Unites States of America v. Apple, Inc. et al

Filing 454

OPINION AND ORDER: Plaintiffs October 11, 2013 motion for class certification is granted; Apples November 15, 2013 motion to exclude Nolls opinions is denied; and plaintiffs January 27, 2014 request to strike portions of Apples sur-reply memorandum of law in opposition to plaintiffs motion for class certification and the sur-reply declarations of Kalt and Orszag is granted in part. (Signed by Judge Denise L. Cote on 3/28/2014) Filed In Associated Cases: 1:11-md-02293-DLC et al.(gr)

UNITED STATES DISTRICT COURT
SOUTHERN DISTRICT OF NEW YORK
--------------------------------------IN RE: ELECTRONIC BOOKS ANTITRUST
LITIGATION
--------------------------------------APPEARANCES:
For class plaintiffs:
Steve W. Berman
George W. Sampson
Sean Matt
Hagens Berman Sobol Shapiro LLP
1918 Eighth Avenue, Suite 3300
Seattle, WA 98101
Jeff D. Friedman
Shana Scarlett
Hagens Berman Sobol Shapiro LLP
715 Hearst Avenue, Suite 202
Berkeley, CA 94710
Kit A. Pierson
Emmy L. Levens
Jeffrey B. Dubner
Cohen Milstein Sellers & Toll PLLC
1100 New York Avenue, N.W.
South Tower, Suite 500
Washington, DC 20005
Douglas Richards
Cohen Milsten Sellers & Toll PLLC
88 Pine Street, 14th Floor
New York, NY 10005
For defendant Apple Inc.:
Theodore J. Boutrous, Jr.
Daniel G. Swanson
Gibson, Dunn & Crutcher, LLP
333 South Grand Ave.

X
:
:
:
:
:
:
X

11 MD 2293 (DLC)
Related to all
matters
OPINION & ORDER

Los Angeles, CA 90071
Cynthia Richman
Gibson, Dunn & Crutcher, LLP
1050 Connecticut Avenue, N.W.
Washington, DC 20036
Howard E. Heiss
Edward Moss
O’Melveny & Myers LLP
1625 Eye Street, NW
Washington, DC 20006
DENISE COTE, District Judge:
After a bench trial in two closely related cases, defendant
Apple Inc. (“Apple”) was found to have colluded with five major
publishers to fix e-book prices, violating Section 1 of the
Sherman Antitrust Act, 15 U.S.C. § 1 (“Sherman Act”).
Plaintiffs in the instant suit now move for class certification
in their action against Apple based on the same conduct.
is a paradigmatic antitrust class action.

This

Virtually all class

members paid inflated prices for e-books as a result of a
centralized price-fixing conspiracy, and they have proffered a
sophisticated damages model to reliably determine damages.

If

certification were not appropriate here, no antitrust class
action could be certified.

For the reasons set out below,

plaintiffs’ motion is granted.
The parties have also moved to exclude the opinions
rendered by each others’ experts.

Plaintiffs move for the

exclusion of the opinions of Apple’s experts, Dr. Joseph Kalt
2

and Mr. Jonathan Orszag.

Because Kalt’s and Orszag’s opinions

would not prevent class certification even if they were
admissible, this Opinion does not decide plaintiffs’ motions to
exclude. 1

Apple’s motion to exclude the opinions of plaintiffs’

expert, Dr. Roger Noll, is denied in this Opinion.
Finally, plaintiffs move to strike portions of Apple’s surreply opposition to the motion for class certification.
Plaintiffs’ motion to strike is granted in part, as described
below.
BACKGROUND
On April 11, 2012, the United States of America (“DOJ”) and
sixteen states filed two antitrust lawsuits alleging that Apple
and five book publishing companies conspired to raise and fix ebook prices in violation of Section 1 of the Sherman Act.
United States v. Apple Inc., 12 Civ. 2826 (S.D.N.Y.) (“DOJ
Action”); State of Texas v. Penguin Grp. (USA) Inc., 12 Civ.
3394 (S.D.N.Y.) (“States’ Action”). 2

The five publishers are

Hachette Book Group, Inc. (“Hachette”), HarperCollins Publishers
LLC (“HarperCollins”), Holtzbrinck Publishers LLC d/b/a
Macmillan (“Macmillan”), Penguin Group (USA), Inc. (“Penguin”),
and Simon & Schuster, Inc. (“Simon & Shuster”) (collectively,
A separate Opinion issued today rules on plaintiffs’ motions to
exclude Apple’s experts.
1

Today, thirty-three states and U.S. territories (the “States”)
are plaintiffs in the States’ Action.
2

3

“Publisher Defendants”).

The Publisher Defendants settled with

the DOJ and the States; Apple alone went to trial.
A bench trial was held in these two actions from June 3 to
20, 2013 to determine liability.

A July 10 Opinion found, inter

alia, that Apple had committed a per se violation of the Sherman
Act.

952 F. Supp. 2d 639, 694 (S.D.N.Y. 2013) (the “Liability

Opinion”).

In particular, the Opinion found that Apple played a

central role in orchestrating a conspiracy among the Publisher
Defendants to raise e-book prices.

Id. at 647.

Even before the DOJ and the States sued Apple, class
actions were filed alleging the same violation of the Sherman
Act.

