In Re Polypropylene Carpet Antitrust Litigation, 996 F. Supp. 18 (N.D. Ga. 1997)

*19 *20 Martin D. Chitwood, Craig G. Harley, John Hinton, Appel Chitwood & Harley, Atlanta, GA; W. Pitts Carr, Carr, Tabb & Pope, Atlanta, GA; Leonard Barrack and Anthony J. Bolognese, Barrack, Rodos, & Bacine, Philadelphia, PA; Richard A. Lockridge, W. Joseph Bruckner, T. Brent Jordan, Lockridge Grindal Nauen & Holstein, Minneapolis, MN, for Plaintiffs.

David R. Aufdenspring, Dean S. Daskal, John M. Gross, Glenn Johnson, Powell Goldstein Frazier & Murphy, Atlanta, GA; Gregory J. Digel, Holland & Knight, Atlanta, GA; Michael Dockterman, Wildman Harrold Allen & Dixon, Chicago, IL; Randall Lee Allen, Teresa Thebaut Bonder, Alston & Bird, Atlanta, GA; Charles Conrow Murphy, Vaughan & Murphy, Atlanta, GA, for Defendants.

HAROLD L. MURPHY, District Judge.

This antitrust case is before the Court on Plaintiffs' Consolidated Motion to Certify *21 Class Action [18]. The procedural history and factual background underlying Plaintiffs' Motion is set forth in an Order entered on June 2, 1997. (Order of June 2, 1997, at 1-4.) After entering the Order, the Court held an evidentiary hearing on July 8, 1997, during which Plaintiffs and Defendants offered the testimony of their respective expert economists, Dr. Martin Asher and Dr. Peter Aranson. (See Tr. of July 8, 1997, Hearing on Class Certification ("Hearing Tr.").) Pursuant to the June 2, 1997, Order, Plaintiffs and Defendants have filed Proposed Findings and Counterfindings of Fact. Plaintiffs' Motion therefore is ripe for disposition by the Court.

In the interest of efficiency, the Court incorporates its discussion of the law governing class certification as set forth in its June 2, 1997, Order. This Order will revisit legal questions only as necessary to dispose of the remaining class certification issues.

In its June 2, 1997, Order, the Court requested the parties to submit findings and counterfindings of fact following the evidentiary hearing. (Order of June 2, 1997, at 52.) Defendants object to a number of proposed conclusions of law contained in Plaintiffs' submission, arguing that the Court's Order requests only proposed factual findings. (Defendant Shaw Industries' Counterfindings of Fact ¶ 1; Defendants Mohawk and Aladdin Mills' Counterfindings at 10; Defendant Beaulieu's Counterfindings ¶¶ 1-2.)

Defendants' objections lose their robustness in light of the proposed conclusions of law that seem to have winnowed their way into Defendants' own submissions. (E.g., Defendants' Joint Proposed Findings of Fact ("Defendants' Findings") ¶¶ 52-54; Defendant Beaulieu's Counterfindings ¶¶ 2-3.) Perhaps the tenor of Defendants' objections is that the legal conclusions in Plaintiffs' submission should have been shorter and less obtrusive. At any rate, the Court believes the law governing this case is sufficiently set forth in the Court's June 2, 1997, Order; additional supplementation is included in this Order as necessary. The Court therefore strikes the proposed conclusions of law contained in paragraphs 47 through 70 of Plaintiffs' Proposed Findings of Fact ("Plaintiffs' Findings") and in paragraphs 52 through 54 of Defendants' Findings.

This action is taken in the interest of fairness to all parties. The Court is confident that its action will not discourage in any way the parties' thorough research of case law in future submissions, nor does it reflect an abatement of the Court's enthusiasm for such efforts. (See Order of June 2, 1997, at 1 n. 1.)

As requested by the Court in its June 2, 1997, Order, Plaintiffs have adduced evidence showing that the named Plaintiffs purchased carpet from the named Defendants during the alleged conspiracy period. (Plaintiffs' Notice of Filing Affidavits, June 17, 1997.) Plaintiffs therefore have satisfied the showing required at this stage of the litigation to establish antitrust standing. (Order of June 2, 1997, at 4-7.)

In response to concerns voiced by the Court in the June 2, 1997, Order and during the July 8, 1997, evidentiary hearing, Plaintiffs have amended the definition of the proposed class by deleting all unnamed co-conspirators from the list of Defendants. (Plaintiffs' Clarification of the Definition of the Class at 1.) In addition, Plaintiffs have excluded from the class any persons and entities whose only purchases of polypropylene carpet were from Defendant Shaw Industries, Inc. through its retail establishments. (Id.) Finally, Plaintiffs have clarified the definition of "polypropylene carpet" to include carpet manufactured with pile consisting of 100 percent polypropylene fibers, or with pile consisting of a blend of fibers containing polypropylene fibers. (Plaintiffs' Findings ¶ 3.)

