Dow response to Comment Letter dated May 2, 2014

CORRESP 1 filename1.htm Dow response to Comment Letter dated May 2, 2014

May 16, 2014

Via EDGAR

Mr. John Cash

Accounting Branch Chief

U.S. Securities and Exchange Commission

Division of Corporation Finance

100 F Street, NE

Washington, D.C. 20549

Response to Comment Letter dated May 2, 2014

The Dow Chemical Company

Form 10-K for the Fiscal Year Ended December 31, 2013

Filed February 14, 2014

File No. 1-3433

Dear Mr. Cash:

The Dow Chemical Company acknowledges receipt of the Staff’s letter dated May 2, 2014, concerning its Annual Report on Form 10-K for the year ended December 31, 2013. We are responding to the Staff’s comment in this letter.

Following is a list of defined terms that are used throughout this letter, a repeat of the comment made by the Staff, and Dow’s response with respect to the comment:

Defined Terms

“2012 10-K” means the Company’s Annual Report on Form 10-K for the year ended

December 31, 2012.

“2013 10-K” means the Company’s Annual Report on Form 10-K for the year ended December 31, 2013.

“Dow” or the “Company” means The Dow Chemical Company and its consolidated subsidiaries, except where otherwise required by the context.

Comment and Response

Form 10-K for the Fiscal Year Ended December 31, 2013

Item 8. Financial Statements and Supplementary Data, page 68

Note 14 – Commitments and Contingent Liabilities, page 101

Litigation, page 102

Urethane Matters, page 106

We note your disclosures regarding litigation related to the Urethane Matters, including that on May 15, 2013, the District Court denied your request to overturn a prior jury verdict and, under antitrust laws, tripled the damages resulting in a $1.2 billion judgment and that on July 26, 2013, the District Court entered an amended judgment in the amount of $1.06 billion. We also note your disclosures that you are appealing the amended judgment and have concluded it is not probable that a loss will occur and, therefore, you have not recorded a liability for this matter. In light of the judgments, please explain to us:


•	How and why you believe a loss is not probable;


•	What specific legal remedies are available to you and how long you anticipate it will take to avail yourself of those remedies and resolve this matter;


•	Any information you have regarding these settlements, including when and why you believe all the other defendants entered into settlement agreements; and


•	What disclosures you provided in your Exchange Act filings, prior to the jury verdict, related to this matter. Specifically, address why it appears you did not disclose or discuss this pending matter in your December 31, 2012 Form 10-K.

RESPONSE

The following statement was included on page 106 of the 2013 10-K:

The Company has concluded it is not probable a loss will occur and, therefore, a liability has not been recorded with respect to these matters.

Why a loss is not probable

The Company’s position that a loss on this matter is not probable, regardless of the District Court jury verdict, is rooted in the arguments included as part of the pending appeals process. The Company believes that the verdict and judgment entered in the urethanes class action case should have been in the Company’s favor, and that the class should have been decertified based upon the law established by the United States Supreme Court. In addition, this same matter was thoroughly investigated by the United States Department of Justice (“DOJ”) from 2005 through 2007. The Company fully cooperated with the DOJ, and the DOJ closed the investigation in December 2007 without filing criminal charges or a civil complaint. Consistent with the outcome in this DOJ investigation, the Company believes that it should not be held liable in the civil case. The Company has always denied plaintiffs' allegations of price fixing, will continue to vigorously defend this litigation and is confident of a favorable ultimate outcome in the appeals process.

The Company’s probable outcome analysis is a comprehensive legal and privileged analysis which holistically takes into account all of the legal and factual circumstances of the case, the trial record and the applicable law. After taking all of these circumstances

into account, the Company has concluded, notwithstanding the outcome in District Court, that a loss is not probable under the FASB Accounting Standards Codification Topic 450, Contingencies (“ASC 450”). The best and most complete description of why the Company is of this view is found in key documents filed with the United States Court of Appeals for the Tenth Circuit (“Court of Appeals”) under Appellate Case number 13-3215. The key documents are listed below and included as attachments to this letter:


•	Urethane Antitrust Litigation, Appellant’s Opening Brief, December 6, 2013 (Attachment A);


•	Urethane Antitrust Litigation, Appellant’s Reply Brief to Plaintiff’s Opposition Brief to Appeal, March 7, 2014 (Attachment B); and,


•	Urethane Antitrust Litigation, Brief for the Chamber of Commerce of the United States as Amicus Curiae in Support of Petitioners Urging Reversal, December 13, 2013 (Attachment C).

Based on the legal and factual circumstances of the case, the trial record and the applicable law included or described in the legal briefings referenced above, the Company has concluded it is not probable that a loss will be incurred.

Legal remedies

The Company’s legal remedies include the pending appeal with the Court of Appeals noted above, any proceedings on remand from the Court of Appeals and, ultimately, if necessary, a writ of certiori to the United States Supreme Court. As noted above and in our 2013 10-K disclosures, the Company has appealed the verdict with the Court of Appeals. The Company expects the verdict and amended judgment will ultimately be vacated on appeal, based on our positions and arguments contained in the legal briefings referenced above. The Company is not able to predict the ultimate timeline for resolution of this matter.

Settlement information

While the Company is not in a position to speculate or comment on the actions of other defendants, public information related to their settlement agreements is provided below:



Party	Settlement Agreement Date	Court Approved Date	Settlement Amount
BASF SE	9/21/2011	12/12/2011	$51,000,000
Bayer AG	1/31/2006	4/06/2006	$55,300,000
Huntsman International, LLC	5/27/2011	9/27/2011	$33,000,000
Lyondell Chemical Co.	3/01/2011	9/27/2011	0 ¹

¹ Lyondell filed for bankruptcy while the litigation was ongoing. Lyondell’s settlement with the class included a $35,000 payment for administrative expenses associated with providing notice of the settlement to the Class and an agreement to cooperate.

Disclosures

The Company did not specifically disclose or discuss the pending Urethane Matters in the 2012 10-K, filed on February 15, 2013, as the Company considered (and still considers) the possibility remote that it would incur a material loss related to this matter. Although the Urethane Matters was not specifically disclosed in the 2012 10-K, the Company did include the following disclosure on page 108, “…the Company is party to a number of other claims and lawsuits arising out of the normal course of business with respect to commercial matters, including product liability, governmental regulation and other actions. Certain of these actions purport to be class actions and seek damages in very large amounts. All such claims are being contested.”

Following the February 20, 2013 jury verdict, disclosures were added to the Company’s Form 10-Q for the period ended March 31, 2013 and for each periodic report thereafter, up to and including the 2013 10-K, to provide an informative disclosure related to the Urethane Matters given the widespread publicity received from the jury verdict and in accordance with ASC 450 disclosure guidance for unrecognized contingencies.

In connection with our responses to the Staff’s comments, we acknowledge that:


•	Dow is responsible for the adequacy and accuracy of the disclosure in its filings;


•	Staff comments or changes to disclosures in response to Staff comments do not foreclose the Commission from taking any action with respect to a filing; and


•	Dow may not assert Staff comments as a defense in any proceeding initiated by the Commission or any person under the federal securities laws of the United States.

We trust that the foregoing responses satisfactorily address the Staff’s comments. If you have questions, please feel free to contact me by telephone at (989) 636-5018 or by facsimile at (989) 638-9723.

Best regards,

/S/ RONALD C. EDMONDS

Ronald C. Edmonds

Vice President and Controller

Attachments



cc:	Dale Welcome, Division of Corporate Finance
	Anne McConnell, Division of Corporate Finance

Attachment A

No. 13-3215

UNITED STATES COURT OF APPEALS

FOR THE TENTH CIRCUIT

___________________________________

IN RE URETHANE ANTITRUST LITIGATION

____________________________________

The Dow Chemical Company,

Appellant

____________________________________

On Appeal from the United States District Court

For the District of Kansas

The Honorable John W. Lungstrum

D.C. No. 04-md-1616-JWL

________________

APPELLANT’S OPENING BRIEF

_______________



Charles J. Kalil	Carter G. Phillips
Executive Vice President, General Counsel	Counsel of Record
And Corporate Secretary	Joseph R. Guerra
Duncan A. Stuart	C. Frederick Beckner III
Associate General Counsel	Kathleen Moriarty Mueller
THE DOW CHEMICAL COMPANY	Jeffrey S. Beelaert
2030 Dow Center	SIDLEY AUSTIN LLP
Midland, MI 48674	1501 K Street, NW
Telephone: (989) 636-1000	Washington, D.C. 20005
	Telephone: (202) 736-8000
	Facsimile: (202) 736-8711

Oral Argument Requested

December 6, 2013

TABLE OF CONTENTS

TABLE OF AUTHORITIES ............................................................................iii

GLOSSARY ...................................................................................................vii

STATEMENT OF PRIOR OR RELATED CASES ........................................viii

CORPORATE DISCLOSURE STATEMENT .................................................ix

JURISDICTIONAL STATEMENT ...................................................................1

ISSUES PRESENTED ......................................................................................1

INTRODUCTION AND STATEMENT OF THE CASE ..................................1

STATEMENT OF FACTS ...............................................................................5

SUMMARY OF ARGUMENT ........................................................................22

STANDARD OF REVIEW .............................................................................25

ARGUMENT .................................................................................................26

I. THE DISTRICT COURT ERRED IN CERTIFYING THE CLASS ...........26

A. Wal-Mart Precludes Class Certification ............................................27

1. Dow Was Entitled To Show, In Individualized

Proceedings, That Particular Class Members Suffered No

Injury Or Damages ................................................................29

2. The District Court Violated Wal-Mart By Allowing The

Class To Proceed On The Basis Of Assumed,

“Extrapolated” Impact And Damages ...................................34

B. Comcast Precludes Class Certification ..............................................40

1. McClave’s Models Do Not Purport To Determine The

Impact And Damages Resulting From The Conspiracy

Plaintiffs Sought To Prove At Trial .....................................41

2. The District Court Erred By Allowing The Class To

Proceed On The Basis Of Regression Models That Do

Not Reliably Predict Prices In A Competitive Market ...........44

3. The Jury’s Verdict Confirms McClave’s Models Predict

Overcharges Where None Exists .........................................50

II. PLAINTIFFS FAILED TO ESTABLISH ANTITRUST LIABILITY

AS A MATTER OF LAW ..........................................................53

A. The Evidence Does Not Support A Finding That Defendants

Implemented Any Conspiracy ......................................................54

B. Plaintiffs Failed As A Matter Of Law To Demonstrate That

Lyondell Participated In The Alleged Conspiracy .........................59

III. DOW IS ENTITLED TO A NEW TRIAL BECAUSE THE

DAMAGES AWARD IS BASED ON SPECULATION, AND THE

RESULTING JUDGMENT VIOLATES THE SEVENTH

AMENDMENT ......................................................................................62

A. The Damages Award Is Based Upon Speculation ...........................63

B. The Judgment Violates The Seventh Amendment ...........................64

CONCLUSION .........................................................................................67

STATEMENT OF COUNSEL AS TO ORAL ARGUMENT ....................68

CERTIFICATE OF COMPLIANCE .........................................................69

CERTIFICATE OF DIGITAL SUBMISSION AND PRIVACY

REDACTIONS .........................................................................................70

CERTIFICATE OF SERVICE .................................................................71

ADDENDUM

TABLE OF AUTHORITIES

Page

CASES

Anixter v. Home-Stake Prod. Co.,

77 F.3d 1215 (10th Cir. 1996) ...................................................................61

In re Baby Food Antitrust Litig.,

166 F.3d 112 (3d Cir. 1999) ................................................................56, 58

Bigelow v. RKO Radio Pictures, Inc.,

327 U.S. 251 (1946) .................................................................................62

Bolden v. Walsh Const. Co.,

688 F.3d 893 (7th Cir. 2012) ....................................................................35

Carrera v. Bayer Corp.,

727 F.3d 300 (3d Cir. 2013) .....................................................................66

Cimino v. Raymark Indus., Inc.,

151 F.3d 297 (5th Cir. 1998) ....................................................................37

In re Citric Acid Litig.,

191 F.3d 1090 (9th Cir. 1999) ..................................................................58

Coleman Motor Co. v. Chrysler Corp.,

525 F.2d 1338 (3d Cir. 1975) .............................................................52, 64

Comcast Corp. v. Behrend,

133 S. Ct. 1426 (2013) ........................................................................passim

Concord Boat Corp. v. Brunswick Corp.,

207 F.3d 1039 (8th Cir. 2000) ..................................................................62

Daubert v. Merrell Dow Pharm., Inc.,

509 U.S. 579 (1993) ................................................................................44

Dimick v. Schiedt,

293 U.S. 474 (1935) ..........................................................................64, 66

iii

Garcia v. Mountain States Tel. & Tel. Co.,

315 F.2d 166 (10th Cir. 1963) .................................................................63

Gates v. Rohm & Haas Co.,

655 F.3d 255 (3d Cir. 2011) ....................................................................35

Guides, Ltd. v. Yarmouth Grp. Prop. Mgmt., Inc.,

295 F.3d 1065 (10th Cir. 2002) ...............................................................26

Hardeman v. City of Albuquerque,

377 F.3d 1106 (10th Cir. 2004) ...............................................................25

Haslund v. Simon Prop. Grp., Inc.,

378 F.3d 653 (7th Cir. 2004) ...................................................................52

J.R. Simplot v. Chevron Pipeline Co.,

563 F.3d 1102 (10th Cir. 2009) ...............................................................26

James River Ins. Co. v. Rapid Funding, LLC,

658 F.3d 1207 (10th Cir. 2011) ...............................................................61

Klehr v. A.O. Smith Corp.,

521 U.S. 179 (1997) ................................................................................66

Kumho Tire Co. v. Carmichael,

526 U.S. 137 (1999) ................................................................................50

M.D. Mark, Inc. v. Kerr-McGee Corp.,

565 F.3d 753 (10th Cir. 2009) .................................................................26

Matsushita Electric Indus. Co. v. Zenith Radio Corp.,

475 U.S. 574 (1986) ................................................................................58

MCI Commc’ns Corp. v. AT&T Co.,

708 F.2d 1081 (7th Cir. 1983) .................................................................51

In re New Motor Vehicles Canadian Export Antitrust Litig.,

522 F.3d 6 (1st Cir. 2008) .......................................................................32

Novell Inc. v. Microsoft Corp.,

731 F.3d 1064 (10th Cir. 2013) ...............................................................26

Phillips Petroleum Co. v. Shutts,

472 U.S. 797 (1985) ...............................................................................66

In re Rail Freight Fuel Surcharge Antitrust Litig.,

725 F.3d 244 (D.C. Cir. 2013) .........................................................passim

Robinson v. Tex. Auto. Dealers Ass’n,

387 F.3d 416 (5th Cir. 2004) ..................................................................32

Rodriquez v. Nat’l City Bank,

726 F.3d 372 (3d Cir. 2013) ...................................................................35

Sloan v. State Farm Mut. Auto. Ins. Co.,

360 F.3d 1220 (10th Cir. 2004) ..............................................................62

U.S. Indus., Inc. v. Blake Const. Co.,

671 F.2d 539 (D.C. Cir. 1982) ................................................................64

United States v. Garcia,

635 F.3d 472 (10th Cir. 2011) .................................................................25

In re Urethane Antitrust Litig.,

No. 08-602 (10th Cir. Sept. 2, 2008) .......................................................11

USFL v. NFL,

842 F.2d 1335 (2d Cir. 1988) ..................................................................52

Wal-Mart Stores, Inc. v. Dukes,

131 S. Ct. 2541 (2011) .....................................................................passim

Wallace B. Roderick Revocable Living Trust v. XTO Energy, Inc.,

725 F.3d 1213 (10th Cir. 2013) .........................................................25, 26

STATUTE

15 U.S.C. § 15 ................................................................................................31

RULE

Fed. R. Evid. 702 ...........................................................................................44

SCHOLARLY AUTHORITIES

2A Phillip E. Areeda & Herbert Hovenkamp, Antitrust Law: An Analysis of

Antitrust Principles and Their Application (2013) .................................32

Franklin Fisher, Multiple Regression in Legal Proceedings, 80 Colum. L.

Rev. 702 (1980) .................................................................................46, 50

Peter Kennedy, A Guide to Econometrics (6th ed. 2008) ........................48, 50

D. Leinweber, Stupid Data Miner Tricks: Overfitting the S&P 500 (1995),

available at http://nerdsonwallstreet.typepad.com/my_weblog/files/

dataminejune_2000.pdf ............................................................................4

Daniel L. Rubinfeld, Reference Guide on Multiple Regression, in Manual on

Scientific Evidence 303 (3d ed. 2011) ......................................................48

GLOSSARY


MDI	Methylene diphenyl diisocyanate. It can be combined with a polyether polyol to produce flexible foam, such as mattresses and automobile seats.


TDI	Toluene diisocyanate. It can be combined with a polyether polyol to produce to produce rigid insulation or structural foam, such as household and appliance insulation, adhesives, and sealants.


Polyether	Chemicals that can be combined with TDI or MDI to make a


Polyols	polyurethane.


Systems	A set of chemicals, most commonly comprised of MDI with a polyether polyol and additives, needed to make a particular polyurethane product.

vii

STATEMENT OF PRIOR OR RELATED CASES

The district court’s initial class certification order was previously before this Court on a petition to appeal pursuant to Rule 23(f) of the Federal Rules of Civil Procedure. In re Urethane Antitrust Litig., No. 08-602 (10th Cir. Sept. 2, 2008). This Court denied the petition. Id.

viii

CORPORATE DISCLOSURE STATEMENT

The Dow Chemical Company (“Dow”) is a publicly held corporation with no parent corporation. No publicly held corporation owns more than 10 percent of Dow’s stock.

JURISDICTIONAL STATEMENT

The district court had jurisdiction under 28 U.S.C. §§1331, 1337(a) and 15 U.S.C. §15(a). The district court entered an amended final judgment on July 26, 2013. Appellant’s Appendix (“AA”) 0566-67. Dow timely filed a notice of appeal on August 22, 2013. AA0568-69. This Court has jurisdiction under 28 U.S.C. §1291.

ISSUES PRESENTED

1. Whether the district court erred in allowing plaintiffs to proceed as a class?

2. Whether Dow is entitled to judgment as a matter of law because plaintiffs did not prove a conspiracy sufficient to support the judgment?

3. Whether the district court erred in admitting the testimony of plaintiffs’ statistical expert?

4. Whether the jury’s damages award is speculative and without record support?

5. Whether the judgment violates the Seventh Amendment by allocating damages among class members in a manner not specified by the jury?

INTRODUCTION AND STATEMENT OF THE CASE

Dow appeals from a $1.06 billion judgment entered following a jury trial in an antitrust class action brought by industrial purchasers of polyether polyol chemicals (commonly known as polyurethane components). Plaintiffs alleged that

Dow and four other manufacturers conspired to announce price increases for hundreds of different chemical products and to try to make those increases “stick” in individual contract negotiations with the approximately 2,400 purchasers in the class. While the question whether there was any agreement to conspire was hotly contested, the evidence before the jury was insufficient to support a finding that any agreement was actually implemented and caused damages in the amount entered.

Prices for the relevant chemicals fell or held steady during the alleged conspiracy period—then spiked after its end. To explain away evidence so at odds with a price-fixing conspiracy, plaintiffs claimed that defendants acted to prevent prices from falling further. It is undisputed, however, that demand was steady and costs were increasing during the alleged conspiracy. In a functioning conspiracy, prices should have increased. The absence of an effective conspiracy is also evident from the fact that the manufacturers’ price increase announcements did not consistently result in actual price increases. Prices did not move in tandem (as would occur in a conspiracy); they moved in different directions and fluctuated month-to-month. In virtually every month, one or more alleged conspirators undercut any price “leader.”

This price variability is consistent with the individualized nature of the polyurethanes industry. There is no one “market price” for the chemicals at issue;

rather, there are hundreds of different prices that vary by product and purchaser. Individual purchasers (many of them companies with substantial purchasing power such as General Electric, Whirlpool, and Johns Manville) negotiate extensively over price and other terms. The evidence included examples of purchasers tossing price increase announcements in the trash and switching from one manufacturer to another to obtain better prices.

To try to bridge the enormous gap between the economic realities of this market and their counter-intuitive claims of a price-maintenance conspiracy, plaintiffs relied on models designed by Dr. James McClave, the statistician whose models the Supreme Court rejected in Comcast Corp. v. Behrend, 133 S. Ct. 1426 (2013). McClave’s models, however, were insufficient to justify class certification.

In deciding the propriety of class certification, a court must consider not only the plaintiffs’ evidence, but the defendants’ rebuttal. Given the nature of the polyurethanes industry, where many customers negotiated away price increases, Dow was entitled to show, in individual proceedings, that particular class members were not in fact injured. See Wal-Mart Stores, Inc. v. Dukes, 131 S. Ct. 2541 (2011). McClave’s models provide no justification for stripping Dow of that right.

Class certification was also improper because McClave’s flawed models cannot reliably show antitrust impact and damages. As he did in Comcast,

McClave failed to tailor his models to the liability theory that plaintiffs pursued at trial. His models do not even purport to show any correlation between price increase announcements and price increases. Instead, they purport to predict competitive prices; plaintiffs then asked the jury to assume that the entire difference between McClave’s predicted “but for” prices and actual prices was caused by a price-fixing conspiracy.

Moreover, like the statistician who linked S&P 500 returns to butter production in Bangladesh, see D. Leinweber, Stupid Data Miner Tricks: Overfitting the S&P 500 (1995), McClave engaged in blatant “variable-shopping.” He included variables in his regression models not because they determine prices in a competitive market, but because they produced a good statistical “fit” with observed prices. When the economically relevant variables are included, however, supra-competitive prices (and the basis for the jury finding of impact and damages) disappear.

But even with all this gerrymandering, McClave’s models did not show plausible supra-competitive prices from 1999 to 2004—the conspiracy period plaintiffs alleged. To solve this problem, McClave made an ad hoc decision (for which he had no economic explanation) to treat 2004 as a “competitive” year. Only with that change were his models able to show damages from 1999 to 2003.

The jury, however, rejected McClave’s models in significant part. It found no overcharge between January 1, 1999 and November 23, 2000—a 23-month period during which McClave’s models found over $600 million in “overcharges.” The jury thus found that McClave’s models did not predict competitive prices before November 2000, and the same models that did not accurately predict “competitive” prices prior to November 2000 could not do so after that date.

In addition, McClave did not even use his models to establish impact and damages for 75% of the class. Instead, McClave used extrapolation techniques that simply assumed these class members were overcharged by the average of the overcharges observed in his modeling of damages for 25% of class members. This is the “Trial by Formula” forbidden by Wal-Mart. 113 S. Ct. at 2561.

These flaws render McClave’s models incapable of establishing antitrust impact and damages on a classwide basis. Without a valid model, plaintiffs here, as in Comcast, “cannot show Rule 23(b)(3) predominance.” 133 S. Ct. at 1433. “No damages model, no predominance, no class certification.” In re Rail Freight Fuel Surcharge Antitrust Litig., 725 F.3d 244, 253 (D.C. Cir. 2013).

STATEMENT OF FACTS

Polyether Polyol Products. This case involves four categories of chemicals: polyether polyols, TDI, MDI, and systems products (collectively “Polyurethane Products”). Polyether polyols can be combined with TDI or MDI to

make a polyurethane. AA0393. Polyurethanes are used in a variety of consumer and industrial applications such as cushion/mattress foams, insulation, adhesives/sealants, footwear, etc. TDI (toluene diisocyanate) can be combined with a polyether polyol to produce flexible foam, such as mattresses and automobile seats. AA0392-93; AA1160. MDI (methylene diphenyl diisocyanate) can be combined with a polyether polyol to produce rigid insulation or structural foam, such as household and appliance insulation, adhesives, and sealants. AA0392-93. Finally, systems are most commonly comprised of MDI with a polyether polyol and additives needed to make a particular polyurethane product. AA0393; AA1232-33. Systems are typically custom-made for a manufacturer of a product (e.g., picnic coolers) that needs a particular type of polyurethane but does not do the chemistry itself. AA914-16; AA1237-41; AA1796.

The Marketplace For Polyurethanes. Plaintiffs initially alleged that five producers—BASF, Bayer, Dow, Huntsman and Lyondell—conspired to fix prices from January 1, 1999 through December 31, 2004. The marketplace in which this alleged conspiracy took place is complex. As the district court observed, it comprises “myriad [] products, pricing structures, individualized negotiations and contracts.” AA0413.

