Siddharth Hariharan, et al v. Adobe Systems, Inc., et al

Filing 1

FILED ON 11/07/2013 PETITION FOR PERMISSION TO APPEAL PURSUANT TO RULE 23(f). SERVED ON 11/07/2013. [8856405] (HC)

Case No. __________
UNITED STATES COURT OF APPEALS
FOR THE NINTH CIRCUIT

IN RE HIGH-TECH EMPLOYEE ANTITRUST LITIGATION

Petition for permission to appeal from the United States District Court
Northern District of California
The Honorable Lucy H. Koh, Presiding
Case No. 5:11-2509-LHK

PETITION FOR LEAVE TO APPEAL A CLASS
CERTIFICATION ORDER PURSUANT TO FEDERAL RULE
OF CIVIL PROCEDURE 23(F)
ROBERT A. VAN NEST, #84065
DANIEL PURCELL, #191424
EUGENE M. PAIGE, #202849
JUSTINA SESSIONS, #270914
KEKER & VAN NEST LLP
633 Battery Street
San Francisco, CA 94111-1809
Telephone: [phone]
Facsimile: 415 397 7188
Attorneys for Defendant and Petitioner
Google Inc.

790064

Additional counsel listed on signature
page

CORPORATE DISCLOSURE STATEMENTS

Adobe Systems, Inc. submits the following corporate disclosure statement
pursuant to Federal Rule of Appellate Procedure 26.1: (1) Adobe is a publicly held
corporation; (2) Adobe does not have any parent corporation, and (3) no publicly
held corporation owns ten percent or more of Adobe’s stock.

Dated: Nov. 7, 2013

JONES DAY

By: /s/ David C. Kiernan
David C. Kiernan
Robert A. Mittelstaedt
Craig A. Waldman
555 California Street, 26th Floor
San Francisco, CA 94104
Telephone: [phone]
Facsimile: [phone]
Attorneys for Defendant and Petitioner Adobe
Systems, Inc.

790064

Apple Inc. submits the following corporate disclosure statement pursuant to
Federal Rule of Appellate Procedure 26.1: (1) Apple is a publicly held corporation; (2) Apple does not have any parent corporation; and (3) no publicly held corporation owns ten percent or more of Apple’s stock.

Dated: Nov. 7, 2013

O’MELVENY & MYERS LLP

By: /s/ Michael F. Tubach
Michael F. Tubach
George Riley
Christina J. Brown
Two Embarcadero Center, 28th Floor
San Francisco, CA 94111
Telephone: [phone]
Facsimile: [phone]
Attorneys For Defendant and Petitioner Apple
Inc.

2
790064

Google Inc. submits the following corporate disclosure statement pursuant to
Federal Rule of Appellate Procedure 26.1: (1) Google Inc. is a publicly held corporation; (2) Google Inc. does not have any parent corporation; and (3) no publicly
held corporation owns ten percent or more of Google Inc.’s stock.

Dated: Nov. 7, 2013

KEKER & VAN NEST LLP
By:

/s/ Robert A. Van Nest
Robert A. Van Nest
Daniel Purcell
Eugene M. Paige
Justina Sessions
633 Battery Street
San Francisco, CA 94111
Telephone: [phone]
Facsimile: [phone]
Edward D. Johnson
Lee H. Rubin
Donald M. Falk
MAYER BROWN LLP
Two Palo Alto Square
3000 El Camino Real, Suite 300
Palo Alto, CA 94306-2112
Telephone: [phone]
Facsimile: [phone]
Attorneys for Defendant and Petitioner
Google Inc.

3
790064

Intel Corporation submits the following corporate disclosure statement pursuant to Federal Rule of Appellate Procedure 26.1: (1) Intel is a publicly held corporation; (2) Intel does not have any parent corporation; and (3) no publicly held
corporation owns 10% or more of Intel’s stock.