Defendants object to the clarified definition of polypropylene carpet, arguing that Plaintiffs' evidence does not address carpets *22 containing polypropylene blends. (Defendants' Findings ¶¶ 4-6.) Plaintiffs' evidence, however, pertains to both 100 percent polypropylene carpet and carpet containing polypropylene blends. (Hearing Tr. at 35-36, 89-90.) The Court therefore concludes that the additional information provided by Plaintiffs satisfies Federal Rule of Civil Procedure 23(a) (1)'s requirement that Plaintiffs propose a legally definable class that can be ascertained through reasonable effort. (Order of June 2, 1997, at 15.)

The Court already has concluded that Plaintiffs' proposed class complies with the remaining requirements contained in Rule 23(a) (2)-(4). (Id. at 15-20.) Plaintiffs therefore have satisfied Rule 23(a), and the Court may proceed to resolve the remaining Rule 23(b) (3) issues.

Stated broadly, Plaintiffs' burden under Federal Rule of Civil Procedure 23(b) (3) is to establish that common or "generalized proof" will predominate at trial with respect to the essential elements of their antitrust claim. (Order of June 2, 1997, at 27 (citing Alabama v. Blue Bird Body Co., Inc., 573 F.2d 309, 317 (5th Cir. 1978); In re Domestic Air Transp. Antitrust Litig., 137 F.R.D. 677, 685 (N.D.Ga.1991))). The Court inquires whether:

the addition or subtraction of any of the plaintiffs to or from the class will have a substantial effect on the substance or quality of the evidence offered. If such addition or subtraction of plaintiffs does affect the substance or quality of evidence offered, then the necessary common question might not be present.

(Order of June 2, 1997, at 28 (quoting Blue Bird, 573 F.2d at 322).)

This test is applied to the three elements required to establish antitrust liability in a class action proceeding: (1) one common conspiracy to fix prices in the relevant industry; (2) antitrust impact as to each member of the proposed class; and (3) damages. (Order of June 2, 1997, at 28-31.) The Court already has concluded that Plaintiffs' evidence satisfies the first element (id. at 33-35); this Order therefore addresses the second and third elements.

Antitrust impact is established in this type of case if Defendants' activities had the effect of stabilizing prices above competitive levels. (Id. at 36 (citing Dry Cleaning & Laundry Inst. of Detroit, Inc. v. Flom's Corp., No. 91-cv-76072-DT, 1993 WL 527928, at *3 (E.D.Mich. Oct.19, 1993) (unpublished)).) To proceed as a class, Plaintiffs must show they plan to use common evidence that reveals impact as to each member of the proposed class without resorting to lengthy individualized examinations. (Id.) Moreover, the common evidence must allow each class member to prove the conspiracy actually was implemented in the class member's relevant market and did in fact cause injury to the class member. (Id.)

In the instant case, Plaintiffs intend to establish impact in three steps. First, Plaintiffs intend to introduce evidence showing a common structure for pricing polypropylene carpet throughout the industry. (Plaintiffs' Proposed Findings of Fact ¶¶ 17-19.) Second, Plaintiffs intend to demonstrate that this structure is achieved primarily through the use of list prices, which exert a significant influence on the transaction prices for polypropylene carpet. (Id. ¶¶ 17-19; 25-26.) Third, Plaintiffs intend to perform an econometric analysis of the list and transaction prices to show that Plaintiffs paid supracompetitive prices for polypropylene carpet and thus suffered antitrust impact. (Id. ¶¶ 17-19, 25-26, 28-29.)

The Court believes sufficient evidence exists to support a reasonable conclusion that Plaintiffs will use common evidence to prove the existence of a pricing structure for the polypropylene carpet market. (Order of June 2, 1997, at 11.) The evidence upon which Plaintiffs intend to rely to prove a pricing structure can be grouped into the four following categories.

*23 First, documents generated by Defendants' executives treat polypropylene carpet as a single market segment when setting and implementing pricing strategies. (Aff. of Dr. Martin Asher Ex. 27 (August 19, 1994, Shaw Industries Letter); Id. Ex. 29 (January 18, 1995, Mohawk Industries Letter); Id. Ex. 30 (April 7, 1994, Mohawk Industries Letter); Id. Ex. 32 (May 20, 1994, Aladdin Mills Letter).)^[1]

Second, price lists generated by Defendants set forth prices that apply broadly to all polypropylene carpet products.^[2] (Dep. of Allen Stein at 30, 31, 33; Written Direct Testimony of Dr. Martin Asher Ex. 2 (July 26, 1991, Shaw Industries Inter-Office Correspondence); Id. Ex. 5 (March 4, 1994, Shaw Industries Memorandum); Id. Ex. 7 (September 27, 1991, Shaw Industries Memorandum); Id. Ex. 12 ¶¶ 3-4 (Aff. of Richard Butler); Id. Ex. 13 ¶ 8 (Aff. of Leon Drozd); Id. Ex. 14 ¶¶ 4, 7 (Aff. of Pam Cherba); Id. Ex. 15 ¶¶ 4-5 (Aff. of Kathleen Keating).)