The defendants were not the only producers of all the products at issue. Several other companies produce polyether polyols. AA1833-38, 1875-76, 1879-

80; AA1639 (Shell gained market share in polyol slab market by pricing below defendants). Hundreds of independent “systems houses,” including plaintiff Industrial Polymers, sell systems. AA0914-16; AA1233-34; AA1836-37. For some applications, customers can switch to non-urethane substitutes. AA1837-38.

Purchasers negotiate individually with producers, sometimes entering into long-term contracts and other times purchasing on a “spot” basis. AA0409-10, 0424-45; AA1165-66. As a result, there is no one “market price” for each urethane product, but a variety of individually negotiated prices that vary based on the bargaining power of the customer and producer. AA0409-10, 0413, 0424-25; AA0850; AA1133; AA1162-63; AA1203-08; AA1221; AA1362-63; AA1457-58; AA1862, 1877-79, 1887-89, 1893-97. Relevant terms of sale include not only price, but features such as payment terms and “price protection” that restrict the manufacturer’s ability to raise prices. AA1224-25, 1451-53, 1678; AA1920-24.

Systems are typically made for a particular customer, and the price and other terms are highly individualized. Systems prices do not necessarily rise and fall with changes in the prices of the component chemicals. Negotiations focus on the system’s value to the customer, AA1197-1200; AA1245-47; AA1797-98; AA1862-66, and Dow sales representatives were authorized to negotiate (within margin limits) the price of the systems they sold, AA1244-45.

For customers without contractually fixed prices, price changes generally take effect on the first day of each business quarter. AA0651-52; AA1136; AA1154-55. Contracts typically require producers to give 30-days notice of price changes, so producers wanting to raise prices typically send letters to affected customers 30-45 days in advance. AA1153-55; AA1746 (Dow contract with Foamex). The letters state the amount of the increase and the effective date, and frequently give a reason. AA1771.

Announced price increases frequently did not result in higher prices. AA1378, 1381-82; AA1217-21. When they received a price increase letter, many customers “threw it in the garbage can,” sometimes using language that you “can’t repeat” in court. AA1218. Named plaintiff Quabaug provides an example (without the colorful language): it “refused to take the price increase” from Huntsman in January 2001 and began “purchasing its system from Bayer at five cents per pound less.” AA0424-25; see also AA1843,n.84; AA1228-30 (GE Appliances “played Bayer and Dow off each other” and ultimately switched to Dow); AA1227 (Bayer lost business to BASF “based on price”); AA1635-38 (Bayer secured additional purchases from JM Huber “at $0.05/lb less at the expense of Huntsman”); AA1646 (describing “competitive jungle” where Dow lowered MDI price and took sales from Bayer); AA1683-85 (customers switching to Huntsman to avoid Bayer MDI price increase). In other instances, negotiations

ensued, customers sought better deals from other suppliers, and the incumbent supplier agreed to eliminate or reduce the price increase. E.g., AA1731-34 (Bayer reduced prices to FFP to meet a “very aggressive offer” from Dow); AA1169-76; AA1843. And sometimes, producers refused to sell at a lower price. E.g., AA0885.

Department of Justice (“DOJ”) Investigation. DOJ conducted a multi-year investigation of the alleged antitrust violations, which included voluntary grand jury testimony and appearances by employees from Dow and the other polyurethanes producers. DOJ closed the investigation without filing criminal charges or a civil complaint. AA2046.

District Court Proceedings. Plaintiffs’ amended complaint alleged that BASF, Bayer, Dow, Huntsman and Lyondell conspired to fix prices and allocate customers and markets from January 1, 1999 through December 31, 2004, and fraudulently concealed their unlawful conduct. AA0369.

Class Certification. Plaintiffs sought class certification. Defendants opposed, arguing, inter alia, that determining whether class members were injured by the conspiracy (an element commonly called “impact”), and, if so, the amount of damages, could only be determined through individual proceedings.¹ AA391- 427. Defendants showed that prices were negotiated individually, and some

^{____________________________________}

¹ Defendants at the class certification stage were BASF, Dow, Huntsman and Lyondell. AA0392. Bayer previously settled.

customers refused to accept announced price increases or obtained other concessions from the manufacturer. AA0403; AA2007, 2011-14, 2017-20; AA2035, 2038-40; AA2048-52, 2056-57, 2059-62. In some instances, the manufacturers did not even attempt to effect price increases on individual customers or were prohibited from doing so by the terms of their contracts. AA2007, 2012-14, 2018-19; AA2050-52, 2056-57, 2059-62. Defendants further showed that there are other producers of polyether polyols and systems that are not alleged to have fixed prices, another reason the alleged conspiracy would not necessarily have raised the price each class member actually paid. AA0402-03; AA1833-37, 1875-76, 1879-80. Finally, they showed that some products faced competition from non-urethane substitutes. AA0402-03; AA1838; AA2010, 2016, 2019-20; AA2049, 2055-56, 2058.

The district court certified a class of those who purchased Polyurethane Products between January 1, 1999 and December 31, 2004. AA0427. It acknowledged that “sales of the basic chemicals were characterized by individual negotiations, variations in contractual relationships and the like.” AA0409. Nevertheless, the court thought common impact could be shown by evidence of defendants’ allegedly “coordinated price increase announcements,” which “presumably establish[] an artificially inflated baseline from which any individualized negotiations would proceed.” AA0409-10.

As to systems, the court agreed that they “are heterogeneous, non-commodity, non-fungible products” that “are not subject to class certification based on the same rationale.” AA0411. The court concluded, however, that there was evidence that systems prices are based “to some extent” on the “costs of the basic chemicals that make up the systems,” and thus “plaintiffs’ proof that systems purchasers were impacted by the alleged price-fixing conspiracy as to the basic chemicals is amenable to class-wide, as opposed to individualized, proof.” AA0412.

The court was “not nearly as persuaded that the issue of damages is amenable to class-wide proof ... in light of the myriad of products, pricing structures, individualized negotiations and contracts at issue.” AA0413. But it believed that the “possibility that individual issues may predominate the issue of damages” could be addressed by bifurcating the damages phase or decertifying the class as to individualized damages. Id.

This Court denied defendants’ motion to appeal the class certification order pursuant to Fed.R.Civ.P. 23(f). In re Urethane Antitrust Litig., No. 08-602 (Sept. 2, 2008).

Daubert. As the case moved toward trial, BASF, Huntsman and Lyondell settled. AA0438-41; AA0442-44; AA0045-47.² Plaintiffs retained a new expert, Dr. McClave, who proffered a statistical analysis purporting to show that virtually all class members were overcharged and to quantify damages. Dow moved to exclude McClave’s testimony on the grounds, inter alia, that McClave’s approach for selecting variables for his regression models was economically improper, and the models could not accurately predict prices in years that plaintiffs agreed were “competitive.” The district court denied that motion. AA0488-0509.

Decertification Motion. Dow subsequently moved to decertify the class, arguing that McClave’s models did not show injury and damages on a classwide basis and that certification was inappropriate under Wal-Mart, 131 S. Ct. 2541, a decision rendered after the initial class certification. The district court deferred consideration until after trial. AA0715-16.

Trial. Although plaintiffs initially alleged two antitrust theories—a conspiracy to fix prices and to allocate customers and markets between January 1, 1999 and December 31, 2004—they sought at trial to prove only a price-fixing conspiracy (not customer or market allocation) and ending in December 2003 (not December 2004). AA0451-52. Plaintiffs claimed that Dow and the other suppliers

__________________

² In contrast to the other defendants (see supra p. 21), Lyondell settled for no liability payment after filing for Chapter 11 bankruptcy, AA0428-37.

“coordinated nearly lockstep price increase announcements” and tried to make the price increases “stick.” AA0452.

Plaintiffs presented the testimony of Larry Stern, head of urethane products for Bayer in the United States from April 2000 to December 2002. AA0591. Stern denied that he agreed to fix prices, but testified that he had discussions with his counterparts at Dow, BASF, and Huntsman about the need for price increases or their intent to raise prices. He also testified that he took steps to conceal some of these conversations. E.g., AA0607-10. The other participants denied that these discussions occurred. AA0723-75, 0728-32, 0745-47, 0749-51, 0778-79, 0782-84, 1254, 1257-61.

Plaintiffs also called two Bayer employees (Michelle Blumberg and Gerald Phelan) who testified that Wolfgang Friedrich, Bayer’s global head of MDI, told them that he “talked to the competition” and had an “understanding” with unspecified competitors about MDI prices. AA0640, 0658-60, 0664-73. Blumberg further testified, however, that she, not Friedrich, negotiated with the customer he mentioned, that she never conspired to fix prices, and that BASF and Huntsman would undercut Bayer’s prices in an attempt to obtain this customer’s business. AA0635-37, 0643-49.

And plaintiffs called Stephanie Barbour, Dow’s global director for MDI and related polyols from 2000 until early 2004. She testified that Marco Levi, Dow’s

global director for TDI and related polyols, told her that he had an “agreement” with Bayer and BASF to make TDI price increases stick in 2003, AA0685-87—an allegation Levi denied, AA904, 1249-51. She also said that Peter Davies, head of Dow’s systems business, once told her that BASF complained because Dow had not raised MDI prices for GE. AA688-91, 0708-09. Barbour testified that she had pricing authority over MDI and denied that she or her subordinates agreed to fix prices for MDI and related polyether polyols; she stated that Bayer, BASF and Huntsman were “aggressive competitors.” AA0696-98, 0700-04, 0712-13.

Almost none of plaintiffs’ evidence pertained to Lyondell. The only Lyondell employee who testified stated without contradiction that Lyondell did not participate in any agreement to coordinate price increase announcements and, in fact, avoided contacts with an executive of another producer that discussed prices on one occasion. AA0757-61, 0767, 0770-72.

Plaintiffs also called Dr. John Solow, who opined that the industry structure was conducive to collusion, and that low margins and excess capacity gave producers incentives to collude. AA0798-99. He characterized the conspiracy as an agreement “to issue announcements of prices by an amount within some range and to try to match those price increases and then to stick to them as best they could to prevent the price from falling to the competitive level.” AA0862. He did not analyze whether there was any relationship between the price increase

announcements and actual prices, instead basing his opinion that the alleged cartel was effective primarily on McClave’s models. AA0831-33, 0849-51.

McClave, in turn, assumed the existence of a conspiracy to fix prices and allocate customers and markets, but did not analyze the price increase announcements and how they impacted prices. Instead, he developed regression models to predict the competitive prices that would have occurred in the absence of conspiracy. AA2079-86. To “specify” those models, he used price data from “benchmark” years outside the alleged conspiracy period. He then applied those models to predict what “competitive” prices would be during the “conspiracy” period. AA2080; AA0992-94.

McClave used his models to calculate injury and damages for approximately 25% of the class members. AA0879, 1033-34, AA2172-73. For the remainder (which included all Lyondell purchasers and all systems purchasers), McClave assumed class members were overcharged and calculated “extrapolated” damages based on the average percentage of the overcharges calculated in his regression models. AA1033-36; AA2087; AA2167, 2172-73. McClave calculated that total damages for the class were $1,125,608,094 between January 1, 1999 and December 31, 2003, and $496,680,046 between November 24, 2000 and December 31, 2003. AA1587-88.

Dow called two expert economists, Dr. Keith Ugone and Dr. Kenneth Elzinga. Elzinga testified that the pricing data did not support plaintiffs’ allegations of a price-fixing conspiracy from 1999 through 2003. He explained that the price increase announcements did not regularly “stick” and, in fact, often dropped as illustrated below.³ AA1279-92, 1296-97, 1303.

AA1527.

_________________

³ The blue line shows actual prices. The red lines show the months where price increase announcements occurred and what prices would have been in the next month if prices went up by the amount of the price increase announcement.

Prices during the alleged conspiracy were basically flat, and then rose after the conspiracy allegedly ended. AA1311-13.

AA1534. This is the “flip side of what you would normally see” occur with a cartel. AA1314.

Elzinga explained that plaintiffs’ theory was contrary to basic economics. Because demand was relatively flat and costs were higher at the end of the alleged conspiracy period than at the beginning, there was no economic reason to think prices would have fallen. AA1315-25.

Elzinga also saw numerous examples of head-to-head competition, something he would not expect to see in a cartel. AA1330-37. There was evidence of purchasers refusing to accept announced price increases and negotiating lower prices. AA1336-37, 1448-71; AA1672; AA1633-34; AA1647-49; AA1653-57; AA1667-71; AA1693-1702; AA1707-37. Foamex repeatedly played the defendants off each other to get better prices. AA1330-36; AA1768-70. Great Dane benefited from a “price war” between Dow and BASF, AA1703-06, while Woodbridge received price protections and offers of “lower prices almost weekly in an attempt to secure” its business, AA1735-37. See also AA1457 (Bayer “rolled over” an $0.08/lb announced price increase so it “didn’t have any impact” on Woodbridge); AA1203-06; AA1678-81; AA1650-52 (detailing lower prices and price protections Woodbridge received). Most vividly, Leggett and Platt sent Dow an email saying that it was throwing Dow’s price increase announcement in the “circle file” and that Dow “should check out who is sending this B S and terminate them immediately—before someone reminds us of announced recent price increases that deteriorate below the existing price.” AA1743; see also AA1686 (Bayer email stating that Huber “is not changing the price in their system” and Bayer “will not be able to get a higher price ... no matter what we invoice them!”).

Ugone testified that McClave’s regression models do not reliably predict “but for” prices. AA1400-19. McClave chose variables not because there was an economic reason to think they drove prices, but because they happened to correlate with actual prices in the 2004-2008 “benchmark” period. AA1393-1405. McClave also decided to move 2004 to the “benchmark” period even though plaintiffs alleged it was a “conspiracy” year. AA2081. Ugone explained that if 2004 is removed from the benchmark period, McClave’s models do not perform well. AA1408-09. If McClave’s models are run using a 2005-2008 benchmark, they do not accurately predict 2004 prices, and they produce nonsensical results for the 1999 to 2003 period. AA1406-09; AA1612-16. Likewise, if 2008 is removed from the benchmark period, the models do not accurately predict 2008 prices. AA1406-09; AA1612-16. Finally, Ugone criticized McClave’s extrapolations because they calculate overcharges for every transaction regardless of whether the customer was actually injured. AA1419-28.

Jury Verdict. The jury returned a split verdict, finding that (1) there was a conspiracy (2) that caused class members to pay more for urethane chemicals, but (3) there was no overcharge before November 24, 2000,⁴ and (4) the damages were $400,049,039. AA0513-15.

___________________

⁴The verdict form asked about pre- and post-November 24, 2000, the date the four-year Clayton Act statute of limitations would run absent fraudulent concealment. AA0513-15.

Post-Trial Motions. Dow moved for judgment on the verdict and as a matter of law or for a new trial and renewed its motion to decertify the class, arguing that the Supreme Court’s recent decision in Comcast, 133 S. Ct. 1426, provided additional grounds for decertification. The district court denied both motions, but modified the class to exclude purchases in 2004 (the year McClave dropped from the alleged conspiracy). AA0522-51.

The court rejected Dow’s argument that McClave’s use of extrapolated damages fails to show impact. It said Dow did not seek to exclude McClave’s testimony on this basis, no expert opinion showed “McClave’s method was unreliable,” and no “relevant precedent” supported Dow’s argument. AA0526. The district court did not discuss or distinguish Wal-Mart.

With respect to Comcast, the court recognized that a damages model must measure damages attributable to the antitrust violation plaintiffs pursued at trial. AA0529. Here, McClave’s models assumed that defendants engaged in price fixing and allocation of customers and markets, but plaintiffs presented only the price-fixing theory at trial. The court refused to decertify, however, because it thought McClave testified at trial that the alleged price-fixing conspiracy itself impacted nearly every class member. AA0529-30.

The district court also rejected Dow’s argument that the jury verdict establishes that McClave’s models do not reliably show impact and damages.

Although the jury found no overcharges before November 2000, the court did not think it necessarily rejected McClave’s opinion that the difference “between actual prices and his but-for model could be attributed to the alleged price-fixing.” AA0535. The court also found that, even if there were insufficient evidence of Lyondell’s participation, the jury “could reasonably have found a conspiracy involving Dow and at least one of the other manufacturers.” AA0537-39. The court also found the jury did not have to speculate to arrive at a damages figure different from that calculated by McClave. AA0537.

Entry of Final Judgment. After denying the post-trial motions, the district court trebled the jury verdict and entered judgment against Dow for $1,200,147,117. AA0552. Both parties moved to amend this judgment because it did not include, as an offset, the $139.3 million paid by the settling defendants. AA0555. Plaintiffs also moved to amend the judgment to permit a claims administrator to allocate the judgment among class members who purchased chemicals after November 23, 2000 in accordance with McClave’s calculations, with a pro rata reduction to reflect the jury’s award of an amount less than McClave calculated. AA0556-57. Dow objected that this violates the Seventh Amendment by allocating damages pursuant to McClave’s models even though the jury declined to follow those models in critical part, and the verdict form did not indicate which transactions the jury thought were overcharged. AA0557-58. The

court found, however, that Dow had no right to complain about “the manner in which the total damages found by the jury are distributed among the class members.” AA0558. The court entered an amended judgment of $1,060,847,117, plus interest, to be distributed in accordance with plaintiffs’ proposed allocation plan. AA0566-67.

SUMMARY OF ARGUMENT

I. The district court erred in certifying (and refusing to decertify) the class.

First, allowing plaintiffs to proceed as a class violated Wal-Mart. The district court erred in finding that common issues predominate because plaintiffs said they could show antitrust impact and damages with McClave’s models. Even if this were true (and it is not), in certifying the class the district court disregarded Dow’s right to show, in individual proceedings, that particular class members could not establish injury and impact based on their relative bargaining power, the availability of substitute products, and other factors that enabled them to avoid price increases. This violates Wal-Mart’s holding that a class cannot be certified where individualized proceedings are necessary to determine the defenses that may be raised to the claims of individual class members. See infra §I.A.1.

The district court also independently violated Wal-Mart by holding that classwide impact could be demonstrated on the basis of extrapolation techniques that simply assume, rather than establish, the fact of injury. Wal-Mart precludes

such extrapolation where, as here, the defendant has evidence that not all class members are “average.” See infra §I.A.2.

Second, allowing plaintiffs to proceed as a class violated Comcast. McClave developed his models on the assumption that plaintiffs would prove a conspiracy to fix prices and allocate customers and markets, but plaintiffs attempted to prove only price-fixing at trial. The damages shown in McClave’s models are thus not attributable solely to the antitrust violation that forms the basis for Dow’s liability. This is the very error that precluded class certification in Comcast. See infra §I.B.1.

In addition, McClave’s could not reliably establish the existence of supra-competitive prices at all. He engaged in blatant “variable-shopping,” ignoring factors that drive the price of polyurethanes, and he had to modify the dates of the conspiracy alleged by plaintffs to get prices that appeared to be supra-competitive. See infra §I.B.2

The jury’s findings confirm that McClave’s models cannot predict “competitive” prices. His models showed over $600 million in overcharges from January 1, 1999 to November 24, 2000, but the jury found no overcharge during that period. The jury thus necessarily rejected McClave’s approach, as there is no basis for concluding that models that could not accurately predict “competitive” prices prior to November 2000 could do so after that date. See infra §I.B.3.

II. The judgment should be reversed and judgment entered for Dow because there is insufficient evidence to sustain the jury’s liability finding. An actionable conspiracy requires actual impact, and there is insufficient evidence to show classwide impact. Average prices remained flat or fell during the class period, most price increase announcements did not lead to price increases, the producers’ pricing was not parallel, and there was evidence of vigorous competition. See infra §II.A.

Dow also is entitled to judgment or, at a minimum, a new trial because there is insufficient evidence that Lyondell (one of the major producers) participated in any conspiracy. Moreover, McClave did not even attempt to “model” any Lyondell prices, and Lyondell’s prices were typically the lowest—often by a wide margin. Plaintiffs offered no theory as to how an agreement without Lyondell could result in classwide anticompetitive impact. See infra §II.B.

III. Finally, because plaintiffs asked the jury to award an aggregate damages figure and did not give the jury information to reduce that figure if they disagreed with any aspect of plaintiffs’ claims, the district court erred in reducing the verdict to a final judgment. The jury reduced by nearly $100,000,000 the damages McClave calculated between November, 24 2000 and December 31, 2003, but there is no evidentiary basis for the amount ultimately awarded. Damages necessarily based on jury speculation cannot be sustained. See infra §III.A.

The district court also violated the Seventh Amendment by entering an amended final judgment that allocates damages based on McClave’s models, with pro rata reduction to account for the jury’s reduction of McClave’s damages figure. Dow has a constitutionally protected right to pay only those class members that the jury found to be injured and only to the extent of the injury the jury found. That right was violated here, because there is no way to determine which class members the jury thought were not injured or were injured less than McClave calculated. Accordingly, a new trial is warranted. See infra §III.B.

STANDARD OF REVIEW

This Court reviews the standard the district court used in certifying a class de novo, “‘and the merits of that determination for an abuse of discretion.’ Material misapplication of the Rule 23 factors constitutes an abuse of discretion.” Wallace B. Roderick Revocable Living Trust v. XTO Energy, Inc., 725 F.3d 1213, 1217 (10th Cir. 2013) (citation omitted).

The denial of Dow’s motion for judgment as a matter of law is reviewed de novo, drawing all reasonable inferences in favor of plaintiffs. Hardeman v. City of Albuquerque, 377 F.3d 1106, 1112 (10th Cir. 2004). This court “review[s] de novo whether the district court applied the proper standard” in admitting expert testimony, but reviews the district court’s performance of its “gatekeeping role” for abuse of discretion. United States v. Garcia, 635 F.3d 472, 476 (10th Cir. 2011).

The denial of a new trial motion is reviewed for abuse of discretion. M.D. Mark, Inc. v. Kerr-McGee Corp., 565 F.3d 753, 762 (10th Cir. 2009). A jury’s damages award will be reversed if it “is clearly erroneous or there is no evidence to support [it].” Guides, Ltd. v. Yarmouth Grp. Prop. Mgmt., Inc., 295 F.3d 1065, 1076 (10th Cir. 2002).

Finally, whether the judgment violates the Seventh Amendment is a question of law reviewed de novo. J.R. Simplot v. Chevron Pipeline Co., 563 F.3d 1102, 1115 (10th Cir. 2009).

ARGUMENT


I.	THE DISTRICT COURT ERRED IN CERTIFYING THE CLASS.

The “antitrust laws don’t turn private parties into bounty hunters entitled to a windfall anytime they can ferret out anticompetitive conduct.” Novell Inc. v. Microsoft Corp., 731 F.3d 1064, 1080 (10th Cir. 2013). To recover damages, plaintiffs must show “impact”—i.e., that the illegal conduct actually injured them. Id. And, to proceed as class, plaintiffs must show that injury is “capable of classwide resolution … in one stroke.” Wal-Mart, 131 S. Ct. at 2545. Following Wal-Mart, courts must conduct a “rigorous analysis before concluding that” Rule 23’s requirements are satisfied. XTO Energy, Inc., 725 F.3d at 1217 (quotations omitted).

The “rigorous analysis” required by Rule 23 demonstrates that class certification was improper. It is undisputed that there was no single “market price” for Polyurethane Products because sales were individually negotiated. Plaintiffs alleged not that defendants conspired to fix prices, but that they conspired to coordinate “price announcements” and “to try to make them stick as much as possible.” AA0866 (emphasis added). The evidence, however, showed that prices varied considerably, moved in different directions, and fluctuated from month-to-month. Lower-priced alternatives were available throughout the alleged “conspiracy,” and price increases did not actually “stick.” To the contrary, despite steady demand and increasing primary input costs for all four product categories, average prices at the end of the alleged conspiracy were lower than or equal to average prices at the beginning, and rose sharply after the alleged conspiracy ended. See infra §II.A.

When given the “hard look” that Rule 23 “commands,” In re Rail Freight Fuel Surcharge Antitrust Litig., 725 F.3d at 255, it is evident that certification of the class violated Wal-Mart and Comcast. This Court should decertify the class and vacate the judgment.