Dated: Nov. 7, 2013

MUNGER TOLLES & OLSON, LLP

By: /s/ Gregory P. Stone
Gregory P. Stone
Bradley S. Phillips
Gregory M. Sergi
John P. Mittelbach
355 South Grand Avenue, 35th Floor
Los Angeles, CA 90071-1560
Telephone: [phone]
Facsimile: [phone]
Attorneys for Defendant and Petitioner Intel
Corporation

4
790064

The class certified in this wage-suppression antitrust case sweeps in 60,000
employees holding 2,400 diverse job titles at seven companies that produce markedly different goods and services. The conduct alleged to have commonly affected
all 60,000 employees consists of six bilateral agreements in which pairs of defendants refrained from one method of recruiting each other’s employees—cold calls.
All other recruiting methods were unaffected. Plaintiffs do not allege that there
was any impact on total hiring by the defendants. Defendants comprise only a tiny
fraction of the employer pool for these diverse jobs; actual hiring from one another
was only 1% of their total hires before, during, and after the agreements.
Only by committing several fundamental legal errors was the district court
able to certify such a heterogeneous class. The district court viewed the standards
governing class certification as “not altogether clear” and “somewhat unsettled.”
ER822(19).1 Yet under Rule 23(b)(3), “[c]ommon questions of fact cannot predominate where there exists no reliable means of proving classwide injury in fact.”
In re Rail Freight Surcharge Antitrust Litig., 725 F.3d 244, 252-53 (D.C. Cir.
2013). To be “common” under Rule 23, an issue must be “capable of classwide
resolution—which means that determination of its truth or falsity will resolve an
issue that is central to the validity of each one of the claims in one stroke.” Wal-

1

Defendants have filed public and sealed versions of the excerpts of record. Citations to materials that are partially under seal will first provide a pinpoint citation
to the sealed materials, then a parenthetical pinpoint citation to the public version.
1
790064

Mart Stores, Inc. v. Dukes, 131 S. Ct. 2541, 2551 (2011).
Plaintiffs claim that the reduction in cold calls prevented employees from
learning information that would have allowed them to bid up their salaries. But the
district court permitted plaintiffs to rely on statistical methods that measured only
average and aggregated effects. The methods were designed to mask critical differences among individuals and cannot possibly “resolve … the validity of each
one of the claims.” Id. The certification order rested on the theory that a raise for
one or some employees would “ripple” throughout the class, but undisputed evidence shows that each class member’s compensation is determined by highly individualized factors unsuited for classwide adjudication and that the compensation of
individuals within a job title (let alone across jobs) did not rise and fall together.
The district court failed to conduct the “rigorous analysis” required before
certifying a class. See Comcast Corp. v. Behrend, 133 S. Ct. 1426, 1432 (2013);
Ellis v. Costco Wholesale Corp., 657 F.3d 970, 980 (9th Cir. 2011). In approving
an arbitrary impact and damages model that violated Comcast, the district court
subjected defendants to a “Trial by Formula” that would deny them individualized
defenses in violation of the Rules Enabling Act. Dukes, 131 S. Ct. at 2561. For
instance, defendants will be prevented from proving that specific individuals would
not have received any different compensation than they did, regardless of the alleged conspiracy, and thus did not suffer any antitrust injury.

2
790064

The district court’s divergence from the Supreme Court’s and this Court’s
precedent is manifestly erroneous. And the errors are important. The certification
order is slated for publication in the Federal Supplement, and has been highly publicized.2 Its approval of the use of averages to substitute for common proof of impact and damages could be followed in almost any class action. And that approach
squarely conflicts with decisions of federal courts in this and other Circuits. As a
result, defendants now confront the imminent prospect of a class action conspiracy
trial where they face more than $9 billion in potential trebled antitrust damages,
which exerts obvious pressure to settle and foreclose appellate review.
QUESTIONS PRESENTED
1.

Whether a class of over 60,000 employees from seven different companies
may be certified based on methods of averaging and aggregation that would
mask rather than resolve the individual issues as to impact and damages that
would overwhelm any purportedly common questions.

2.

Whether the formula approved by the district court impermissibly precludes
defendants from presenting individualized defenses as to the existence of
impact and the amount of damages, abridging defendants’ substantive rights
in violation of the Rules Enabling Act and Due Process Clause.

2

See, e.g., Sakthi Prasad, Lawsuit against Silicon Valley hiring practices gets
class action status, Reuters, Oct. 25, 2013; Joel Rosenblatt & Karen Gullo, Apple,
Google Must Face Group Antitrust Suit, Judge Rules, Washington Post, Oct. 25,
2013.
3
790064

BACKGROUND
This action followed a Department of Justice consent decree addressing six
bilateral agreements by pairs of defendants not to cold call each other’s employees.
ER808(5). Plaintiffs seek damages for the same six bilateral agreements, which
they, unlike the DOJ, claim were linked by an “overarching” conspiracy.
A.