Third, deposition testimony by Defendants' executives indicates that the prices of different polypropylene carpet products generally moved together; that is, when the price of one style of polypropylene carpet increased, the prices of other polypropylene carpet styles also increased. (Dep. of Marion McAbee at 62-63; Dep. of Gary Hollowell at 105; Dep. of Clay Miller at 119, 134-35.)

Fourth, empirical studies by Plaintiffs' expert economist, Dr. Martin Asher, supports an inference that the listed prices of different polypropylene carpet products moved together in this fashion. (Asher Aff.App. B.)

Defendants respond to this evidence in four ways, only three of which merit discussion here.^[3] First, Defendants point to other evidence in the record that purportedly repudiates the express statements contained in the documents cited above. (Defendants' Findings ¶ 10; Defendant Shaw Industries' Counterfindings ¶ 2.) Second, Defendants offer differing interpretations of the statements. (Defendant Shaw Industries' Counterfindings ¶¶ 6, 8, 9.) Third, Defendants suggest that, even if the documents support an inference that a pricing structure existed, the inference must be cabined within the narrow context of the individual document; in other words, Plaintiffs' evidence cannot be extrapolated to establish a pricing structure that involved all Defendants and the entire polypropylene carpet market. (Defendants' Findings ¶ 10; Defendant Shaw Industries' Counterfindings ¶¶ 2, 3; Defendants Mohawk and Aladdin Mills' Counterfindings at 4, 5.)

These arguments miss the point of the Court's June 2, 1997, Order. Plaintiffs' evidence need not conclusively prove the existence of a pricing structure that involved all Defendants and all polypropylene carpet products manufactured by Defendants. At this stage of the proceedings, "the Court does not inquire whether Plaintiffs can prove the existence of a [pricing structure], but how Plaintiffs will prove the existence of a [pricing structure]." (Order of June 2, 1997, at 34) (emphasis in original); see also id. at 31 ("Plaintiffs must show that antitrust impact can be proven with common evidence on a classwide basis; Plaintiffs need not show antitrust impact in fact occurred on a classwide basis."). To proceed as a class, Plaintiffs need not eliminate the possibility that, "in various instances," prices moved in different directions. (Defendants' Findings ¶ 10.) Nor must Dr. Asher's empirical study, in its present form, prove with statistical significance the existence of a pricing structure. If *24 such things were required, Plaintiffs could bypass summary judgment and proceed directly to trial on this claim.

The Court therefore believes the evidence to be relied upon by Plaintiffs: (1) is common to all members of the proposed class, and (2) supports an inference that a pricing structure existed during the alleged conspiracy period. The Court therefore concludes that Plaintiffs' evidence satisfies the predominance requirement of Rule 23(b) (3) with respect to this issue.

The Court believes sufficient evidence exists to support a reasonable conclusion that Plaintiffs will use common evidence to prove the existence of a relationship between list prices and transaction prices for polypropylene carpet. (Order of June 2, 1997, at 11.) The evidence upon which Plaintiffs intend to rely to prove this relationship can be grouped into the five following categories.

First, deposition testimony by at least one of Defendants' executives indicates that Defendants' salesmen quoted the price lists to buyers of polypropylene carpet. (Dep. of Allen Stein at 30-33.)

Second, documents created by Defendants' executives encourage Defendants' salesmen to conform transaction prices to the list prices when selling polypropylene carpet. (Asher Written Direct Ex. 2 (July 26, 1991, Shaw Inter-Office Correspondence); Id. Ex. 5 (March 4, 1994, Shaw Memorandum).)

Third, affidavits by Plaintiffs' executives describe how the prices they paid for Defendants' polypropylene carpet matched or closely approximated the list prices promulgated by Defendants. (Asher Written Direct Ex. 12 ¶¶ 3-4 (Aff. of Richard Butler); Id. Ex. 13 ¶ 8 (Aff. of Leon Drozd); Id. Ex. 14 ¶¶ 4, 7 (Aff. of Pam Cherba); Id. Ex. 15 ¶¶ 4-5 (Aff. of Kathleen Keating).) These affidavits further state that, when Plaintiffs' executives were able to negotiate lower prices than those on the price lists, the transaction prices took the form of a percentage or dollar discount off the list prices. (Id.)

Fourth, an empirical study performed by Dr. Asher suggests that a relationship existed between the list prices promulgated by one Defendant for a particular brand of carpet and the transaction prices paid by Plaintiff B & B Quick Bros. for this brand of carpet in April 1992. (Asher Written Direct ¶ 11.)