A.	Wal-Mart Precludes Class Certification.

To meet the predominance requirement of Rule 23, plaintiffs must be able to prove through common evidence a conspiracy to raise prices and “that all class

members were in fact injured by the alleged conspiracy.” In re Rail Freight Fuel Surcharge Antitrust Litig., 725 F.3d at 252. The district court thought plaintiffs could make that showing with McClave’s statistical models. Even if these models were otherwise valid—and as explained in §I.B below, they are not—they cannot support class certification for two reasons. First, Wal-Mart makes it clear that plaintiffs cannot obtain class certification based on statistical models without regard to the defendant’s contrary individualized proof. 131 S. Ct. at 2561. Dow was entitled to defend itself by showing that particular class members could not have been injured by any conspiracy in light of the specific negotiations, prices, contract terms and products involved. Individual trials are therefore necessary “to establish whether a particular [class member] suffered harm from the price-fixing scheme.” In re Rail Freight Fuel Surcharge Antitrust Litig., 725 F.3d at 252.

Second, plaintiffs used McClave’s models to determine impact and calculate damages for only about 25% of class members. AA0879-80,1034. For the remaining 75%, plaintiffs relied on McClave’s “extrapolations” that assumed class members were overcharged by the average percentage of overcharges observed in the transactions McClave modeled. AA1035. Wal-Mart makes clear that extrapolated injury is not a valid basis for class certification.


1.	Dow Was Entitled To Show, In Individualized Proceedings, That Particular Class Members Suffered No Injury Or Damages.

The district court erred in certifying the class because classwide adjudication deprived Dow of its right to establish that particular class members were not overcharged by any conspiracy. There is no single “market price” for Polyurethane Products. AA0850, 1133,1162; AA2007; AA2048; AA2035. Prices are determined through individualized negotiations between manufacturers and purchasers, many of whom were large, sophisticated entities. AA0849-50, 0886, 1137-42, 1166, 1362-63; AA1658-60; AA1696-702; AA1703-06; AA1707-30; AA1862-66, 1877-80, 1887-89, 1891, 1893-97; AA2007; AA2048; AA2035;. Once a price increase is announced, “the customer begins the process of defeating” it. AA1139; see also AA0850, 0886, 1161-66, 1179, 1188-89, 1221, 1303; AA1862, 1877-79. Indeed, for some customers, the possibility of obtaining a price increase was so remote, that a manufacturer did not even send a price increase announcement. AA2007.

Many individual customers had, and used, “ buying power” to “leverage ... [a] lower price,” AA1129, frequently by playing one supplier off another, e.g., AA1658 (Bayer lowered price for FFP to ward off competition from Dow); AA1696-97 (Bayer dropped base price for Vita to meet competition from Dow, BASF and Shell, and dropped prices for Carpenter, Vita and Hickory Springs due

to competition from BASF and Lyondell); AA1707-30 (detailing competition among suppliers and price cuts for over 15 companies); AA1703-06 (same for MDI and systems). Some customers obtained concessions such as rebates, most-favored-nation clauses, early payment discounts and investments from the manufacturer to develop new products. AA0650-51, 0702, 0910-11, 1138, 1348; AA1919-20; AA2012-14; AA2050-52; AA2038-39; see also AA1672 (Dow offered GE Appliances $3 million to switch business from Bayer); AA1742 (Foamex invoking clause in Dow contract to obtain lower prices and more favorable payment terms that other purchasers received). Some customers negotiated “price protections” insulating them from price increases for specific periods. AA0738; AA1919-20; AA2050-51, 2056, 2059-60; AA2012-14, 2019; see also, e.g., AA1707-30 (Bayer lost MDI and systems business to Dow and BASF, which offered lower prices and guaranteed those prices for all of 2001); AA1664-66 (“Long term guaranteed prices” are “[b]arriers” to effective price increases); AA1683-85 (MDI prices for the appliance segment of the market “are guaranteed by contract by all suppliers for the 2002 calendar year”).

Others had options that gave them leverage to reject a proposed price increase. For example, some customers could purchase from providers outside the alleged conspiracy, while others could purchase non-Polyurethane Products as

substitutes. AA1833-39, 1866-69, 1875-76, 1879, 1883-84; AA2010, 2016, 2019-20; AA2049-50, 2055, 2058.

The district court acknowledged that the class involves a “myriad of products, pricing structures, individual negotiations, and contracts,” AA0413, and that systems are customized and thus “heterogeneous, non-commodity, non-fungible products.” AA0411. The court initially ruled, however, that plaintiffs could show “widespread injury” under the theory that “coordinated price … announcements” increased the “baseline from which any individualized negotiations would proceed.” AA0409-10. In denying Dow’s motion to decertify based on Wal-Mart (and later Comcast), the court added that “all members of the class may be shown to have been impacted by a conspiracy that elevates prices above the competitive level, even if some members” did not pay the elevated prices “or otherwise did not suffer damages that may be quantified.” AA0525. That is wrong.

To prove injury-in-fact under the antitrust laws, a plaintiff must demonstrate that it paid more than it would have paid in the absence of the conspiracy. See 15 U.S.C. §15(a) (only a claimant who was “injured in his business or property by reason of” a violation may recover damages under the antitrust laws); Novell, 731 F.3d at 1070 (“a private plaintiff must show that its injuries were caused by the defendant’s anticompetitive conduct”). And where, as here, prices are individually

negotiated and there is evidence that individual class members negotiated away announced price increases, a court cannot simply presume that an allegedly conspiratorial price increase announcement caused all class members to pay supra-competitive prices.⁵

The district court also erred in certifying the class based on McClave’s models, which plaintiffs claimed proved impact and damages to virtually all class members. AA0528-31. That ruling fails to account for defendants’ right to show that individual class members were not injured. Wal-Mart and the Rules Enabling Act “forbid[] interpreting Rule 23 to ‘abridge, enlarge or modify any substantive right.’” Wal-Mart, 131 S. Ct. at 2561 (quoting 28 U.S.C. §2072(b)). Therefore, a class “cannot be certified on the premise that [the defendant] will not be entitled to litigate its statutory defenses to individual claims.” Id. Given the realities and complexities of this market, Dow had a right to defend itself by showing, in individualized proceedings, that particular class members suffered no injury

___________________

⁵ See In re New Motor Vehicles Canadian Export Antitrust Litig., 522 F.3d 6, 29 (1st Cir. 2008) (class could not be certified on the assumption that a conspiracy that increased MSRPs for automobiles caused “all consumers” to pay more); Robinson v. Tex. Auto. Dealers Ass’n, 387 F.3d 416, 423 (5th Cir. 2004) (assumption that raising base price for cars increased the purchase price for every class member “defies the realities of the haggling that ensues in the American market”); 2A, Phillip E. Areeda & Herbert Hovenkamp, Antitrust Law: An Analysis of Antitrust Principles and Their Application § 398(c), at 423, n.14 (2013) (“When transaction prices are negotiated,” “proof of antitrust injury is bound to be individualized.”).

because their specific circumstances enabled them to avoid a price increase. McClave’s models provide no justification for denying Dow that right.

In a class action of 2,400 businesses that purchased hundreds of different chemicals pursuant to agreements that were individually negotiated with five manufacturers over five years, it is not possible for Dow to discover and present evidence of all the individualized transactions where a class member played one defendant off another to avoid a price increase, switched to an alternative source of supply, never even received a price increase announcement because it had a fixed-price contract, or otherwise avoided injury. Nor is it feasible for the jury to weigh that individualized evidence and compare it to the damages predicted by plaintiffs’ model and decide whether particular class members were, in fact, injured in the amounts specified in the model. Indeed, plaintiffs consciously sought to avoid such individualized inquiry: they did not present the jury with McClave’s damages calculations for individual class members (except to give aggregate damages figures for the named plaintiffs), and they successfully requested that the jury be instructed to provide only a lump-sum damages award. See infra pp. 62-63.

Class certification thus impaired Dow’s substantive right to show that individual class members were not injured. The district court erred as a matter of law in refusing to decertify the class.


2.	The District Court Violated Wal-Mart By Allowing The Class To Proceed On The Basis Of Assumed, “Extrapolated” Impact And Damages.

The district court violated Wal-Mart for another, independent reason: It allowed plaintiffs to model injury and damages for one-fourth of the class and rely on extrapolations that assumed the remainder were injured by the modeled average. This violates Wal-Mart’s injunction against “Trial by Formula.” Wal-Mart, 131 S. Ct. at 2561.

To estimate impact and damages by extrapolation, McClave first calculated the average “overcharges” observed in the transactions used in his regression models. AA1032-41, 1419-28; AA1569; AA2167. Claiming that he lacked the data to apply his models for most class members, McClave applied average overcharges determined from his regression models (14% for TDI, 15.6% for MDI, and 14.9% for polyols) to the hundreds of thousands of transactions he did not model. AA1033-41, 1419-28; AA2167, 2173; AA2271-72, 2362. For systems (which he did not model at all), McClave assumed that an increase in the price of the basic chemicals caused a proportional increase in the price of the system. AA1036-40; AA2087-88. He estimated that the basic chemicals comprise approximately 74% of a system, by weight, and assumed that all systems were subject to an overcharge equal to approximately 74% of the average overcharge for MDI, resulting in a 7.2% overcharge. AA1036-40.

Thus, McClave assumed that each class member whose transactions were not modeled was injured and suffered damages in each transaction equal to the average overcharge for that product category. AA1426-27. Wal-Mart prohibits such assumptions. The Court held that the Rules Enabling Act forbids certifying a class on the theory that liability and damages will be determined for a sample set of class members and the average damages applied by extrapolation to the rest of the class “without further individualized proceedings.” Wal-Mart, 131 S. Ct. at 2561; see also, e.g., Rodriquez v. Nat’l City Bank, 726 F.3d 372, 385 (3d Cir. 2013) (regression showing disparity in “average” lending rates does not show that “the policy affected each individual plaintiff in the same general fashion”); Bolden v. Walsh Const. Co., 688 F.3d 893, 896 (7th Cir. 2012) (data “show[ing] that black workers did worse than white workers” simply does “not imply that all” supervisors discriminated as necessary to establish classwide liability); Gates v. Rohm and Haas Co., 655 F.3d 255, 266 (3d Cir. 2011) (rejecting class certification because “[a]verage … estimations would not be probative of any individual’s claim”).

a. Although ignored in his reports, McClave acknowledged that under his regression models, many customers were not harmed by the alleged conspiracy. AA2221-22; AA2232; AA2441, 2445. Approximately 10 percent of McClave’s “modeled” calculations produced negative damages—i.e., actual prices paid were

less than the “competitive” prices predicted by McClave’s regression models. Compare AA2415, with AA2441. And at least seven percent of class members whose damages were determined only by McClave’s regression models (meaning that damages were not extrapolated for any of their purchases) had zero or negative damages.⁶ Yet McClave’s extrapolation assumes that, in every extrapolated transaction, the customer was injured by the conspiracy and should be awarded damages equal to the average overcharge for that product category.

McClave’s extrapolations are also contradicted by evidence in the record. For example, Quabaug purchased only systems, AA0424, so McClave assumed all of its purchases during the class period were subject to a 7.2% overcharge. But when Quabaug’s then-supplier announced a price increase in February 2001, Quabaug switched to Bayer and purchased a system “at five cents per pound less than it had been purchasing from Huntsman.” AA0425. See supra p. 8.

Similarly, uncontroverted testimony and supporting documents showed that Whirlpool switched its business from Bayer to BASF because Bayer announced a price increase, and BASF countered with a lower price. AA1445; AA1703-06.

____________________

⁶Of the approximately 2400 class members, only 192 had their damages determined solely by McClave’s models with no extrapolation. Compare AA2415, with AA2445 (McClave not disputing this figure). Dow produced evidence that 24 of these “modeled-only” customers had zero or negative damages. AA2415. McClave says the number is 14. See AA2445. Using McClave’s number, at least seven percent (14/192) of “modeled-only” customers were not impacted by the conspiracy.

Yet McClave’s extrapolations assume that Whirlpool was injured and entitled to damages on these purchases equal to the average shown on McClave’s models for purchases by other class members from all defendants. AA2364; AA2424.

b. McClave’s use of extrapolation was particularly inappropriate because he wrongly assumed that the transactions for which he extrapolated damages were comparable to the transactions he modeled. In fact, there were significant differences between modeled and extrapolated transactions. AA1431, 1436-38. McClave erred in assuming away those differences. See Cimino v. Raymark Indus., Inc., 151 F.3d 297, 320 (5th Cir. 1998) (rejecting use of extrapolation inter alia where there was no showing sample was “representative” of “extrapolation plaintiffs”). The two starkest examples involve Lyondell and systems transactions.

First, McClave did not model any Lyondell transactions. AA1055,1431; AA2172-73. He simply assumed that Lyondell was in the conspiracy and charged an anticompetitive markup of the average of the other four manufacturers on every transaction. Id.; see also AA1427,1437. These assumptions are unwarranted. As Ugone explained, McClave did “no testing” for Lyondell. AA1437. McClave did not analyze differences between Lyondell’s products, customers, or pricing decisions and the ones he modeled. He cannot assume the company participated in any conspiracy, much less overcharged on every transaction to the same extent as the other defendants. See infra §II.B. In addition, McClave’s regression models

showed a range of overcharges for TDI and polyether polyols, AA2167, and McClave had no basis for assuming that Lyondell imposed the “average” overcharge rather than a lower one or none at all. Lyondell frequently charged lower prices than the other manufacturers, see infra §II.B, and pursued a different business strategy than the other manufacturers (indeed, one that ultimately led to its bankruptcy). AA1348.

Second, as previously explained (supra pp. 15, 34), McClave did not model any systems transactions. He simply assumed that “because systems are comprised overwhelmingly of basic chemicals,” an increase in the price of the basic chemicals caused a corresponding increase in the price of systems. AA2087-88. McClave conceded that he undertook no statistical analysis to support that assumption, AA2227-28, 2232, and the record belies it.

Defendants did not attempt to raise the systems prices each time they attempted to raise the price of basic chemicals. Between December 2000 and December 2003 (when the jury found impact), plaintiffs alleged there were ten collusive MDI or polyol price increase announcements. AA1772-92. For the majority of these announcements, at least one or two of the defendants did not announce a corresponding price increase for systems. Id. Indeed, on three occasions, Dow, Bayer, BASF and Huntsman each announced price increases for MDI and/or polyols but did not announce any systems increases. Id. On only two

occasions (March/April 2003 and October 2003) did all defendants announce price increases for both basic chemicals and systems. Id.

This disconnect between systems and component chemical pricing is unsurprising because systems are, as the district court recognized, specialized products “specifically engineered to meet the performance requirements of individual customers” with completely different supply and demand” considerations. AA0411; see also AA1852-71. There are “different market dynamics” between “systems” and the “raw material side of the business.” AA1198. “Defendants routinely charge prices for systems that reflect the value that those systems provide to customers, not just simple markup over production costs for the systems.” AA1862; see also AA1433; AA2010-11.

The record also shows that Dow and the other manufacturers faced price competition from independent systems houses not part of the alleged conspiracy. AA1836-37, 1866-69; see also AA1687-92 (Bayer detailing business lost to systems houses that did not raise prices); AA1683-85 (Dow and BASF announced systems price increases, but Carpenter did not raise systems prices); supra p. 8 (Quabaug obtained competitive prices for systems). This competition prevented defendants from implementing uniform systems price increases even if they conspired to do so. AA1836-37, 1866-71. In addition, many systems products face competition from non-Polyurethane Products not involved in the alleged

conspiracy, so any attempt to raise prices could have been defeated by “substitution.” AA1838; AA2010. Finally, as noted, many systems customers had price protections preventing a price increase even if the price of the components increased. AA1683-85; AA1707-30; AA1923-24; AA2012-14.

Notwithstanding these economic realities, McClave made “no empirical effort” to justify his across-the-board 7.2% overcharges for systems sales. AA1438-39. Given the economic realities of the marketplace, and the direct evidence of competition from independent systems houses, McClave’s assumption was plainly unwarranted and his extrapolations untenable.


B.	Comcast Precludes Class Certification.

The district court also erred in holding that plaintiffs could use McClave’s models to prove antitrust impact on a classwide basis. McClave made no attempt to relate a particular anticompetitive act with a price effect. Price increase announcements are not a variable in his models, AA1116, and his report “doesn’t contain a word” about the “relationship between price announcements and price” or “the relationship between communications suggesting coordination and price.” AA1117. Indeed, he conceded his models are “not supposed to capture cartel events.” AA1115; see also AA0853,0855 (Solow agreeing there is “no relationship” between price-increase announcements and actual prices in

McClave’s models); AA0853 (Solow stating McClave did not “quantif[y] any relationship between announced prices and actual prices”).

As explained below, McClave’s models cannot support class certification under Comcast. As in Comcast, McClave claimed that his model reflected overcharges attributable to multiple antitrust violations, and failed to adjust for the fact that plaintiffs pursued only one antitrust theory at trial. His gerrymandered models, moreover, cannot reliably show the existence of supra-competitive prices, much less supra-competitive prices caused by the antitrust violation plaintiffs sought to prove. That is confirmed by the jury’s verdict, which found no overcharges between 1999 and November 2000, a period in which McClave found over $600,000,000 in overcharges.


1.	McClave’s Models Do Not Purport To Determine The Impact And Damages Resulting From The Conspiracy Plaintiffs Sought To Prove At Trial.

McClave developed regression models that purported to model the impact of the two types of anticompetitive conduct alleged in plaintiffs’ complaint: fixing prices by coordinating pricing announcements, and “allocat[ing] customers and markets.” AA0368; AA2709. In theory, an agreement to allocate customers and markets could have allowed suppliers to increase prices to their allocated customers and markets and could have enhanced the effectiveness of an agreement to fix prices through coordinated price increase announcements by making it more

difficult for class members to switch suppliers to avoid price increases. McClave’s expert reports stated that McClave assumed defendants engaged in both types of anticompetitive conduct, and he developed a single measure of damages that reflected the combined effect of both. AA2079-80; AA2147, 2180.

At trial, plaintiffs pursued only the “price fixing,” not the “allocation” claim. AA0451-52, 0462; AA0573-74, 0836, 1477-1517. But McClave did not adjust his models. Instead, he presented at trial the damages figures from his rebuttal report. AA0919-1014, 1018-73; AA1569, 1576-78, 1582, 1585.

McClave’s failure to distinguish between the impact and damages attributable to the liability theory plaintiffs pursued at trial and another liability theory that they dropped is the precise flaw that precluded class certification in Comcast. There, as here, plaintiffs alleged multiple antitrust theories, and McClave developed a regression to predict the prices that would have been charged in the “but for” world in which no such violations occurred. Comcast, 133 S. Ct. at 1431-32. There, as here, plaintiffs ultimately pursued only one of those theories, so McClave’s “model failed to measure damages resulting from the particular antitrust injury” on which defendants’ liability was premised. Id. at 1433-34. A “model purporting to serve as evidence of damages in this class action,” the Supreme Court explained, “must measure only those damages attributable” to the liability theory plaintiffs pursued. Id. at 1433 (emphasis

added). Because McClave’s model failed to do so, it could not “bridge the differences between supra-competitive prices in general and supra-competitive prices attributable” to the antitrust violation plaintiffs alleged. Id. at 1435. “Prices whose level above what an expert deems ‘competitive’ has been caused by factors unrelated to an accepted theory of antitrust harm are not ‘anticompetitive’ in any sense relevant here.” Id. at 1435.

The district court rejected this argument, stating that Dow waived it by failing to seek the exclusion of McClave’s testimony on this ground. AA0530. But Comcast makes clear that the failure to object to the admission of expert testimony under Federal Rule of Evidence 702 does not preclude a showing that the testimony is insufficient to satisfy Rule 23’s predominance requirement for class certification. 133 S. Ct. at 1431 n.4. Indeed, the Supreme Court found McClave’s model insufficient even though Comcast had never moved to exclude McClave’s testimony. Id.

The district court also thought Comcast was inapplicable because here McClave testified at trial that the “horizontal price-fixing conspiracy … impacted nearly every class member.” AA0530. That is wrong. McClave never opined that the impact was due only to price-fixing. He testified that “given the persistence and size” of the damages estimated by his models, he concluded that “nearly all class members had been impacted or overcharged.” AA0940. Because the models

that generated those sizable damages assumed there was an allocation agreement and a price-fixing agreement, McClave’s trial testimony makes clear that his impact opinion was based on an assumption that the jury would find defendants engaged in both types of anticompetitive conduct.

Finally, the district court deemed it significant that no party “presented any evidence at trial of any illegal customer allocation.” AA0530. But it is McClave who constructed models that assumed plaintiffs would prove a customer allocation conspiracy. Plaintiffs’ failure to prove an allocation scheme demonstrates that McClave’s methodology measures damages that are “not a result of the wrong” Dow was found to have committed. Comcast, 133 S. Ct. at 1434 (emphasis added). As in Comcast, “[t]here is no question that the model failed to measure damages resulting from the particular antitrust injury on which [defendant’s] liability in this action is premised.” Id. at 1433-34.


2.	The District Court Erred By Allowing The Class To Proceed On The Basis Of Regression Models That Do Not Reliably Predict Prices In A Competitive Market.

“[A]ll scientific testimony or evidence admitted” at trial must be “relevant” and “reliable.” Daubert v. Merrell Dow Pharm, Inc., 509 U.S. 579, 589 (1993); see also Fed. R. Evid. 702. Moreover, to justify class certification, a model must “‘translat[e] … the legal theory of the harmful event into an analysis of the economic impact of that event.” Comcast, 133 S. Ct. at 1435 (emphasis in

original). “It is not enough [for plaintiffs] to submit a questionable model whose unsubstantiated claims cannot be refuted through a priori analysis. Otherwise, ‘at the class-certification stage any method of measurement is acceptable as long as it can be applied classwide, no matter how arbitrary [its] measurements ….’” In re Rail Freight Fuel Surcharge Antitrust Litig., 725 F.3d at 254 (quoting Comcast, 133 S. Ct. at 1433).

McClave violated both principles. Even assuming class certification could be predicated on models that do not measure damages caused by the specific conduct plaintiffs alleged to be illegal, McClave engaged in blatant “variable shopping” and other data manipulation to manufacture “supracompetitive prices” during the alleged conspiracy. His models thus provide “no reliable means of proving classwide injury in fact.” Id. at 253.

McClave’s Variable Shopping. McClave sponsored models that purported to explain the relationship between prices (the dependent variable) and independent variables (such as proxies for supply and demand) that affect price in a market unaffected by collusion. AA2081-82. He ran the models during the alleged conspiracy period, observed that the prices predicted by his models were (usually) lower than actual prices, and said the jury could infer that the difference was overcharges caused by the conspiracy. AA2080, 2089. In constructing such models, one should “include[] all the variables likely to have a large effect on the

dependent variable” and be able to “safely assume that the remaining effects are not correlated with the independent variables included.” Franklin Fisher, Multiple Regression in Legal Proceedings, 80 Colum. L. Rev. 702, 713 (1980). It is therefore important to specify and include, in advance, the variables that one might think would have a significant effect on price. “To proceed by first looking at the data and then including those factors that appear correlated with the dependent variable is a recipe for spurious results,” id. at 714, such as linking S&P 500 returns to butter production in Bangladesh, see Leinweber, supra .

Yet McClave engaged in precisely such backward-looking variable shopping. He did not select variables based on a priori assumptions or economic theory about the variables that could independently explain price changes. Instead, he selected and rejected variables for his models based on whether they appeared to correlate with Polyurethane Product prices without regard to whether the selections made economic sense or were consistent. For example, he considered “at least a dozen” demand variables for the TDI model, but included only one—TDI exports. AA1069; AA2085. The model thus assumes that foreign demand for TDI affects domestic prices and that domestic demand is irrelevant—an irrational assumption, since the majority of TDI is produced to meet domestic demand. AA1068; AA2085 n.18. In specifying his MDI and polyols models, in contrast, McClave used only domestic demand variables and did not include a variable for

exports. AA1394-99; AA1597; AA2267. He continued his results-oriented approach in the MDI and polyols models by selectively picking and choosing among the variables used as a proxy for domestic demand, AA1394-99; AA2083, 2086, as the following chart illustrates:

AA1596. McClave provided no reason for believing that the demand variables he excluded do not affect price, AA1394-98, even though it is well-established that

“[b]ias caused by the omission of an important variable” can be “a serious problem.”⁷ Daniel L. Rubinfeld, Reference Guide on Multiple Regression, in Reference Manual on Scientific Evidence 303, 315 (3d ed. 2011).