The 60,000 Absent Class Members Hold More Than 2,400 Different Job
Titles Across Seven Disparate Technology Companies And Receive
Highly Variable Compensation
Defendants are technology companies with employees throughout the coun-

try and world. Each defendant’s business is distinct: semiconductors (Intel); digital media and marketing software (Adobe); visual effects, video games and sound
for films (Lucasfilm); financial and tax preparation software (Intuit); web search
and information organization technologies (Google); hardware, software and related services (Apple); and animated films (Pixar).
Absent class members include IT personnel, hardware engineers, web designers, safety engineers, graphic designers and customer support managers,
among many other jobs. More than half of the class comprises Intel employees,
most of whom work outside the Bay Area and whose compensation was set in relation to hundreds of companies based outside Silicon Valley. ER1453, 1464(675,
697).
Defendants hired at most 1% of their employees from one another (and, for

4
790064

some pairs of defendants, as little as 0.01%) before, during, and after the class period. ER954, 1095(150, 291). Defendants compete for employees with scores of
other name-brand companies—such as Microsoft, IBM, Hewlett-Packard, Cisco,
Amazon, Oracle, Yahoo!, Electronic Arts, and Zynga—as well as start-ups, universities, and government agencies. ER231, 1455. While certain broad job categories
may be common to defendants (i.e., software engineer), many others are not (i.e., a
Lucasfilm sound engineer or Intel semiconductor-fabrication engineers). More
than 33% of the absent class members are Intel semiconductor workers, whose
skill sets would be of no interest to any other defendant. ER1366, 1452, 1492(588,
674, 685).
Defendants’ compensation decisions are individualized. Each defendant set
each class member’s pay on a case-by-case basis, taking into account a variety of
factors

that

center on individual performance, skills, and

experience.

ER1395(617). To varying degrees, those criteria were assessed within the context
of base-salary ranges that guided managers’ discretion. To the extent they used
salary ranges for individual positions, those base-salary range were set independently for each individual job. ER1369-70(591-92). The ranges were broad—
often over $60,000 and sometimes over $100,000—leaving plenty of room for differentiation among employees. ER1454. Defendants further differentiated among
employees by awarding bonuses and equity grants. In 2009, for example, total an-

5
790064

nual compensation for one Google job title varied by up to $640,000. ER1398,
1488(620, 721). Defendants also promoted strong performers into jobs with higher
base-pay ranges.
B.

The Alleged Agreements
Plaintiffs challenge six no-cold-call agreements that arose out of particular

collaborative relationships between companies, such as overlapping board membership or joint product development. Plaintiffs concede every defendant was free
to hire employees from every other defendant. Each defendant also was free to
cold call all defendants with which it did not have an agreement, and to call or hire
from the thousands of other non-defendant sources in the market. Class members,
in turn, had many sources of information about the job market other than cold
calls, such as information from new employees at their company, personal networks, job fairs, and Internet job sites. ER687-89.
C.

Plaintiffs’ Theory Of Classwide Impact
Plaintiffs moved to certify a nationwide class of nearly all of defendants’

employees or, in the alternative, employees with “technical, creative, or researchand-development” job titles. Plaintiffs’ theory of classwide impact was that the six
bilateral agreements restricted some undetermined amount of information to all
employees regarding job opportunities and market compensation. Absent these
agreements, plaintiffs speculated, the recipients of hypothetical cold calls would

6
790064

have spread the information to other employees via “water cooler chatter”—as
plaintiffs’ expert put it—ultimately causing defendants to raise compensation for
everyone through supposedly rigid wage structures. Wage increases for some employees would purportedly translate into increases for all. ER906-08(103-05).
Plaintiffs supported their motion with a “conduct regression” that aggregated
all defendants’ compensation data, assumed the alleged agreements had the same
effect at each defendant, and assigned to the agreements any unaccounted-for difference in compensation between the class period and the time before and afterward. Then, having calculated a purported aggregate damages number, plaintiffs
presented a “common impact” model that eliminated individual variation in compensation through averaging. ER1151(347). Plaintiffs’ expert admitted he had no
basis to opine that the overall “information flow” to defendants’ employees was
reduced, ER677-78, or that any firm’s compensation structure was “so rigid that
raises for one or a few employees would necessarily propagate into raises for all or
nearly all of the technical employees, absent the agreements.” ER681-85.
D.

