that throughout the Class Period there were communications between foam industry competitors “to coordinate the percentage amount and timing of price increases for foam” so that individual Defendants could enter account negotiations “knowing that [] competitors had also issued the same or a nearly identical increase,” limiting a customer’s ability to “play [competitors] against [ ] competitors and either negotiate a much lower increase or no increase at all” (Doc. 584-11 at 14-15). See also Doc. 584-9 at 8 (describing Defendants’ discussions about, and the exchange of, price increase letters as aimed at helping “push through price increases” for foam products); Doc. 584-15 at 39; id. at 10-11 (noting that “purchasing agents” for customers are “paid to tell stories” about a Defendant’s competitors’ pricing decisions).

Direct Purchasers argue that five categories of documentary or deposition evidence establish the patterns in price increase letters resulted from the price-fixing conspiracy. First, Direct Purchasers offer “direct evidence” of communications between Defendants about foam pricing (see, e.g., 584-2 at 58-59; Doc. 584-3 at 2 (Leggett & Platt email noting Carpenter, Foamex, and Future Foam price increases and Leggett & Platt’s intention to “have our letters ready to mail to selected accounts”); Doc. 584-4 at 11; id. at 59 (Leggett & Platt email recounting sender’s price discussions with Vitafoam and Hickory Springs, and noting “Carpenter wants Foamex to lead”)). This “direct evi *242 dence” includes materials that Direct Purchasers claim show “advance knowledge” of a competitor’s impending price increases (see, e.g., Doc. 584-5 at 91 (Leggett & Platt employee e-mailing to Flexible Foam employee an “initial draft” price increase letter, expected to be mailed to customers one or two business days later); id. at 82 (draft Foamex letter with “tracked changes,” notifying customers of Foamex plans to “allocate products to [Foamex] customers,” produced in discovery in this matter by Carpenter)). Direct Purchasers also offer documents in which the employees of Carpenter and Woodbridge make none-too-subtle allusions to price-fixing (see, e.g., id. at 124 (email of Woodbridge employee, concluding with the observation that “I think we should go strong and tr[y] to keep Vita and [Foamex] on board. Forget [market] share for now — fix pricing.” (line break omitted))

Direct Purchasers also collect a series of faxed price increase letters, in which the letter’s header shows the transmitting Defendant, while the Bates stamp shows the Defendant that (directly or indirectly) received the letter and later produced the document in discovery. For instance, Vitafoam produced a March 2000 Foamex price increase letter (Doc. 584-6 at 6), while Hickory Springs produced a June 2004 Future Foam price increase letter (id. at 12). Direct Purchasers offer a second compilation of price increase letters, sent by fax or e-mail, except this collection shows one Defendant transmitting to another Defendant the price increase letters of a third Defendant or group of Defendants (see, e.g., id. at 123-24 (Future Foam employee sending “Carpenter Price Increase Letter Third one” to Vitafoam employee)). Defendants also exchanged price increase letters in face-to-face meetings (Doc. 584-15 at 67).

Second, Direct Purchasers point to “indirect communications” between Defendants. Specifically, Direct Purchasers offer evidence that Defendants communicated about price coordination through third-parties. See, e.g., Doc. 584-8 at 38 (scrap broker sending Carpenter a Leggett & Platt price increase letter); id. at 42-44 (scrap broker sending Carpenter the price increase letters of Hickory Springs, Scottdel, and Flexible Foam); Doc. 584-8 at 71-72; Doc. 584-9 at 139 (email from Woodbridge employee to WoodbridgeHickory Springs joint venture employee, asking for, and later receiving, price increase information on “what Hickory is doing”).

Third, Direct Purchasers offer evidence that, during periods in which price increase letters were issued, Defendants refused new customers because a new “customer would only come[ ] to [a Defendant] in an attempt to beat the price increase” and the price increase was “more important to [a Defendant] than a little additional business” (Doc. 584-8 at 75).

Fourth, Direct Purchasers emphasize the “admissions” of Vitafoam, made during Vita-foam’s 30(b)(6) deposition in related litigation. Specifically, Vitafoam’s corporate witness agreed that the Vitafoam Defendants— Vitafoam Canada and Vitafoam, Inc. — had “communicated and reached understandings on the percentage amount and timing of price increases and market allocation in the sale of polyurethane foam” (Doe. 584-10 at 13). Vitafoam’s witness specifically identified FXI, Woodbridge, Carpenter, Hickory Springs, and Future Foam as the Vitafoam partners in these “understandings” (Doe. 584-1 at 29).

Finally, Direct Purchasers note that twenty employees of four Defendants have invoked their Fifth Amendment privileges against self-incrimination in the course of depositions in this matter (id. at 30 n. 1 (cataloguing witness who invoked the privilege as of summer 2013)).

Direct Purchasers argue all of this evidence — the price increase letters themselves and patterns among Defendants in issuing price increase letters, direct and indirect communications between Defendants about pricing, certain Defendants’ stated refusal to pursue new customers through price competition during a period in which a price increase letter was in effect, the Vitafoam “admissions,” and deponents’ Fifth Amendment invocations — establish that Direct Purchasers “will seek to show through common proof at trial that Defendants coordinated the substance and timing of price increase letters on a national level to all customers for all flexi *243 ble polyurethane foam products each defendant sold.” Direct Purchasers will use the same proof to answer for each class member questions like “did the conspiracy exist” and “for how long did it exist?” (see Doe. 584-1 at 42), 5

Defendants counter with two related arguments. First, Defendants argue the price-fixing conspiracy alleged in the Complaint did not exist, a fact borne out by the failure of discovery to “yield[] evidence of any agreement among foam manufacturers in the [United States] to fix the timing or content of price increase” letters (Doc. 682 at 24). Defendants characterize the declarations of the Domfoam and Valle Foam witnesses as expressly disclaiming “that they ... ever entered into an agreement with any Defendant to fix prices or allocate customers” (id.). See also Doc. 682-6 at 68 (“Q: So ... [was] your decision to pass on a price increase [letter] ... the result of an agreement with another company ... ? [objection omitted] A: No, [it was not result of an agreement] with anybody”); Doe. 682-7 at 4 (same); Doc. 682-8 at 3 (same)). But see Doc. 744-2 at 17 (testifying that in the aftermath of chemical price increases “[m]eetings weren’t held by all the foamers saying, ‘Okay. Now it is time. We’ve got to put letters out.’ But on an informal basis, ‘I’m raising prices, and its going to be on this date and this percent. And here’s evidence of what I’m going to do [in the form of exchanged price increase letters]’____ So that’s as sophisticated as we got in assuring that everybody was on the same page in terms of getting prices up when there was a chemical price increase.”) (paragraph break omitted)).

Defendants cast Vitafoam’s corporate depositions, taken in related litigation, in the same light (Doc. 682-5 at 4). But see Doc. 584-10 at 13 (relating testimony of Vita-foam’s corporate witness that she was “struggling with, ... the word ‘agreement,’ ” but agreeing that Vitafoam reached “understandings” with other Defendants as to the “percentage amount and timing of price increases”). In the absence of specific evidence of express agreements to fix prices, Defendants argue that Direct Purchasers merely attempt — unsuccessfully—to transform “field chatter and attempts to gather competitive intelligence into” the conspiracy (Doc. 682 at 25) (quotation marks omitted).

Second, and similarly, Defendants argue market reality, consistent with legal and economic theory predictions of behavior in concentrated markets, shows only independent behavior by firms faced with the same supply and demand conditions. Defendants argue these independent decisions are bare conscious parallelism, or “the process, not in itself unlawful, by which firms in a concentrated market might in effect share monopoly power, setting their prices at a profit-maximizing, supracompetitive level by recognizing their shared economic interests and their interdependence with respect to price and output decisions.” Brooke Grp. Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 227 , 113 S.Ct. 2578 , 125 L.Ed.2d 168 (1993).

It would, of course, be inappropriate for this Court to decide, on the basis of an incomplete discovery record and in the context of a motion for class certification, that Defendants are correct in arguing the conspiracy did not in fact exist. This Court’s *244 inquiry focuses instead on whether, on the basis of this record, the element of antitrust liability is susceptible of proof through evidence common to the class.

Defendants may be correct in the significance they attach to Direct Purchasers’ liability evidence. But if so, that fact simply would mean summary judgment would be appropriate as to the Direct Purchaser class, or that a trier of fact should reject Direct Purchasers’ liability arguments. Defendants do not succeed in showing liability questions — however answered — cannot be answered thi-ough common proof. Like the defendant in In re Whirlpool Corp., Defendants simply point to “a fatal similarity!!,] an alleged failure of proof as to an element of the [Direct Purchasers’] cause of action,” that this Court may not yet resolve. In re Whirlpool Corp., 722 F.3d at 859 . This Court concludes that, considered as a whole, the Direct Purchaser class certification evidentiary record does support such a finding.

Direct Purchasers’ Impact Proof

“[T]he task for plaintiffs at class certification is to demonstrate that the element of antitrust impact is capable of proof at trial through evidence that is common to the class rather than individual to its members.” In re Hydrogen Peroxide Antitrust Litig., 552 F.3d at 311-12 . Direct Purchasers need only produce a method of proof capable of showing “some damage flowing from the unlawful conspiracy; inquiry beyond this minimum point goes only to the amount and not the fact of damage. It is enough that the illegality is shown to be a material cause of the injury.” Zenith Radio Corp. v. Hazeltine Research, Inc., 395 U.S. 100 , 114 n. 9, 89 S.Ct. 1562 , 23 L.Ed.2d 129 (1969) (emphasis added). Direct Purchasers must show “that the class members paid a higher price for [foam products] purchased from Defendants than they would have absent the existence of a conspiracy.” In re Titanium Dioxide Antitrust Litig., 284 F.R.D. 328, 340 (D.Md.2012) amended, 962 F.Supp.2d 840 (D.Md.2013). To show impact is susceptible of proof on a elasswide basis, Direct Purchasers must show “widespread impact,” or that all or nearly all class members suffered injury:

What is true is that a class will often include persons who have not been injured by the defendant’s conduct; indeed this is almost inevitable because at the outset of the case many of the members of the class may be unknown, or if they are known still the facts bearing on their claims may be unknown. Such a possibility or indeed inevitability does not preclude class certification, despite statements in some cases that it must be reasonably clear at the outset that all class members were injured by the defendant’s conduct. Those cases focus on the class definition; if the definition is so broad that it sweeps within it persons who could not have been injured by the defendant’s conduct, it is too broad.

Kohen, 571 F.3d at 677 (emphasis added) (citations omitted).

Direct Purchasers’ proposed proof on impact comes in four forms. First, Direct Purchasers refer back to evidence, discussed above in the context of antitrust liability, purporting to show Defendants “systematically and collusively coordinated ... nearly-simultaneous price increase” letters that applied to all of a Defendant’s flexible foam products (see Doc. 584-7 at 42), or to all products in a foam sub-market (see id. at 87). Second, Direct Purchasers state the price increase letters did not just announce price increases, but paved the way for actual price increases after individual negotiations with customers. Third, Direct Purchasers point to foam’s commodity nature, which dictates that competition occur primarily on the basis of price. And finally, Direct Purchasers offer Dr. Leitzinger’s analysis of market structures and his impact models (Doc. 584-1 at 44-47). Defendants vigorously dispute that any of these facts can support classwide proof of impact, relying heavily on dueling expert reports, which are examined next. See In re Initial Pub. Offerings Sec. Litig., 471 F.3d 24, 42 (2d Cir.2006) decision clarified on denial of reh’g., 483 F.3d 70 (2d Cir.2007). Leitzinger’s assertion that there is evidence “common to members of the proposed class ... which leads to the conclusion that all or virtually all members of the proposed class have been impacted by the alleged conspiracy” rests on three analytical components, summarized in turn below: (1) *245 inferences drawn from market structure; (2) Defendants’ pricing behavior; and (3) statistical evidence of impact (Doc. 584-14 at 42).