McClave’s sole reason was that the statistical “signs went bad” in his model when these economically sensible variables were included. AA1068. For example, when he included proxies for domestic demand for TDI, his model generated a nonsensical result—i.e., that the price for TDI decreases as domestic demand for TDI increases. AA1069-70, 1395-97; AA2323-25. But this kind of “wrong sign” is “a friendly message that ... there is undoubtedly some shortcoming in one’s theory, data, specification, or estimation procedure.” Peter Kennedy, A Guide to Econometrics 368 (6th ed. 2008). Rather than uncover and fix that shortcoming, however, McClave discarded variables with “wrong” signs and shopped for new variables that produced the “right” signs in his models. See AA1068-70; AA2220.

The impropriety—and resulting unreliability—of such data manipulation is confirmed by the fact that McClave’s models do not show credible damages when other economically sensible variables are included. For example, when domestic

________________________

⁷Likewise, on the cost side of the equation, McClave inconsistently used 12-month moving averages for input costs in the MDI model, but 6-month moving averages for input costs in the TDI and polyols models. He provided no economic reason for thinking that prices would be driven by annual costs in one instance but not the others. AA1399-400; AA1597; AA2082-86 & nn.16,22.

demand is included in McClave’s TDI model, it shows negative damages of more than $49 million, AA2325—meaning no impact or damages.

McClave’s Benchmark Shopping. McClave’s cherry-picking did not stop with variables. It extended to the selection of the benchmark competitive period used to specify his regressions. At the time McClave submitted his report, plaintiffs alleged a conspiracy from 1999-2004. AA2136, 2139-45. Yet McClave unilaterally moved 2004 to the competitive benchmark period in his models. AA2081.

McClave admitted that he moved 2004 based on his “analyses testing 2004 as either part of [the] benchmark [period] or collusive period.” AA2215-16, 2219; AA0953-54. McClave did not explain what “analysis testing” he did, and he not disclose the results. Ugone therefore tested what happens if 2004 is included in the “conspiracy” period and found that McClave’s models show negative damages for MDI, polyols and systems and improbably high damages for TDI. AA2330 (models show 34% undercharge for polyols, 2.3% undercharge for MDI, and negative extrapolated damages for systems but a 34.2% overcharge for TDI during a period when TDI prices fell); see also AA1405, 1408-09. In other words, McClave had to include 2004 in the benchmark period to generate supra-competitive prices for all products covered by the alleged conspiracy.

* * *

The district court should have excluded McClave’s testimony. The court’s essential gatekeeping function “is to make certain that an expert … employs in the courtroom the same level of intellectual rigor that characterizes the practice of an expert in the relevant field.” Kumho Tire Co. v. Carmichael, 526 U.S. 137, 152 (1999). An econometrician seeking to model prices in an industry would study the industry and determine ex ante what demand and supply factors drive price, and would reject models that are based on nonsensical assumptions, such as the assumption that domestic demand does not affect price. Fisher, supra, at 713; Kennedy, supra, at 368. An econometrician would not gerrymander the conspiracy and benchmark periods to maximize damages and attempt to rationalize these choices ex post. McClave’s testimony cannot support class certification, much less the jury’s verdict.


3.	The Jury’s Verdict Confirms McClave’s Models Predict Overcharges Where None Exists.

The jury’s verdict confirms that McClave’s models did not reliably demonstrate antitrust impact and damages. As noted, McClave said that his models show “a variance that is not explained by th[e] [competitive] factors [he controlled for],” AA1089, and opined that “something other than competition” was responsible for the “variance” between actual prices and those predicted by his models, AA1101,1119. The jury, however, rejected that conclusion. It found no overcharge prior to November 24, 2000, AA513-15 —a 23-month period during

which, as the chart below shows, McClave’s models found over $620 million in overcharges. AA1587-88.

As the D.C. Circuit explained in a similar case, this type of flaw “shred[s] the plaintiffs’ case” for class certification. In re Rail Freight Fuel Surcharge Antitrust Litig., 725 F.3d at 252. There, plaintiffs’ expert proffered a model that generated damages for shippers that could not have been harmed by the alleged conspiracy because their rates were set by contracts negotiated prior to the alleged conspiracy. The D.C. Circuit held that no weight can be given to a “methodology [that] detects injury where none could exist.” Id.; see also MCI Commc’ns Corp.

v. AT&T Co., 708 F.2d 1081, 1161 (7th Cir. 1983) (“It is essential ... that damages reflect only the losses directly attributable to unlawful competition.”); Coleman Motor Co. v. Chrysler Corp., 525 F.2d 1338, 1353 (3d Cir. 1975) (vacating judgment where damages expert failed to distinguish between “losses resulting from unlawful, as opposed to lawful, competition.”)

The district court, by contrast, asserted that plaintiffs’ failure “to sustain their burden of proof with respect to one period of time does not necessarily mean that the evidence was not sufficient to support the jury’s finding of liability with respect to another period.” AA0535. But the models were identical in both periods, and there was no basis for the jury to reject them in one period and accept them in the other. When a plaintiff claims to have suffered substantial injury and the defendant claims the plaintiff was not injured at all, the jury may not split the difference if there is “no basis in the evidence” for doing so. Haslund v. Simon Prop. Grp., Inc., 378 F.3d 653, 658 (7th Cir. 2004); see also USFL v. NFL, 842 F.2d 1335, 1378-79 (2d Cir. 1988) (jury must have a “reasonable basis” for distinguishing between lawful and unlawful acts that cause damages).

Here, nothing in the record provides a non-speculative basis for concluding that McClave’s models could accurately predicate “competitive” prices after November 24, 2000 was the result of conspiracy but not before that date. McClave assumed a conspiracy existed from 1999 to 20003, AA2078-132; AA2146-211,

and did not consider how his models would have to change if one assumed that prices between 1999 and November 24,2000 were “competitive,” as the jury found. AA1112. For McClave’s models, November 24, 2000 was an irrelevant date; he tallied damages for the two periods separately only because of Dow’s statute of limitations defense.⁸

Nor did the district court identify any criticism of McClave’s models that applied only to the pre-November 2000 period but not the post-November 2000 period. Dow’s critique of McClave’s variable shopping (supra pp. 44-50) applied equally to both periods.

Of course, given the weakness of plaintiffs’ evidence, the jury could have concluded that the alleged conspiracy did not exist before November 2000. See infra pp. 54-62. But that only underscores that McClave’s models cannot distinguish between competition and conspiracy because they would find overcharges when the jury found no conspiracy.


II.	PLAINTIFFS FAILED TO ESTABLISH ANTITRUST LIABILITY AS A MATTER OF LAW.

For the reasons just explained, McClave’s models are insufficient to demonstrate antitrust impact. Accordingly, this Court should not only decertify the class, it should enter judgment in favor of Dow as to the named plaintiffs. Should

_________________

⁸ McClave’s expert report does not mention November 24, 2000 or specify damages before or after this date. Compare AA2090, with AA1007.

this Court instead find that the class was properly certified, Dow is entitled to judgment as a matter of law as to the entire class, because plaintiffs failed to prove an antitrust violation as a matter of law. Whether or not there was evidence from which the jury could reasonably find an agreement to fix prices, the undisputed economic evidence shows that this alleged agreement did not succeed. Likewise, plaintiffs offered no evidence from which a reasonable jury could find that Lyondell participated in any conspiracy.


A.	The Evidence Does Not Support A Finding That Defendants Implemented Any Conspiracy.

Plaintiffs do not allege defendants conspired to set particular prices; rather, they claim that defendants conspired to “announce[] increases in prices” and “to try to make them stick as much as possible,” recognizing “that there were always incentives for firms to ... cheat” when negotiating actual purchases with customers. AA0866; see also AA0862 (conspiracy was to issue price increase “announcements” and “then to stick to them as best they could”). But price increase announcements that can be withdrawn are not alone evidence of an effective cartel. Areeda & Hovencamp, supra, §1435d . Indeed, a similar pattern of price increase announcements occurred in 2004, a year plaintiffs now say was competitive. AA1375-78, 1386-88; AA0451; AA2018.

As Solow conceded, if “there’s no agreement with respect to the actual prices, that wouldn’t be much of a cartel.” AA0865. Here, plaintiffs offered no

evidence on actual pricing after the announcements or on parallel pricing. The undisputed evidence shows that there was no discernible pattern between price announcements and subsequent prices, and, frequently, prices remained flat or decreased. AA1275-87; AA1756, 1761-67. Likewise, rather than being parallel, prices varied, moved in different directions, and fluctuated from month-to-month. AA1275-305. The following chart illustrates the price variation.

AA1762-66; AA0518-20.

Defendants’ failure to maintain consistent, parallel prices shows that members of the “cartel” did not follow through on any alleged “agreement.” Fluctuating prices are what one would expect in a competitive market, not a

conspiracy. Moreover, there were always lower-priced competitive alternatives to the “price leader.” In re Baby Food Antitrust Litig., 166 F.3d 112, 132 (3d Cir. 1999). No cartel can function where only a few (or one) of the manufacturers raised prices because customers could switch to the lower priced alternatives, just as in a competitive market. AA1267-72. Not surprisingly, the evidence shows that customers took advantage of these fluctuating, nonuniform prices and switched providers where a better deal could be secured. E.g., AA1330-38; AA1768-70; supra pp. 29-31.

The economic evidence is also inconsistent with a finding of an effective conspiracy because prices at the end of the alleged conspiracy were lower than or roughly equal to prices at the beginning, and rose sharply after the conspiracy allegedly ended.

AA1756.

Plaintiffs speculate that, but for the alleged cartel, prices would have been even lower. But the undisputed evidence shows that demand was steady and primary input costs for all four product categories were higher at the end of the alleged conspiracy period than at the beginning. AA1315-29; AA1757-60. If there were a functioning conspiracy, the combination of steady demand and increased costs would have led to higher prices. Instead, steady market prices and lack of parallel pricing is consistent with extensive testimony that the companies

engaged in vigorous competition, “stealing” business by undercutting a competitor’s prices and retaining business by matching or beating a competitor’s price.⁹ AA1282-92,1330-34,1340-44,1128-30,1137-42,1197-98; see In re Baby Food Antitrust Litig., 166 F.3d at 128-32 (no inference of price-fixing where there is no parallel pricing and other evidence “reveal[ed] that the defendants engaged in independent pricing determined by market conditions”); In re Citric Acid Litig., 191 F.3d 1090, 1102-03 (9th Cir. 1999) (evidence of “nearly identical” list prices insufficient where other evidence showed that defendant “did price aggressively in actual contracts”). Thus, evidence of discussions about price increase announcements cannot support the verdict.

Nor can McClave’s models fill the evidentiary void. Even if they were valid—and they are not, see supra §I.B.2.—McClave admitted that his models did not prove causation or impact; McClave simply invited the jury to infer that differences between the prices he predicted and actual prices resulted from a conspiracy. See, e.g., AA1104 (“it’s for the jury to decide whether or not [any variation between predicted and actual prices is] a result of noncooperative or cooperative” actions by the defendants); AA1097 (“that’s ultimately for the jury to

___________________

⁹Solow characterized such competition as attempts to punish “cheater[s],” but cutting prices to win business cannot support the inference of a conspiracy (unless perhaps it results in below-cost sales, which plaintiffs made no attempt to show). AA0871,0875-76. To the contrary, “cutting prices in order to increase business is the very essence of competition.” Matsushita Electric Indus., Co. v. Zenith Radio Corp., 475 U.S. 574, 594 (1986).

decide”); AA1106 (“jury has to answer … why”). An inference drawn from damages models cannot sustain a finding of antitrust impact where, as here, actual prices were not parallel, there is evidence the alleged conspirators were competing against each other, and their average prices were falling despite steady demand and rising costs.


B.	Plaintiffs Failed As A Matter Of Law To Demonstrate That Lyondell Participated In The Alleged Conspiracy.

In addition, plaintiffs failed to present any evidence that Lyondell participated in any conspiracy. Plaintiffs’ evidence with respect to Lyondell consisted almost entirely of a single dinner attended by Edward Dineen (of Lyondell) in February 1999—22 months before November 2000, the month before which the jury found no antitrust impact. AA0513-15; AA0767-70. At the dinner, Mr. Dineen became “uncomfortable” when Mr. Dhanis (of BASF) discussed an industry price increase and sought assurances that Dow and Lyondell would participate. AA0757-61. However, Dineen testified that he was not aware of any agreement “to coordinate the price increase for polyols or TDI that was announced effective on April 1, 1999.” AA0771. He further confirmed that the dinner played no “role whatsoever in any subsequent price increase announcement that Lyondell issued,” AA0770, and that following the meeting he resolved he would “steer clear of Mr. Dhanis” and did not meet with him “for several years on any matters.” AA0771-72.

The only evidence plaintiffs offered regarding Lyondell during the period where the jury actually found overcharges was a single two-minute phone call on August 5, 2002 between Dow and Lyondell—over three years after the 1999 dinner. AA0892. Plaintiffs offered no evidence of what was discussed during the call, id., and at the time Dow and Lyondell had a supply agreement that required communication between the two companies, AA0764, 0899-901. Price increase announcements were to take effect for TDI in September 2002, but Lyondell increased TDI prices by only $0.01, one-sixth of Dow’s $0.06 increase. AA1764, 1766. Moreover, from November 2000 until the end of the alleged conspiracy in December 2003, Lyondell’s average price for TDI was the lowest of all five defendants in every quarter but two. See supra p. 55.

This evidence cannot support a finding that Lyondell participated in a conspiracy. There is no evidence that Lyondell—the largest or second largest TDI producer in the market during the relevant years, AA2282—ever sought “to coordinate near identical price announcements” and to “work[] together [with the other defendants] to try to stick to those announced prices.” AA0578. Because plaintiffs sought to prove a “single overarching conspiracy” among Dow and the other four manufacturers, AA2380, this failure of proof as to Lyondell means Dow is entitled to judgment as a matter of law.

At a minimum, a new trial is necessary. Plaintiffs offered no theory under which a conspiracy could be viable without Lyondell’s participation. See AA0819 (Solow testifying his opinion supported by the fact “all five firms” engaged in “price over volume strategy” to effectuate price-fixing agreement); AA0820 (Solow testifying that “the industry” coordinated price announcements and tried to make them stick); see also AA1415 (Ugone testifying that it “make[s] a difference” that the assumed cartel included five companies). That Lyondell did not participate in a conspiracy bears on the question of antitrust impact, because customers could have purchased TDI from Lyondell and avoided overcharges. And Lyondell’s ability to undercut attempted overcharges calls into question the incentive for the remaining manufacturers to collude. In this regard, neither of plaintiffs’ experts expressed the opinion that a conspiracy could succeed if Lyondell did not participate. AA0796-98, AA0939-41. Accord, Anixter v. Home-Stake Prod. Co., 77 F.3d 1215, 1229 (10th Cir. 1996) (“Generally, where a jury has returned a general verdict and one theory of liability upon which the verdict may have rested was erroneous, the verdict cannot stand because one cannot determine whether the jury relied on the improper ground.”). Finally, a new trial is required because plaintiffs offered no evidence of the damages attributed to Lyondell, and remittitur is not appropriate. See James River Ins. Co. v. Rapid Funding, LLC, 658

F.3d 1207, 1221-22 (10th Cir. 2011); Concord Boat Corp. v. Brunswick Corp., 207 F.3d 1039, 1054 (8th Cir. 2000).


III.	DOW IS ENTITLED TO A NEW TRIAL BECAUSE THE DAMAGES AWARD IS BASED ON SPECULATION, AND THE RESULTING JUDGMENT VIOLATES THE SEVENTH AMENDMENT.

Dow is entitled to a new trial because: (1) the damages award is based on “speculation,” Bigelow v. RKO Radio Pictures, Inc., 327 U.S. 251, 264 (1946), and (2) the district court violated the Seventh Amendment by replacing the jury’s judgment with its own–one that allocated the damages award pursuant to McClave’s models that the jury refused fully to follow. See Sloan v. State Farm Mut. Auto. Ins. Co., 360 F.3d 1220, 1225 (10th Cir. 2004) (court may not “substitute[e] its judgment for that of the jury’s regarding any issue of fact”). These defects—though independent legal grounds for reversal—both stem from plaintiffs’ calculated decision to present the jury with the total damages McClave calculated (within and outside the limitations period), but to withhold McClave’s actual calculations. As a result, when the jury rejected McClave’s pre-November 2000 damages and reduced his post-November 2000 damages figure by approximately 20%, it had no evidentiary basis for these results. The district court nevertheless upheld the damages award because it thought the jury could have concluded that Lyondell was not in the conspiracy, that the conspiracy did not exist for the entire post-November 2000 period, or that it did not cause damages to

systems purchasers. AA0537. Regardless of what surmise is made, the damages figure is speculative and not supported by the evidence, and the judgment violates the Seventh Amendment.

A. The Damages Award Is Based Upon Speculation.

McClave explained the general structure of his models, the assumptions underlying his extrapolations, and the total damages he calculated for each product in and outside the limitations period. AA0934-1009,1018-73; AA1563-88. But plaintiffs did not introduce the underlying calculations or provide the jury with the information necessary to adjust McClave’s post-November 2000 damages figures if they disagreed with any of his assumptions. This was a critical omission because McClave’s models are non-linear—i.e., the damages they calculate are not distributed equally but varied throughout the 1999-2003 period. AA1587-88. Plaintiffs likewise did not break the damages down by month, manufacturer or (except for the named plaintiffs) by purchaser.

Thus, although the record evidence would support a finding that Lyondell was not involved in any conspiracy, or that any conspiracy had a shorter duration or did not harm class members that negotiated away the announced price increases, the jury had no evidentiary basis for reducing McClave’s damages to account for these or similar deficiencies in plaintiffs’ proof. “Instead, the jury was left to pure guess and speculation,” and the damage award must be reversed. Garcia v.

Mountain States Tel. & Tel. Co., 315 F.2d 166, 169 (10th Cir. 1963); accord U.S. Indus., Inc. v. Blake Const. Co., 671 F.2d 539, 550 (D.C. Cir. 1982) (reversing award of costs as “based … upon ‘conjecture or speculation’” where plaintiff provided evidence of the amount “it incurred each day” but not “a reliable estimate of the number of days for which it incurred those ... costs”) (quotations omitted); Coleman Motor Co., 525 F.2d at 1353 (reversing verdict even though it was within the range of plaintiff’s expert because the range included damages “attributable at least in part to lawful competition” and there was no evidence from which the jury rationally could have “reduced these figures to reflect only losses due to unlawful competition”). Indeed, because plaintiffs forced the jury to speculate by not providing the jury with the tools to adjust the damages calculations if it disagreed with McClave’s assumptions, judgment should be entered in Dow’s favor (or, at a minimum, a new trial should be granted). See U.S. Indus., 671 F.2d at 550; Garcia, 315 F.2d at 169.

B. The Judgment Violates The Seventh Amendment.

The judgment must also be reversed because it violates the Seventh Amendment, which guarantees the right to have a jury determine “the question of liability and the extent of the injury by an assessment of damages.” Dimick v. Schiedt, 293 U.S. 474, 486 (1935). In reducing McClave’s post-November 2000 damages figure by almost $100,000,000, the jury departed from McClave’s models

and extrapolations. But, because it awarded a lump-sum award, there is no way to know which class members the jury found were uninjured or were injured less than McClave calculated.

The district court never acknowledged these ambiguities. It instead entered a judgment allocating damages according to McClave’s models and extrapolations, with pro rata reduction for the jury’s reduction of McClave’s post-November 2000 damages figure. AA0562-63. The court did not even attempt to explain how the judgment is consistent with the Seventh Amendment, and it could not have done so given the jury’s unexplained modification of McClave’s damages figures. Instead, the court said Dow’s Seventh Amendment objection was “untimely” and Dow has no interest in the distribution of the damages among the class. AA0557-58. Neither is correct.

Dow’s objection was timely. Dow argued in the motion for judgment as a matter of law and the motion to amend the judgment that the verdict cannot be reduced to a judgment without violating the Seventh Amendment because it is impossible to tell from the verdict which transactions the jury found were overcharged and by how much. AA0517; AA0552.4-52.5. The issue could not have been raised earlier.¹⁰

^{_________________________}

¹⁰ Dow also requested a verdict form asking the jury to identify the companies, products and time frame of any conspiracy they found. AA0510-12. When the verdict was read, Dow asked the court to ask the jury to clarify the time frame of

Moreover, Dow has an interest in ensuring that damages are properly allocated so all class members are bound by the judgment. Accord, Phillips Petroleum Co. v. Shutts, 472 U.S. 797, 804-05 (1985) (defendant had “standing” to assert arguments concerning the “rights” and “interests” of class members in order to ensure “entire plaintiff class is bound”); Carrera v. Bayer Corp., 727 F.3d 300, 310 (3d Cir. 2013) (defendant has “an interest in ensuring it pays only legitimate claims” because if payment of “inaccurate claims” causes some class members to “get less than full relief,” those class members may claim they are not bound by the judgment). The Seventh Amendment affords “both parties” the right to have liability and damages determined by a jury. Dimick, 293 U.S. at 486 (emphasis added). Thus, Dow has a right to pay only those class members that the jury found were injured, and only to the extent of the injury the jury found. That right was violated here, so a new trial is warranted.

If a new trial is granted, it should be limited to sales on or after November 24, 2000. Each overpayment gives rise to a separate cause of action, Klehr v. A.O. Smith Corp., 521 U.S. 179, 190 (1997), and the jury’s finding of no overpayment before that date is a verdict in Dow’s favor on all causes of action accruing before

__________________________________________________________________

the conspiracy they found. AA1522-23. Had Dow’s requests been granted, the court would know which companies, products, and time frames formed the basis for the award.

November 24, 2000. The district court erred in not granting Dow judgment as to those claims. AA0552.2-52.3.

CONCLUSION

For the reasons stated above, the judgment entered by the district court should be vacated and judgment entered for Dow or in the alternative, the class should be decertified and Dow should be granted a new trial.

STATEMENT OF COUNSEL AS TO ORAL ARGUMENT

Because of the important issues presented, counsel thinks oral argument may be helpful to the court.



	Respectfully submitted,

	/s/ Carter G. Phillips
Charles J. Kalil	Carter G. Phillips
Executive Vice President, General Counsel	Counsel of Record
And Corporate Secretary	Joseph R. Guerra
Duncan A. Stuart	C. Frederick Beckner III
Associate General Counsel	Kathleen Moriarty Mueller
THE DOW CHEMICAL COMPANY	Jeffrey S. Beelaert
2030 Dow Center	SIDLEY AUSTIN LLP
Midland, MI 48674	1501 K Street, NW
Telephone: (989) 636-1000	Washington, D.C. 20005
	Telephone: (202) 736-8000
	Facsimile: (202) 736-8711

December 6, 2013

CERTIFICATE OF COMPLIANCE

Section 1. Word count

As required by Fed. R. App. P. 32(a)(7)(c), I certify that this brief is proportionally spaced and contains 13,769 words.

Complete one of the following:

X I relied on my word processor to obtain the count and it is Word version 2007:

I counted all characters including citations and numerals.

I certify that the information on this form is true and correct to the best of my knowledge and belief formed after a reasonably inquiry.

By: /s/Carter G. Phillips

Attorney for The Dow Chemical Company

CERTIFICATE OF DIGITAL SUBMISSION AND PRIVACY REDACTIONS

I hereby certify that a copy of the foregoing APPELLANT’S OPENING BRIEF, as submitted in Digital Form via the court’s ECF system, is an exact copy of the written document filed with the Clerk and has been scanned for viruses with the Symantec AntiVirus Corporate Edition version 11, Virus Definition File Dated: 7/31/11 rev. 3 and, according to the program, is free of viruses. In addition, I certify all required privacy redactions have been made.

By: /s/ Carter G. Phillips

Attorney for The Dow Chemical Company

CERTIFICATE OF SERVICE

I hereby certify that on this 6th day of December 2013, a copy of Appellant The Dow Chemical Company’s Opening Brief was filed with the Court through the ECF system, which provides electronic service of the filing to all counsel of record who have registered for ECF notification in this matter.