The Certification Order
The district court initially granted the class-certification motion in part and

denied it in part. The court refused to resolve disputed issues about plaintiffs’ conduct regression, ruling that plaintiffs had provided a “plausible methodology for
showing generalized harm to the Class as well as estimating class-wide damages”

7
790064

and, therefore, that “Plaintiffs have satisfied their burden, for the purpose of Rule
23(b)(3), on the issue of damages.” ER933(130). But the court declined to certify
any class because nothing in the conduct regression or any of plaintiffs’ other evidence supported the “theory that there was a rigid wage structure such that an impact to some of Defendants’ employees would necessarily have resulted in an impact to all or nearly all employees.” ER932(129).
Plaintiffs then filed a renewed certification motion. They offered a new statistical analysis that averaged compensation across job titles, thereby eliminating
the wide variations in pay reflecting individualized factors. The court certified the
class, concluding that documentary evidence of individual compensation decisions
“paints a picture … that suggests that common proof could be used to demonstrate
the impact of defendants’ actions on technical class members.” ER834(31) (emphases added). In light of this anecdotal evidence, the court found the deeply
flawed statistical evidence was of “diminished” importance. ER872(69).
Intuit, Lucasfilm, and Pixar have settled with plaintiffs.
ARGUMENT
Interlocutory review of a class certification order is warranted if “the certification decision presents an unsettled and fundamental issue of law relating to class
actions, important both to the specific litigation and generally, that is likely to
evade end-of-the-case review,” or the decision is “manifestly erroneous.” Cham-

8
790064

berlan v. Ford Motor Co., 402 F.3d 952, 959 (9th Cir. 2005) (per curiam). As explained below, those criteria are met here. A class certification order “premised on
legal error” is an abuse of discretion. Hawkins v. Comparet-Cassani, 251 F.3d
1230, 1237 (9th Cir. 2001). The certification order here—which, like other applications of Rule 23, must be rigorously scrutinized (Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 612-19 (1997))—rests on several fundamental errors.
A.

The District Court Improperly Relied On Anecdotes And Averaging Rather Than Requiring A Method Of Common Proof of Antitrust Impact
And Damages.
Rule 23(b)(3), which must be “satisf[ied] through evidentiary proof,” Com-

cast, 133 S. Ct. at 1432, imposes “stringent requirements for certification that in
practice exclude most claims.” Am. Express Co. v. Italian Colors Rest., 133 S. Ct.
2304, 2310 (2013). The predominance requirement of Rule 23(b)(3) is a “vital
prescription” for “assur[ing] the class cohesion that legitimizes representative action in the first place.” Amchem, 521 U.S. at 623. Thus, district courts have a “duty to take a ‘close look’ at whether common questions predominate over individual
ones.” Comcast, 133 S. Ct. at 1432.
The district court recognized the need for further appellate guidance, noting
that “the legal standards with respect to predominance are not altogether clear.”
ER822(19). Because the certification order suggests district courts in this Circuit
remain unclear about the practical application of the Supreme Court’s Comcast and

9
790064

Dukes decisions, and because that order cannot be reconciled with those decisions
and rulings of the other courts of appeals, this Court should grant review.
1.

The district court applied a Rule 23(b)(3) predominance standard that

is less stringent than the standard governing the commonality requirement under
Rule 23(a)(2), finding that a question could be “common” for predominance purposes even if it could not generate a common answer.3 ER825, 827, 887(22, 24,
84). But a question cannot be common within the meaning of Rule 23(a)(2) unless
the question can generate an answer common to the class. See Dukes, 131 S. Ct. at
2550-51. And the Supreme Court has repeatedly instructed that predominance is a
“far more demanding” inquiry than commonality. Amchem, 521 U.S. at 623–24;
Comcast, 133 S. Ct. at 1432. The district court gave the words “questions of law
or fact common to the class” less weight under Rule 23(b)(3) than Rule 23(a)(2),
contrary to Comcast.
The district court also appeared to believe that a class could be certified so
long as the alleged agreement alone could be determined in common, because it
would be so “central” at trial. ER886(83). But “[t]he main concern of the predom3

The district court relied (ER824-26(21-23)) in substantial part on two decisions
holding that a purportedly common issue as to the nature of an alleged product defect satisfied the predominance requirement—despite intensely individualized issues as to the manifestation of the defect, damages, and other issues. See Butler v.
Sears, Roebuck & Co., 727 F.3d 796 (7th Cir. 2013), pet for cert. filed (U.S. Oct. 7,
2013); In re Whirlpool Corp. Front–Loading Washer Prods. Liab. Litig., 722 F.3d
838 (6th Cir. 2013), pet. for cert. filed (U.S. Oct. 7, 2013).