Market Structure

Leitzinger’s review of the discovery to date leads him to conclude that there are five characteristics of the flexible foam market that make this case susceptible to a showing that all or nearly all direct purchasers suffered antitrust injury. Fust, throughout the Class Period Defendants controlled the vast majority of the slabstock and underlay markets. Defendants’ own internal estimates of concentration in the relevant sub-markets bear this out. A series of Foamex and FXI estimates show Defendants, as a group, gradually increasing their control of the slabstock market until achieving almost total control of that market: in 2000, Defendants controlled 85 percent of the slabstock market; that number rose to 89 percent in 2003; and to 96 percent in 2010 (id. at 26-27). Market concentration is actually more pronounced, because Foamex estimated in 2009 that just four Defendants accounted for 87 percent of domestic slabstock production (id. at 27-28). Leggett & Platt estimated in 2001 that 91 percent of the underlay market rested in the hands of Defendants, while a 2009 FXI document estimated Defendants held 94 percent market share in that industry (id. at 27).

The upshot of all these concentration figures is straightforward. With such a large portion of the relevant sub-markets controlled by Defendants, it is unlikely that a substantial number of direct purchasers would have avoided the alleged conspiracy’s price effects. And, high market concentration, particularly with certain Defendants’ focus in one sub-market, reduces coordination costs, increases the likelihood that a price-fixing conspiracy would have been successful, and increases the scope of the conspiracy’s likely price effects.

Second, several barriers to entry substantially impede firms from entering the slab-stock or underlay markets. A Foamex investment analysis concludes barriers to entry are “relatively high” in the industry, specifically observing that to be competitive with existing market participants, new entrants must: be able to purchase foam inputs on a large scale to take advantage of purchasing economies; make substantial capital investments before beginning foam production; possess certain expertise in large-scale foam production processes; and gain access to large foam direct purchasers, like Wal-Mart. Moreover, foam inputs, including TDI, are heavily-regulated hazardous materials, the proper handling of which may be cost-prohibitive for smaller firms looking to penetrate the foam market. Producing foam itself subjects a firm to still more regulatory requirements, such as federal and state clean air standards (id. at 29-30).

Even if a firm could overcome these initial barriers with respect to either slabstock or underlay, Defendants’ substantial excess capacity in each sub-market would allow Defendants to flood either market with increased production, causing slabstock or underlay prices to plummet, thus preventing the new entrant from recouping on the investment needed to enter the market (id. at 31-35). These various barriers block a non-conspirator firm from newly entering the slabstock or underlay markets in hopes of undercutting the alleged cartel’s supracompetitive pricing decisions, and support a finding that common evidence exists to show that all or nearly all direct purchasers suffered antitrust injury (id. at 28-29 (“[ejntry barriers discourage new competitors (who might otherwise be attracted by inflated prices) who could in turn destabilize the conspiracy or create pockets of competitive pricing that would limit the scope of the conspiracy’s impact.”)).

Third, demand for flexible foam products is inelastic. For the effect of a price conspiracy to be widespread, a price-increase driven decline in sales volume must “not outweigh the benefit of higher prices” (id. at 35). Put differently, collusion is more likely to occur, and its effects are more likely to be widespread, when conspirators can profit more from selling less of a product at a higher, supracompetitive price than by selling more of a product at a lower, competitive price. A cartel’s chance of fixing a supracompetitive price that pulls off that balancing act increases in tandem with the inelasticity of the product’s demand. And that inelasticity depends, in part, on the existence (or not) of *246 acceptable substitutes for the product (id. at 35-36), or the product’s share of end-use product costs (id. at 39).

Leitzinger bases his conclusions that demand for flexible foam is inelastic on discovery materials that describe flexible foam gradually supplanting alternative cushioning materials, like polyester fibers, natural latex, and resinated cotton. A 1996 “Economic Impact Analysis,” authored by the U.S. Environmental Protection Agency, compares flexible foam to its “primary substitute for slabstock in cushioning applications,” polyester fiber, noting flexible foam’s superior product qualities (id. at 37-38 (citing similar comparisons generated by the Polyurethane Foam Association and La-Z-Boy, Inc.)). Leitzinger offers similar observations with respect to bedding applications (id, at 38) and underlay (id. at 39). Because slabstock, as an intermediate material, comprises a relatively small share of end-use product costs, furniture and bedding manufacturers are relatively insensitive to flexible foam price increases (id. at 39-40).

Fourth, there exists evidence establishing that slabstock and underlay are “commodities,” as economic literature defines that term. The more closely flexible foam products resemble a prototypical commodity, like gasoline of a given grade, the more cohesive an industry cartel will be and the more widespread the price effects. This is so because cartel members would need to reach agreement, and then police that agreement, only with respect to one product characteristic: price (id. at 43). Though Defendants produce a wide range of slabstock grades, varying according to product characteristics like foam density or IFD, discovery materials show Defendants and their customers could identify and compare each Defendant’s version of a given slabstock grade (see, e.g., id. at 40-41 (citing Flexible Foam documents comparing equivalent but differently-named slabstock foam grades for bedding uses between Flexible Foam, Foamex, and Carpenter)). An FXI employee testified that foam fabricators would buy buns “from anyone who had the lowest price that day” (id. at 41). Similar evidence is offered with respect to underlay, which Hickory Springs, Leggett & Platt, Foamex, and Carpenter all described as commodity-like products (id. at 41-42), with Mohawk making similar statements (id. at 42). Here, too, variability exists among underlay grades (see id. at 42 n.216 (explaining underlay grades as largely a function of density or specific product qualities like antimicrobial properties)), but again Defendants and their customers were able to compare these various grades on a price basis (id. at 42).

Fifth, because TDIs and polyols comprise such a large portion of flexible foam’s input costs, and because chemical manufacturers noticed similar price increases for these key inputs at similar times, there exists further evidence that the cartel’s price effects likely were widespread. All Defendants faced similar price increases from the chemical manufacturers. Therefore, Defendants had “ready focal point[s]” to use “as triggers for jointly established announcements regarding slabstock price increases,” further reducing the difficulties the alleged cartel would have faced in its efforts to agree on pricing (id. at 44).

In sum, then, Leitzinger draws inferences from the flexible foam industry’s market structure to paint that industry as one in which a few dominant firms, protected from outside competition by high barriers to market entry and using common foam input price announcements, succeeded in fixing the price of a commodity with an inelastic demand function. All those qualities provide common evidence that all or nearly all direct purchasers would have been impacted by the alleged cartel.

Pricing Behavior

Leitzinger further supports his common impact conclusion by pointing to the manner in which prices are set in the foam industry. Three aspects of Defendants’ pricing behavior would allow the conspiratorial conduct— fixing prices — to impact all or virtually all direct purchasers. First, price increase letters were not tailored to any particular customer or to any particular subset of foam products. In Carpenter’s case, central management drafted slabstock and underlay price increase letters. Once drafted, letter copies would be sent to local sales staff, who *247 would add a customer’s contact information and the local salesman’s signature (id, at 45). Each such letter contained some version of the salutation “Dear Customer” (id.). Moreover, with few exceptions, the percentage price increase reflected in each letter was uniform across Defendants’ foam product lines (id. at 46-47 (noting Carpenter, Flexible Foam, FXI, Future Foam, Hickory Springs, Leggett & Platt, Mohawk, Vitafoam, and Woodbridge letters signaled price increases across all slabstock or underlay products)).

Second, Defendants typically (though not always) were able to transform these generally-applicable price increase announcements into effective price increases. One indicator of this success, according to Leitzinger, is the repeated insistence of pricing executives that no customer be excepted from the price increase percentage reflected in a given price increase letter (see id. at 48-50). When exceptions were permitted — typically following negotiations that used the announced price increase percentage as the jumping-off point for discussions — a price increase letter could still be “partially successful” by leading to a percentage price increase, even if the effective price increase was less than the amount reflected in the price increase letter or had a later effective date than was announced in a price increase letter (id. at 50-51 (recounting certain Defendants accepting a lower percentage price increase or delaying the price increase after receiving customer pushback following release of a letter)). Slabstock and underlay price indexes, constructed using Defendants’ transactional data, show Defendants’ effective price increases generally closely track one another (id. at 123-24).

Third, Defendants set prices in such a way as to limit customer-specific deviations from generally-enforced profit margins. This aspect of Defendants’ pricing behavior reflects institutional pricing controls, and differs from the “no-exceptions” instructions, noted above, that pricing executives sent along with price increase letters (see id at 48-49). For example, Carpenter uses a “centrally administered minimum pricing system” for underlay, tied to profit margins over costs, which was aimed at “removing most of the [sales force’s] pricing flexibility” through a tiered system that accorded less price discretion to less senior sales officials (see id. at 52 (describing four levels of pricing minimum margins, which could only be disregarded with approval from Carpenter’s Vice President for the underlay division)). Likewise, Carpenter used “branch prices” for slabstock price quotes that could be tailored to specific grades (id at 53-54 & 54 n.293 (noting standard “upcharges” above branch prices for convolution)), while Foamex used a system for slabstock pricing similar to Carpenter’s tiered limits on minimum margins for underlay (id. at 54-55). See also id. at 55-57 (describing similar minimum price systems employed by Flexible Foam, Future Foam, and Leggett & Platt).

Leitzinger thus paints a pricing system among Defendants that results in all or nearly all direct purchasers suffering antitrust impact. That process began with the publication of price increase letters that, with few exceptions, purported to apply to all customers and to all foam products. Price increase letter publication would then trigger negotiations with customers, who would attempt to drive down the quoted foam price. Based on Defendants’ pricing behavior, those attempts would have generally been unsuccessful as customers (1) would have to negotiate with a Defendant salesperson who was limited, by various minimum price or margin systems, in his or her ability to accept a lower sales price, or (2) would see similar benchmarks offered by all Defendants, owing to the coordination of price increase levels. “Given these [pricing] mechanisms,” Leitzinger concludes, “I find it implausible that customers in any significant number would have escaped the effects of repeated price increase announcements over a period of more than ten years” (id. at 57).

Statistical Evidence of Impact

Finally, Leitzinger presents the results of a regression analysis that isolates from other factors and then measures the impact of Defendants’ price increase letters on actual prices. He describes this mode of statistical inquiry as “particularly useful in separating the impact of an alleged anticompetitive act on market outcomes (such as pricing) from the impact of other influences” (id. at 61 *248 n.334 (quoting ABA Section of Antitrust Law, Proving Antitrust Damages 125-26 (2010)). For each Defendant, Leitzinger creates three separate models for each category of slabstoek — buns, rolls, and fabricated products — and for each grade — conventional, viscoelastic, and high resilience. He also estimates models for prime, standard, and premium underlay products (Doc. 584-14 at 61-62). All told, he runs 73 regression models (Doc. 967 at 29).