/s/ Carter G. Phillips_____________

Attorney for The Dow Chemical Company

ADDENDUM (Pertinent Orders)

STATUTORY ADDENDUM

Attachment B

No. 13-3215

UNITED STATES COURT OF APPEALS

FOR THE TENTH CIRCUIT

___________________________________

IN RE URETHANE ANTITRUST LITIGATION

____________________________________

The Dow Chemical Company,

Appellant

_____________________________________

On Appeal from the United States District Court

For the District of Kansas

The Honorable John W. Lungstrum

D.C. No. 04-md-1616-JWL

________________

APPELLANT’S REPLY BRIEF

________________



Charles J. Kalil	Carter G. Phillips
Executive Vice President, General Counsel	Counsel of Record
And Corporate Secretary	Joseph R. Guerra
Duncan A. Stuart	C. Frederick Beckner III
Associate General Counsel	Kathleen Moriarty Mueller
THE DOW CHEMICAL COMPANY	Jeffrey S. Beelaert
2030 Dow Center	SIDLEY AUSTIN LLP
Midland, MI 48674	1501 K Street, NW
Telephone: (989) 636-1000	Washington, D.C. 20005
	Telephone: (202) 736-8000
	Facsimile: (202) 736-8711

Oral Argument Requested

March 7, 2014

TABLE OF CONTENTS

TABLE OF AUTHORITIES ........................................................................... iii

GLOSSARY .................................................................................................... vi

INTRODUCTION ............................................................................................ 1

ARGUMENT .................................................................................................... 4

I. THE DISTRICT COURT ERRED IN CERTIFYING THE CLASS ............ 4

A. Wal-Mart Precludes Class Certification ............................................. 4

1. Dow Was Entitled To Show, In Individualized

Proceedings, That Particular Class Members Suffered No

Injury Or Damages .......................................................................4

2. The District Court Violated Wal-Mart By Allowing The

Class To Proceed On The Basis Of Assumed,

“Extrapolated” Impact And Damages ..........................................7

B. Comcast Precludes Class Certification ............................................ 11

1. McClave Failed To Model Injury Resulting From The

Only The Violation On Which Dow’s Liability Is Based...........11

2. Class Certification Was Improper Because McClave’s

Regression Models Do Not Reliably Measure Impact

And Damages Caused By The Alleged Conspiracy ..................14

3. The Jury Verdict Confirms McClave’s Models Were

Unreliable ...................................................................................18

C. Dow’s Challenges To Class Certification Were Preserved ............... 20

II. PLAINTIFFS FAILED TO ESTABLISH ANTITRUST LIABILITY AS A

MATTER OF LAW ................................................................................ 22

A. The Evidence Does Not Support A Finding That Defendants

Implemented Any Conspiracy .......................................................... 22

B. Plaintiffs Failed As A Matter Of Law To Demonstrate That Lyondell

Participated In The Alleged Conspiracy ........................................... 27

III. DOW IS ENTITLED TO A NEW TRIAL BECAUSE THE DAMAGES

AWARD IS BASED ON SPECULATION, AND THE RESULTING

JUDGMENT VIOLATES THE SEVENTH AMENDMENT .............. 30

CONCLUSION ............................................................................................... 33

CERTIFICATE OF COMPLIANCE .............................................................. 34

CERTIFICATE OF DIGITAL SUBMISSION AND PRIVACY

REDACTIONS ............................................................................................... 35

CERTIFICATE OF SERVICE ........................................................................ 36

TABLE OF AUTHORITIES

Page(s)

CASES

Am. Chiropractic Ass’n v. Trigon Healthcare,

367 F.3d 212 (4th Cir. 2004) ................................................................................28

In re Baby Food Antitrust Litig.,

166 F.3d 112 (3d Cir. 1999) .................................................................................29

Bazemore v. Friday,

478 U.S. 385 (1986) .......................................................................................16, 17

Bickerstaff v. Vassar Coll.,

196 F.3d 436 (2d Cir. 1999) .................................................................................16

Blades v. Monsanto Co.,

400 F.3d 562 (8th Cir. 2005) ..................................................................................4

Broussard v. Meineke Discount Muffler Shops, Inc.,

155 F.3d 331 (4th Cir. 1998) ......................................................................7, 21, 22

Comcast Corp. v. Behrend,

133 S. Ct. 1426 (2013) ...............................................................................2, 13, 14

Cordes & Co. Fin. Servs., Inc. v. A.G. Edwards & Sons, Inc.,

502 F.3d 91 (2d Cir. 2007) .....................................................................................4

Coward v. ADT Sec. Sys.,

140 F.3d 271 (D.C. Cir. 1988) .............................................................................16

DG ex rel. Stricklin v. Devaughn,

594 F.3d 1188 (10th Cir. 2010) ............................................................................20

Dimick v. Schiedt,

293 U.S. 474 (1935) .............................................................................................33

In re Flat Glass Antitrust Litig.,

385 F.3d 350 (3d Cir. 2004) .................................................................................29

Forehand v. Fla. State Hosp. at Chattahoochee,

89 F.3d 1562 (11th Cir. 1996) ..............................................................................22

iii

Gen. Tel. Co. of Sw. v. Falcon,

457 U.S. 147 (1982) .............................................................................................20

Haslund v. Simon Prop. Grp., Inc.,

378 F.3d 653 (7th Cir. 2004) ................................................................................31

In re Hydrogen Peroxide Antitrust Litig.,

552 F.3d 305 (3d Cir. 2008) ...................................................................................4

In re Live Concert Antitrust Litig.,

863 F. Supp. 2d 966 (C.D. Cal. 2012) .................................................................17

Martinez v. Wyo. Dep’t of Family Servs.,

218 F.3d 1133 (10th Cir. 2000) ............................................................................16

McLaughlin v. Am. Tobacco Co.,

522 F.3d 215 (2d Cir. 2008), abrogated on other grounds by Bridge v.

Phx. Bond & Indem. Co., 553 U.S. 639 (2008) .....................................................6

Mitchael v. Intracorp, Inc.,

179 F.3d 847 (10th Cir. 1999) ..............................................................................25

Ortiz v. Jordan,

131 S. Ct. 884 (2011) ...........................................................................................24

Pelt v. Utah,

539 F.3d 1271 (10th Cir. 2008) ............................................................................32

Phillips Petroleum Co. v. Shutts,

472 U.S. 797 (1985) .............................................................................................33

In re Publ’n Paper Antitrust Litig.,

690 F.3d 51 (2d Cir. 2012), cert. denied sub nom. 133 S. Ct. 940 (2013)............29

In re Rail Freight Fuel Surcharge Antitrust Litig.,

725 F.3d 244 (D.C. Cir. 2013) .....................................................................4, 6, 19

In re Scrap Metal Antitrust Litig.,

527 F.3d 517 (6th Cir. 2008) ..................................................................................4

United States v. Suntar Roofing, Inc.,

897 F.2d 469 (10th Cir. 1990) ..............................................................................30

Wal-Mart Stores, Inc. v. Dukes,

131 S. Ct. 2541 (2011) .............................................................................2, 5, 6, 22

Wallace B. Roderick Rovocable Living Trust v. XTO Energy, Inc.,

725 F.3d 1213 (10th Cir. 2013) ........................................................................4, 21

Windham v. Am. Brands, Inc.,

565 F.2d 59 (4th Cir. 1977) ................................................................................4, 5

STATUTE

15 U.S.C. §15(a) ......................................................................................................22

OTHER AUTHORITIES

Paul Johnson, The Economics of Common Impact in Antitrust Class

Certification, 77 Antitrust L.J. 533 (2011) ............................................................9

Peter Kennedy, A Guide to Econometrics (6th ed. 2008) ........................................15

David Leinweber, Stupid Data Miner Tricks: Overfitting the S&P 500

(1995) ...................................................................................................................15

James McClave et al., Statistics for Business and Economics (12th ed. 2014).........8

John Neter et al., Applied Linear Regression Models (3d ed. 1996) .......................15

Robert Pindyck & Daniel Rubinfeld, Econometric Models and Economic

Forecasts (4th ed. 1998) ......................................................................................14

GLOSSARY


AA	Appellant’s appendix


AMIBA	American Independent Business Alliance


ASA	Appellant’s supplemental appendix


Dow.Br.	Appellant’s opening brief


MDI	Methylene diphenyl diisocyanate. It can be combined with a polyether polyol to produce flexible foam, such as mattresses and automobile seats.


TDI	Toluene diisocyanate. It can be combined with a polyether polyol to produce rigid insulation or structural foam, such as household and appliance insulation, adhesives, and sealants.


Pltfs.Br.	Class plaintiffs’ response brief


Polyether	Chemicals that can be combined with TDI or MDI to make a


Polyols	polyurethane.


SA	Appellees’ supplemental appendix


Systems	A set of chemicals, most commonly comprised of MDI with a polyether polyol and additives, needed to make a particular polyurethane product.

INTRODUCTION

As Dow demonstrated in its opening brief, class certification—and the $1.06 billion judgment—must be reversed because individualized evidence is needed to determine injury and damages. Plaintiffs respond that the existence of the conspiracy is a common question, and they emphasize evidence of communications between some executives about plans to announce price increases. Common issues do not predominate in antitrust actions, however, unless both conspiracy and injury can be resolved through common evidence. Here, plaintiffs alleged not an agreement as to actual prices, but an agreement to issue price-increase announcements. Those announcements did not raise actual prices, let alone cause injury, to all members of the class. It is undisputed that defendants often failed to make the increases “stick” with individual customers, many of whom are large multi-national companies.

To avoid individualized inquiries into how these announcements affected each customer’s prices, plaintiffs hired Dr. McClave, a statistician who attempted to show that virtually all class members were injured and to quantify damages. Plaintiffs say that, with McClave’s testimony and the “common” conspiracy evidence, the “trial was eminently manageable.” Pltfs.Br.3. They urge deference to the verdict and claim decertification would be “unprecedented.” Id. Plaintiffs ignore, however, both the text of Rule 23 and cases holding that a class may be

decertified prior to judgment even after a jury verdict. And the trial was manageable only because the court overrode Dow’s right to contest injury on an individualized basis.

Moreover, McClave purported to prove classwide impact only by using extrapolation techniques that impermissibly assumed that 75% of the class was overcharged. This violated Wal-Mart’s prohibition on “Trial By Formula”—a prohibition not limited to class actions under Rule 23(b)(2), as plaintiffs claim. See Wal-Mart Stores, Inc. v. Dukes, 131 S. Ct. 2541 (2011). McClave’s models also contained the same fatal defect condemned in Comcast Corp. v. Behrend, 133 S. Ct. 1426 (2013). To show otherwise, plaintiffs mischaracterize McClave’s testimony, claiming he testified that the “price-fixing conspiracy” harmed nearly all class members. McClave said the size of the damages predicted by his models showed classwide injury, and those models indisputably assumed more than one antitrust violation—just as in Comcast.

In fact, McClave’s models could not prove injury and damages at all. Plaintiffs simply repeat McClave’s results-driven justification for his benchmark-shopping. As to his variable-shopping, they argue that one (and only one) of the demand variables he chose—TDI exports—is relevant to TDI prices. TDI domestic demand, however, was also relevant to TDI prices, and the omission of

this major variable rendered his models inadmissible. Nor can plaintiffs explain how McClave’s models were reliable when they predicted damages during a period when the jury found none.

Tacitly conceding the deficiencies in McClave’s models, plaintiffs claim other evidence supports the verdict. But plaintiffs’ other expert, Dr. Solow, acknowledged the he relied on McClave in forming his conclusions about impact. And evidence of communication among senior executives about price-increase announcements cannot alter the undisputed economic evidence that prices were not parallel and many customers used their bargaining power to defeat price increases. Overall, prices stayed flat or fell despite rising demand and steady costs.

Because this undisputed economic evidence renders their allegations of an effective conspiracy untenable, plaintiffs claim prices would have fallen further due to “excess capacity.” But at trial, McClave said excess capacity was not an important driver of price. Plaintiffs cannot simultaneously assert that defendants conspired to counteract downward price pressure from excess capacity, then “prove” the impact of that conspiracy based on models that deem excess capacity unimportant to price.

These and other flaws detailed below require reversal of the class certification rulings and entry of judgment for Dow.

ARGUMENT


I.	THE DISTRICT COURT ERRED IN CERTIFYING THE CLASS.


A.	Wal-Mart Precludes Class Certification.


1.	Dow Was Entitled To Show, In Individualized Proceedings, That Particular Class Members Suffered No Injury Or Damages.

Plaintiffs argue that class certification is frequently granted in antitrust cases. Pltfs.Br.37-40. But “[e]very proposed class action must be decided on its own facts.” Wallace B. Roderick Revocable Living Trust v. XTO Energy, Inc., 725 F.3d 1213, 1220 (10th Cir. 2013) (quotations omitted). Individual issues predominate and a class cannot be certified in an antitrust case unless both the conspiracy and injury can be determined with common evidence. See In re Rail Freight Fuel Surcharge Antitrust Litig., 725 F.3d 244, 252 (D.C. Cir. 2013); In re Hydrogen Peroxide Antitrust Litig., 552 F.3d 305, 311-12 (3d Cir. 2008); Cordes & Co. Fin. Servs., Inc. v. A.G. Edwards & Sons, Inc., 502 F.3d 91, 105 (2d Cir. 2007); Blades v. Monsanto Co., 400 F.3d 562, 566 (8th Cir. 2005); Windham v. Am. Brands, Inc., 565 F.2d 59, 65-66 (4th Cir. 1977) (en banc).¹ Wal-Mart holds that, for injury to

____________________

¹ In re Scrap Metal Antitrust Litigation, 527 F.3d 517 (6th Cir. 2008), is not to the contrary. See Pltfs.Br.39, 42. There, the court upheld class certification where there was evidence that the conspiracy uniformly impacted all class members. 527 F.3d at 534. There is no discussion in the opinion that any class members were able to negotiate competitive prices.

be a common question, it must be “capable of classwide resolution” with common evidence, “in one stroke.” 131 S. Ct. at 2551. Here, because prices were set through individual negotiations, whether all class members were injured cannot be resolved with common evidence “in one stroke.”

Plaintiffs say “negotiations affect the quantum of damages, not the fact of injury.” Pltfs.Br.40 (emphases in original). Not surprisingly, they cite no case for this illogical proposition. When a purchaser throws the price-increase announcement in the trash, buys from a supplier outside the alleged “cartel,” or bargains to obtain a price at or below the price it had been paying regardless of the announcement, Dow.Br.6-9, 29-31, it is not injured.

Plaintiffs cite cases where district courts certified classes on the assumption that a conspiracy caused classwide injury by raising the baseline price from which negotiations began. Pltfs.Br.40. But plaintiffs cannot explain how a customer who negotiates away a baseline increase is “‘injured in his business or property.’” Windham, 565 F.2d at 65-66 (quoting 15 U.S.C. §15). Nor do plaintiffs distinguish the appellate decisions holding that a class cannot be certified on the assumption that an increase in the baseline for negotiations caused all class members to pay supracompetive prices. Dow.Br.31-32 & n.5.

Plaintiffs are thus left arguing that McClave provided “compelling proof of damages” and classwide impact, and Dow had no right to individualized proceedings. Pltfs.Br.41-42. Wal-Mart forecloses both arguments.²

Wal-Mart’s prohibition on “Trial by Formula” is not limited to class actions under Rule 23(b)(2). Plaintiffs note that, because individual proceedings were needed to adjudicate Wal-Mart’s defenses to class members’ backpay claims, damages were not “incidental” to equitable relief and thus could not be awarded in a Rule 23(b)(2) class action. Pls.Br.44-45. But individual proceedings were necessary because due process and the Rules Enabling Act protect a defendant’s right to litigate defenses. Wal-Mart, 131 S. Ct. at 2561. That right cannot be defeated because plaintiffs offer a statistical model to show injury and damages for all class members; a defendant is entitled to rebut that showing with individualized proof. Id. That same reasoning applies here, as courts, both before and after Wal-Mart, have recognized. See, e.g., McLaughlin v. Am. Tobacco Co., 522 F.3d 215, 231-32 (2d Cir. 2008) (plaintiffs’ plan to show injury with expert testimony about

____________________

² Plaintiffs contend that Wal-Mart (and Comcast) did not establish any “new” class certification law. Pltfs.Br.37. That assertion was rejected in In re Rail Freight Fuel Surcharge Antitrust Litigation, 725 F.3d at 255, and is contradicted by arguments plaintiffs’ counsel is advancing elsewhere, see Br. of Applleants at 22, 28, Sykes v. Mel S. Harris & Assocs., No. 13-2742(L) (2d Cir. Sept. 25, 2013) (arguing that Wal-Mart and Comcast ended lax approach to class certification taken by many courts).

“the average loss for each plaintiff” without allowing defendants “to challenge the allegations of individual plaintiffs” is “a due process violation” and “offends the Rules Enabling Act”); Broussard v. Meineke Discount Muffler Shops, Inc., 155 F.3d 331, 333-34 (4th Cir. 1998) (reversing $390 million jury verdict where plaintiffs’ expert based his damages calculations on “averages,” and “classwide relief was awarded without ... any necessary connection to the merits of each individual claim”) (quotations omitted); Dow.Br.35 (citing cases). The district court erred in not following that precedent and decertifying the class here.


2.	The District Court Violated Wal-Mart By Allowing The Class To Proceed On The Basis Of Assumed, “Extrapolated” Impact And Damages.

Plaintiffs do not dispute that many customers rejected the price increases in the allegedly collusive announcements, Dow.Br.6-10, or that prices did not typically or uniformly increase after those announcements, Dow.Br.54-56. Quite the contrary, Solow testified that “collusion and competition are not an all-or-nothing thing,” and admitted that defendants sometimes competed for class members’ business. SA2723-24; see also SA2883-85. It is therefore undisputed that class members were not injured on every transaction.

Indeed, even McClave’s gerrymandered models show periods when actual prices were at or below his predicted “but-for prices.” AA1576-78. McClave’s

models also show zero or “negative” overcharges for about 10% of all transactions he modeled.³ Dow.Br.35-36. Yet in the extrapolations plaintiffs used to show injury and damages for 75% of the class, McClave assumed that the average overcharge observed in his models applied to every transaction. Dow.Br.34-40. Not only is this approach forbidden by Wal-Mart (see supra at 4-7), it also is not “sampling” in accordance with “‘Statistics 101,’” as plaintiffs claim. Pltfs.Br.49 (quoting SA3535).

As Dow’s expert Dr. Ugone explained, to be statistically valid, a sample must be “representative.” AA1437-38. That is confirmed by the Reference Guide on Statistics cited in plaintiffs’ brief (at 48), and McClave’s own textbook. James McClave et al., Statistics for Business and Economics 15 (12th ed. 2014) (“to apply inferential statistics,” one “must obtain a representative sample”). McClave’s models, however, are not based on a “simple random sample.” See id.;

____________________

³Additionally, at least 7% of class members whose damages were determined only by McClave’s models had zero or negative damages. Dow.Br.35-36 & n.36. Plaintiffs claim the relevant number of class members with “zero overcharges” is 2%-a number they get by counting class members whose damages were calculated in part by extrapolation and in part by modeling as “modeled class members.” Pltfs.Br.48-49 & n.8. That explanation proves Dow’s point: because McClave’s extrapolations assumed an overcharge on every transaction, they artificially inflated the percentage of “injured” class members. Had McClave not added the positive extrapolated damages to the zero or negative damages estimated by the models, the number of uninjured class members he identified would have been higher.

AA1433. He did not model “all products,” “all producers,” or “all customers large and small.” Pltfs.Br. 49. Indeed, he did not model any Lyondell or systems transactions. Nor did he establish through any other form of analysis that his modeled transactions were representative of Lyondell or systems transactions. Dow.Br.37-40. It was therefore improper for McClave to apply the average overcharges from his models to every Lyondell and systems transaction. See Paul Johnson, The Economics of Common Impact in Antitrust Class Certification, 77 Antitrust L.J. 533, 548-49 n.43 (2011) (where data exist for only some defendants, the data “clearly are not a representative sample, and any conclusion based on such data may not extend to the population”).

Plaintiffs attempt to excuse these violations of basic statistics by saying that Lyondell data “was missing” due to Lyondell’s bankruptcy. Pltfs.Br.50. In fact, McClave had Lyondell data for transactions through 2006. He simply said he could not model it because “Lyondell was not available to respond to data questions.” AA2173. But Lyondell agreed in its settlement to make “reasonable efforts” to help plaintiffs, including with respect to former employees plaintiffs might want to interview. AA0434-35. In any event, any “limitations” McClave perceived in the data do not result from wrongdoing by Lyondell, Dow or the other defendants, and cannot excuse the improper use of extrapolation.

Although Ugone “‘understood’” McClave’s reasoning for not modeling the Lyondell transactions, Plts.Br.50 (quoting SA5553-54), he did not endorse McClave’s extrapolations. Rather, Ugone testified that McClave’s extrapolations were unreliable because “you can’t just take the [modeled] results and apply it to ... all of the systems where there was no modeling at all” or to Lyondell “where there was no testing at all.” AA1437; see also AA1427. Plaintiffs are therefore wrong to say that McClave’s extrapolations “rest[] upon a reliable foundation” and Dow did not marshal any “precedent” or “expert opinion to the contrary.” Pltfs.Br.48 (internal quotation marks omitted).

Plaintiffs try to defend McClave’s systems extrapolations by citing the district court’s reference to unspecified “internal documents” that allegedly show that defendants viewed price increases for basic chemicals as “‘helping them increase systems prices.’” Pltfs.Br.50 (quoting AA0412). Even if true, that would not support the conclusion that systems prices always were elevated by 74% of the average overcharge calculated by McClave’s MDI model, as McClave simplistically assumed.

Plaintiffs ignore that Dow’s systems prices were individually negotiated by sales representatives based on the system’s value to the customer, so different systems containing the same MDI and polyols were sold at different prices.

AA1241-46; AA2020-21. Plaintiffs also ignore that systems tended to be “higher margin products” sold under “longer term contracts” so manufacturers “tended not to move the price up and down in the way [they] would with a commodity.” SA5667. Consequently, there was “more stability in systems pricing,” and a manufacturer “did not necessarily increase its systems prices even when it increased its MDI pricing.” SA5667-68. Plaintiffs’ own charts show that for a majority of the MDI and polyols price-increase announcements, at least one or two defendants did not announce a corresponding price increase for systems. Dow.Br.38.

In short, certification of a class based on McClave’s extrapolations was improper.


B.	Comcast Precludes Class Certification.


1.	McClave Failed To Model Injury Resulting From The Only Violation On Which Dow’s Liability Is Based.

McClave’s models here have the same flaw that precluded class certification in Comcast. Dow.Br.41-44. Plaintiffs’ attempts to distinguish Comcast are unavailing.

It is undisputed that (1) McClave’s models assumed defendants engaged in a conspiracy to fix prices and to allocate customers and markets, and (2) plaintiffs did not prove the latter violation. Dow.Br.41-42. McClave thus did not model

injury and damages resulting from the only antitrust violation on which Dow’s liability is based. Plaintiffs claim that McClave testified at trial that “Dow’s price-fixing conspiracy ‘impacted nearly every class member’” because prices exceeded those predicted by his models. Pltfs.Br.53 (quoting AA0530 (emphasis added)). But McClave testified only that, “given the persistence and size” of the damages “estimate[ed]” by his models, he concluded that “nearly all class members had been impacted or overcharged.” AA0940. The “persistence and size” of damages estimated by models that assume both a price-fixing agreement and an allocation agreement do not estimate injury from a price-fixing agreement alone.

Alternatively, plaintiffs contend that Comcast is distinguishable because they “abandoned” their allocation theory before trial. Specifically, plaintiffs contend that McClave proffered a “but for” model that determined the difference between prices that would have existed in a “competitive” marketplace and observed prices. Pltfs.Br.54. Plaintiffs do not dispute that, if there had been an allocation conspiracy, it would have affected the “difference” McClave calculated. They say this is not a concern because no one contends there was an allocation conspiracy. Pltfs.Br.55-56.

But the Comcast defendants likewise did not agree that plaintiffs’ additional theories were valid; they disputed plaintiffs’ antitrust allegations. See Behrend v.