10
790064

inance inquiry under Rule 23(b)(3) is the balance between individual and common
issues.” Wang v. Chinese Daily News, Inc., — F.3d —, 2013 WL 4712728, at *5
(9th Cir. 2013) (internal quotation omitted). The district court abused its discretion
in relying on the commonality of the violation issue “to the near exclusion of other
factors relevant to the predominance inquiry.” Id. (internal quotation omitted).
Indeed, antitrust impact, or “injury in fact,” is an essential element of an antitrust claim, and the ability to prove antitrust impact in common is a prerequisite
to predominance in an antitrust case: “[c]ommon questions of fact cannot predominate where there exists no reliable means of proving classwide injury in fact.” Rail
Freight, 725 F.3d at 252-53. Even if liability issues were common, individualized
issues of antitrust impact and damages would “inevitably overwhelm questions
common to the class.” Comcast, 133 S. Ct. at 1433.
Establishing a common injury through classwide proof for a broad employee class may be difficult or impossible: employees’ circumstances tend to be individualized, as this Court recently recognized in remanding a class certification order for more rigorous consideration of “potentially significant differences among
the class members” of a 200-employee, single-employer class. Wang, 2013 WL
4712728, at *3.4 Yet the district court did not require plaintiffs to show how they

4

Substantial variation in among employees has repeatedly precluded certification
even of more cohesive putative classes in wage-suppression antitrust cases. See
Weisfeld v. Sun Chemical Corp., 210 F.R.D. 136 (D.N.J. 2002), aff’d, 84 F. App’x
11
790064

could prove with classwide evidence that each member of the certified class “suffered the same injury.” Dukes, 131 S. Ct. at 2551 (quoting Gen. Tel. Co. of Sw. v.
Falcon, 457 U.S. 147, 157 (1982)). Without a common injury, class proceedings
cannot “‘drive the resolution of the litigation’”; they merely necessitate future litigation to answer the individualized injury questions that inevitably remain. Id.
(quoting Richard A. Nagareda, Class Certification in the Age of Aggregate Proof,
84 N.Y.U. L. Rev. 97, 132 (2009)).
2.

The district court allowed plaintiffs to attempt to show predominance

for a sprawling putative class largely through the use of isolated anecdotal evidence, coupled with meaningless aggregated and averaged statistical analyses. In
Dukes, the plaintiffs attempted to prove commonality for a sprawling class of employees through a combination of isolated anecdotes, meaningless statistics, and a
novel theory of liability. See 131 S. Ct. at 2555. The Court flatly rejected that effort, holding that plaintiffs’ evidence could not prove there was “a common answer
to the crucial question” why each class member was “disfavored.” Id. at 2552.
An average answer is not a common answer applicable to each individual
class member. An average reflects the same impact and the same damages for top
performers, average performers, people who were promoted or fired, and incoming
257 (3d Cir. 2004); Reed v. Advocate Health Care, 268 F.R.D. 573 (N.D. Ill.
2009); Fleischman v. Albany Med. Ctr., 2008 WL 2945993 (N.D.N.Y. July 28,
2008); In re Comp. of Managerial, Prof’l, & Technical Emps. Antitrust Litig., 2003
WL 26115698 (D.N.J. May 27, 2003).
12
790064

new hires who found the defendants’ actual compensation sufficient to entice them
to move. Yet the district court dismissed “the importance of [] statistical models . . . in light of the extensive documentary evidence.” ER872(69). But a few
anecdotes relevant to the impact of the agreements on a few employees among tens
of thousands “prove nothing at all.” Dukes, 131 S. Ct. at 2556 n.9.
In fact, none of the “extensive documentary evidence” cited by the district
court addresses, much less resolves, the crucial question whether plaintiffs can establish classwide impact by common evidence.5 Much of the evidence the court
cited relates to the existence of the alleged no-cold-call agreements, which is irrelevant to impact. And as to impact, the district court cited unremarkable evidence
that defendants generally paid employees within base-salary ranges and tried to
compensate similarly performing employees similarly under “internal equity” policies. ER887(84). That evidence does not address plaintiffs’ contention that
changes to some employees’ compensation would cause changes to the pay of others in the same job title, let alone across very different seniority levels or job functions at different companies. And those anecdotes plainly provide no means of
proving that purported ripple effect with common evidence.