Leitzinger inputs several data sources into these models. First, he uses Defendants’ transactional data, which contain more than 50 million “transaction records” (excluding Woodbridge and Mohawk) consisting of some or all of the following information fields: product ID; foam grade; sales volume for a given transaction; price; rebates; “other price adjustments”; billing ID; and transaction date (Doc. 584-14 at 61). A billing ID is essentially an account number. Leitzinger synthesizes certain parts of this transactional data — billing ID, customer name, and location — to create a “master billing ID,” reflecting the purchases of “the same physical purchasing facility,” which may be represented in the transactional data by one or more product IDs (id. at 62 n,337).

Second, Leitzinger assembles a group of explanatory variables or regressors. Leitzinger’s first set of regressors represent those he seeks to isolate from his measurement of the effect on price of Defendants’ price increase letters. He compiles one regressor from a U.S. Bureau of Labor Statistics (“BLS”) producer price index (“PPI”) reflecting quarterly prices of TDI and polyols (id. at 57-58, 58 n.315, 62). Another regressor incorporates a “durable goods index,” a measure of demand for durable goods like furniture; demand for slabstoek is tied to demand for durable goods (id. at 58 & n.316, 62). A third regressor measures housing starts, described as a “driver of sales for both slabstoek and carpet cushion” in light of the demand new housing creates for furniture and carpeting (id, at 58-59, 58 ns.31718, 62). Finally, Leitzinger includes regressors representing imports of foam-containing products, like furniture and mattresses, and imports of underlay (id. at 59-60, 62).

Together, these various regressors create a “regression model [that] provides a series of rolling quarterly benchmarks for prices changes[ ] based on normal market effects on prices” (id. at 63). Then, Leitzinger adds a third data source, also as a regressor: Defendants’ allegedly conspiratorial price increase letters. He claims that these models make it “possible to see statistically the extent to which those [price increase letters] caused prices paid under a given Master Billing ID for a particular product type to increase more than otherwise would have been the case” (id.). See also ABA Section of Antitrust Law, Proving Antitrust Damages 125-26 (2010) (“An explanatory variable’s partial effect is the change in the dependent variable that would result from a change in the explanatory variable, holding all of the other explanatory variables constant”). The price increase letter variable is used to determine whether “price increase [letters] independently dr[o]ve prices higher” than the models would predict using only supply and demand variables (Doc. 967 at 104-05). 6

Leitzinger’s models report a supracompetitive premium for a given transaction whenever it returns a positive coefficient between a price increase letter and actual prices. A positive coefficient associated with a price increase letter for a given transaction means “it is more likely that the true effects of a price increase [letter] were positive than ... *249 zero or negative” (Doc. 584-14 at 63-64 n.344). The magnitude of that coefficient represents the “average percentage of the Defendant’s price announcements that Dr. Leitzinger interprets as anticompetitive” (Doc. 682-1 at 125). For instance, a coefficient estimate of 0.1657 indicates that roughly 17 percent of the price increase announced in a price increase letter is attributable to anticompetitive effects, according to a formula for converting the coefficient estimate into a percentage (id. at 125 & 127 fig. 23). See also id. at 101 n.216 (describing deposition testimony in which Leitzinger confirms this method of interpreting the magnitude of his estimated coefficients). In the context of a specific price increase letter, the .1657 impact coefficient means that if a Defendant announced a 10 percent price increase, 1.7 percent of that increase would represent antitrust impact. All told, Leitzinger’s framework yields over “32,000 separate Defendant, product[,] and customer-specific estimates” of the anticompetitive impact of the price increase letters, if any, before incorporating Mohawk data (id. at 124).

Leitzinger finds antitrust impact with respect to a given master billing ID whenever his models estimates for at least one of that master billing ID’s purchases a positive coefficient for a price increase letter. Leitzinger claims to be able to do so for more than 14,000 master billing IDs that “collectively accounted for roughly $18.1 billion of the $18.3 billion” of foam product purchases recorded in Defendants’ transactional data (Doc. 584-14 at 64) (emphasis added). In percentage terms, Leitzinger’s models find antitrust impact for customers who account for 99 percent of “purchases for which [he] could identify product form, foam type, and transactional details from Defendants’ produced data” (id.) Leitzinger also reports results showing the probability of his models’ generating a random effect for a given sales volume. With respect to 82 percent of the sales volume reflected in Defendants’ transactional data, Leitzinger finds a less than five percent likelihood that his models produced a spurious impact finding (id. at 65).

Leitzinger’s models can be adjusted so that if a particular set of price increase letters were deemed not the product of a conspiracy, those letters could be removed from the models, and thus “remove any potential effect of those announcements in terms of [generating] impact estimates” (Doc. 954-1 at 48). Leitzinger concludes his impact regression analysis by effectively “double-checking” his models’ results, using three alternative sets of assumptions to conduct sensitivity analyses. All three alternative sets of assumptions confirm his base models’ finding of antitrust impact for the overwhelming majority of revenues grouped by master billing IDs (see id. at 65-66 (using Carpenter’s price increase letters as an indicator of Defendants’ conspiratorial activity because Carpenter often led price increases, assuming the conspiracy ended slightly earlier than alleged, and gauging impact only with respect to large buyers)).

Ordover Challenges Leitzinger’s Analysis

Ordover and Burtis respond to Leitzinger’s proposed common-impact method of proof, with the latter expert focusing her efforts on the Direct Purchasers’ burden as it relates to the underlay market. Leitzinger then responds to both experts’ critiques. This Court turns first to the Ordover-Leitzinger disputes, and then to the Burtis-Leitzinger disputes.

As an initial matter, this Court finds portions of Ordover’s Report proceed along a line of reasoning that is inappropriate at this stage of the proceedings, in view of his claim to only testify as to impact and damages issues. Ordover contests the notion that Leitzinger’s inferences from market structure can be used as common evidence to support a claim that all or nearly all direct purchasers suffered antitrust impact. Some of Ordover’s justifications for that line of argument, discussed below, are properly addressed now. But at other times, Ordover bolsters his conclusions about the lack of common evidence demonstrating impact by pointing to the lack of evidence that collusion actually took place (see, e.g., Doc. 682-1 at 58 (“[T]he industry characteristics adduced by Dr. Leitzinger cannot be used to prove the existence of a cartel, as none of these bear on the question of whether the Defendants *250 reached an actual agreement to fix prices.”)) (emphasis added). Of course, if Direct Purchasers cannot prove an antitrust conspiracy existed, it follows that they cannot prove Direct Purchasers suffered common antitrust impact. But proof of a conspiracy is not required at this stage of the litigation. Therefore, this Court does not address merits-based arguments, like these, unrelated to the Rule 23 requirements.

Product Variability

Ordover begins his substantive critique of Leitzinger’s methods with a series of arguments focusing on the degree of product heterogeneity in the flexible foam industry. First, Ordover argues that Leitzinger errs in failing to recognize “undisputed heterogeneity of [flexible foam] products.” These products vary according to the particular product being manufactured, that product’s specific attributes (e.g., the product’s IFD) and end uses (Doc. 682-1 at 59; Doc. 682 at 43-46). As a result, transaction prices vary across the product range that, Defendants argue, Direct Purchasers inappropriately lump together.

Ordover claims that product heterogeneity has a few consequences for the Direct Purchasers’ case. First, this heterogeneity contradicts the notion that an agreement to fix and raise prices in the flexible foam industry could have existed at all, or, if it did exist, would have been successful. Defendants would have had to agree on prices for a range of products “subject to different demand conditions, different competitive condition, and different costs,” rendering the fact of agreement implausible. Further, even if an agreement had been reached with respect to products that varied to this extent, the agreement would have had “variegated effects on prices and price changes across different products and customers” (Doc. 682-1 at 60).

Ordover claims to demonstrate a wide range of price variance by producing dispersion charts reflecting actual prices charged by each Defendant over the Class Period. Ordover uses the same foam product and grade classifications that Leitzinger uses to estimate his models (i.e., conventional slab-stock, rebond underlay, etc.) (see id. at 61-66 & 62 fig. 6, 64 fig. 7, 66 fig. 8 (displaying price dispersion for each non-settling Defendant for conventional rolls, conventional buns, and rebond underlay)). For Ordover, these price dispersion charts show not just the infeasibility of a price-fixing conspiracy existing, but further question “how common impact is possible in light of the market facts” (id. at 67).

Leitzinger successfully refutes this aspect of Ordover’s analysis. Specifically, Leitzinger explains that Ordover’s price dispersion charts do not control for the various attributes that account for two foam products of the same category and type — for instance, conventional slabstock — selling at different prices for the same board feet of foam (or other relevant unit of measure). By way of example, Leitzinger observes that “one would expect that the observed price of a product with twice the density of another product” — meaning the product contains “twice as much foam per board feet” as the less dense product — “would be (all else equal) twice the price of the less dense product” (Doc. 744-49 at 14). Selecting specific plot points from Ordover’s graphs, Leitzinger shows that product characteristics, like density, account for that price variance (id. at 14 n.40 (noting that a conventional bun product with a very high board-foot price was also very dense and had high IFD)).

Leitzinger also performed a “hedonic analysis,” a standard statistical analysis for analyzing whether differences in product characteristics can explain price variance. That analysis constructs a regression model with price as a dependent variable, and the various product characteristics as independent variables (id. at 16-17 & 16 n.52). Using Carpenter data, Leitzinger’s model finds that product characteristics account for 90 percent of the variance in slabstock pricing, and 80 percent of the variance in underlay pricing (id. at 17). Ordover disputes this tight correlation (Doc. 892-1 at 56-59, 58 fig. 20, 59 fig. 21 (showing price dispersion for Carpenter underlay and bun products, controlling for differences in three attributes)), but when shown one of Ordover’s controlled price dispersions, Leitzinger explained one cannot tie that analysis, controlling for three (albeit im *251 portant) factors, to the output of his original analysis (Doc, 954-1 at 238-39). Leitzinger’s more comprehensive analysis is persuasive.

Moreover, Defendants knew their product pricing would vary in this way and could easily (and in fact did) compare the prices of foam with varying characteristics (Doc. 744-49 at 14-15). Additionally, Ordoveris claim that “[a]ny attempt to analyze impact on prices paid by the putative class members from the alleged cartel must account for this product heterogeneity” is true enough (Doc. 682-1 at 67), but Leitzinger’s proposed models attempt to do just that (see Doc. 584-14 at 61 (noting that Leitzinger’s models employ a transactional dataset that includes product IDs and foam grades, among other data-points)).

Ordover similarly argues that by including in one proposed class the purchasers of slab-stock and underlay, Direct Purchasers have made it impossible to demonstrate “the alleged price fixing conspiracy could ... have had a common impact on prices between these two product groups” (Doc. 682-1 at 67). This is so, Ordover argues, because the two product categories bear different costs and demonstrate variances in price that do not track one another. He demonstrates that point by producing two line graphs, one for slabstock and one for underlay, that plot each Defendant’s average price for that product (id. at 67-70 & 69 fig. 9). These charts show Defendants’ average prices varied with respect to other Defendants’ prices for the same foam product group over the course of the Class Period (though not by much). Ordover argues this variance refutes the notion that the alleged conspiracy would have resulted “in antitrust injury to all or virtually [all] members of the proposed class that could be established using evidence that is common across poured foam and rebond carpet cushion products” (id. at 70).