Comcast Corp., No. 2:03-cv-06604-JP (E.D. Pa.) (Dkt. Entry 138) (answer). Rather, what was dispositive in Comcast was that McClave himself developed models that “assumed the validity of all four theories of antitrust impact initially advanced by [plaintiffs],” but those models could not “attribute damages to any one particular theory of anticompetitive impact.” 133 S. Ct. at 1434. The same is true here.

Plaintiffs are also wrong in suggesting that, because they eventually abandoned their allocation theory, McClave’s damages calculations could only have measured the effect of the price-fixing conspiracy. Pltfs.Br.56 n.10. Under Comcast, the relevant issue is what the models were designed to measure. Plaintiffs’ subsequent abandonment of their allocation conspiracy claims cannot alter that McClave assumed an allocation conspiracy and sought to measure its effects.

Moreover, plaintiffs’ suggestion that they abandoned their allocation conspiracy because there was no evidence to support it, Pltfs.Br.56 n.10, is a convenient, post hoc assertion that contradicts positions they advanced below. Plaintiffs not only alleged an allocation conspiracy, see AA0368, AA0374, AA0378-80; they also served interrogatory responses before McClave filed his report claiming to have substantial supporting evidence, see ASA140, ASA148,

ASA158-59 (describing alleged meetings and discussions of “customer allocation agreement”).

Having alleged the existence of an allocation conspiracy, claimed there was evidence to support that conspiracy, and submitted models premised on its existence, plaintiffs were obligated to proffer a revised “‘damages case … consistent with [their] liability case.’” Comcast, 133 S. Ct. at 1433. Their failure to do so requires decertification of the class.


2.	Class Certification Was Improper Because McClave’s Regression Models Do Not Reliably Measure Impact And Damages Caused By The Alleged Conspiracy.

McClave’s models do not reliably measure impact and damages because he engaged in variable-shopping: he chose variables not because they were likely, according to sound economic principles, to influence price in a competitive market, but because they appeared to correlate with actual prices in the period he was examining. Dow.Br.45-49. Plaintiffs do not claim otherwise. Instead, seeking to make a virtue of vice, plaintiffs tout that McClave picked variables to achieve a high R-squared. Pltfs.Br.57. But that is Dow’s point. A high R-squared only means the statistician developed a “regression line” that is a reasonably good fit for the data being modeled. Robert Pindyck & Daniel Rubinfeld, Econometric Models and Economic Forecasts 73 (4th ed. 1998). It does not establish that the model

includes economically relevant variables or produces reliable predictions. See John Neter et al., Applied Linear Regression Models 82-83 (3d ed. 1996) (discussing “common misunderstanding” that high R-squared “indicates that useful predictions can be made”). To the contrary, adopting regression variables based on whether they correlate with prices (as McClave did) “runs the real danger of finding, through perseverance, an equation that fits the data well but is incorrect because it captures accidental features of the particular data set at hand (called ‘capitalizing on chance’) rather than the true underlying relationship.” Peter Kennedy, A Guide to Econometrics 89 (6th ed. 2008). It is the kind of spurious approach that enabled statisticians to link S&P 500 returns to butter production and sheep population in Bangladesh in a model with a high R-squared. David Leinweber, Stupid Data Miner Tricks: Overfitting the S&P 500 (1995). Such manipulated results are “utterly useless” for predicting “anything outside the fitted period.” Id. at 5.

Plaintiffs offer an after-the-fact, economic justification for one (but only one) instance of McClave’s variable shopping—namely, McClave’s decision to use TDI exports as the only demand variable. Plaintiffs argue that TDI exports are “an important driver for TDI demand,” because TDI exports “account[] for 40% of domestic production.” Pltfs.Br.57. The problem, however, is not the inclusion of

TDI exports as a demand variable, but the economically irrational exclusion of domestic demand, which accounts for the remaining 60% of domestic TDI production. Indeed, McClave acknowledged that domestic demand was a critically important variable because he used only domestic demand variables in his models for MDI and polyols. AA1394-99, AA1597, AA2267. Plaintiffs provide no principled reason, much less any learned treatises, for believing that domestic TDI prices can be accurately predicted by models that exclude domestic demand.⁴

This is not a quibble that merely implicates the “‘probativeness’” of the analysis. Pltfs.Br.57 (quoting Bazemore v. Friday, 478 U.S. 385, 400 (1986)). A regression analysis can be so “incomplete to be inadmissible as irrelevant,” Bazemore, 478 U.S. at 400 n.10, and courts have excluded flawed regression analyses that omit a “major variable” relevant to the case, see, e.g., Bickerstaff v. Vassar Coll., 196 F.3d 435, 449 (2d Cir. 1999); Coward v. ADT Sec. Sys., 140 F.3d 271, 274 (D.C. Cir. 1988); see also Martinez v. Wyo. Dep’t of Family Servs., 218 F.3d 1133, 1138-39 (10th Cir. 2000) (“[S]tatistical evidence … may be so flawed as to render it insufficient to raise a jury question.”). Inclusion of all major

^{____________________________________}

⁴ Plaintiffs also concede that McClave used 12-month moving averages for input costs in the MDI model, but 6-month moving averages in the TDI and polyols models based solely on “fit.” Pltfs.Br.57 n.11. They claim this was “sensible,” but, like McClave, provide no economic justification for why the effect of cost on prices lingers for a year in one instance but for only 6 months in others.

variables is “particularly important in the context of antitrust litigation,” where “‘[d]amage estimates ... hinge on careful statistical analysis, reasonable assumptions, [and] reliable data,’” and errors in these areas can “‘provide faulty conclusions as to the existence or the amount of damages.’” In re Live Concert Antitrust Litig., 863 F. Supp. 2d 966, 973-74 (C.D. Cal. 2012) (quoting 2A Areeda & Hovenkamp, Antitrust Law ¶399c, at 447 (3d ed. 2006)).

Here, plaintiffs do not dispute the commonsense economic proposition that domestic demand is a major variable affecting domestic prices. See AA2323-24. And when domestic demand is included in McClave’s TDI model, it predicts negative damages of more than $49 million. Dow.Br.48-49. This shows that his flawed method for selecting variables resulted in models that are “so incomplete as to be inadmissible as irrelevant.” Bazemore, 478 U.S. at 400 n.10.

Nor have plaintiffs provided any justification for McClave’s benchmark-shopping. Although plaintiffs contended that the conspiracy continued through 2004, McClave moved 2004 to the “competitive benchmark” period in his models. Dow.Br.49. Plaintiffs say this was appropriate because “McClave studied the data and found that 2004 prices were more consistent with competition than collusion.” Pltfs.Br.58. But the entirety of McClave’s “analysis” was that including 2004 in

the “competitive benchmark” generated favorable results.⁵ Treating 2004 as a “conspiracy” year resulted in negative damages for MDI, polyols, and systems, and improbably high damages for TDI. Dow.Br.49. McClave’s reasons for including 2004 in the benchmark period were thus circular and results-oriented. That the model can show plausible damages only if 2004 is moved to the benchmark period is not a justification for doing so.


3.	The Jury Verdict Confirms McClave’s Models Were Unreliable.

The jury verdict confirms that McClave’s models do not reliably distinguish competitive prices from collusive prices. He testified that his models accounted for competitive factors, so “[s]omething other than competition” caused prices to be higher between 1999 and 2003 than the “but-for prices” his models predicted. AA1101. The jury, however, found no overcharges before November 24, 2000. Plaintiffs provide no reason for believing that McClave’s models were reliable after that date when they were not but before that date.

___________________

⁵ Contrary to their suggestions, Pltfs.Br.58, plaintiffs moved 2004 only after McClave determined he could not produce a model showing plausible damages with 2004 in the conspiracy period, see Dow.Br.49. McClave conceded he moved 2004 solely because of this “analysis” testing, AA0953-54, AA2215-16, AA2219, AA2231, not because he determined there was no evidence the conspiracy continued through 2004.

Plaintiffs note that the jury could have found the conspiracy was narrower or shorter than they claimed. Pltfs.Br.59. If so, that would prove Dow’s point: If the conspiracy were narrower or shorter, McClave’s models are unreliable because they showed damages over the broader or longer period. In any event, the models were identical in both periods, and there is no rational basis upon which the jury could reject them in one period but accept them in the other. Dow.Br.52. A jury may not “split-the-baby” without an evidentiary basis for doing so. Id. (citing cases).

Plaintiffs speculate that the models properly could have shown overcharges prior to November 24, 2000, because the “conspiracy was operational yet not proved.” Pltfs.Br.61. But if the “conspiracy was operational” and inflated prices as McClave claimed, then the conspiracy would have been “proved.” The jury, however, rejected the notion that there was an “operational” conspiracy; it found no overcharges prior to November 24, 2000. AA513-15. Thus, McClave’s models “detect[] injury where none could exist” and cannot support class certification. Rail Freight Fuel Surcharge Antitrust Litig., 725 F.3d at 252.

C. Dow’s Challenges To Class Certification Were Preserved.

Plaintiffs ask this Court to ignore the defects in the class certification ruling on the grounds that Dow’s challenges were waived or untimely. These arguments are groundless.

Defendants initially opposed class certification on the grounds that impact and damages are not common questions. ASA079-83, ASA086-109, ASA113-34. Dow raised the same objections in the motion to decertify the class, this time focusing on intervening Supreme Court precedent (Wal-Mart) and plaintiffs’ new expert (McClave). Dow.Br.12.

That decertification motion, moreover, was not properly denied as untimely simply because it was filed “on the eve of trial.” Pltfs.Br.27. Under Rule 23(c)(1)(C), an order granting class certification may be amended any time “before final judgment,” so courts are “free to modify” the certification order in “light of subsequent developments in the litigation.” Gen. Tel. Co. of Sw. v. Falcon, 457 U.S. 147, 160 (1982); see also, e.g., DG ex rel. Stricklin v. Devaughn, 594 F.3d 1188, 1201 (10th Cir. 2010) (if the court subsequently finds that the requirements of Rule 23 are not met, it should “amend its certification order to reflect its findings or decertify the class altogether prior to final judgment”). Indeed, the district court amended the class certification order post-trial to exclude from the

class those who purchased polyurethane products in 2004, a year plaintiffs initially alleged to be part of the conspiracy but then dropped after McClave moved it from the conspiracy period to the benchmark period in his models. Dow.Br.20.

Nor was Dow’s decertification motion properly denied because its timing prejudiced plaintiffs. Whatever difficulties plaintiffs might have faced in “‘assert[ing] individual claims at [that] time,’” Pltfs.Br.44 (quoting AA0523-24), were attributable to plaintiffs’ litigation strategy. Plaintiffs successfully opposed defendants’ motion for interlocutory review to resolve the propriety of class certification before the parties conducted merits discovery and prepared for trial. Dow.Br.11. Plaintiffs then discarded the expert they used to obtain class certification and retained McClave to attempt to show classwide injury and damages at trial. Moreover, the Supreme Court issued important decisions interpreting Rule 23 after McClave issued his report (Wal-Mart) and after trial but before entry of judgment (Comcast). This Court has recognized that Wal-Mart and Comcast must be considered and may require the reversal of a class certification order previously entered. See XTO Energy, 725 F.3d at 1221. Any suggestion that arguments based on those decisions were untimely is unfounded.

Plaintiffs claim that class decertification “on the posture here would be unprecedented,” Pltfs.Br.3, 28, but this, too, is untrue, see, e.g., Broussard, 155

F.3d at 334 (decertifying class after jury trial); Forehand v. Fla. State Hosp. at Chattahoochee, 89 F.3d 1562, 1566 (11th Cir. 1996) (same). A class should not be certified and a jury verdict cannot be affirmed where plaintiffs are permitted to portray the class as a “unified group that suffered a uniform, collective injury” based on the testimony of an expert who calculated damages using “averages” extrapolated from a “sample” of class members the defendant could not cross-examine. Broussard, 155 F.3d at 343, 345. Wal-Mart confirms that such a “Trial by Formula” violates the Rules Enabling Act by allowing the procedural device of a class action to “‘abridge, enlarge or modify ... substantive right[s].’” 131 S. Ct. at 2561 (quoting 28 U.S.C. §2702(b)); see also Broussard, 155 F.3d at 345 (same). As Dow showed in its opening brief (at 27-53) and above, that is precisely what occurred here.⁶


II.	PLAINTIFFS FAILED TO ESTABLISH ANTITRUST LIABILITY AS A MATTER OF LAW.


A.	The Evidence Does Not Support A Finding That Defendants Implemented Any Conspiracy.

The evidence before the jury was insufficient to support a finding that any conspiracy was implemented and caused over $400 million in overcharges.

^{____________________________________}

⁶ Faithful application of Wal-Mart and Comcast will not preclude civil antitrust claims by private parties, as amicus AMIBA claims. The Clayton Act provides both for treble damages and attorneys’ fees, 15 U.S.C. §15(a), and many plaintiffs opted out of the class to pursue individual claims.

Dow.Br.54-59. Demand was relatively constant and costs were rising. In these circumstances, a functioning conspiracy should have increased prices. Yet, actual prices held steady or declined, and prices of the various defendants moved in different directions and fluctuated month-to-month. Dow Br. 53-59.

Plaintiffs do not dispute this, and reviewing courts owe no deference to an economically incoherent jury finding of harm. Recognizing this, plaintiffs claim that “price stability” is “consistent with” an effective conspiracy because there was “excess capacity,” which exerted “downward pressure on price, motivating formation of the cartel to arrest that decline.” Pltfs.Br.36. But under plaintiffs’ theory, only an increase in excess capacity during the class period could have caused prices to fall below where they were when the “cartel” allegedly began. Solow did not say excess capacity increased in the class period. He testified only that “there was excess capacity for these three products throughout much of this time period.” AA0811; see also AA0809.

More fundamentally, Solow’s assertion that excess capacity would have driven prices down is at odds with one of McClave’s key modeling decisions. McClave did not include “capacity utilization” in his models (i.e., how much capacity is being used and thus “how tight capacity is”), because he said it did “not … prove to be an important ... price driver.” SA3685-86. Instead, McClave

considered only total capacity without regard to how much of that capacity was “excess.” See SA3485-87. Plaintiffs cannot have it both ways. They cannot claim that defendants conspired to counteract downward price pressure from excess capacity, then “prove” the impact of that conspiracy based on McClave’s models and methodology that rejected excess capacity as an important driver of price.

Plaintiffs’ other efforts to divert attention from the undisputed and dispositive economic facts are equally unavailing. They claim that a jury could reasonably infer classwide injury from the “scope and nature of the conspiracy”; the “top-down” involvement of senior executives; the oligopolistic nature of the polyurethanes industry, which has high barriers to entry and “homogenous commodity products without close substitutes”; and a few documents and statements indicating that some believed price-increase announcements “worked” for some products or customers at some times. Pltfs.Br.32-34. Many of these claims are wrong; all of them are irrelevant.

In claiming that polyurethane products are “‘homogenous commodity products without close substitutes,’” Pltfs.Br. (quoting SA9), plaintiffs cite the district court’s summary judgment opinion that is irrelevant. See Ortiz v. Jordan, 131 S. Ct. 884, 889 (2011) (trial record “supersedes the record existing at the time

of the summary judgment motion.”). Certainly, systems are not commodity products, but are custom-designed for a particular customer, Dow.Br.7—as the district court acknowledged in connection with class certification. See AA0411. And courts have rejected the notion that a market structure conducive to oligopolistic pricing is sufficient to permit an inference of anticompetitive pricing. See, e.g., Mitchael v. Intracorp, Inc., 179 F.3d 847, 859 (10th Cir. 1999).

More fundamentally, plaintiffs’ purported inferences cannot obscure the economic incoherence of an alleged conspiracy in which prices held steady while demand was relatively constant and costs were rising in a market where price-increase announcements frequently did not lead to increases in actual prices. See Dow.Br. 8-9, 16-18, 29-31, 54-58. Plaintiffs say Dow’s expert Professor Elzinga “conceded that actual price increases followed the conspiratorial announcements more than half the time.” Pltfs.Br.34. But this one statement is cherry-picked. Elzinga admitted that prices sometimes increased in the month or two after an announced price increase was supposed to take effect, but he also noted that there was frequently no price increase or, if there was, prices dropped in the following months. E.g., SA5258, SA5298-5300, SA5053-64. In the end, Elzinga concluded that “the actual prices paid in the marketplace did not follow the price-increase announcements in a lockstep pattern, to use Professor Solow’s words, nor in most

cases did they stick, to use Professor Solow’s terms.” SA5075 (emphasis added). This shows that many class members were not injured by the announcements.

Finally, plaintiffs cite Solow’s opinion that nearly all class members were injured. Pltfs.Br.34. Solow admitted, however, that there were significant instances of “cheating” and periods when the “cartel” broke down and defendants competed. SA2723-24; cf. SA5122-24 (Elzinga testifying that he found over 1000 internal documents from defendants discussing how “one of their competitors was poaching” and they risked losing business); AA1540-42 (competition for Foamex); AA1560 (competition for Firestone). Class members who were beneficiaries of periods of “hard fought competition,” SA788, were not injured.

Moreover, Solow admitted that he relied on McClave’s analysis in reaching his conclusion about impact. AA0825. He testified that his analysis would be “incomplete” if he “didn’t have Dr. McClave’s work.” SA2791; see also AA0831-33. Solow’s testimony thus provides no independent basis for concluding that any cartel impacted all class members.

Plaintiffs are therefore wrong to claim that, even without McClave’s testimony, there is “[e]xtensive evidence support[ing] the finding that the cartel caused class-wide injury.” Pltfs.Br.32. Without McClave’s flawed models and extrapolations, nothing supports a finding that the alleged conspiracy was

implemented and thus injured all class members who would recover under the judgment.


B.	Plaintiffs Failed As A Matter Of Law To Demonstrate That Lyondell Participated In The Alleged Conspiracy.

At a minimum, a new trial should be granted because there is no evidence that Lyondell participated in any conspiracy. Dow.Br.59-62. Plaintiffs say this argument was waived because Dow did not raise it until its post-trial reply brief. Pltfs.Br.30. Not so. In its post-trial motions, Dow argued that plaintiffs failed to prove a conspiracy as to any defendants—including Lyondell—and that Dow was therefore entitled to judgment as a matter of law. ASA046-50, ASA060-73. Dow specifically pointed to the evidence that Lyondell did not conspire to fix prices. ASA049-50, ASA062-63, ASA066.

Nor have plaintiffs identified evidence sufficient to support a finding that Lyondell agreed to coordinate price-increase announcements. The only relevant testimony is that of Lyondell’s Dineen, who testified that he did not agree to anything at the Swan Restaurant, and that the meeting there played no role in Lyondell’s price-increase announcements. AA0757-61, AA0771-72. Plaintiffs assert that the jury could discount Dineen’s undisputed testimony, Pltfs.Br.31, but the jury necessarily rejected plaintiffs’ view that the “lockstep” price-increase announcements after the Swan dinner caused any injury. The meeting occurred

during the pre-November 2000 period when the jury found no overcharges. AA0513-15; AA0767-70.

Plaintiffs also say that a conspiracy can be inferred because three, two-minute phone calls between Levi (Dow) and Portella (Lyondell) in 2002 and early 2003 preceded some price-increase announcements. Pltfs.Br.30. Massive antitrust liability cannot be imposed on the basis of such “evidence.” The price-increase announcements cannot establish liability—such announcements were commonplace during periods where plaintiffs themselves acknowledge the marketplace was “competitive.” AA0451; AA1375-78, AA1386-88; AA2143-45; see also, e.g., Am. Chiropractic Ass’n v. Trigon Healthcare, 367 F.3d 212, 227 (4th Cir. 2004) (“mere contacts and communications … among antitrust defendants is insufficient evidence [of] an anticompetitive conspiracy”) (quotations omitted). There is no evidence that the calls involved prices, SA489-97, SA3150-52, SA3232-34, SA3277-80, and undisputed evidence of legitimate reasons for the calls, AA0764, A0899-901 (supply agreement between Dow and Lyondell). Moreover, plaintiffs have no theory how these brief calls could have affected price-increase announcements Lyondell made before 2002.

Finally, the inference plaintiffs seek to draw presupposes that Lyondell “mirrored” the “parallel” prices of Dow and the other manufacturers, Pltfs.Br.30—

but the evidence demonstrates the opposite. See In re Baby Food Antitrust Litig., 166 F.3d 112, 125 (3d Cir. 1999) (“[T]here must be evidence that the exchanges of information had an impact on pricing decisions.”). The undisputed evidence is that: (1) Lyondell’s prices did not follow Dow’s prices, AA0520, AA1528, AA1533, AA1764, AA1766; (2) even after Lyondell announced a price increase, its prices frequently stayed flat or fell, AA1533, AA1766; (3) Lyondell generally remained the lowest-priced manufacturer, AA0520, AA1528-33, AA1762-66; and (4) rather than being parallel, prices of all of the defendants moved in different directions and fluctuated substantially from month-to-month, AA0520, AA1528-33, AA1762-66.⁷

Plaintiffs also say the verdict can stand even if the evidence is insufficient as to Lyondell. Pltfs.Br.31. But a finding that Lyondell was not in the conspiracy calls into question whether an operational conspiracy actually existed. In any

____________________

⁷ Plaintiffs’ cases are distinguishable. See Pls.Br.30. In In re Publication Paper Antitrust Litigation, 690 F.3d 51, 57-59, 64 (2d Cir. 2012), there was direct testimony that defendants had an express “agreement” to coordinate prices and coordinated pricing consistent with that agreement. In In re Flat Glass Antitrust Litigation, 385 F.3d 350, 363, 368-69 (3d Cir. 2004), there was “undoubted[] evidence” that numerous manufacturers had an express agreement to fix prices, that one conspirator had acknowledged an “across-the-board” agreement to fix prices, the defendant had been faxed non-public information about planned price increases by a member of the conspiracy, the defendant raised prices that were identical to the planned, non-public price increase, and the other manufacturers “followed with identical price increases.”

event, even if it may be true in a criminal case where proof of injury is not required, cf. United States v. Suntar Roofing, Inc., 897 F.2d 469, 476 (10th Cir. 1990), it is not true in a civil lawsuit by private plaintiffs seeking damages. If Lyondell did not conspire, Dow cannot be held jointly and severally liable for overcharges to Lyondell’s customers—as the jury instructions provided. ASA027. Because the damages attributable only to Lyondell cannot be determined, Dow is at least entitled to a new trial. See Dow.Br.62.


III.	DOW IS ENTITLED TO A NEW TRIAL BECAUSE THE DAMAGES AWARD IS BASED ON SPECULATION, AND THE RESULTING JUDGMENT VIOLATES THE SEVENTH AMENDMENT.

There was no evidentiary basis for the jury’s damages award, which deviated from McClave’s damages calculation by approximately 20%. Dow.Br.63-64. The verdict must be set aside, and further cannot be reduced to a judgment consistent with the Seventh Amendment, because there is no way of knowing which class members the jury thought were injured. Id. at 64-67. Plaintiffs fail to show otherwise.

1. Although the jury never received McClave’s calculations or any breakdown of damages (except for the named plaintiffs), plaintiffs note that McClave testified that, if the jury accepted Ugone’s criticisms of his variable selection, the overall, average overcharge would be reduced to approximately 10%

from the 13.4% McClave calculated—“that is, overall overcharges might be reduced approximately 25%.” Pltfs.Br.62. McClave’s testimony, however, relates only to the effect of changes in overcharges over the entire 1999-2003 class period, not to the post-November 2000 period in which the jury found liability. AA1586-87. McClave’s damages estimates are not distributed equally over time. After November 24, 2000, the overcharge was only 9.8%—lower than McClave said would result if Ugone’s criticisms were accepted for the entire class period. See AA1586-88; SA3556-71. McClave’s statement thus did not provide a basis for re-calculating damages in the later period. And because McClave’s models are non-linear, the jury had no rational way to reduce his damages estimate if it found the conspiracy was of a shorter duration, as the district court speculated. AA0537.⁸

Thus, the situation here is nothing like the cases relied upon by plaintiffs. See Pltfs.Br.62-63. Whatever latitude exists for a jury to split-the-baby, it may not do so where there is “no basis in the evidence” for the compromise. Haslund v. Simon Prop. Grp., Inc., 378 F.3d 653, 658 (7th Cir. 2004). Here, the jury necessarily had to speculate to reduce McClave’s damages.