5

If anything, the anecdotes offered by plaintiffs disprove common impact. For
example, that Adobe decided to give a pay raise to a “‘star performer’” who could
“‘easily get a great job elsewhere’” (ER841(38)) says nothing about whether the
effect on that employee was accompanied by raises to any other employee, let
alone the entire Adobe workforce.
13
790064

And, to the extent the district court did credit plaintiffs’ expert statistical evidence, the court failed to carefully scrutinize that evidence to determine whether it
could establish that all class members suffered the “same injury.” Dukes, 131 S.
Ct. at 2551. The district court acknowledged that it had “concerns about the probativeness” of at least one of the statistical models it relied on to find predominance,
but found that the evidence was not “so methodologically flawed as to warrant exclusion.” ER872(69) (citations omitted). Mere admissibility is not enough to satisfy Rule 23. Ellis, 657 F.3d at 982 (limiting Rule 23 scrutiny “to a determination
of whether Plaintiffs’ evidence on that point was admissible” is “error.”); see also
In re Hydrogen Peroxide Antitrust Litig., 552 F.3d 305, 315 n.13, 323 (3d Cir.
2008).
Rather, before certifying a class, the district court must resolve evidentiary
disputes—including challenges to expert opinion testimony—and determine
whether plaintiffs’ evidence, in fact, establishes predominance. Dukes, 131 S. Ct.
at 2552 n.6; Comcast, 133 S. Ct. at 1432-33. “Rule 23 not only authorizes a hard
look at the soundness of statistical models that purport to show predominance—the
rule commands it.” Rail Freight, 725 F.3d at 255.
Had the district court taken the requisite “hard look” at plaintiffs’ statistics,
it could not have found that they could establish that the no-cold-call agreements
commonly injured all class members. Plaintiffs first purported to show aggregate

14
790064

harm to the class through their “conduct regression” by improperly aggregating all
defendants’ compensation data, but when the same model is run defendant by defendant, it falls apart, showing overcompensation by various defendants. ER101012(206-08). Plaintiffs next sought to show such a common injury by proving that
compensation was rigidly interlinked across job titles and localities, meaning that
the suppression of one class member’s salary would result in a “ripple” effect that
suppressed the salaries of all class members. See ER1149(345).
Accordingly, plaintiffs’ statistical models did not try to prove that individual
compensation was linked; rather, their analyses rely on averaging that masks, rather than accounts for, differences among individuals. The district court thus permitted plaintiffs to assume away the key predominance questions they were required to prove. Plaintiffs’ expert Dr. Edward Leamer admitted he used averages
“because the individual data is likely to be dominated by forces that operate at the
individual level.” ER1151(347). The Supreme Court has held this type of statistical analysis does not establish common injury. Dukes, 131 S. Ct. at 2555.
Indeed, Dr. Leamer admitted that “‘the inherent noise in the individual level
data tends to drown out the signal of the internal pay structures [Plaintiffs] are trying to detect.’” ER874(71) (alteration in original). And the district court acknowledged that plaintiffs’ method “may have masked some of the individual variations
within each job title”; but the court then accepted the tautology that plaintiffs need-

15
790064

ed to use averages to prove what they were trying to prove. Id. This was manifest
error. As Judge Alsup has recognized, plaintiffs have the “burden to show that individual differences … could be accounted for, not that individual differences
could be ignored.” In re Graphics Processing Units Antitrust Litig., 253 F.R.D.
478, 494 (N.D. Cal. 2008) (emphasis omitted); see also generally Dukes, 131 S.
Ct. at 2550-51.
The actual data show there is no rigid salary linkage even for employees
within a given job title at a single defendant, much less across job titles or companies. Even within job titles, the raw data clearly showed that compensation of employees trended in opposite directions in any given year. The chart below is one
example from Intuit, (ER1217(461)), but defendants calculated compensation distributions for each job title at issue (ER1250(472)):

These data conclusively disprove plaintiffs’ theory of a “rigid” wage structure.
3.