Defendants misunderstand the type of “impact” that must be capable of proof using evidence common to the class. Ordover’s discussion of variance in average price between the relevant sub-markets would only be relevant if that difference in price were relevant to the concept of “impact.” But it is not. Direct Purchaser’s Rule 23 burden with respect to impact does not require proof of “identical damages” or “common results.” Butler, 727 F.3d at 801 . See also In re Electronic Books Antitrust Litig., 11-md-02293-DLC, at *74, 2014 WL 1282293 (S.D.N.Y Mar. 28, 2014) (“[T]he relevant question is not whether the model can explain variances within the pricing data for a given [product], but rather whether the model can reliably explain the differences between collusive [product] prices and competitive prices so that it can disentangle the effects of collusion.”) (footnote omitted); In re Blood Reagents Antitrust Litig., 283 F.R.D. 222, 239 (E.D.Pa.2012) (rejecting the proposition that “it is not enough for plaintiffs to show that a customer paid more than the but-for price for at least one item in at least one transaction”).

Rather, the impact burden requires a method of proof, using evidence common to the class, that can establish all or nearly all class members incurred an antitrust overcharge of some amount. See Hanover Shoe, Inc., 392 U.S. at 491 , 88 S.Ct. 2224 . Leitzinger explains that is precisely what his models would prove — the fact of an overcharge with respect to all or virtually all class members, not identical overcharges. His regression models measure overcharges with respect to specific master billing IDs which, again, are the closest identifiers for discrete customers that can be culled from Defendants’ transactional data. And so what matters is “whether the pattern of price changes for given [master billing ID] changed in connection with the price increase announcements” — that is, whether the price increase letters led to a specific master billing ID paying more for a given product than market forces, represented in Leitzinger’s models by his supply and demand regressors, would have demanded — “not whether the pattern of price changes across [master billing IDs] or products was the same over a given interval of time.” (Doc. 744-49 at 17).

Market Composition

Ordover next turns his focus to the manner in which Leitzinger describes the relevant flexible foam markets. First, Ordover disputes that Defendants control the overwhelming majority of the slabstock and un *252 derlay markets, pointing to a January 2010 IBISWorld survey of the “Urethane Foam Product Manufacturing” market, which claims that the “variety of foam products produced in this industry as well as the diverse downstream markets makes it difficult for individual firms to grab a large portion of industry market share” (see Doc. 682-1 at 70 & n.120). But as Lamb explains, that market survey “accounts for a broader market” — specifically, one that includes rigid and molded foam products, neither of which are included in either proposed class definition— “with more firms than the [flexible foam] market, thus artificially reducing the level of market concentration applied to the [flexible foam] market” (Doc. 743 at 16). See also id. at 16 n.71 (quoting from the IBISWorld survey’s description of the contours of the market it measured). As previously explained, Leitzinger bases his market concentration calculations on Defendants’ own calculations for those figures. Those documents do not include market share calculations for the bi-oader swath of urethane foam products included in the IBISWorld Report. This Court credits Defendants’ own contemporaneous calculations of market concentration.

Second, Ordover claims there exist other firms “in this industry” — an ambiguous phrase, which may refer to the overbroad IBISWorld definition discussed in the same paragraph — that “serve as potential entrants into the manufacturing of polyurethane products relevant to this litigation” (Doc. 682-1 at 70-71) (emphasis added). This argument is underwhelming. In effect, Ordover merely names a handful of firms who produce urethane foam products of some sort, and concludes (without further grappling with how these various firms would confront and overcome the industry barriers to entry and the threat posed by Defendants’ excess capacity) that these firms could enter, or expand their presence in, the relevant foam markets in some unspecified period of time (Doc. 744-49 at 31). His deposition testimony demonstrates a lack of any particular understanding of these other non-defendant firms’ production capabilities (see Doc. 744-25 at 31-33), further undercutting his claim that they serve as ready entrants to the relevant flexible foam markets. Of course, Defendants bear no particular burden of proof on this (or any) point, but to effectively counter Plaintiffs’ substantial proof, Defendants must do more than merely point to a list of possible market entrants assembled from a survey of websites or 10-K filings.

Third, Ordover searches Defendants’ transactional data, using the terms “foam” and “fabricators” to identify a set of firms he believes to be foam fabricators who purchase buns and rolls from Defendants. He posits these non-defendant firms would serve as a further competitive constraint on Defendants with respect to those customers who might purchase fabricated foam products from Defendants (Doc. 682-1 at 75-77, 75 n.137, 76 fig. 11). These top-25 non-defendant foam fabricators “have the ability to absorb some portion of the anticompetitive effect from the alleged price fixing conspiracy” and provide possible cover for Defendant firms cheating on price with respect to fabricated foam, all necessitating “individualized inquiries to determine the extent to which these foam fabricators provide competitive alternatives” to Defendants (id. at 77). But Ordover fails to examine any particular non-defendant fabricator’s sales behavior to show they have in fact undercut Defendants by accepting lower margins on their sale of fabricated foam products.

Leitzinger concludes that scenario “makes no economic sense,” unless Defendants also have monopoly power with respect to fabricated foam — a claim no party to this litigation or party expert has explored (Doe. 744-49 at 32). Even assuming these non-defendant foam fabricators have some potential to serve as competitive checks on Defendant firms, that fact alone does not occlude the analysis with respect to whether Direct Purchasers have demonstrated common issues with respect to impact which predominate over individualized inquiry.

Fourth, Ordover identifies Defendants’ largest customers, and argues that two features of Defendants’ relationships with these large customers undercut Direct Purchasers’ ability to demonstrate through common proof that all or nearly all direct purchasers suffered antitrust injury. Ordover identifies the *253 fifty largest customers during the latter half of the Class Period; these large customers account for $6.044 billion in Defendant sales, or 46.7 percent of total Defendant sales from 2005 through 2010; fourteen of those large firms account for just over 30 percent of total Defendant sales during the same period (Doc. 682-1 at 78 fig. 12). In addition, Ordover displays, for each Defendant, the percentage of Defendant’s 2005-10 sales that were claimed by that Defendant’s top 50 customers. Only Leggett & Platt sells more than 50 percent of its slabstock and underlay products to non-top 50 customers (see id. at 79 fig. 13).

Ordover concludes that such large buyers likely would have wielded negotiating power of varying degrees. Also, these large purchasers obtained product from Defendants “under long term contracts or other agreements that place limits on price changes” (id. at 79). As a result, “the effect on prices from the alleged price fixing conspiracy would have been different for these [large] customers than for other customers” (id. at 80). Individualized inquiry would be required to gauge impact, if any, on these large customers. For his part, Leitzinger conducts a sensitivity analysis, limiting a data sample to only large buyers, and finds impact at conventional significance levels for customers who account for over 95 percent of purchase activity (Doc. 744-49 at 55 & n.244).

This Court reserves for discussion below Ordover’s reference to the effect of purchasing contracts on Direct Purchaser’s impact burden, for that argument is more fully-developed by the experts in the course of discussing Leitzinger’s observations regarding the existence and functioning of pricing structures in the flexible foam industry. However, this Court concludes the possibility that large purchasers generally exercise bargaining power in the flexible foam markets does not preclude Direct Purchasers from establishing impact using a common proof. Assuming Leitzinger’s regression models properly function as Leitzinger describes them — an assumption critically examined below — the models necessarily would take into account these large buyers’ negotiating power, if any. After all, the models contain price as the dependent variable. Ordover’s line of argument is that negotiating power would permit these large buyers to avoid any price increase attributable to the price fixing conspiracy. If that is so, Leitzinger’s regression models should estimate a negative coefficient with respect to a given transaction and master billing ID, indicating the lack of antitrust impact (see Doc. 744-26 at 2) (“[Ordover;] ____ But ultimately, to the extent that such contracts exist and that they affect prices, it’s going to be picked up in Leitzinger’s regression”). If the bargaining power only allows some, but not all, of the supra-competitive premium to be avoided, the regression models should estimate a positive, but relatively low, coefficient with respect to a given transaction and master billing ID, indicating that the price increase letter explains a smaller portion of actual price increases.

Fifth, Ordover notes that foam has a high volume-to-weight ratio, making it expensive to transport long distances. Therefore, Defendants sell foam products into local geographic markets where prices are not always the same (see Doc. 682 at 47-48). For instance, Carpenter plants located in California and Texas charged different prices for re-bond underlay of the same thickness and density. In Ordover’s view, these local geographic markets mean antitrust injury cannot be examined using proof common to the class, but instead can be gauged only through individualized analysis that accounts for these differences in geography (Doc. 682-1 at 80-81). It is difficult to understand why this is so, in light of the manner in which Leitzinger constructed his models. Again, those models’ dependent variable are prices charged with respect to a specific master billing ID. A master billing ID represents Leitzinger’s efforts at identifying a single physical purchasing location (Doc. 584-14 at 62 n.337). Just as Defendants sell only to purchasers located within some given distance from a plant, a single physical purchasing location would likely only purchase product sold by plants in that local market. If a master billing ID purchased foam from a plant in another geographic region, that difference in price as compared to products sold by a nearer plant would be reflected in De *254 fendants’ transactional data and Leitzinger’s models (see Doc. 744-49 at 26) (“[I]f [master billing] IDs and markets are localized — precisely what Dr. Ordover points to as problematic here — then the regression results for each [master billing] ID will reflect those localized competitive circumstances.”). Moreover, as noted above, Defendants themselves did not distinguish between local markets when announcing price increases.

Sixth, Ordover identifies instances of apparent competition between Defendants, asserting that this competition is “contrary to [Direct Purchasers’] allegations of price fixing” and is “consistent with a diversity of competitive conditions faced by individual” direct purchasers (Doc. 682-1 at 82). Ordover points to Defendant documents that show Defendants discussing accounts lost to, or gained from, other Defendants, as well as examples from Defendants’ transactional data that show certain of Defendants’ customers shifting their buying habits over the course of the Class Period (id. at 82-84).

But Direct Purchasers’ theory is not inconsistent with these instances of competition (Doc. 744-49 at 10-11). Ordover intimates that Direct Purchasers charge Defendants with conspiring to eliminate competition within the cartel. That is not so. Direct Purchasers allege that price increase letters were a part of the conspiracy (Doc. 46 at ¶ 94). But price increase letters were not issued quarterly — instead, some Defendants issued price increase letters, on average, once every three quarters (Doc. 584-14 at 68). Moreover, these very competitive conditions served as the impetus for the alleged price fixing arrangement in this case, according to a Defendant’s senior executive: “I cannot think of a single account at Domfoam and Valle Foam where we felt confident that there was no way we could lose [the account] to a competitor. In fact, it was for this reason that we were agreeable to coordinating price increases with our competitors. Coordination benefitted everyone because it minimized the degree to which we undercut each other on common accounts and gave confidence that all foamers would raise prices at about the same time” (Doc. 584-11 at 14).

Finally, Ordover argues that conditions in the flexible foam industry are such that Defendants would not have been able to monitor and compare actual prices charged for their products, citing the testimony of a Leggett & Platt underlay employee who asserted that “it would have been totally impossible” to collect information on prices charged, given the wide dispersion in underlay prices (Doe. 682-1 at 86). From that assertion about price visibility, Ordover argues that “the inability of the Defendants to monitor each other’s prices indicates that the alleged price fixing conspiracy would likely not have resulted in antitrust harm to all or virtually all members of the proposed class” (id.). “Monitoring prices is important to the success of the alleged conspiracy,” he earlier explains, “[but] it is difficult to accomplish” (id.).

This argument must be rejected for three reasons. First, to the extent Ordover again strays into opining on merits questions not bound up with the requirements of Rule 23, his argument is improper. Of course, if the antitrust conspiracy alleged in this case failed owing to the inability to monitor pricing, then necessarily there would be fewer impacted Direct Purchasers as compared to a better coordinated price fixing conspiracy. But Direct Purchasers need only demonstrate at this stage of the proceedings that they offer a method that is capable of establishing antitrust impact using common evidence, assuming the conspiracy existed.