____________________

⁸ Likewise, there was no way for the jury to have adjusted McClave’s damages if it found Lyondell did not participate in the conspiracy, cf. AA0537, because McClave never provided the damages attributable to Lyondell, AA1563-89.

2. Plaintiffs assert that there is no Seventh Amendment problem because aggregated damage calculations are common in class actions; Dow did not request a customer-by-customer allocation of damages; and Dow has no interest in the distribution of the award among the class members. Pltfs.Br.63-65.

Dow is not arguing, however, that the jury was required to award individualized damages for every class member. Cf. Pltfs.Br.64-65. Had the jury returned a verdict in the amount McClave requested, there would be no Seventh Amendment problem. The problem arises here because there is no way to tell whether the jury reduced the damages because it concluded that certain class members (e.g., Lyondell customers) suffered no injury. Dow.Br.64-65. Plaintiffs cite no case rejecting a Seventh Amendment challenge to a judgment under these circumstances. Nor have they explained how the district court could mandate a pro rata reduction without substituting its judgment for the jury’s.

Dow, moreover, does have an “interest” in the allocation of the damages award. While class members are ordinarily bound by a properly entertained class action, this protection does not apply against class members who “were not accorded due process of the law.” Pelt v. Utah, 539 F.3d 1271, 1284 (10th Cir. 2008). As Dow showed, the courts have thus recognized a defendant may assert

arguments advancing the “rights” of class members to ensure the “‘entire plaintiff class is bound.’” Dow.Br.66 (citing cases).⁹

CONCLUSION

The class should be decertified and judgment entered for Dow or in the alternative, the class should be decertified and Dow should be granted a new trial.

Respectfully submitted,



	/s/ Carter G. Phillips
Charles J. Kalil	Carter G. Phillips
Executive Vice President, General Counsel	Counsel of Record
And Corporate Secretary	Joseph R. Guerra
Duncan A. Stuart	C. Frederick Beckner III
Associate General Counsel	Kathleen Moriarty Mueller
THE DOW CHEMICAL COMPANY	Jeffrey S. Beelaert
2030 Dow Center	SIDLEY AUSTIN LLP
Midland, MI 48674	1501 K Street, NW
Telephone: (989) 636-1000	Washington, D.C. 20005
	Telephone: (202) 736-8000
	Facsimile: (202) 736-8711

March 7, 2014

__________________

⁹ Plaintiffs’ cases are irrelevant. See Pltfs.Br.64. None involved the Seventh Amendment, which affords “both parties” the right to have liability and damages determined by a jury. E.g., Dimick v. Schiedt, 293 U.S. 474, 486 (1935) (emphasis added). And none involved the situation where a class defendant seeks to “vindicate its own interests,” Phillips Petroleum Co. v. Shutts, 472 U.S. 797, 804-

05 (1985), in ensuring that damages are properly allocated so that all class members are bound.

CERTIFICATE OF COMPLIANCE

Section 1. Word count

As required by Fed. R. App. P. 32(a)(7)(c), I certify that this brief is proportionally spaced and contains 6,990 words.

Complete one of the following:

X I relied on my word processor to obtain the count and it is Word version 2007:

I counted all characters including citations and numerals.

I certify that the information on this form is true and correct to the best of my knowledge and belief formed after a reasonably inquiry.

By: /s/Carter G. Phillips

Attorney for The Dow Chemical Company

CERTIFICATE OF DIGITAL SUBMISSION AND PRIVACY

REDACTIONS

I hereby certify that a copy of the foregoing APPELLANT’S REPLY BRIEF, as submitted in Digital Form via the court’s ECF system, is an exact copy of the written document filed with the Clerk and has been scanned for viruses with McAffee AntiVirus software version 8.0.0.3709, virus definition file dated 3/7/2014 and, according to the program, is free of viruses. In addition, I certify all required privacy redactions have been made.

By: /s/ Carter G. Phillips

Attorney for The Dow Chemical Company

CERTIFICATE OF SERVICE

I hereby certify that on this 7th day of March 2014, a copy of APPELLANT’S REPLY BRIEF was filed with the Court through the ECF system, which provides electronic service of the filing to all counsel of record who have registered for ECF notification in this matter.

By: /s/ Carter G. Phillips

Attorney for The Dow Chemical Company

Attachment C

No. 13-3215

[ORAL ARGUMENT NOT YET SCHEDULED]

______________________________________________

IN THE UNITED STATES COURT OF APPEALS

FOR THE TENTH CIRCUIT

______________________________

IN RE: URETHANE ANTITRUST LITIGATION

______________

On Petition for Appeal from the United States District Court for the District of Kansas

______________________________

BRIEF FOR THE CHAMBER OF COMMERCE OF THE UNITED STATES AS AMICUS CURIAE IN SUPPORT OF PETITIONERS URGING REVERSAL

______________________________________________



Kathryn Comerford Todd	Gene C. Schaerr
Tyler R. Green	Robert F. Ruyak
NATIONAL CHAMBER	William A. Roach, Jr.
LITIGATION CENTER, INC.	WINSTON & STRAWN LLP
1615 H Street, N.W.	1700 K Street, N.W.
Washington, D.C. 20062	Washington, D.C. 20006
(202) 463-5337	(202) 282-5000

Jeffrey L. Kessler
George E. Mastoris
WINSTON & STRAWN LLP
200 Park Avenue
New York, New York 10166
(212) 294-6700

Counsel for Amicus Curiae

Chamber of Commerce of the United States of America

CORPORATE DISCLOSURE STATEMENT

The Chamber of Commerce of the United States of America is a nonprofit corporation organized under the laws of the District of Columbia. It has no parent company and has issued no stock.

CERTIFICATE AS TO PARTIES, RULINGS, AND RELATED CASES

Pursuant to this Circuit’s Rule 28(a)(1), amicus curiae certifies:

(A) Parties and Amici. Except for the following, all parties, intervenors, and amici appearing before the district court are listed in the Opening Brief of Petitioners: Amicus curiae Chamber of Commerce of the United States of America.

(B) Rulings Under Review. References to the rulings at issue appear in the Opening Brief of Petitioners.

(C) Related Cases. Amicus curiae is aware of no related cases pending in this Court or any other Court. The class certification order at issue in this case was previously before this Court on a petition to appeal pursuant to Rule 23(f) of the Federal Rules of Civil Procedure. In re Urethane Antitrust Litig., No. 08-602 (10th Cir. Sept. 2, 2008). This Court denied the petition. Id.

-i-

TABLE OF CONTENTS

Page

TABLE OF AUTHORITIES..................................................................................iii

INTRODUCTION AND INTEREST OF AMICUS................................................1

PERTINENT FACTS...............................................................................................2

SUMMARY OF ARGUMENT................................................................................5

ARGUMENT...........................................................................................................7

I. The District Court’s Certification Decisions Violated Three

Core Legal Principles Governing The Use Of Statistical Models

In Class Litigation...............................................................................................7

A. Statistical Models that Fail Fairly to Account for

Relevant Differences Among Individual Class Members

Cannot Satisfy Rule 23(b)(3)’s “Predominance”

Requirement...............................................................................................8

B. Any Statistical Model Purporting To Establish Classwide

Injury Must Show That All Members Of The Class Were

Injured, In The Same Way........................................................................14

C. The Abandonment of a Foundational Element of

Plaintiffs’ Liability Case Requires Decertification..................................16

II. Strict Adherence to Rule 23’s Requirements Is Essential to

Ensure that Class Actions Do Not Impose Unwarranted

Burdens on the National Economy..................................................................21

A. The Ruling Below Will Exacerbate the Problem of

Defendants Being Effectively Forced to Settle Meritless

Claims......................................................................................................21

B. Improper Class Actions Levy a Substantial and

Unjustified Burden on the National Economy.........................................26

CONCLUSION......................................................................................................27

-ii-

TABLE OF AUTHORITIES

Page(s)

CASES

Amchem Prods., Inc. v. Windsor,

521 U.S. 591 (1997)............................................................................................9

Bell Atl. Corp. v. Twombly,

550 U.S. 544 (2007)..........................................................................................23

Blue Chip Stamps v. Manor Drug Stores,

421 U.S. 723 (1975)..........................................................................................26

Carpenter v. Boeing Co.,

456 F.3d 1183 (10th Cir. 2006)...........................................................................8

Cohlmia v. St. John Med. Ctr.,

693 F.3d 1269 (10th Cir. 2012)...........................................................................9

Comcast Corp. v. Behrend,

133 S. Ct. 1426 (2013) (“Comcast”).........................................................passim

Coopers & Lybrand v. Livesay,

437 U.S. 463 (1978)....................................................................................18, 22

DG ex rel. Stricklin v. Devaughn,

594 F.3d 1188 (10th Cir. 2010).........................................................................18

Gen. Tel. Co. of Sw. v. Falcon,

457 U.S. 147 (1982)..........................................................................................18

In re Linerboard Antitrust Litig.,

296 F. Supp. 2d 568 (E.D. Pa. 2003)................................................................23

In re Lorazepam & Clorazepate Antitrust Litig.,

289 F.3d 98 (D.C. Cir. 2002)............................................................................22

In re Rail Freight Fuel Surcharge Antitrust Litigation,

725 F.3d 244 (D.C. Cir. 2013)...................................................................passim

In re Rhone-Poulenc Rorer, Inc.,

51 F.3d 1293 (7th Cir. 1995).............................................................................25

-iii-

In re Zurn Pex Plumbing Prods. Liab. Litig.,

644 F.3d 604 (8th Cir. 2011).............................................................................18

Jacob v. Duane Reade, Inc.,

– F.R.D. –, 2013 WL 4028147 (S.D.N.Y. Aug. 8, 2013)............................13, 16

SEC v. Tambone,

597 F.3d 436 (1st Cir. 2010).............................................................................26

Shumate & Co. v. Nat’l Ass’n of Sec. Dealers, Inc.,

509 F.2d 147 (5th Cir. 1975)...............................................................................9

Spagnola v. Chubb Corp.,

264 F.R.D. 76 (S.D.N.Y. 2010).........................................................................19

Thorogood v. Sears, Roebuck and Co.,

624 F.3d 842 (7th Cir. 2010).............................................................................25

Wal-Mart Stores, Inc. v. Dukes,

131 S. Ct. 2541 (2011) (“Wal-Mart”)........................................................passim

Wallace B. Roderick Revocable Living Trust v. XTO Energy, Inc.,

725 F.3d 1213 (10th Cir. 2013)...........................................................................8

STATUTES

15 U.S.C. § 15(a).....................................................................................................9

Sherman Act, 15 U.S.C. § 1.....................................................................................2

OTHER AUTHORITIES

Fed. R. Civ. P. 23(c)(1)(C)...............................................................................18, 20

http://www.merriam- webster.com/dictionary/

extrapolate?show=0&t=1386951357 (last visited Dec. 13, 2013).....................4

Manual for Complex Litigation § 30 (4th ed. 2004)..............................................23

Phillip E. Areeda & Herbert Hovenkamp, Antitrust Law: An Analysis of

Antitrust Principles and Their Application § 398(c) (2013).............................10

Robert G. Bone & David S. Evans, Class Certification and the Substantive

Merits, 51 Duke L.J. 1251 (2002).....................................................................21

-iv-

Robert H. Mnookin & Robert B. Wilson, Rational Bargaining and Market

Efficiency: Understanding Pennzoil v. Texaco, 75 Va. L. Rev. 29 (1989).........24

Richard A. Nagareda, Class Certification in the Age of Aggregate Proof, 84

N.Y.U. L. Rev. 97 (2009)....................................................................................25

-v-

INTRODUCTION AND INTEREST OF AMICUS¹

The Chamber of Commerce of the United States of America (“the Chamber”) is the world’s largest business federation, representing 300,000 direct members, and indirectly representing more than three million businesses and trade and professional organizations of every size, sector, and geographic region. One of the Chamber’s most important functions is to represent the interests of its members in matters before the courts, Congress, and the Executive Branch. To that end, the Chamber regularly files amicus curiae briefs in cases that present issues of vital concern to the Nation’s business community.

Because class certification is one of those issues, the Chamber has a substantial interest in the questions presented in this case. Those questions are governed by the Supreme Court’s teachings in Wal-Mart Stores, Inc. v. Dukes, 131 S. Ct. 2541 (2011) (“Wal-Mart”) and Comcast Corp. v. Behrend, 133 S. Ct. 1426 (2013) (“Comcast”), but were decided below in a manner that contravenes those teachings. The Chamber participated as an amicus in both Wal-Mart and Comcast to protect the interests of its members, who are frequent targets of class actions and who thus bear the substantial burdens that improper class certification inevitably imposes. For the same reason, the Chamber also participated as an amicus in In re

¹ All parties have consented to the filing of this brief. No counsel for a party authored this brief in whole or in part; no party or its counsel made a monetary contribution intended to fund the preparation or submission of this brief; and no person other than amicus curiae, its members, or its counsel made a monetary contribution intended to fund its preparation or submission.

Rail Freight Fuel Surcharge Antitrust Litigation, 725 F.3d 244, 253 (D.C. Cir. 2013) (“In re Rail Freight”), which applied Wal-Mart’s and Comcast’s teachings and in so doing vacated the district court’s improper decision to certify the class. Given the Chamber’s well-established history of participating in cases that clarify the standards for class certification under Rule 23, it has an evident interest in advocating for the proper application of those standards in this context—a rare antitrust case that went to trial after class certification.

PERTINENT FACTS

The plaintiffs in this case sued The Dow Chemical Company and four other manufacturers alleging that they conspired over a six-year period to fix the prices for urethane chemical products and to allocate customers, both in violation of the Sherman Act, 15 U.S.C. § 1. Plaintiffs sought class certification on the basis that they could show both antitrust impact and damages resulting from these two theories of liability on a classwide basis.

At the certification stage, the defendants introduced unrebutted evidence that prices in the urethane chemical products industry were negotiated individually, and therefore that impact and damages could only be determined on an individualized basis. Despite acknowledging the pertinent market facts, the district court certified the class. Defendants unsuccessfully sought interlocutory review under Rule 23(f).

- 2 -

Faced with joint-and-several liability and potentially billions of dollars in damages, three defendants (BASF, Bayer, and Huntsman) settled.² Dow forged ahead. Plaintiffs retained a new expert, Dr. James McClave, who constructed a statistical regression model that he claimed could (i) prove impact—specifically, that “virtually all” class members were overcharged by defendants’ conduct—and (ii) calculate damages for all injured class members. Dow moved to exclude Dr. McClave’s testimony, arguing that his regression model was fundamentally flawed. The district court denied the motion.

Before trial, Dow also moved to decertify the class. Dow argued that, even if it was admissible evidence, Dr. McClave’s model was incapable of showing injury and damages on a classwide basis, and that certification was no longer permissible in light of the Supreme Court’s recent decision in Wal-Mart. The district court deferred consideration of the motion, and the parties proceeded to trial.

At trial, plaintiffs introduced evidence of a price-fixing conspiracy but abandoned their customer-allocation theory. Dr. McClave, however, based his impact and damages testimony on his regression models, which assumed both the price-fixing and customer-allocation conspiracies. Dr. McClave’s models, moreover, found impact in the form of (and calculated damages on the basis of)

² After declaring bankruptcy, Lyondell settled without any payment on liability. AA0428-37.

- 3 -

certain “overcharges” that he found to exist for approximately 25% of the class. For the remaining 75%, he assumed the same overcharges as those he claimed to have found for the first 25%, and then used what he called “extrapolation” to find injury and damages as to most of the 75% for which he expressly did not analyze specific overcharges. Dr. McClave’s use of the term “extrapolation” was itself a concession that, at least as to the 75% of customers, his results were speculative.³

Critically, Dr. McClave’s model did not find that each class member suffered injury or damages. His model showed that at least 7% of class members for whom he modeled damages had zero or negative damages. Appellant’s Br. at 35-36. On the other hand, as a matter of mathematics, Dr. McClave assumed that every one of his extrapolated transactions was subject to an overcharge as a result of the alleged conspiracy. Id. at 15, 19, 34-35.

The jury ultimately found liability (including injury) and awarded damages of over $400 million, which the district court trebled.

After trial, Dow renewed its motion for decertification. Dow argued that Dr. McClave’s use of extrapolated damages calculations prevented plaintiffs from proving impact and damages as to all class members “in one stroke.” Wal-Mart, 131 S. Ct. at 2551. Dow also challenged Dr. McClave’s model under the

³ To “extrapolate” means “to project, extend, or expand into an area not known or experienced so as to arrive at a usually conjectural knowledge of the unknown area.” See Extrapolate Definition, Merriam-Webster.com, http://www.merriam- webster.com/dictionary/ extrapolate?show=0&t=1386951357 (last visited Dec. 13, 2013).

- 4 -

intervening decision in Comcast, showing that it mismatched liability and damages because Dr. McClave assumed the existence of both a price-fixing conspiracy and a customer allocation conspiracy, while the jury’s verdict was based on only price-fixing.

The district court denied Dow’s motions. It suggested that Dow had not challenged the admissibility of Dr. McClave’s testimony on these same grounds at trial and suggested that such a challenge was a prerequisite to Dow’s challenging the use of Dr. McClave’s model for class-certification purposes.

SUMMARY OF ARGUMENT

I. The district court’s certification decision—and its refusal to decertify—violated three central and independent legal principles that (among other things) govern the use of statistical models in class litigation: first, a class cannot be certified if a defendant must forfeit its right under the Rules Enabling Act—which is designed to protect litigants’ “substantive right[s]”—to maintain its defenses as to individual class members, Wal-Mart, 131 S. Ct. at 2561; second, a properly certified class must be able to prove “in one stroke,” id. at 2551, that all class members were impacted; and third, if a plaintiff attempts to rely on a statistical model to show impact and damages on a class-wide basis, its impact/damages model must match its ultimate theory of liability. See, e.g., Comcast, 133 S. Ct. at 1433.

- 5 -

Here, the certification decisions violated the first principle by effectively preventing the defendant from contesting antitrust impact and damages as to all class members—particularly the 75% of class members whose injury was simply assumed by Dr. McClave’s model, and whose putative damages he “extrapolated” from the other 25% of class members. Those decisions also violated the second principle, both because Dr. McClave’s analysis showed that some portion of class members—even of the 25% he claimed to have actually analyzed—were not injured, and because his “extrapolation” methodology introduced a forbidden “second stroke” into the certification analysis. And those decisions violated the third principle because of the mismatch between Dr. McClave’s injury/damages analysis—which assumed price-fixing and customer-allocation conspiracies—and plaintiffs’ ultimate proof of liability, which rested upon alleged price-fixing alone. Thus, the district court’s decision deprived Dow of its substantive right under the Rules Enabling Act, as explained in Wal-Mart and Comcast, to defend itself against each plaintiff’s unique claims in “individualized proceedings.” Wal-Mart, 131 S.Ct. at 2561.

II. Enforcement of these principles in this case is particularly important, given the drag that improperly certified class actions impose on the entire economy. The high costs of class action litigation are passed along to consumers in the form of higher prices. And defendants faced with the burdensome costs of class action litigation may be forced to reduce operations, curtail capital

- 6 -

investment, and in extreme cases, forego entering new markets and developing new products—all of which will result in lower employment.

Such costs can be warranted only if courts faithfully apply Rule 23’s requirements to certification (or decertification) motions. Indeed, as this case illustrates, the pressure to enter an unjustified settlement becomes enormous whenever a defendant is forced to go to trial against a certified class. It is thus essential to the strength of our Nation’s economy—and to all who invest or are employed in it—that the requirements articulated in Rule 23, as interpreted in Wal-Mart and Comcast, be rigorously and faithfully applied, both in the initial certification decision, and on any subsequent motion for decertification.

ARGUMENT


I.	The District Court's Certification Decisions Violated Three Core Legal Principles Governing the Use of Statistical Models In Class Litigation.

Wal-Mart and Comcast clarified three critical legal principles governing class certification and, specifically, the use of statistical models in certification decisions. First, under Rule 23, “a class cannot be certified on the premise that [a defendant] will not be entitled to litigate its statutory defenses to individual claims.” Wal-Mart, 131 S. Ct. at 2561. Accordingly, a class cannot be certified—or maintain its certification—based on a statistical model that fails to account for relevant differences among individual class members. Second, the plaintiff seeking certification must provide common evidence of injury that will resolve that

- 7 -

question “in one stroke.” Id. at 2551. Comcast additionally requires that any statistical model of injury must demonstrate with “evidentiary proof” that all members of the class were injured. Comcast, 133 S. Ct. at 1432. Third, “any model supporting a plaintiff’s damages case” that is used as a basis for certification under Rule 23 “must be consistent with its liability case.” Id. at 1433.

These independent requirements are all necessary to ensure that classes are certified only where permitted by Rule 23. Yet, as we now show, the district court in this case violated all three of these core principles. Because each of these violations was an error of law, the pertinent standard of review as to each is de novo. See, e.g., Carpenter v. Boeing Co., 456 F.3d 1183, 1187 (10th Cir. 2006); Wallace B. Roderick Revocable Living Trust v. XTO Energy, Inc., 725 F.3d 1213, 1217 (10th Cir. 2013). And the district court’s certification decisions should be reversed on any or all of these bases.


A.	Statistical Models that Fail Fairly to Account for Relevant Differences Among Individual Class Members Cannot Satisfy Rule 23(b)(3)’s “Predominance” Requirement.

Summarizing Wal-Mart, the D.C. Circuit has recently observed that “Rule 23 not only authorizes a hard look at the soundness of statistical models that purport to show predominance—the rule commands it.” In re Rail Freight, 725 F.3d at 255. That command is reinforced by the Rules Enabling Act, which reflects underlying due process concerns and “forbids [courts from] interpreting Rule 23 to ‘abridge, enlarge or modify any substantive right.’” Wal-Mart, 131 S.

- 8 -

Ct. at 2561 (quoting 28 U.S.C. § 2072(b)); accord Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 613 (1997). If courts permitted class plaintiffs to rely on statistical models to prove classwide impact and damages when there are material differences among class members—like the “Trial by Formula” rejected in Wal-Mart, 131 S. Ct. at 2561—defendants would thereby forfeit a “substantive right,” namely, the right to effectively defend against every distinct claim. Id.

1. Under the Clayton Act—which governs private litigation of substantive claims under both the Sherman and Clayton Acts—impact, or “antitrust injury,” is an essential element of a plaintiff’s federal antitrust claim. See 15 U.S.C. § 15(a) (limiting recovery to “any person . . . injured in his business or property”) (emphasis added). Therefore, to establish antitrust liability, plaintiffs bear the burden of showing not only the alleged violation—i.e., the defendant’s alleged act—but also “injury” to their “business or property” before they can recover. Cohlmia v. St. John Med. Ctr., 693 F.3d 1269, 1280 (10th Cir. 2012).

This burden of proof applies whether a plaintiff proceeds as an individual or as part of a class. Indeed, “proof of injury to . . . each class member is critical for the determination of defendants’ liability to any individual.” Shumate & Co. v. Nat’l Ass’n of Sec. Dealers, Inc., 509 F.2d 147, 155 (5th Cir. 1975) (emphasis added). It follows that, to satisfy Rule 23’s requirements, any statistical model used to establish the elements of an antitrust claim on a classwide basis must account for potentially relevant differences among class members as to the fact and

- 9 -

mechanism of injury. Otherwise use of the model would violate a defendant’s “substantive right” to “litigate its statutory defenses to individual claims.” Wal-Mart, 131 S. Ct. at 2561.

2. Here, the complexities and economic realities of the marketplace in which the alleged conspiracy took place precluded a showing of classwide impact by a single statistical model. Undisputed evidence shows that there is no baseline market price for chemicals in the polyurethanes industry. See Appellant’s. Br. at 2. Rather, individual purchasers negotiate extensively over price and other terms. Id. The district court recognized these economic realities, finding that the market comprises “myriad [] products, pricing structures, individualized negotiations and contracts.” Appellant’s Br. at 6; AA0413. It also found that “sales of the basic chemicals were characterized by individual negotiations, variations in contractual relationships and the like.” AA0409.