Plaintiffs’ failure to prove that there is a common question as to im-

pact is compounded by their failure to show that “damages are capable of meas-

16
790064

urement on a classwide basis.” Comcast, 133 S. Ct. at 1432-33. As the Supreme
Court held in Comcast, if the plaintiff provides no valid method for accurately calculating each class member’s damages, “[q]uestions of individual damage calculations will inevitably overwhelm questions common to the class.” Id. at 1433. District courts must take a “close look” at a damages model before certifying a class, a
“close look” that models relying on “arbitrary” methods cannot survive. Id. at
1432. In contrast with Leyva v. Medline Industries, Inc., 716 F.3d 510 (9th Cir.
2013), plaintiffs here offered no method that “would enable the court to accurately
calculate damages … for each claim.” Id. at 514. Instead, the district court accepted a damages methodology that improperly aggregated all defendants’ compensation data to estimate classwide damages, then used formulaic averages that
assumed harm to all individuals. ER1007-12(203-08). That “‘rough justice’” approach (Dukes v. Wal-Mart Stores, Inc., 222 F.R.D. 137, 177 (N.D. Cal. 2004),
aff’d in part 603 F.3d 571 (9th Cir. 2010) (en banc), rev’d 131 S. Ct. 2541 (2011))
ignored dozens of variables that affect the actual economic impact of the alleged
agreements on any individual employee’s salary, and conflicts with Comcast.
B.

In Violation of the Rules Enabling Act and Due Process, The Certification Order Abridges Defendants’ Substantive Rights To Present Individualized Defenses And Permits Uninjured Parties To Recover.
The district court further fundamentally erred in holding that, because plain-

tiffs invoked Rule 23(b)(3)—not (b)(2)—it did not need to consider whether certi-

17
790064

fication would abridge the defendants’ substantive rights in violation of the Rules
Enabling Act. ER889(86). The district court’s order does exactly that, preventing
defendants from showing that particular class members were not injured at all, or
that they were damaged by a less-than-average amount. By contrast, permitting
plaintiffs to show impact and damages based on averages would permit uninjured
parties to recover compensation they could never recover individually.
A class action is “a procedural right only,” Deposit Guar. Nat’l Bank v.
Roper, 445 U.S. 326, 332 (1980), which “leaves the parties’ legal rights and duties
intact and the rules of decision unchanged.” Shady Grove Orthopedic Assocs.,
P.A. v. Allstate Ins. Co., 559 U.S. 393, 408 (2010) (plurality opinion). Because the
class action is a procedural device, “Rule 23’s requirements must be interpreted in
keeping with Article III constraints, and with the Rules Enabling Act, which instructs that rules of procedure ‘shall not abridge, enlarge or modify any substantive
right[.]’” Amchem, 521 U.S. at 613 (quoting 28 U.S.C. § 2072(b)). For these reasons, as the Third Circuit recently explained, “[a] defendant in a class action has a
due process right to raise individual challenges and defenses to claims, and a class
action cannot be certified in a way that eviscerates this right[.]” Carrera v. Bayer
Corp., 727 F.3d 300, 307 (3d Cir. 2013) (citing Dukes, 131 S. Ct. at 2561).
There is no way to square the district court’s interpretation of Rule 23 with
the Rules Enabling Act. The district court misconstrued Dukes as holding that the

18
790064

Rules Enabling Act acts as a substantive constraint to certification of only a Rule
23(b)(2) class, with no effect on certification of (b)(1), (b)(3), or (c)(4) classes.
See ER889(86). But the Supreme Court has repeatedly made clear that the general
principles limiting Rule 23 apply to all class actions—no matter which subsection
is involved. See Ortiz v. Fibreboard Corp., 527 U.S. 815, 845 (1999). “Rule 23
provides a one-size-fits-all formula for deciding the class-action question,” Shady
Grove, 559 U.S. at 399, and a Rule 23(b)(3) analysis “turns on the straightforward
application of class-certification principles.” Comcast, 133 S. Ct. at 1433.
The district court manifestly erred by openly “interpreting Rule 23 to
‘abridge, enlarge or modify any substantive right.’” Dukes, 131 S. Ct. at 2561
(quoting 28 U.S.C. § 2072(b)). Due process entitles defendants to “litigate the issues raised” (United States v. Armour & Co., 402 U.S. 673, 682 (1971)), which includes the opportunity to “present every available defense.” Lindsey v. Normet,
405 U.S. 56, 66 (1972); accord Carrera, 727 F.3d at 307; McLaughlin v. Am. Tobacco Co., 522 F.3d 215, 231–32 (2d Cir. 2008). Here, defendants are entitled to
present evidence that class members were not injured (or were damaged less)
through evidence of class members’ individual circumstances, such as tenure, skill
set, and job performance.
Moreover, the district court wholly ignored the critical due-process question
of how classwide damages could be accurately calculated. As a classwide meas-