Second, and again setting aside critiques of Leitzinger’s model design for present purposes, the number of positive coefficients his models are capable of producing would be a reflection on the alleged conspiracy’s success which, as Ordover notes, is partly dependent on price transparency. If, as Ordover contends, Defendants were substantially unable to monitor the prices they actually charged, then Leitzinger’s models would be unable to predict impact to any significant degree. That is, Defendants could have agreed to issue coordinated price increase letters containing prices set at a supracompetitive level. When it came to actually concluding deals with respect to specific customers, individual Defendants would then cheat on the price *255 fixing agreement in hopes of undercutting coconspirators. And all throughout, Ordover’s description of price opaqueness would conceal this cheating, presenting Defendants with strong incentives to cheat on the agreement, lest they become the only party not to cheat. As a result, no antitrust impact would be demonstrable under Leitzinger’s models. But in fact, Leitzinger finds impact with respect to customer who account for the overwhelming majority of Class Period revenues, and of master billing IDs in their own right. That result contradicts the notion that widespread cheating, borne of price opaqueness, led to Defendants paying lip service to agreed prices.

Finally, Direct Purchasers offer what is comparatively a much weightier body of prelitigation evidence suggesting that in fact Defendants could, and did, compare actual prices charged, and with great granularity to boot (Doc. 744-49 at 20-22) (noting, for instance, that employees of Defendant Vita-foam were able to compare its prices to Carpenter pricing on the same product “to the cent”)).

Price Increase Letters and Actual Prices

Ordover next attempts to demonstrate that Defendants were unable to implement actual price increases in the full amount mentioned in price letters (see Doc. 682 at 52-53). Ordover begins by examining in two ways the relationship between announced price increases and actual prices charged. For both examinations, Ordover uses Leitzinger’s “impact” regression dataset (Doc. 682-1 at 98), and assumes “that the impact of [a] price [increase letter] is on that quarter’s price change [i.e., the quarter in which the price increase letter is issued] from the prior quarter” (id. at 99).

First, Ordover compares the announced price increase to the average increase in prices charged in that quarter showing, at least with respect to these identified quarters, Defendants failed to implement the full amount of the announced price increase (id.). Ordover also shows the variance in actual price changes, reporting, for a quarter, the percentage of customer-product observations that show prices increases, or price declines, or no change in price from the prior quarter. Ordover finds, for example, while “Leitzinger’s data indicates an 11 percent announced price increase for Future Foam, the average price increase in 2003 Q2 was 3 percent” and that in the same quarter for the same Defendant “18 percent of prices changes are negative, while 67 percent of price changes are positive, and there are zero price changes in 15 percent of the observations” ( id. at 100). See also id. at 98-106,101 fig. 16,103 fig. 17, 105 fig. 18, 106 fig. 19 (describing Ordover’s methodology for his actual-price-effects analysis and presenting results for slabstock price increase letters issued in the Second Quarter of 2003 and the Second Quarter of 2008, as well as results for rebond underlay price increase letters issued in the Third Quarter of 2001 and the First Quarter of 2009); Doc. 892-1 at 16-19, 17 fig. 2, 19 fig. 3.

Second, Ordover analyzes “the patterns of price changes for all Defendants in Dr. Leitzinger’s regression dataset for all periods in which Dr. Leitzinger’s data indicate that Defendants issued price announcements” for both slabstock and underlay products (Doc. 682-1 at 107 & 108 fig. 20). That analysis shows that, on average, a substantial portion of customers did not experience an increase in prices during quarters in which price increase letters were issued as compared to the prior quarter. Ordover claims that for both poured foam and underlay products, 60 percent of customers experienced price increases. However, 40 percent of customers who purchased slabstock or underlay did not experience a price increase (see id. at 107 (noting that 15 percent of slabstock customers saw a decline in price, while 25 percent saw no price change; 21 percent of underlay customers saw a price decline, while 19 percent saw no change in price)).

The results of this actual-price-effects analysis is significant to Ordover because if prices did not increase for a particular customer during a quarter in which Defendants issued a price increase letter, “there was no antitrust impact” for those customers (id. at 98). See also id. at 109 (arguing that with respect to the 40 percent of slabstock and underlay customers who, on average, did not experience price increases “there was no im *256 pact from these price announcements”); Doe. 892-1 at 10 (urging an impact analysis with respect to a customer’s “entire basket” of purchases). Ordover concludes that because such a large portion of the Direct Purchaser class suffered no “impact,” as he defines that term, Direct Purchasers necessarily cannot demonstrate through common evidence that all or nearly all class members suffered antitrust impact.

Ordover, like Defendants, again misunderstands what it means for a customer to suffer antitrust impact. Ordover’s actual-price-effects analysis does not consider the price increase letter’s effect in the so-called “but-for” world, as Leitzinger’s impact models do. Under Ordover’s view (and despite protests to the contrary (see Doc. 892-1 at 8-10)), the lack of an actual price increase indicates an absence of antitrust'impact. On that line of reasoning, an agreement among competitors that serves to slow, but not reverse, a decline in price does not allow antitrust recovery. This cannot be the law.

If market forces would lead to a decline in competitive prices, a group of defendants cannot collude to slow that decline, just as participants in more vibrant industries may not collude to generate more pronounced increases in prices than market forces would otherwise produce. See, e.g., In re Wellbutrin XL Antitrust Litig., 282 F.R.D. 126, 140 (E.D.Pa.2011). Defendants’ alleged unlawful coordination of price increase letters set the benchmark for negotiations with customers over actual prices. Leitzinger claims those “benchmark” levels were higher than they would have been absent the alleged agreement. See, e.g., In re Dynamic Random Access Memory (DRAM) Antitrust Litig., 2006 WL 1530166 , at *9, 2006 U.S. Dist. LEXIS 39841 , at *47 (N.D.Cal.); In re Vitamins Antitrust Litig., 209 F.R.D. 251, 266 (D.D.C.2002). It stands to reason, then, that if because of an antitrust conspiracy a seller and buyer’s jumping-off point for negotiations is higher than market forces would otherwise determine, that buyer suffers injury.

Setting aside Defendants’ questionable definition of “impact,” this Court cannot even draw any firm conclusions from Ordover’s analysis regarding the number of class members who did not experience an increase in price in the same quarter in which a price increase was issued. This is so because, as Leitzinger explains, Ordover identifies impact with respect to “customer-product observations” for purposes of his actual-price-effects analysis. But Defendants’ transactional dataset includes purchases from “customers” under product IDs; each entry represents one item purchased by a customer under what, in effect, is a given account number. Because some customers can purchase foam product under multiple product IDs, Leitzinger constructs the master billing ID datapoint. By disregarding that unit of aggregation, Ordover concludes that a customer is not impacted if it did not experience a price increase with respect to a specific purchase, even if, with respect to other purchases, the same customer did experience a price increase driven by the alleged collusive agreement (see Doc. 744-49 at 23 (“By conflating impact to a customer with the extent to which there were overcharges on every product the customer purchased, Dr. Ordover misses the full extent of common impact across members of the proposed class. To illustrate, suppose ten members of the proposed class each purchased ten products” but each experienced price increases with respect to only six of those purchases. “From that fact pattern Dr. Ordover’s method for analyzing impact would lead to the conclusion that, 40 percent of the time, there is no impact. In fact, however, 100 percent of those customers were impacted.”) (emphasis added)). See also id. at 24 (noting other difficulties in Ordover’s analysis on this point); ABA Section on Antitrust Law, Econometrics 210-11 (2005) (“Generally, when the prices for some customers are going up while the prices of other customers are not, there is reason to doubt that the different customers (class members) are experiencing a common impact. However, if the observed differences are due to measurable and systematic factors that can be controlled for in a regression, a common impact may be shown and class treatment may still be appropriate.”).

Ordover next discusses the fact that Defendants entered into long-term purchase *257 contracts with customers (see Doc. 682 at 41-43). Some of those long-term contracts included clauses that limited the extent to which Defendants could seek price increases (“price escalation” clauses). These price escalation clauses typically limited price increases according to a formula tied to the price of foam inputs (see Doe. 682-1 at 110 (describing a 2006 Foamex-Tempur-Pedic contract that, in part, “specified prices would be governed by a formula that weighted changes in TDI and polyol prices by specified cost shares and then reduced by 15 percent to calculate the resulting percentage change in foam prices.”)). Other such contracts prohibited price increases of any kind during certain periods (id. at 111), or determined prices at the outset (id. at 112 (discussing a Carpenter-Sealy contract)). These contracts, in Ordover’s view, mean “the price announcement letters would not have increased prices already specified in these contracts and therefore the alleged conspiracy would not result in antitrust injury to these customers” (id. at 114). Determining which customers had such contracts, and whether the contracts operated as Ordover describes, would require “individualized analysis” to identify those customers who, because of such contracts, “could not have been impacted by the allegedly coordinated price announcements” (id. at 110) (emphasis added).

From the observation that a customer may have had a purchase contract with one Defendant for a limited period of time (which would have protected that customer, in part or in full, from the effects of price increase letters issued during the life of the contract) Ordover concludes that holders of such long term contracts could not have been impacted by the conspiracy at all. Missing in this line of reasoning are two critical considerations. First, to suffer “impact,” a customer need only be able to point to having paid a supraeompetitive premium as a result of one price increase letter. Second, the customer need not have suffered this adverse impact with respect to a price increase letter issued by each Defendant.

Even if the purchase contracts operated as Ordover claims, the “protected” customer could still have been affected by a price increase letter issued by the same Defendant when no such price protections existed, or by a different Defendant at a point in time when that price increase letter would have generated antitrust impact. Direct Purchasers are not litigating 21 cases alleging that 21 conspiracies each issued a single price increase letter that carried a supracompetitive premium; they allege that one conspiracy issued 21 sets of unlawful price increase letters. They need only offer a workable method of proof with respect to that one case. In re TFT-LCD (Flat Panel) Antitrust Litig., 267 F.R.D. 583, 607 (N.D.Cal.2010) amended in part, 2011 WL 3268649 (N.D.Cal.2011) (“The Court agrees with plaintiffs that defendants may not recast plaintiffs’ allegations [regarding the nature of the conspiracy], and plaintiffs have consistently alleged a single, overriding conspiracy spanning the entire class period.”).

Direct Purchasers’ proof on this point is, again, Leitzinger’s regression models. Leitzinger claims “if a customer had a contract in place which effectively insulated it from the effects of one or more price increase [letters] from a specific defendant, then presumably the data feeding into my regression would reflect that fact” (Doc. 744-49 at 66) (emphasis added). In other words, Leitzinger’s models would predict no impact with respect to each purchase entry made under a “protective” agreement if Ordover’s description of those contracts is accurate. Ordover describes these purchase agreements as closely tied to the cost of inputs (Doe. 682-1 at 110 (describing the 2006 Foamex Intemational-Tempur-Pedic purchase contract as including a provision that allowed “price change[s] ... only if [the weighted-by-inputs price formula] exceeded +/-3 percent, and prices were permitted to change quarterly.”)). Those close ties would mean that Leitzinger’s regressors fully explain the actual price charged, as Ordover himself admits. But the models may well disclose impact with respect to the same master billing ID for purchases from another Defendant in the same quarter or from purchases in other quarters in which price protections do not exist.