Fifth, and perhaps most persuasive of all Plaintiffs' evidence, is a logical argument that proceeds as follows: if list prices bore absolutely no relationship to transaction prices, Defendants logically would not have spent their time and effort preparing the price lists and publicizing them to their sales staff and customers. The Court cannot perceive why Defendants would send their customers a price list unless Defendants intended that the transaction price be related in at least some degree to the list price.^[4]Cf. In re Industrial Diamonds Antitrust Litig., 167 F.R.D. 374, 383 (S.D.N.Y.1996) (argument that list prices have no relationship to transaction prices "is inherently implausible, because if ... correct, there would be no reason for defendants to raise list prices").

Defendants' response to this evidence once again takes three familiar forms: (1) evidence in the record purportedly repudiates the express statements contained in the documents cited above (Defendants' Findings ¶¶ 14, 15, 16, 24, 29, 30, 31; Defendant Shaw Industries' Counterfindings ¶¶ 13-14); (2) Defendants' own interpretations of the evidence negate the probative value of this evidence (Defendants' Proposed Findings ¶ 27; Defendant Shaw Industries' Counterfindings ¶¶ 12, 14, 17); and (3) the evidence fails to establish a relationship between list and transaction prices with respect to all Defendants, the entire duration of the alleged conspiracy, and all polypropylene carpet products manufactured by Defendants (Defendants' Findings ¶¶ 22, 23, 26; Defendant Shaw Industries' Counterfindings ¶¶ 11, 12, 16).

The Court readily agrees that Plaintiffs' evidence by no means establishes conclusively *25 that a relationship existed between list and transaction prices that lasted throughout the entire conspiracy period and involved all products manufactured by Defendants. The evidence instead serves as examples of the type of evidence upon which Plaintiffs intend to rely at trial. See In re Corrugated Container Antitrust Litig., 80 F.R.D. 244, 252 (S.D.Tex.1978) ("[i]t is certainly true that the plaintiffs may not ultimately be successful in demonstrating all they hope to show, but the threshold showing they have made here is sufficient to convince the court that what proof they offer will be sufficiently generalized in nature"). The Court therefore cannot ignore Plaintiffs' evidence simply because the list price "does not necessarily" play the primary role in determining transaction prices. (Defendants' Findings ¶ 14.) Nor is Plaintiffs' evidence worthless in light of the opinion of Defendants' expert economist that "the transaction price is not likely to be related to the list price." Defendants may explore these arguments in great detail at trial or, if warranted, in a motion for summary judgment.

At this stage of the proceedings, the Court finds that Plaintiffs' evidence: (1) is common to all members of the proposed class, and (2) supports an inference that a relationship existed between list and transaction prices. The Court therefore concludes that this evidence satisfies the predominance requirement of Rule 23(b) (3) with respect to this issue. Cf. Industrial Diamonds, 167 F.R.D. at 382 ("we are persuaded that, despite the wide range of products and prices involved, common proof of impact is possible on behalf of purchasers who bought list-price products").

This finding does not hold true, however, for custom-made polypropylene carpet products manufactured by Defendants. Dr. Asher acknowledged that, without evidence showing the existence of price lists for custom-made products, Plaintiffs may not be able to use the same analytical framework as proposed above for non-custom products. (Hearing Tr. at 74-75; Asher Direct at ¶ 27.) Plaintiffs have not adduced any such evidence. The record shows only that Defendants created price lists for custom-made carpet containing nylon fiber, not polypropylene fiber. (Asher Direct at ¶ 27.) The Court therefore concludes that Plaintiffs' evidence with respect to custom-made polypropylene carpet fails to satisfy the predominance requirement with respect to this issue. See Industrial Diamonds, 167 F.R.D. at 383 ("Common proof of impact is not possible, however, on behalf of those purchasers who bought non-list price products.").

The Court believes sufficient evidence exists to conclude Plaintiffs will use common evidence to show that Plaintiffs paid supracompetitive prices for polypropylene carpet, thus establishing antitrust impact. (Order of June 2, 1997, at 11.)

To make this showing, Plaintiffs propose to rely on an econometric study performed by Dr. Asher that will examine the relationship between list prices and transaction prices. Stated simply, Dr. Asher intends to use an econometric technique known as multiple regression analysis to (1) estimate what transaction prices for polypropylene carpet would have been in the absence of the alleged antitrust conspiracy, and (2) compare the estimated transaction prices to actual transaction prices paid by the proposed class members during the alleged conspiracy period. (Asher Written Direct ¶¶ 16-18.) If the estimated transaction prices deviate from the actual transaction prices in a manner that is practically significant,^[5] Plaintiffs contend *26 that the finder of fact may reasonably infer the existence of antitrust impact. (Id.)