19
790064

ure of damages, the court accepted Dr. Leamer’s “formulaic method [for] quantifying the amount of suppressed compensation suffered by each class member.”
ER855(36). But “[r]oughly estimating the gross damages to the class as a whole
and only subsequently allowing for the processing of individual claims would inevitably alter defendants’ substantive right to pay damages reflective of their actual
liability,” McLaughlin, 522 F.3d at 231 (collecting cases). This is exactly what Dr.
Leamer did here. This Court agrees that “allowing gross damages by treating unsubstantiated claims of class members collectively significantly alters substantive
rights under the antitrust statutes,” in violation of the Rules Enabling Act. In re
Hotel Tel. Charges, 500 F.2d 86, 90 (9th Cir. 1974). The district court thus endorsed the “novel project” unanimously disapproved in Dukes: a resort to “Trial
by Formula” to calculate an “entire class recovery[] without further individualized
proceedings[.]” 131 S. Ct. at 2561. That approach also deepened a conflict among
this Circuit’s district courts.6
This Court should correct the district court’s fundamental error.
CONCLUSION
The petition should be granted and the class certification order reversed.
6

See Brown v. Wal-Mart Stores, Inc., No. 5:09-CV-03339 EJD, 2012 WL
5818300, at *3 (N.D. Cal. Nov. 15, 2012) (after Dukes, “the district courts of the
Ninth Circuit have split on the issue of utilizing statistical sampling” and collecting
cases). This Court withdrew an opinion resolving the issue. See Wang v. Chinese
Daily News, Inc., 709 F.3d 829, 836 (9th Cir.), superseded. 2013 WL 4712728, at
*6 (9th Cir. 2013).
20
790064

Dated: November 7, 2013

KEKER & VAN NEST LLP
By:

/s/ Robert A. Van Nest
Robert A. Van Nest
Daniel Purcell
Eugene M. Paige
Justina Sessions
633 Battery Street
San Francisco, CA 94111
Telephone: [phone]
Facsimile: [phone]
Edward D. Johnson
Lee H. Rubin
Donald M. Falk
MAYER BROWN LLP
Two Palo Alto Square
3000 El Camino Real, Suite 300
Palo Alto, CA 94306-2112
Telephone: [phone]
Facsimile: [phone]
Attorneys for Defendant and Petitioner
GOOGLE INC.

Dated: November 7, 2013

O’MELVENY & MYERS LLP

By: /s/ Michael F. Tubach
Michael F. Tubach
George Riley
Christina J. Brown
Two Embarcadero Center, 28th Floor
San Francisco, CA 94111
Telephone: [phone]
Facsimile: [phone]
Attorneys For Defendant and Petitioner APPLE INC.

21
790064

Dated: November 7, 2013

JONES DAY

By: /s/ David C. Kiernan
David C. Kiernan
Robert A. Mittelstaedt
Craig A. Waldman
555 California Street, 26th Floor
San Francisco, CA 94104
Telephone: [phone]
Facsimile: [phone]
Attorneys for Defendant and Petitioner ADOBE SYSTEMS, INC.

Dated: November 7, 2013

MUNGER TOLLES & OLSON, LLP

By: /s/ Gregory P. Stone
Gregory P. Stone
Bradley S. Phillips
Gregory M. Sergi
John P. Mittelbach
355 South Grand Avenue, 35th Floor
Los Angeles, CA 90071-1560
Telephone: [phone]
Facsimile: [phone]
Attorneys for Defendant and Petitioner INTEL
CORPORATION.
ATTESTATION: The filer attests that concurrence in the filing of this document
has been obtained from all signatories.

22
790064