Further, the record reveals long-term contract holders very likely suffered some im *258 pact. None of the long-term agreements extended throughout the Class Period (see Docs. 682-40 through 682-44). For example, Tempur-Pedic purchased foam for roughly seven years outside of a longterm contract, while Comfor-Products did the same for roughly six years (Doc. 682-40-41). Contract negotiations thus took place in the context of artificially inflated baseline pricing, effects which likely became “baked into” the contracts. See In re Urethane Antitrust Litig., 251 F.R.D. 629, 644 (D.Kan.2008). See also Doc. 682-43 at 8 (Sealy contract pegging “baseline price” to price prior to contract effective date); Doc. 682-41 at 3 (noting Comfor-Products pricing “shall be adjusted at the Review,” which review occurs only biennially absent substantial increases or decreases in chemical pricing); Docs. 744-32 through 744-47 (collecting contracts executed within the Class Period). To the extent a contract pegs increases to input cost increases, the record reveals that actual costs were closely guarded secrets (see Doc. 954 at 76 (“The prices that we paid for our chemicals are strategic to Hickory Springs. Company President and CEO []and head of purchasing [ ] are the only two in the company that knew what the actual neLnet-net prices were, and we set the standard price to ensure that.”) (alterations omitted)).

Moreover, a purchaser who negotiated such a contract would have been at least partly affected by the alleged industry-wide artificial price inflation. The seller could justify a price increase by reference to similar price increases issued by other Defendants (or resist a call for a price decrease by referencing other Defendants’ pricing). See In re Urethane Antitrust Litig., 251 F.R.D. at 637-38 . See also Doc. 682-44 at 12 (requiring Carpenter’s pricing to be “competitive,” meaning “equal to the average of any two competitors’ quotes for identical products”); Doe. 744-35 at 5 (price warranty tied to competitor foam pricing). Finally, at points in the Class Period, long-term contracts were simply disregarded (Doc. 744-49 at 67-69).

Leitzinger’s Impact Model Design

So far, Ordover has pointed to factors that cause individual Direct Purchasers’ interactions with individual Defendants to vary, while Leitzinger finds that many of the varying individual factors are accounted for in his transactional data and, thus, the regression results. But for that to be so, Leitzingeris regression models must themselves be designed appropriately. Ordover argues Leitzinger’s models are flawed with respect to impact for six reasons, addressed below in turn.

Multicollinearity

Leitzinger agrees there may well be some relationship between two of his models’ regressors: the challenged price increase letters and the BLS cost index. In other words, not only is there a relationship between the dependent variable, actual prices charged, and each of the regressors, there is also a relationship between the models’ regressors. Ordover explains that multicollinearity can result in parameter estimates which are not “precise,” or reliable, and so the models do not provide powerful hypothesis testing (Doc. 682-1 at 123).

After carefully considering all positions on this point, this Court agrees with Leitzinger. Any input cost regressor would inject this quality into Leitzinger’s models. Direct Purchasers claim that increases in the price of inputs like TDIs and polyols served as a screen for conspiratorial price increases. While Direct Purchasers do allege that each and every price increase letter was a product of the conspiracy, they do not allege that each and every percentage of the price increase reflected in those letters was in excess of what market forces would have otherwise demanded. In other words, some portion of the price increase letters were legitimate, and the magnitude of the estimated coefficients bears this point out. One legitimate driver of price increases would be input costs. Any reasonably well-designed model must account for that variable, and so it must include an input cost index as one of its regressors. See ABA Section of Antitrust Law, Proving Antitrust Damages: Legal and Economic Issues 151 (2010) (noting that in the presence of multicollinearity “the best course of action” is to retain collinear variables that represent economically significant factors in explaining some phenomenon).

*259 But that simple fact does not doom the models. “That many of our explanatory variables are highly collinear is a fact of life” (Doc. 744-49 at 48). This “fact of life” does not ultimately affect the magnitude of the estimated coefficient, which in turn determines the key inquiry here: whether impact can be proven with respect to all or nearly all class members (see id. at 48-49 (“[E]ven where multieollinearity is present, the resulting coefficients are still the best unbiased estimates of the underlying relationship.”)). See also id. at 49 n.219 (citing the same economist and textbook as Ordover does in explaining the impact of multieollinearity for the proposition that the “ordinary least squares estimator in the presence of multicollinearity remains unbiased”); Federal Judicial Center, Reference Manual on Scientific Evidence 324 (3d. ed.2011). Multicollinearity does ultimately affect the statistical significance of those estimated coefficients (Doc. 744-49 at 49). Therefore, if Ordover’s critiques with respect to statistical significance level prove fatal for Leitzinger’s models, then multieollinearity contributes to the models’ demise on that count. Without more though, the unavoidable existence of multieollinearity in Leitzinger’s models is not grounds for holding it may not be used as common proof of impact.

Statistical Signifícance

Ordover argues Leitzinger ignores “the conventional and regularly accepted approach that coefficients are statistically significant if they are distinguishable from zero at the 95 percent confidence level [ (“the conventional level”) ]” because Leitzinger notes impact for each master billing ID for which one purchase from one Defendant shows a positive impact coefficient, regardless of significance level (Doe. 682-1 at 128-29; Doc. 682 at 62-63). Leitzinger’s approach is flawed, Ordover argues, because Leitzinger cannot conclude that the coefficient estímate is statistically different from zero according to the “conventional” confidence level (Doc. 682-1 at 129 n.220). “Statistical significance” in this context means “the likelihood that a coefficient result would have occurred [by chance, or spuriously,] notwithstanding the absence of any real relationship linking the variables in question” (Doc. 744-49 at 51). 7 When Leitzinger’s coefficient estimates are examined according to that “conventional” confidence level, “over 65 percent of the estimated ‘impact’ coefficients are not statistically different from zero” (Doc. 682-1 at 130 & fig. 25).

It is undisputed that generating statistically significant positive coefficients turns on sample size (Doc. 744-49 at 54). Here, the relevant “sample size” with respect to estimating a statistically significant impact coefficient for a given master billing ID is not the very extensive all-Defendants transactional data set, which records some 50 million transactions (minus Woodbridge and Mohawk sales) (Doc. 584-14 at 61). Instead, “sample size” for statistical significance purposes is the number of transactions appearing in Defendants’ transactional data that specifically relate to a given master billing ID. Not surprisingly, the transactional data reveal large and frequent buyers (see Doc. 682-1 at 78 fig. 12). But the transactional data also reveal a substantial number of low-volume, infrequent customers, and Leitzinger must attempt to produce impact results for each of these small buyers. Because the sample size is small for the small buyer, it is not surprising to see the models produce the results that they do. The sample size problem with respect to small buyers, borne of industry realities, is further compounded by the fact that Leitzinger’s models are capable of estimating a coefficient for a specific master billing ID only when the data contain “[a] minimum of three quarters of consecutive *260 purchases under a given [master billing ID] from a particular defendant for a particular product type”; the models measure price changes from quarter to quarter, and requires three observations to produce a result (Doe. 744-49 at 54).

To show the sample size problem is in fact at work in skewing significance results, Leitzinger groups master billing IDs by number of associated observations (e.g., 3-5 observations, 30-50 observations), He then identifies, with respect to each grouping of master billing IDs, the proportion of master billing IDs or purchase activity that show statistically significant impact results at the “conventional” level. When only 3-5 observations exist, just over 11 percent of the 1,138 estimated coefficients for that data grouping show impact at the “conventional” level. When more than 125 observations exist, more than 83 percent of the estimated coefficients have p-values of .05 or less. And master billing IDs with that amount of data represent the vast majority of “purchase dollars” (id. at 78). More broadly, Leitzinger reports “there is a high correlation (0.74) between the amount of transactional activity under each [master billing ID] and the significance level associated with the corresponding impact coefficient” (id. at 54). See also Doc. 886-1 at 123.

Of course, Ordover’s critique assumes the “conventional” significance level is required to analyze the particular phenomena in this case. That proposition is an oversimplification as a statistical matter, and not compelled as a legal matter. See Matrixx Initiatives, Inc. v. Siracusano, 563 U.S. 27 , 131 S.Ct. 1309, 1319 , 179 L.Ed.2d 398 (2011) (rejecting the premise that “statistical significance is the only reliable indication of causation” because “[statistically significant data are not always available” and a phenomenon being examined can be “subtle or rare” such that experts in the relevant field must rely on other tools); In re High-Tech Employee Antitrust Litig., 289 F.R.D. at 581 . Because, as described above, statistical significance is so closely tied to sample size, statisticians employ statistical significance levels less restrictive than the “conventional” level when examining phenomena for which data is limited, reaching the 10 percent — 20 percent levels (Doc. 744-49 at 56). Moreover, Krieger, a 40-year faculty member at the Wharton School’s Department of Statistics, confirms that it would be “erroneous” to reject a particular model result which does not attain statistical significance at the conventional level (Doc. 886-1 at 113 (distinguishing material significance)). See also ABA Section op Antitrust Law, Econometrics 15-16 (2005) (same); New Wigmore: Expert Evidence § 12.8.2 (“In short, the p-value does not measure the strength or importance of an association”).

Direct Purchasers show high correlations between sample size and statistical significance (Doc. 744^49 at 54); high proportions of large buyers, responsible for the overwhelming majority of Class Period sales, suffering impact at conventional levels (id, at 78); the smallest of the “small” buyers with impact test results at 48 standard deviations from what one would expect if there were no independent underlying relationship between price increase letters and actual prices (id. at 53; Doc. 967 at 10; Doc. 954-1 at 179); impressive results from a random coefficients model with all 73 regressions showing an “average implied share of negative coefficients [of] roughly two percent” (Doc. 744-49 at 60; Doc. 954-1 at 201); conclusions that can be drawn regarding the likely impact suffered by certain master billing IDs based on a different level of aggregation, affected purchase dollars, which is discussed below; and qualitative evidence showing the likelihood of common impact.

This Court will not adopt what is in essence a “blinders” approach to examining impact (Doc. 967 at 66), nor accept the proposition that “susceptibility” of classwide proof at this stage of the proceedings requires an impossibly high standard of impact results, recorded at the conventional level, for the entire universe of customers contained in Defendants’ transactional data, no matter the data limitations. Schumacher v. Tyson Fresh Meats, Inc., 2006 WL 47504 , at *7 (D.S.D.2006) (“Even the best regression equation cannot prove causation. The most it can show is a correlation that can give rise to an inference that causation exists.”). *261 Rather, Direct Purchasers must show (and have shown) that the central impact issue in this case is susceptible of class-wide proof. In re Whirlpool Corp., 722 F.3d at 860 . Should this case reach a factfinder, Defendants are free to argue that despite Direct Purchasers having shown common questions are susceptible of classwide proof, Direct Purchasers do not, at the end of the day, succeed in using this common evidence to show impact in fact due to (for example) a lack of a sufficient number of coefficients coming in at the conventional significance level. See In re Ethylene Propylene Diene Monomer (EPDM) Antitrust Litig., 256 F.R.D. 82, 96 (D.Conn.2009) (noting that in the context of contesting a motion for class certification “defendants should be focused on disputing the use of the methodology itself, not the results of the methodology”). This Court concludes that imposing the conventional significance level as a necessary condition for certification is not warranted in the context of this case and this industry.