Multiple regression analysis is a statistical tool for understanding the relationships among two or more "variables," which are defined as "anything that can take on two or more values." Daniel L. Rubinfeld, Reference Guide on Multiple Regression, Reference Manual on Scientific Evidence 419 & n. 1 (Federal Judicial Center 1994). Use of a regression analysis is:

valuable in numerous situations in which simple statistics will not suffice, including many ... antitrust suits.... Applied to the theory of the case, multiple regression allows one to choose among alternative hypotheses and to sort out those correlations that are spurious from those that are not.

Daniel L. Rubinfeld, Econometrics in the Courtroom, 85 Colum.L .Rev. 1048, 1065-66 (1985). Thus, multiple regression analysis is "a device designed to sift through various factors in order to assess as accurately as possible the influence of any one of them." Sobel v. Yeshiva Univ., 839 F.2d 18, 35 (2d Cir.1988).

As a general rule, a properly constructed regression analysis "can play a vital role in legal proceedings. Used properly, it is an accurate and reliable method of determining the relationships between two or more variables, and it can be a valuable tool for resolving factual disputes." Franklin M. Fisher, Multiple Regression in Legal Proceedings, 80 Colum.L.Rev. 702, 735 (1980). The caveat, of course, is that the study must be performed correctly to have any probative value. The Court therefore agrees with Defendant Beaulieu that a multiple regression analysis is a scientific endeavor whose admissibility in court proceedings must be determined using the test set forth in Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S. Ct. 2786, 125 L. Ed. 2d 469 (1993). See Estate of Bud Hill v. ConAgra Poultry Co., No. 4:94-CV-0198-HLM, 1997 WL 538887 (N.D.Ga. August 25, 1997) (finding regression study is admissible under Daubert).

The question whether Dr. Asher's study is admissible under Daubert, however, is not presently before the Court. At the class certification stage, the Court simply examines whether Dr. Asher's methodology, as proposed, will comport with the basic principles of econometric theory, will have any probative value, and will primarily use evidence that is common to all members of the proposed class.^[6] (Order of June 2, 1997, at 40-41.)

Before examining whether Dr. Asher's proposed regression analysis satisfies these three criteria, the Court offers an observation regarding the burden of production underlying the Court's analysis. Plaintiffs and Defendants have raised quite a ruckus concerning which side has the burden of production regarding the validity of Dr. Asher's proposed methodology. The Court observes that, as the party seeking class certification, Plaintiffs have the initial burden of production to show the type of evidence with which they will attempt to prove their case.^[7] (See Order of June 2, 1997, at 32.) This burden requires Plaintiffs to provide sufficient details about Dr. Asher's proposed regression analysis to enable the Court to perform an initial review of the study's methodology. (Id.)

Once Plaintiffs have satisfied this burden of production, the Court believes that Defendants assume a burden to produce tangible evidence, beyond mere speculation and inference, *27 to support their claims that Dr. Asher's analysis reflects flawed econometric techniques. Cases that have ruled on the admissibility of regression analyses are instructive in this regard, because they shed light on the burdens that are imposed on parties in a dispute over a regression study's validity. See, e.g., Sobel, 839 F.2d at 34 (requiring a party "challenging the validity of a multiple regression analysis to make a showing that the factors it contends ought to have been included would weaken the showing ... made by the analysis"); Palmer v. Shultz, 815 F.2d 84, 101 (D.C.Cir.1987) (a party cannot rebut a regression analysis without introducing evidence to support the contention that an alleged flaw creates fatal bias in the analysis); Smith v. Virginia Commonwealth Univ., 84 F.3d 672, 687 (4th Cir.1996) (when an expert "offers only his opinion, unsupported by any evidence of statistical significance, it is pure speculation for him to say that the outcome of ... [the] regression analysis would have been different") (Michael, J., dissenting). The Court realizes that, at this stage of the litigation, the data underlying Dr. Asher's proposed regression analysis is not fully developed, and neither party can establish with absolute certainty whether the study eventually will provide reliable results. Even so, the Court cannot rely simply on inference and speculation to reject Dr. Asher's proposed methodology. Rather, Defendants must proffer evidence to support their claims that Dr. Asher's regression analysis is doomed to fail. Cf. Estate of Bud Hill, No. 4:94-CV-0198-HLM, 1997 WL 538887, at 19-20 (applying the cases cited above).

With these burdens in mind, the Court now examines whether Dr. Asher's proposed regression analysis will follow valid econometric principles, will possess probative value, and will primarily use common evidence. Dr. Asher proposes to construct a regression formula that includes as its explanatory variables: (1) the cost of polypropylene yarn; (2) the cost of latex backing; (3) labor costs; (4) business cycles and general economic conditions; (5) the volume of purchases by a customer; and (6) the degree to which payment terms may vary among customers. (Asher Written Direct ¶¶ 19-25.) Dr. Asher acknowledges that, after completion of discovery, he may be required to add other variables or to control for certain effects on an existing variable. (Id.) Nonetheless, Dr. Asher believes the analysis will employ valid econometric techniques, thus providing a reliable result. (Id.)