In a case like this, involving a complex industry in which customers negotiate individual agreements that directly affect price and non-price terms, antitrust injury is almost always individualized-and liability cannot be shown in one stroke. As a prominent treatise puts it, “[w]hen transaction prices are negotiated, the actual price paid will be determined at least in part by the negotiating styles of the customers. As a result, proof of antitrust injury is bound to be individualized.” 2A, Phillip E. Areeda & Herbert Hovenkamp, Antitrust Law: An Analysis of Antitrust Principles and Their Application § 398(c), at 423

- 10 -

n.14 (2013). In such a market, therefore, predominance is impossible to establish. And therefore, consistent with Wal-Mart and Comcast, certification in such a market is erroneous as a matter of law. The district court’s refusal to decertify the class in this case can and should be reversed on this basis alone.

3. Even if predominance could be established here on the basis of a statistical model, Dr. McClave’s model was inadequate for that purpose. That is because, as to the majority of purchasers in the class (75%), Dr. McClave did not conduct any genuine modeling at all. Instead, Dr. McClave determined their antitrust impact and damages by “extrapolating” from his analysis of the impact and damages of the remaining 25% of purchasers. See supra at 5-6. Dr. McClave simply assumed that all customers for whom he extrapolated damages suffered antitrust impact and damages equal to the “average” customer he modeled. The district court thus allowed plaintiffs to establish impact and damages for three-quarters of the class by extrapolation-that is, without even requiring proof of impact for any one of them, let alone all class members “in one stroke.” Wal-Mart, 131 S. Ct. at 2551.

Dr. McClave’s model also violated a core principle of Comcast, which demands that plaintiffs come forward with “evidentiary proof” that will “plausibly show[]”any claim elements relied upon to establish the existence of common issues-in this case both impact and damages. 133 S. Ct. 1432, 1436 n.6. But here, the record conclusively shows the absence of “evidentiary proof” in Dr.

- 11 -

McClave’s model on the common issues of impact and damages. His model purported to directly assess impact and damages for only one-quarter of the class, and offered only “arbitrary . . . measurements” and extrapolation, id., for the remaining three-fourths. Accepting Dr. McClave’s model as establishing predominance even though it speaks to only 25% of the putative class “would reduce Rule 23(b)’s predominance requirements to a nullity.” Comcast, 133 S. Ct. at 1433. This defective methodology should have “shred[ded] the plaintiffs’ case for certification.” In re Rail Freight, 725 F.3d at 252.

Indeed, in that respect, Dr. McClave’s model is analogous to the model used by the plaintiffs in Wal-Mart in their effort to bridge the gap among putative class members who experienced different treatment, lived in different parts of the country, or were otherwise dissimilarly situated. 131 S. Ct. at 2550. Here, as in Wal-Mart, the plaintiffs’ expert “extrapolat[ed] the validity and value of the untested claims”-i.e., the 75% of claims subject to extrapolation-“from the sample set,” i.e., the 25% for which damages were calculated. Wal-Mart, 131 S. Ct. at 2550. But Wal-Mart rejected the plaintiffs’ statistical methodology as a basis for class certification because the defendant would have been forced to abdicate its “statutory right” to “litigate its statutory defenses to individual claims.” 131 S. Ct. at 2561.

So too here. Allowing the class plaintiffs to prove injury and damages through a classwide “Trial by Formula”—as Dr. McClave’s model required—

- 12 -

violated the Rules Enabling Act (and the due process principles that underlie it) because it necessarily “abridged” the defendant’s “substantive right” to fully litigate its defenses. Id. at 2561; see also, e.g., Jacob v. Duane Reade, Inc., – F.R.D. –, 2013 WL 4028147, at *10 (S.D.N.Y. Aug. 8, 2013) (“[R]eading [Wal-Mart] and Comcast together, . . . there are due process implications for defendants, which render the so-called ‘trial by formula’ approach, whereby representative testimony is utilized to determine damages for an entire class, inappropriate where individualized issues of proof overwhelm damages calculations.”). For example, it would have been effectively impossible for Dow to take discovery of absent class members to refute the extrapolated impact determinations. It would likewise have been impossible for the jury to determine liability when faced with the countless mini-trials that Wal-Mart would require.⁴

Accordingly, when given the “hard look” that Rule 23 “commands,” In re Rail Freight, 725 F.3d at 255, extrapolation is not a valid basis for class

⁴ Indeed, the facts here are even more egregious than those in Wal-Mart. First, the trial plan proposed in Wal-Mart would have allowed Wal-Mart to present individualized defenses-and the court to adjudicate those defenses-in “randomly selected sample cases.” Id. at 2550. Here, there were no individual adjudications. Second, the district court’s plan in Wal-Mart was to select sample cases for trial “at random,” in an attempt to rid the trial-by-formula of the unfairness that would result if plaintiffs were permitted to present only their strongest claims to the jury. Id. Here, the district court made no attempt to provide even these inadequate protections. On the contrary, it allowed plaintiffs to prove impact and damages for many transactions of a major purchaser and all systems transactions by extrapolation, even though Dr. McClave did not model any of these transactions. See Appellant’s Br. at 15, 37-40.

- 13 -

certification. It violates Wal-Mart’s core legal principle that plaintiffs cannot proceed under Rule 23 if to do would “abridge” a defendant’s “substantive right” to fully litigate its defenses as to each class member. Wal-Mart, 131 S. Ct. at 2561. It also allows class plaintiffs to skirt the Clayton Act’s mandate by recovering antitrust damages without first establishing proof of individual injury.


B.	Any Statistical Model Purporting To Establish Classwide Injury Must Show That All Members Of The Class Were Injured, In The Same Way.

Though Dr. McClave’s model supposedly finds and measures impact on a classwide basis, what it actually shows is that at least seven percent of the class members he modeled, see supra at 4, suffered no antitrust injury at all. This admission also should have “shred[ded] the plaintiffs’ case for certification.” In re Rail Freight, 725 F.3d at 252.

That is because, as decisions interpreting Wal-Mart and Comcast have confirmed, to establish predominance “[t]he plaintiffs must . . . show that they can prove, through common evidence, that all class members were in fact injured by the alleged conspiracy.” Id. (emphasis added). Absent such a showing, “individual trials are necessary to establish whether a particular [plaintiff] suffered harm from the price-fixing scheme” at all. Id. Moreover, courts must first make this assessment at the class certification stage, before trial, because it is the only way to ensure that plaintiffs can carry their ultimate burden of proving antitrust injury—a key element of liability—“in one stroke.” Wal-Mart, 131 S. Ct. at 2545,

- 14 -

2551. And of course that assessment must be repeated if a defendant moves to decertify the class.

Because the Clayton Act requires each plaintiff to prove that it was injured by the defendant’s allegedly anticompetitive conduct, a statistical model that relies on extrapolation to show impact and calculate classwide damages cannot justify certification—especially where, as here, there are distinct, identifiable differences among class plaintiffs’ actual claims. Indeed, one of the preeminent shortcomings of an extrapolation model is that it assumes the same or similar impact across the class, which in turn subverts the required proof that class members in fact “suffered the same injury.” Wal-Mart, 131 S. Ct. at 2551. A model that assumes what it must demonstrate—i.e., that all plaintiffs were injured by the defendant’s conduct—has no probative value. And again, forcing a defendant to defend against dissimilar claims in a Rule 23 trial where it will not have the right to defend against each plaintiff’s unique claims through “individualized proceedings” violates the defendant’s “substantive right[s].” Id. at 2561.

In short, when plaintiffs rely on extrapolated calculations to prove antitrust impact, they cannot prove injury to all class members “in one stroke.” Id. at 2550. Accordingly, the district court’s decision to allow the plaintiffs to establish predominance on the basis of extrapolations from a small subset of the class requires reversal. Dr. McClave’s extrapolation methodology was simply incapable of establishing that all class members were injured, or that they were injured in the

- 15 -

same way. And in any event, his conclusion that a portion of the class members were not injured foreclosed certification under Wal-Mart and Comcast.


C.	The Abandonment of a Foundational Element of Plaintiffs’ Liability Case Requires Decertification.

The district court’s certification decision also violated Comcast’s holding that a plaintiff’s “model purporting to serve as evidence of damages in [a] class action” “must measure only those damages attributable” to the plaintiffs’ liability theory. 133 S. Ct. at 1433. Comcast thus prohibits “a mismatch between the injury and the remedy,” Duane Reade, Inc., 2013 WL 4028147, at *10, or, for that matter, a “mismatch” between the injury and the alleged violation. By so doing, Comcast precludes plaintiffs from playing a game of “bait and switch” in which the evidence they present at trial—and the statistical model they use to support that evidence—depart dramatically from the liability theory they advance as a basis for certification. And if the plaintiffs’ case at trial departs from that theory, the consequence is plain: decertification is required.

1. In Comcast, the plaintiffs attempted to prove classwide impact and damages through a regression model built by the same Dr. McClave who testified here. Because the plaintiffs there had originally advanced four legal theories of antitrust impact, Dr. McClave built his regression model on all four theories. But the district court held that only one of those theories could be proved with classwide evidence. Comcast, 133 S. Ct. at 1426, 1431-32. Dr. McClave’s model

- 16 -

thus “failed to measure damages resulting from the” only “particular antitrust injury on which petitioners’ liability . . . [was] premised.” Id. at 1433. And that, the Court held, meant that continued certification in Comcast was improper. Id. at 1435.

2. Here, Dr. McClave’s initial report estimated damages based on two theories of liability: a price-fixing conspiracy and customer allocation conspiracy. Dr. McClave thus constructed his “common impact” and damages model on the assumption that plaintiffs would prove that the defendant conspired to allocate certain customers. See Appellant’s Br. at 44. Yet at trial “[n]either side presented any evidence . . . of any illegal customer allocation.” Dkt. 2879 at 9. Plaintiffs’ failure to pursue this theory and offer supporting evidence at trial thus resulted in Dr. McClave’s repeating the very error for which he was chastised in Comcast: failing to provide a model that matched the purported impact and damages to “the wrong” that the jury found. Comcast, 133 S. Ct. at 1434.

Her too, under Comcast, the plaintiffs’ abandonment of one of the theories on which Dr. McClave built his damages/impact model—i.e., a conspiracy as to both pricing and customer allocation— fatally undermined the plaintiffs’ argument that Dr. McClave’s model established predominance. Without evidence of a customer allocation conspiracy, Dr. McClave’s damages/impact model in this case failed to match the plaintiffs’ liability model—and was thus unable to support the district court’s initial finding of predominance: “no predominance, no class

- 17 -

certification.” In re Rail Freight, 725 F.3d at 253. Accordingly, under Comcast, plaintiffs’ later abandonment of the customer allocation theory retroactively rendered the earlier class certification invalid, and therefore required decertification.

This is true, moreover, whether or not Dr. McClave’s model was the linchpin of the district court’s initial class certification decision: The core principle of Comcast is that plaintiffs’ case for liability, both in theory and as shown by the evidence, must match its impact and damages model “at trial,” id. at 1433. But here it clearly did not match.

Given that the trial evidence confirmed that certification was improper, it was incumbent upon the district court to decertify the class. Indeed, the Federal Rules permit courts to “alter[] or amend[]” class certification decisions “before final judgment.” Fed. R. Civ. P. 23(c)(1)(C); see also DG ex rel. Stricklin v. Devaughn, 594 F.3d 1188, 1201 (10th Cir. 2010) (district court “possesses the discretion . . . [to] decertify the class altogether prior to final judgment”). Thus, “a district court’s order denying or granting class status is inherently tentative,” Coopers & Lybrand v. Livesay, 437 U.S. 463, 469 n.11 (1978), “and subject to change,” In re Zurn Pex Plumbing Prods. Liab. Litig., 644 F.3d 604, 613 (8th Cir. 2011), “in light of subsequent developments in the litigation.” Gen. Tel. Co. of Sw. v. Falcon, 457 U.S. 147, 160 (1982). And Wal-Mart and Comcast establish that the court has a duty to do so if it becomes clear that the plaintiffs’ proof at trial

- 18 -

departs from the basis on which certification has been granted. E.g., Comcast, 133 S. Ct. at 1433 (stating that Rule 23’s requirements apply equally “at trial”).

3. The district court discounted these problems on two grounds, neither of which is plausible. First, it stated that “Dow did not provide[] any expert opinion . . . to show that Dr. McClave’s method was unreliable.” Id. But this turns class-action precedent on its head. As the defendant, Dow never had the burden to show that the plaintiffs’ evidence failed to meet Rule 23’s requirements. Rather, the burden always falls on the plaintiffs to show that their damages model establishes predominance, even when “the issue . . . comes before the Court on Defendants’ motion.” Spagnola v. Chubb Corp., 264 F.R.D. 76, 92 (S.D.N.Y. 2010) (emphasis added).

Second, the district court stated that, because “Dow did not seek to exclude Dr. McClave’s testimony on this basis before trial,” he “was . . . permitted to testify that such members did suffer impact and damages.” Dkt. 2879 at 5. But that ignores the fact that Rule 702’s admissibility hurdle is distinct from Rule 23’s predominance inquiry. Thus, even assuming Dr. McClave’s expert testimony were admissible, it could be and was insufficient to satisfy plaintiffs’ burden under Rule 23. The court thus erred by conflating the two inquiries and assuming that Dr. McClave’s testimony—once admitted and heard by the jury—was unimpeachable, even for the purposes of Rule 23.

- 19 -

Comcast, moreover, directly addressed this point when it held that the defendant’s decision there not to “make an objection to the admission of Dr. McClave’s testimony under the Federal Rules of Evidence . . . does not make it impossible for them to argue that the evidence failed to show that the case is susceptible to awarding damages on a class-wide basis.” Comcast, 133 S. Ct. at 1431 n.4 (quotation omitted). Nor can the district court’s view be reconciled with Wal-Mart, which requires decertification if the plaintiffs fail to establish predominance (or Rule 23’s other requirements) at any later point in the litigation. 131 S. Ct. at 2561. Nor does it comport with the plain language of Rule 23, which contemplates that a court may determine after trial—such as on defendant’s motion to decertify—that an expert’s trial testimony failed to establish predominance. See Fed. R. Civ. P. 23(c)(1)(C) (permitting court to “alter[] or amend[]” class certification “before final judgment”). Thus, the admissibility vel non of Dr. McClave’s testimony does not alter the plaintiffs’ burden to show that Rule 23’s requirements are met.

In sum, in its treatment of Dr. McClave’s model, the district court violated three principles governing the use of statistical models in certifying class actions: (1) the model must fairly account for differences among class members; (2) it must establish that all class members were injured by the defendant’s challenged conduct; and (3) it must be consistent with the remainder of the plaintiffs’ case, both during pretrial proceedings and at trial. For each of these independent

- 20 -

reasons, the district court’s decision to certify a class, and its decision not to decertify it after trial, must be reversed.


II.	Strict Adherence to Rule 23's Requirements Is Essential to Ensure that Class Actions Do Not Impose Unwarranted Burdens on the National Economy

Considerations of judicial administration compel the same result. Here, the defendant bucked the overwhelming trend in which defendants settle class actions once certification is granted, to avoid the substantial costs and potentially catastrophic risks associated with litigating a class action to verdict. In trying this case, moreover, Dow asserted its due process and other rights at every turn. This case therefore presents an extremely rare opportunity—in the context of an appeal from a class action tried to a jury verdict—to address how lower courts should apply the core principles of Wal-Mart and Comcast at trial and when deciding motions for decertification. Indeed, faithful adherence to those principles is essential to ensure that class actions do not impose substantial, unjustified burdens on the national economy.


A.	The Ruling Below Will Exacerbate the Problem of Defendants Being Effectively Forced to Settle Meritless Claims.

Perhaps most important, careful adherence to these principles is necessary to prevent the socially undesirable assertion and settlement of meritless claims.

1. Class certification is almost invariably followed by settlement. See Robert G. Bone & David S. Evans, Class Certification and the Substantive Merits,

- 21 -

51 Duke L.J. 1251, 1292 (2002) (“[E]mpirical studies . . . confirm what most class action lawyers know to be true: almost all class actions settle.”). Even the Advisory Comments to Rule 23 expressly recognize the enormous risks—and attendant settlement pressure—that a class-certification decision imposes on defendants. See Fed. R. Civ. P. 23(f) Advisory Comm.’s Notes to 1998 Amendments (“An order granting certification . . . may force a defendant to settle rather than incur the costs of defending a class action and run the risk of potentially ruinous liability.”). Most notably, “certification ‘generate[s] unwarranted pressure to settle nonmeritorious or marginal claims.’” In re Rail Freight, 725 F.3d at 252 (quoting Newton v. Merrill Lynch, Pierce, Fenner & Smith, Inc., 259 F.3d 154, 168 (3d Cir. 2001)).

Indeed, “[c]ertification of a large class may so increase the defendant’s potential liability and litigation costs that he may find it economically prudent to settle and abandon a meritorious defense.” Coopers & Lybrand, 437 U.S. at 476; accord In re Lorazepam & Clorazepate Antitrust Litig., 289 F.3d 98, 102 (D.C. Cir. 2002) (“[T]he grant of class status can put substantial pressure on the defendant to settle independent of the merits of the plaintiffs’ claims.”).

2. These coercive-settlement problems are magnified in the antitrust arena. Antitrust defendants risk not only damages arising from their own conduct, but also joint-and-several liability and treble damages.

- 22 -

Moreover, the sheer cost of litigating an antitrust class action is particularly pronounced because these cases are “arguably the most complex action[s]” for a defendant to litigate. In re Linerboard Antitrust Litig., 296 F. Supp. 2d 568, 577 (E.D. Pa. 2003). Given that defendants may incur many of these costs after a certification decision, including costs associated with taking “testimonial evidence” and conducting “extensive discovery,” Manual for Complex Litigation § 30, at 519 (4th ed. 2004), the pressure to settle becomes even more immense after a class is certified. See, e.g., Bell Atl. Corp. v. Twombly, 550 U.S. 544, 559 (2007) (recognizing the “potentially enormous expense of discovery”).

Faced with these staggering costs, many defendants—especially in antitrust actions—will have no practical option but to settle before trial.

3. The financial risks of class-action litigation—including antitrust litigation—are magnified still further when courts allow dissimilar plaintiffs to create for a jury the false impression that they have proven liability “in one stroke.” Wal-Mart, 131 S. Ct. at 2551. And that risk becomes particularly acute in antitrust class actions if the plaintiff class is permitted to rely on aggregate-impact “extrapolations” of the kind Dr. McClave performed.

Indeed, a statistician can use regression software to run many hundreds (even thousands) of different model permutations with different variables and benchmark periods, and then see the damages figure that each model generates. Such results-oriented data mining can always be manipulated to generate a

- 23 -

favorable model that produces both adequate statistical fit and damages for plaintiffs.

Plaintiffs, moreover, without providing a basis in the evidence, can often frame the agreement among the alleged conspirators simply as an agreement to keep prices as high as possible, see, e.g., App. 862, then point the jury to actual prices above the modeled, so-called “competitive” prices as confirmation of the conspiracy. See, e.g., App. 925. They can then ask the jury, again without basis in the evidence, to ignore the periods in which the model fails to show overcharges as times when the conspiracy participants cheated. See, e.g., App. 873-74. And where this is allowed—as it was below—antitrust liability is driven entirely by manipulated modeling, not by underlying economic and market realities.

4. The fact that most class actions settle before trial does not reduce the importance of policing district courts’ procedures in class actions that, like this one, do proceed to trial. See Robert H. Mnookin & Robert B. Wilson, Rational Bargaining and Market Efficiency: Understanding Pennzoil v. Texaco, 75 Va. L. Rev. 295, 295 (1989). The opposite is true: Litigants “bargain with the knowledge that if they cannot strike a deal, a court ultimately may impose a resolution.” Id. Thus, “the rules and procedures used inside of court” may “substantially affect[] the bargains reached outside the court.” Id.

Accordingly, if the effect of class certification is to preclude examination at trial of the weaknesses in the plaintiffs’ individual claims, or even to preclude the

- 24 -

defendant from presenting individualized defenses, the case’s “blackmail” value will skyrocket. See, e.g., In re Rhone-Poulenc Rorer, Inc., 51 F.3d 1293, 1298 (7th Cir. 1995) (finding that the rate of “blackmail settlements” likely will increase exponentially once a district court certifies a class). Correlatively, the defendant’s bargaining position will crumble, and settlement pressure will, in almost all cases, become overwhelming. See Richard A. Nagareda, Class Certification in the Age of Aggregate Proof, 84 N.Y.U. L. Rev. 97, 103 (2009) (“If a cohesive class can be created through . . . savvy crafting of the evidence,” then “[t]he law [will] run a considerable risk of unleashing the settlement-inducing capacity of class certification based simply on the say-so of one side.”).

Moreover, if class certification comes to be viewed as a way to preclude at trial strong defenses to weak individual claims, the sheer number of class actions will surely increase. Indeed, when class trials become “a roll of the dice” capable of “determin[ing] the outcome of an immense number of separate claims,” Thorogood v. Sears, Roebuck and Co., 624 F.3d 842, 849 (7th Cir. 2010), defendants face an increased risk of a catastrophic judgment—which in turn dramatically increases plaintiffs’ settlement leverage.

Courts thus should strictly adhere to Wal-Mart’s and Comcast’s teachings to give defendants a fair chance to defend against class actions before certification, during trial, and after trial.

- 25 -


B.	Improper Class Actions Levy a Substantial and Unjustified Burden on the National Economy

Beyond the unfairness they visit upon individual defendants, improperly certified class actions impose a drag on the entire economy. The high costs of class action litigation are, at least in part, “passed along to the public,” SEC v. Tambone, 597 F.3d 436, 453 (1st Cir. 2010), most recognizably in the form of higher prices to consumers. In addition, defendants faced with the burdensome costs of class action litigation may be forced to reduce operations, curtail capital investment, and in extreme cases forego entering new markets and developing new products—all of which will curtail employment.

These risks infect not just businesses that are currently defending against class actions; it infects virtually all businesses. Indeed, the prospect of excessive liability may cause businesses to alter their operations in ways that deprive consumers of the benefits of lawful, procompetitive conduct. Even worse for consumers is that they may end up footing the bill for the economic toll that class actions take. As the Supreme Court has noted, the costs associated with class actions are paid by both consumers and “innocent investors for the benefit of speculators and their lawyers.” Blue Chip Stamps v. Manor Drug Stores, 421 U.S. 723, 739 (1975) (quoting SEC v. Texas Gulf Sulphur Co., 401 F.2d 833, 867 (2d Cir. 1968)).

- 26 -

For all these reasons, it is essential to the strength of our Nation’s economy—and to all who invest or are employed in it—that the requirements articulated in Rule 23, as interpreted in Wal-Mart and Comcast, be rigorously and faithfully applied.

CONCLUSION

The judgment of the district court should be reversed and the class decertified.

Dated: December 13, 2013 Respectfully Submitted.



	/s/ Gene C. Schaerr
Kathryn Comerford Todd	Gene C. Schaerr
Tyler R. Green	Robert F. Ruyak
NATIONAL CHAMBER	William A. Roach, Jr.
LITIGATION CENTER, INC.	WINSTON & STRAWN LLP
1615 H Street, N.W.	1700 K Street, N.W.
Washington, D.C. 20062	Washington, D.C. 20006
(202) 463-5337	(202) 282-5000

Jeffrey L. Kessler
George E. Mastoris
WINSTON & STRAWN LLP
200 Park Avenue
New York, New York 10166
(212) 294-6700

Counsel for Amicus Curiae

Chamber of Commerce of the United States of America

- 27 -

CERTIFICATE OF COMPLIANCE

1. This brief complies with the type-volume limitations set forth in Fed. R. App. P. 29(d) and 32(a)(7)(B)(i), because this brief contains 6433 words, excluding the parts of the brief exempted by Fed. R. App. P. 32(a)(7)(B)(iii).

2. This brief complies with the typeface requirements of Fed. R. App. P. 32(a)(5) and the type style requirements of Fed. R. App. P. 32(a)(6) because this brief has been prepared in a proportionally spaced typeface using Microsoft Word 2010 in Times New Roman 14-point font.

/s/ Gene C. Schaerr

Gene C. Schaerr

CERTIFICATE OF SERVICE

I hereby certify that, on December 13, 2013, I electronically filed the foregoing with the Clerk of Court for the U.S. Court of Appeals for the Tenth Circuit by using the appellate CM/ECF system. Participants in the case who are registered CM/ECF users will be served by the appellate CM/ECF system.

/s/ Gene C. Schaerr

Gene C. Schaerr