Measuring Antitrust Impact by Master Billing IDs vs. Purchase Dollars

Leitzinger asserts that his models demonstrate impact for customers who account for 99 percent of sales, or $18.1 billion out of $18.3 billion (Doc. 584-14 at 64). In addition, he reports that he can estimate positive coefficients for over 14,000 master billing IDs, but at this point in his analysis does not identify results for the universe of master billing IDs that appear in Defendants’ transactional data. Ordover does. He correctly asserts that Leitzinger’s impacted-purchase dollars metric “does not look at the number of customers with at least one positive ‘impact’ coefficient, but rather the amount of purchases associated with customers with at least one positive ‘impact’ coefficient” (Doc. 682-1 at 131) (emphasis in original)). See also Doe. 892-1 at 9-11. To show antitrust impact with respect to all or nearly all class members, Ordover argues, Leitzinger should focus on customers, not purchase dollars. See also Doc. 682 at 59-60.

Ordover proceeds to show the full range of coefficient estimates for all “customers” in Leitzinger’s dataset (excluding Mohawk and Woodbridge). Ordover understands a master billing ID to be a “customer.” He finds a total of 18,668 master billing IDs in Leitzinger’s dataset. He further finds that Leitzinger estimates positive impact coefficients for 14,396 master billing IDs, regardless of significance level, and negative impact coefficients for 68 master billing IDs. Finally, Ordover shows that no coefficients are estimable for 4,204 master billing IDs (Doc. 682-1 at 134 fig. 26). That is so, Ordover explains, either because these non-estimable master billing IDs: did not purchase foam products in consecutive quarters; or did not purchase foam in a quarter in which a price increase letter had not been issued by the seller-Defendant and in a quarter in which a price increase letter had been issued by the same Defendant (id. at 132). Because 22.5 percent of the master billing IDs do not return an estimated coefficient, Ordover concludes that Leitzinger’s models necessarily cannot show that all or nearly all direct purchasers suffered antitrust impact (id.).

One could equate master billing IDs with class members if each class member purchased foam from Defendants from only one physical purchasing location. But Ordover provides no evidence to support that assumption, and Leitzinger disproves it (Doc. 744^49 at 61-62).

For example, L & L Carpet is represented in Defendants’ transactional data by 11 different master billing IDs, and therefore using master billing IDs as a proxy for class members would tend to underreport the actual impact rate among class members (id. at 61-62; Doc. 954-1 at 102-04). Leitzinger deliberately and candidly chose to report his models’ results on the basis of impacted purchase dollars, and not solely on the percentage of master billing IDs for which positive coefficients are estimable (see Doc. 584-14 at 65 (“Inasmuch as a given member of the proposed class can be associated with multiple [m]aster [b]illing IDs (because of multiple locations with multiple account numbers, multiple customer names with multiple account numbers, or differing terminology used by different Defendants to describe the same customer), this level of aggregation is more detailed than the Class member level” and is not subject to over-aggregation)). And he *262 chose the quarter-by-quarter approach— which leads to “missing” customers — because it avoided certain other difficulties that would have arose had he estimated coefficients based on a longer time period (Doc. 954-1 at 205-07 (explaining that by using annual time periods, the longer period would have created “great damage” to the models’ ability to control for supply and demand factors)).

In an ideal world, the regression models would estimate positive coefficients for the required number of class members as class members, but this shortcoming in the models is borne of data shortfalls and the type of conduct being measured, not a demonstrable absence of relationship between price increase letters and actual prices charged that defeats other inferences that can be drawn from the models. And the models are not the sole sources of evidence advanced by Direct Purchasers to show impact in this case is susceptible of class-wide proof.

Regressor Coefficient Estimates that Do Not Conform to Economic Theory

Ordover’s fourth criticism of Leitzinger’s models is that they are unreliable because they produce relationships between different variables that do not conform to economic theory (see Doc. 682 at 64). Ordover explains that if a regressor representing Defendants’ costs (e.g., Leitzinger’s BLS index) or consumer demand (e.g., the durable goods and housing starts indices) increases in value, price should also increase, and vice versa (Doc. 682-1 at 135). Ordover notes this is not always the case with Leitzinger’s models, reflected in a chart showing the models produce counterintuitive coefficients in a large number of cases (see id. at 136 (noting “44 percent of Dr. Leitzinger’s estimates of the relationship between costs and prices are negative”)). See also id. at 137 fig. 27.

Leitzinger explains he employed a “reduced form equation,” often used in antitrust litigation. Such a model captures “the joint operation of supply and demand factors on price,” but in the process produces these “counterintuitive” coefficients even though supply and demand variables in fact interact with price in the expected manner (Doc. 744-49 at 47-48). So, while Ordover does not criticize the use of a reduced form equation, he attempts to undermine Leitzinger’s models by pointing to a predictable result of this equation, an equation endorsed by the American Bar Association Antitrust Law Section. The model is appropriate in this context, where all parties agree that supply and demand factors play a role in determining actual prices. See ABA Section op Antitrust Law, Proving Antitrust Damages: Legal and Economic Issues 156-157 (2010).

Input Cost Index

Ordover next argues that Leitzinger’s models wrongly select one cost index rather than another (see Doe. 682 at 63-64). As noted above, Leitzinger uses a BLS PPI as a proxy for TDI and polyol costs. That choice is challenged by Ordover because (1) Leitzinger uses this index as his cost regressor for both slabstoek and underlay models, even though slabstoek and underlay costs do not track each other, and (2) “estimates for prices for these products themselves are readily available from chemical industry analyst ICIS,” a measure used by some Defendants to gauge costs in some long-term purchase contracts, and by Lamb in designing his Direct Purchaser Regression (Doe. 682-1 at 138-139).

ICIS data, Ordover argues, more closely tracks the “average price of poured polyurethane foam” and “a measure of scrap foam costs appears to track the prices of rebond carpet cushion more closely” than the BLS index (Doc. 682-1 at 140). See also id. at 141 fig. 28; Doc. 892-1 at 39-43. When these “better” cost indices are incorporated into Leitzinger’s models, the master billing IDs’ estimated coefficients change in value and significance (compare id. at 134 fig. 26 (showing results of Leitzinger’s models using the BLS index), with id at 142 fig. 29 (showing results of Leitzinger’s models when the ICIS and scrap foam costs indices are included)).

Leitzinger’s justifications for his chosen input cost indexes are persuasive. Ordover points to only two long-term purchase contracts in which the ICIS index is used; ICIS data does not account for volume or customer discounts, which Defendants enjoyed owing to their status as the largest domestic buyers of TDIs and polyols; the price Defendants *263 paid for these chemicals under purchase contracts were not included in the ICIS index because those prices were not transparent enough to meet ICIS’s criteria for inclusion; employees or agents for four Defendants testified that ICIS data was unreliable or was not consulted by that individual in the ordinary course of business; and the ICIS index excludes diphenylmethane diisoeyanate (“MDI”), an important ingredient in viscoelastic foam (Doc. 744-49 at 34-38). See also Doc. 954 at 75 (noting no Defendant witness testified that ICIS is a reliable cost measure); Doc. 954-1 at 153-55 (explaining BLS more accurately gauges surveyed firms’ long-term costs). At a more basic level, Leitzinger explains (Doc. 967 at 21-22):

I think sometimes you have the comment made: Well, ICIS correlates better with prices. Well, that’s not the right test. We’re trying to use the regressions to find the role that costs played. We don’t presume that the answer, the best answer[,] is the highest correlation we can find. We want to let the data — I want to let the data tell me what that relationship looks like. And for that purpose I’m going to pick what I think is the most carefully put together, the most authoritative index. And I think the BLS does a better job.

Moreover, ICIS and BLS are, in fact, highly correlated with one another (Doc. 967 at 156; Doc. 954 at 71 (citing Doc. 954-1 at 151)). That fact is borne out in the impact results generated when the ICIS measure is incorporated into the models. Specifically, using the BLS index, 14,396 master billing IDs show impact, while use of the ICIS index results in 14,257 such impact results. Granted, inferences with respect to classwide impact become less strong, but the difference is not so substantial as to make the models unworkable.

Leitzinger’s use of BLS data for underlay regressions is rooted in the particulars of this case. Ordover’s preferred index is determined by third-party prices for scrap, a measure that is perhaps relevant if most of the Defendant underlay manufacturers purchased scrap on the open market. But, only Mohawk did for the entire Class Period. Instead, all other firms are or were “integrated producers,” meaning that Defendants use cast off from their own slabstock production to source some or all of the trim used in producing underlay (Doc. 744-49 at 39). With respect to those firms, Leitzinger’s BLS index is an appropriate measure — the amount a Defendant pays for the primary inputs of the slabstock that, in turn, generates the scrap used in its underlay processes would be more closely tied to actual underlay input cost prices than the price of scrap selling on the open market (id. at 41). Defendants themselves made this connection— between slabstock input costs and scrap costs — when justifying underlay price increases to customers (id. at 41-42).

Finally, Ordover claims the BLS index Leitzinger used does not include polyols, pointing to a different BLS PPI index as including specific chemicals that fall within the “large class of chemicals” to which “polyols” refers (Doe. 892-1 at 33). BLS confirms that the polyols Leitzinger measures are included in his BLS index (Doc. 584-14 at 58 n.315; Doc. 1085). The BLS PPI index is an appropriate proxy for Defendants’ chemical costs; Defendants may continue to dispute the appropriateness of the chemical costs captured by that index if and when the ease reaches a factfinder.

Failure to Include a Hurricane Dummy Variable

Finally, Ordover advocates the use of a hurricane dummy variable in the Fourth Quarter of 2005 when Hurricanes Katrina and Rita resulted in foam input cost increases and, at times, order allocations. Order allocations mean “chemical manufactures did not allow the market to fully clear through prices, but instead through rationing,” so that “changes in chemical prices did not fully reflect changes in Defendants’ costs” (Doc. 682-1 at 148). Ordover calculates the effect of including this dummy variable, after substituting his preferred input cost indices into the models (see id. at 149 fig. 31). On that basis, the dummy variable causes the total initial quarterly overcharges for slabstock to drop from $107.7 million to a negative $87.7 million (id.).

With one exception, customer allocation in this context meant that Defendants were lim *264 ited to their average monthly chemical purchases, during a quarter when no party describes any reason why that limitation would have led to an input cost index not fully reflecting the price of those inputs (i.e., that chemical orders during this period would have exceeded the average monthly order) (see Doc. 744-49 at 43-44). This limitation would prevent Defendants from engaging in arbitrage or hoarding chemicals during a period in which future chemical pricing was highly uncertain (Doc. 967 at 111). The record does reflect, however, that beginning in October 2005, Leggett & Platt saw allocation below these average-month levels (Doc. 744-49 at 44 (noting BASF limited Leggett & Platt to 70 percent of its average monthly orders)).

Leitzinger explains that even if it were proper to include a dummy variable in his models, that change has only a marginal effect on his models’ ability to demonstrate common impact with respect to all or nearly all class members (id. at 45 & n.203 (“[T]here is still evidence of antitrust impact for [master billing] IDs involving nearly 99 percent of the purchases under the [master billing] IDs for which I was able to estimate a relationship” in that the base models’ impact finding decline from customers who account for 99.2 percent of revenues to customers who account for 98.7 percent of revenues)). The particular quarter that would be affected by the dummy variable’s operation also included price increase letters announcing the largest percentage slabstock price increase of the Class Period (see Doc. 584-14 at 121). The same period, of course, saw input cost shocks. But Defendants’ documents show at least some Defendants saw increased profit margins as compared to past quarters (Doc. 744-49 at 49). The dummy variable would be too blunt in accounting for these price increases beyond increased cost (id. at 46 (“Dr. Ordoveris preferred approach would be to essentially ascribe the entirety of the late 2005 price increases to the allocated-related effects of the hurricanes,” understating both antitrust impact and damages should those price increase letters be determined to have in fact been the result of an antitrust conspiracy)).