Defendants offer three objections to Dr. Asher's proposed study.^[8] First, Defendants argue that the polypropylene carpet market inherently is too diverse to permit Dr. Asher to construct a reliable regression formula. (Defendants' Findings ¶¶ 37, 42; Defendant Shaw Industries' Counterfindings ¶¶ 21, 22.) Dr. Aranson, Defendants' expert economist, testified that as many as 140 variables might need to be included in the regression formula to account for this diversity, and construction of a formula that reliably accounts for so many variables would take an economist and an accountant as long as two years to complete. (Id.)

The Court's initial response to this argument is that the potential enormity of discovery is not, to the Court's knowledge, a factor that counsels against class certification. Plaintiffs have the ability and the obligation to recruit the battalion of lawyers, economists, and accountants necessary to complete discovery during the determined period.^[9] Furthermore, Dr. Asher's initial analysis suggests that far fewer variables will be included in the formula than the 140 variables predicted by Dr. Aranson. (Asher Written Direct ¶ 19; Hearing Tr. at 42.) As stated by one well-known econometrics practitioner:

Without some theory about which variables are likely to matter, throwing a great number of variables into the hopper is likely to lead to spurious results.... [W]hen having a study done by an expert, one should not *28 be too insistent about covering every possibility at once. Rather, one should make sure that the expert proceeds by estimating a reasonable model including the major variables and then goes on to test other possibilities. If one insists that all possible variables are likely to be of equal importance, one is likely to end up with a rather doubtful result.

Fisher, supra, at 715; see also Rubinfeld, Econometrics in the Courtroom, supra, at 1072-73 (same). Defendants have no evidence, other than the speculation of Dr. Aranson, that the 140 unidentified variables mentioned by Dr. Aranson will be crucial to a reliable regression analysis. Consequently, the Court cannot conclude that the number of variables is prohibitively high.

Second, Defendants object to the fact that Dr. Asher has failed to specify all the variables to be included in the regression formula. (Defendants' Findings ¶¶ 36, 41; Defendant Shaw Industries' Counterfindings of Fact ¶¶ 22-24; Defendants Mohawk and Aladdin Mills' Counterfindings at 3, 7; Defendant Beaulieu's Counterfindings ¶ 2.) Such an objection, however, imposes too high an evidentiary burden on Plaintiffs at this stage of the litigation. Dr. Asher has identified six variables that likely will be included in the formula, and thus has demonstrated to the Court's satisfaction how he intends to proceed with his regression analysis. (See Order of June 2, 1997, at 40-41.) A reluctance to specify all the potential variables to be included in the formula ex ante is by no means irrational, as economists often withhold a final determination of which variables to be included in a regression analysis until all the relevant data has been compiled and various models have been tested. Hearing Tr. at 41 ("the regression itself helps you determine what factors should be in" the final regression formula); see also Rubinfeld, Econometrics in the Courtroom, supra, at 1073 (same); Fisher, supra, at 713-14 (same).

Third, Defendants argue that the variables Dr. Asher has proposed for inclusion in the formula will require the use of data that varies widely among the proposed class members, thus preventing the use of common evidence when performing the regression analysis. (Defendants' Findings ¶¶ 45-50; Defendant Shaw Industries' Counterfindings ¶¶ 19-21; Defendants Mohawk and Aladdin Mills' Counterfindings at 2, 3, 8-9; Defendant Beaulieu's Counterfindings ¶ 4.) Defendants further contend that Dr. Asher cannot control for these variations by adjusting his regression formula. (Id.) In support of this argument, Defendants point to: (1) the testimony of their expert economist, Dr. Aranson, and(2) specific data points regarding list prices, shipping costs, and other economic factors that allegedly reveal variations specific to the individual members of the proposed class. (Id.)

Dr. Aranson's testimony suggests that regional variations inherent in the variables specified by Dr. Asher will require a prohibitively large number of adjustments to the regression formula to account for individual class members. (Hearing, Tr. at 149-50.) On cross-examination, however, Dr. Aranson admitted that his testimony is based solely on speculation:

Q: What evidence do you have to suggest that state level economic conditions affect pricing in the carpet industry, specific to the carpet industry?

[Dr. ARANSON]: I don't. I have only inference at this point.

(Hearing Tr. at 150-51.) On the other hand, Dr. Asher testified that adjustments to the regression analysis will allow him to control for these variations in one regression formula, or in a small group of regression formulas, thus using evidence that is common to all members of the proposed class:

THE COURT: Well, if regional factors show up during discovery, ... they are bound to play a significant role. Can you account for those regional differences in your regression?