This Court agrees that, regardless of whether a dummy variable should be included in Leitzinger’s models to prove damages in fact, including or omitting that variable at this stage of the litigation does not alter this Court’s predominance analysis.

Burtis Challenges Leitzinger’s Models

Defendants Mohawk and Leggett & Platt offer Burtis’ testimony to oppose class certification. As noted above, these two Defendants produce underlay: exclusively so in Mohawk’s case, and likewise for Leggett & Platt since its 2007 sale of poured foam capabilities (Doc. 679-1 at 6-7).

Certain critiques by Burtis proceed along lines similar to those raised by Ordover. This Court will not rehash the particulars of those arguments, because at best they only provide additional detail as to Mohawk and Leggett & Platt (see id. at 13-19 (discussing underlay product characteristics, local geographic markets, and variations in product prices across customers, all of which are incorporated into Defendants’ transactional dataset and thus accounted for in Leitzinger’s models if they operate as described with respect to Mohawk and Leggett & Platt)). Burtis also advances additional critiques of Leitzinger’s models not previously discussed, or versions of similar arguments (e.g., Leitzinger’s choice of cost inputs) that warrant exploration: specifically, Burtis’ views that Leitzinger fails to establish the alleged conspiracy impacted all or nearly all Mohawk or Leggett & Platt customers, and fails to gauge input costs according to an appropriate scrap index.

Impact for Mohawk or Leggett & Platt Customers

Leitzinger “did not have usable transactional data for Mohawk” to incorporate into his initial models (Doc. 744-49 at 71). Discussion with respect to impact as to Mohawk customers begins then with Burtis who purports to apply Leitzinger’s models to the Mohawk transactional data. She finds 20,368 “Mohawk carpet underlayment customers” with results for 10,890 of those customers, only 5,288 of which were impacted at statistically significant levels (Doc. 679-1 at 25, 26 tbl. 4). 47 percent of Mohawk “customers” *265 fall out of the models because they did not make the minimum number and type of quarterly purchases of Mohawk products required by Leitzinger’s models, and customers who account for 74 percent of Mohawk sales were impacted “after adjusting for statistical significance” (id. at 24-25, 25 n.52). See also Doc. 893-1 at 13-24.

Leitzinger responds by first providing his own impacted-dollar amount results based on Mohawk data alone, but this time does not limit that figure to purchase activity showing impact at conventional significance levels. Under that approach, impact is shown “as to 97 percent of the purchase activity,” mirroring impacted-dollar estimates for other Defendants (Doc. 744-49 at 71). In addition, Leitzinger ran Mohawk’s data through his models and combines Mohawk-specific results with the outputs described in his initial report. He reports a total of 36,798 master billing IDs, 21,598 of which have sufficient data for his models to operate. 19,322 of those master billing IDs (or 88 percent) show impact, with more than half at conventional significance levels. On an impacted-dollars basis, customers accounting for 99 percent of purchase activity are impacted, with customers accounting for 94 percent of that purchase activity showing impact at statistically significant levels (Doc. 954 at 68).

The Burtis model results emphasize why the conventional level should not be made the sine qua non of Direct Purchasers’ predominance proof in the context of this case. As noted above, the purchase activity of 9,478 “customers” fall out of the model because they do not meet the model’s minimum data requirements. But, as Leitzinger reports, customers who account for 97 percent of all Mohawk purchase activity show impact (equating to 9,834 customers), and customers who account for 74 percent of all Mohawk purchase activity show statistically significant impact (equating to 5,288 customers) (Doc. 679-1 at 26 tbl. 24; Doc. 744-49 at 71).

One could draw two inferences as to the “small” buyers not captured by the models. First, one could conclude that if high volume purchasers of Mohawk’s products were impacted at the conventional confidence level, small buyers were even less likely to have escaped impact. Or one could conclude that, despite the models’ impact findings, small buyers, with no bargaining power or a continuous relationship with Mohawk avoided the antitrust injury inflicted on larger buyers with relatively more bargaining power and who are repeat Mohawk customers. That second inference turns Ordover’s argument regarding the top-50 buyers list on its head — again, Ordover claims that if anyone could avoid injury from collusive price increase letters, it would be large buyers — and is, to put it bluntly, implausible (Doc. 954-1 at 118-19). Again, Leitzinger’s base models identify just over 7,000 master billing IDs for which only one impact coefficient could be estimated. He finds impact with respect to 5,589 of these master billing IDs. In the absence of any independent relationship between price increase letters and actual prices, one would expect a 50/50 split in positive and negative impact results. But instead, he generates impact results for the smallest of the “small” buyers that have a “less than one in a billion” chance of occurring by chance (Doc. 744-49 at 51).

The Burtis discussion with respect to Leggett & Platt is similar. Unlike Mohawk, Leitzinger’s initial report analyzed Leggett & Platt transactional data. Leitzinger’s models show 7,683 Leggett & Platt customers, but shows results for only 4,282 of those customers, and positive impact for 3,726 (or 87 percent of customers for whom impact coefficients are estimable) (Doe. 679-1 at 19-20 & 20 tbl. 1). See also id. at 21-22, 21 n,48, 22 tbl. 2 (adjusting Leitzinger’s Leggett & Platt model results for the conventional statistical significance level and finding the model can estimate impact results for 28 percent of the 7,683 customers, and for customers who account for 80 percent of purchase activity).

Burtis also identifies “proposed class members who only purchased polyurethane carpet underlayment products from Leggett & Platt” and were not impacted under the Leitzinger models. She further dices the data to identify 4,807 such customers, running a “unique-customer” regression that shows Leitzinger estimating impact results of any kind for half of those unique customers, and *266 positive impact results for 21 percent of unique customers at conventional significance levels (id at 24 tbl. 3). By subtraction, 2,876 Leggett & Platt customers (or 37 percent of all Leggett & Platt customers) also purchased from other Defendants, and are excluded from the unique-customer regression. And as Leitzinger explains, this 37 percent of Leggett & Platt’s customer base accounts for three-fourths of Leggett & Platt sales. What is left, then, is 62.5 percent of Leggett & Platt’s customer base, who account for one-quarter of Leggett & Platt sales. It is therefore not surprising to see a substantial portion of master billing IDs fall out of the unique-customer results, and to see the proportion of impact coefficients that are statistically significant drop when Leggett & Platt’s largest (and non-unique) customers, who show impact at rates consistent with the base models’ aggregate results, are removed (Doc. 744-49 at 55). In sum, Burtis’ specific examination of Mohawk and Leggett & Platt’s impact results is consistent with Leib zinger’s results and his models are workable for showing classwide impact.

Costs According to a Scrap Index

As noted earlier, the Mohawk and Leggett & Platt underlay operations differ from other underlay producers. While other Defendants generate scrap in-house, Mohawk purchased scrap on the open market throughout the Class Period, while Leggett & Platt has done the same since shedding its pouring capabilities in a March 2007 sale. Still, Leitzinger uses the BLS index as a supply variable in all his models, even as to Mohawk and Leggett & Platt. Burtis notes that data showing the prices Leggett & Platt paid for scrap are available throughout the Class Period, while Mohawk’s scrap cost data begins in January 2005 (Doc. 679-1 at 29 n.63). She also notes that because Leitzinger runs regression models Defendanb-by-Defendant, he presumably could include a scrap cost index only as to Leggett & Platt or Mohawk, while retaining the BLS index for other Defendants (id. at 27 n.60). She swaps out the BLS index, uses the scrap cost index in its place, and notes significant declines in the number of impacted customers (id. at 30 n.62, 31 tbl. 5 & ns.64-65, 30 tbl.6 (reporting impact according to the conventional significance level and, for Mohawk, reporting results only from January 2005 on)). See also Doc. 893-1 at 9-12.

Leitzinger explained his choice not to use a scrap cost index. He frames the key question as “what cost would the market have passed through in the absence of the conspiracy?” (Doc. 967 at 76). He concludes it would not be the cost reflected in the scrap index:

[Generally in economics we expect the price of the primary product [i.e., slab-stock] to affect the price of the byproduct [i.e., scrap]. And my concern is if, as is alleged, there was a conspiracy that’s inflating [slabstoek] foam prices, the effects of that very likely show up in scrap prices as well. And if I put a scrap price index into the regression, I have the very real possibility that what I am capturing in that variable is part of the conspiracy effect. And so as Dr. Burtis describes, you may get — you do get some difference. It’s not as dramatic in my view as she describes, [but] you get some difference in the coefficient estimates. But that may very well simply be that what’s happening when you put that variable into the model is you’re capturing the conspiracy in part over in scrap prices because higher foam prices are drawing higher scrap prices. And that’s why I followed the course that I did.

(id. at 74-75). That conclusion is reasonable, and Leitzinger is entitled to rely on it in designing his models. Recall, Leitzinger forms his models on the assumption that the conspiracy operated as Direct Purchasers allege; that structure includes scrap brokers who participated in the conspiracy (see Doc. 584-7 at 206-07; Doc. 584-8 at 2, 4-23, 25-27, 29-30, 32, 34-35, 38-40). He therefore should design his regression models with that assumed conspiracy structure in mind. See ABA Section Op Antitrust Law, Proving Antitrust Damages: Legal And Economic Issues 213 (2010) (noting that in a horizontal price-fixing case “the values of some independent variables may have been influenced by the conspiracy” and using the example of “conspirators [who] agree on price, but compete on advertising and promotion” *267 so that “expenditures on promotion will raise above the level that [promotion prices] would have assumed but for the collusion.”). Moreover, Leitzinger notes that for most of the Class Period, integrated producers sold the vast bulk of underlay, such that the market price for scrap, as opposed to an individual manufacturer’s costs, would be better represented by integrated producers’ costs. He concludes the “true” market price for scrap cannot be determined in any event (Doc. 954-1 at 147), and Direct Purchasers show that certain of Defendants’ cost data are themselves misleading, and represent standard costs or internal transfer prices (Doc. 954 at 76).

Leitzinger does not attempt to identify a cost index that most closely aligns with a particular Defendant’s costs, but rather picks a variable that most accurately gauges the cost of the relevant input in the but-for world. With respect to most Defendants, the BLS index is reasonably understood to do both — it reasonably gauges actual costs for slabstock and, thus, internally-sourced scrap, and it reflects a but-for world cost because Defendants’ conspiratorial conduct is not alleged to have affected the price of polyols or TDIs. Actual costs, then, align with the but-for costs. Not so for Mohawk and Leggett & Platt — actual costs (e.g., the scrap index) do not equal the but-for cost for the two firms’ primary input. Leitzinger adequately addresses that difference — between the self-sourcing Defendants on the one hand, and Mohawk and Leggett & Platt on the other— even though he uses the same index for all Defendants. Burtis’ view that Leitzinger is required to select a different cost measure is a non sequitur (Doe. 967 at 76 (“So if what Dr. Leitzinger is worried about, this taint that is the scrap price was affected by the alleged conspiracy being too high, then the analysis that he needs in order to determine whether or not [Mohawk and Leggett & Platt’s] customers were affected has to be different than it is for the other [Defendants.”)). Of course, Mohawk and Leggett

This text is long and has been trimmed here. Open the source document for the complete record.