[Dr. ASHER]: Absolutely, with what are called I think [Defendants' expert] labeled them so qualitative or categorical variables. So you can control for differences.

(Hearing Tr. at 85-86.) The Court therefore is presented with a classic battle of experts, in which two experienced economists stand firmly on opposite poles. Dr. Aranson's testimony, *29 however, is based solely on inference, rather than any tangible evidence. Given these circumstances, the Court cannot find that the proposed regression analysis is inherently faulty.

To the extent Defendants proffer specific data points culled from various documents to support their argument, the Court observes that Dr. Asher readily concedes that the data for certain variables may fluctuate in a manner that requires the addition of other variables to control for the fluctuations. (Asher written Direct at 18-19 (business cycles), 21-22 (payment terms), and 16-18 (input costs).) If these fluctuations cannot be controlled in a manner that allows the use of common evidence, Dr. Asher intends to break the data down into subgroups and perform regression analyses of these subgroups. (Id. at 19-20.) At that time, the Court certainly may revisit arguments regarding certification of sub-classes, or, if necessary, decertification of the class. Presently, however, the Court is satisfied that Dr. Asher intends to control for the fluctuations identified by Defendants in a manner that employs evidence that is common to all members of the proposed class.

The Court therefore concludes that Plaintiffs have satisfied their burden of showing that they will rely predominantly on evidence that is common to all members of the proposed class to prove antitrust impact.

An antitrust plaintiff need only introduce sufficient evidence of damages to allow a jury to estimate the amount of damages. Domestic Air, 137 F.R.D. at 692. This showing thus is far less demanding than the requirement for antitrust impact. Id. In a class action context, the plaintiffs must "show they will compute damages through the use of common proof." (Order of June 2, 1997, at 32.)

Plaintiffs intend to establish damages in this case by relying on the regression analysis to be performed by Dr. Asher. (Plaintiffs' Findings ¶ 35-36.) Using the regression analysis, Dr. Asher will attempt to measure the difference between the actual transaction prices during the alleged conspiracy period and estimates of what the prices would have been absent the conspiracy. (Hearing Tr. at 42-43; Asher Written Direct ¶ 18.) In other words, at the same time the regression analysis examines whether antitrust impact exists, it also will measure the magnitude of this impact. (Id.)

To arrive at a damage figure for each class member, Plaintiffs intend to multiply to amount of any price increase estimated by the regression analysis by the number of products purchased by the class members. (Asher Written Direct ¶ 18.) Plaintiffs also may use an alternative method that multiplies the estimated price increase by the total dollar purchases made by the class members during the alleged conspiracy period. (Id.)

Use of this technique to estimate damages in a price-fixing case is accepted by many practitioners of econometric techniques. "Based on theoretical consideration [of this technique], the academic verdict is generally favorable." Michael O. Finkelstein & Hans Levenbach, Regression Estimates of Damages in Price-Fixing Cases, 46 Law & Contemp. Probs. 145, 146, 156 (Autumn 1983). This view is not universally accepted, as at least one author believes that the technique, although feasible, is highly susceptible to challenge:

Although it will be possible to test whether the difference in price is significant, it will probably be very hard to decide how much of that difference is due to random error.... [I]f what is involved is prediction over a long time, this forecasting may be worth trying, but it is not likely to be useful.

Fisher, supra, at 729. Despite its flaws, however, the technique is recognized by at least some authorities to provide probative evidence of damages. Domestic Air, 137 F.R.D. at 692-93; Rubinfeld, Econometrics in the Courtroom, supra, at 1087 ("in a price-fixing case ... a regression model that explains price in a period of nonviolation can be used to predict what the price would have been during the period of violation").

Defendants' objections to Plaintiffs' reliance on the regression analysis to prove damages are identical to the objections raised with respect to antitrust impact. The *30 Court already has found these objections to be unpersuasive.

The Court therefore concludes that Plaintiffs have adduced sufficient evidence to show they intend to rely on evidence that is common to all members of the proposed class to prove damages.

In summary, the Court concludes that, with respect to carpets that are not custom-made, Plaintiffs have satisfied their burden of production that, at trial, they will rely predominantly on evidence that is common to the class. With respect to "mass production" polypropylene carpet, then, this case is appropriate for class certification under Federal Rule of Civil Procedure 23(b) (3), in that Plaintiffs have demonstrated that evidence common to all members of the proposed class will predominate at trial. In re Domestic Air Transp. Litig., 137 F.R.D. 677, 697 (N.D.Ga.1991). With respect to custom-made polypropylene carpet, the Court concludes that class certification is not appropriate unless Plaintiffs can proffer evidence showing that the transaction prices of such products were related to list prices. (See Part I.D.1.b., supra.)

The Court notes that the evidence relied upon by the Court in this Order has not been subjected to the adjudicative process, and will in no way "color the subsequent proceedings and place an unfair burden on the defendant[s]." Eisen v. Carlisle & Jacquelin, 417 U.S. 156, 179, 94 S. Ct. 2140, 40 L. Ed. 2d 732 (1974). Accordingly, the Court's decision to certify a class in this case "should not be viewed as a prediction that Plaintiffs will ultimately prevail on the merits of their action, but simply that they have met their burden of establishing the requirements for class certification pursuant to Rule 23." Diaz v. Hillsborough County Hosp. Auth., 165 F.R.D. 689, 692 (M.D.Fla.1996).

Furthermore, the Court will revisit the conclusions set forth in this Order as necessary after Plaintiffs have conducted the necessary discovery and have completed the econometric analysis to be relied upon at trial. "In certifying the class, the Court, recognizes its right, at a later date, to create subclasses to enhance manageability and to appoint special masters. In addition, the Court will continue to review and monitor the manageability of this action in light of evidentiary developments throughout the course of this litigation." Domestic Air, 137 F.R.D. at 696.

ACCORDINGLY, the Court GRANTS Plaintiffs' Consolidated Motion to Certify Class Action [18] and ORDERS that these consolidated actions shall proceed as a class action under Federal Rule of Civil Procedure 23(b) (3) on behalf of the following class:

Plaintiffs and all other purchasers of "mass production" polypropylene carpet in the United States (excluding federal, state, and local governmental entities and political subdivisions, and excluding Defendants, their co-conspirators, and their respective parents, subsidiaries, and affiliates) that purchased "mass production" polypropylene carpet directly from Defendants, or any parents, subsidiaries, or affiliates thereof, at any time from June 1, 1991, through and including June 30, 1995. Also excluded from the Class are any persons and entities whose only purchases of polypropylene carpet were from Defendant Shaw Industries, Inc., through its retail establishments.

The Court further ORDERS the parties to submit within 10 days a joint proposed schedule for notification of the class and discovery. The parties should be mindful of the Court's preference that this case proceed without unreasonable delay. If necessary, the Court will schedule a telephone conference or hearing to resolve these matters.

[1] Most of the documents cited in this section are filed under seal. The Court has endeavored to protect the confidentiality of these documents by omitting direct quotations from the sealed documents. If presented with the opportunity in the form of a motion to reconsider the instant ruling, the Court will revisit this decision. In any event, this evidence will become public should this case proceed to trial.

[2] An important exception in this category of evidence involves custom-made polypropylene carpet products. Plaintiffs have adduced no evidence suggesting that Defendants generated price lists for these products. The absence of this evidence is discussed in greater detail in Part I.D.1.b. infra.

[3] Defendants' fourth objection is to evidence that involves carpet manufacturers other than Defendants, or that involves conduct that occurred outside the alleged conspiracy period. This objection is well taken, and the Court will not consider this evidence.

[4] Of course, this deductive exercise by the Court in no way precludes Defendants from seeking to show, at the appropriate time, that transaction prices in fact were unrelated to the list prices.

[5] "Practical significance means that the magnitude of the effect being studied is not de minimis it is sufficiently important substantively for the court to be concerned.... While an evaluation of the practical significance of regression analysis is almost always relevant in the courtroom, tests of statistical significance are appropriate only in particular circumstances." Daniel L. Rubinfeld, Reference Guide on Multiple Regression, Reference Manual on Scientific Evidence 429 (Federal Judicial Center 1994) (emphasis in original). See also Daniel L. Rubinfeld, Econometrics in the Courtroom, 85 Colum.L.Rev. 1048, 1067-68 (1985) ("Statistical significance and practical significance ... are two different things. Presumably, the court would be, and ought to be, more concerned with a ... differential of $2.00 ... that is not significant due to limited data, than with a ... differential of $.02 that is statistically significant.... Ideally, findings should allow the trier of fact to make independent inferences based on information relating to both statistical and practical significance.").

[6] The Court therefore postpones a full analysis of the admissibility of Dr. Asher's study under Daubert until the study is completed. In the likely event that the admissibility or validity of the study becomes a question for the Court to resolve, the Court will schedule a conference with the parties to initiate the appointment of an independent, court-appointed expert on the issue of econometric analysis.

[7] The burden of persuasion regarding class action issues rests at all times on Plaintiffs. Heaven v. Trust Co. Bank, 118 F.3d 735, 737 (11th Cir.1997).

[8] Defendant Beaulieu objects to the fact that Dr. Asher's proposed regression analysis is not based on transaction prices. However, this objection evaporates in light of Dr. Asher's testimony that the regression analysis in fact will be based on transaction prices. (Hearing Tr. at 110.)

[9] Implicit in this conclusion is a message to the parties that the Court likely will deny requests to extend the discovery period longer than is reasonable for Plaintiffs to prepare their case.