Following the appointment of lead counsel for the class, a

consolidated amended complaint was filed on January 20, 2012.
While fact discovery in all the actions had concluded before the
June 2013 trial on liability, expert discovery on damages in the
class action and the States’ Action was concluded after that
trial. 3

On October 11, 2013, class plaintiffs moved for

certification of a class in advance of a damages trial to be
held later this year.
The class plaintiffs expect to rely on the doctrine of
collateral estoppel and do not intend to retry Apple’s liability

The DOJ Action only sought declaratory and injunctive relief.
Judgment was entered in the DOJ action on September 5, 2013, and
an appeal is pending.
3

4

for violating the antitrust laws. 4

In arguing against

certification of a class here, Apple does not contend that any
dispute over its liability precludes certification.

To

understand the parties’ arguments concerning certification,
however, it is necessary to understand the history and context
of the price-fixing scheme.

Accordingly, although familiarity

with the Liability Opinion is assumed, findings relevant to the
resolution of the motions addressed in this Opinion are set out
below. 5
I.

Publishers’ Discontent with the $9.99 Price Point
The background to this conspiracy begins with Amazon’s

introduction of the first e-reader to gain widespread commercial
acceptance.

When Amazon’s Kindle was launched in 2007, Amazon

quickly became the market leader in the sale of e-books and ebook readers.

Through 2009, Amazon dominated the e-book retail

market, selling nearly 90% of all e-books.

Id. at 648-49.

Amazon utilized a discount pricing strategy through which
it charged $9.99 for e-book versions of certain newly released
hardcover books (“New Releases”) and New York Times bestselling
books (“NYT Bestsellers”).

Amazon was staunchly committed to

A pending motion for summary judgment is addressed to issues of
collateral estoppel.
4

To the extent that this recitation of the Liability Opinion’s
findings differs in any way from the findings as set out in the
Liability Opinion itself, the Liability Opinion controls.
5

5

its $9.99 price point and believed it would have long-term
benefits for its consumers.

In order to compete with Amazon,

other e-book retailers also adopted a $9.99 or lower retail
price for many e-book titles.

Id. at 649.

The “Big Six” of United States publishing -- the Publisher
Defendants and Random House (collectively, the “Publishers”) -were unhappy with Amazon’s $9.99 price point.

They determined

that they needed to force Amazon to abandon its discount pricing
model.

Id. at 649-50.

The Publisher Defendants did not believe, however, that any
one of them acting alone could convince Amazon to change its
pricing policy.

They also feared that if they did not act as a

group, Amazon would use its ever-growing power in the book
distribution business to retaliate against them.

As a result,

the Publisher Defendants conferred about their need to act
collectively if they were to have any impact on Amazon’s
pricing.

Id. at 650.

Beginning in at least early 2009, the Publisher Defendants
began testing ways to get Amazon to move off its $9.99 price
point.

One of the strategies that they adopted in 2009 to

combat Amazon’s $9.99 pricing was the delayed release or
“withholding” of the e-book versions of New Releases, a practice
that was also called “windowing.”

By the end of 2009, four of

the Publisher Defendants -- Macmillan, Simon & Schuster,
6

Hachette, and HarperCollins -- had announced or implemented a
policy of windowing some of their most popular e-book titles
sold on Amazon.

By making the more expensive hardcover version

available to the public before the lower priced e-book, the
Publisher Defendants hoped to protect the sales of New Release
hardcover books and to pressure Amazon to raise its e-book
prices.

Id. at 651-52.

Even though by the Winter of 2009 four of the Publisher
Defendants had delayed the release of some e-books or announced
an intention to so, they knew that windowing was not a long-term
solution to Amazon’s $9.99 pricing model.

It was in this

context that Apple arrived on the scene and provided the
Publisher Defendants with the means to achieve their shared
goal.
II.

Id. at 653-54.

Apple and the Agency Model
In 2009, Apple was close to unveiling the iPad.

With this

revolutionary tablet, Apple was able to contemplate the arrival
of its first great device for reading e-books.

Therefore, under

the direction of Apple’s Eddy Cue (“Cue”), Senior Vice President
of Internet Software and Services, Apple began studying the
e-book industry.

Id. at 654.

By November 2009, Apple had concluded that selling e-books
as individual apps was “flawed,” and Apple’s founder Steve Jobs
(“Jobs”) authorized Cue to pursue the development of a dedicated
7

Apple e-bookstore (the “iBookstore”) for the iPad.

Apple

planned to demonstrate the iPad to the public at its launch on
January 27, 2010 (the “Launch”), and to ship the devices to
stores in early April 2010.

Even though the iPad Launch would

happen with or without an iBookstore, Apple did hope to announce
its new iBookstore at the Launch.

This left Cue with less than

two months for Apple to acquire enough content to create a
viable Apple e-bookstore, and that period included the Christmas
and New Year holidays.

As a result, Apple streamlined its

efforts and concentrated on executing agreements with the Big
Six Publishers for e-books.

Id. at 654-55.

Cue met with the Publishers, separately, on December 15 and
16, 2009.

Id. at 656.

Hachette and later HarperCollins

surprised Apple with their suggestion that, instead of a
wholesale model, Apple adopt an agency model for the
distribution of e-books. 6

Id. at 657.

Days later, Apple decided to embrace the agency model, and
settled on an agency model with a 30% commission, the same
commission it was using in its App Store.

Apple realized,

however, that in handing over pricing decisions to the

Under a wholesale model, a retailer such as Amazon purchases
e-books from publishers and resells them to consumers at a price
Amazon sets. As an agent, Amazon does not set retail prices for
e-books; the publishers sell the e-books through their agent at
a price that the publishers set.
6

8

Publishers, it needed to restrain their desire to raise e-book
prices sky high.

It decided to restrain retail prices through

the use of pricing tiers with caps.

While Apple was willing to

raise retail e-book prices by as much as 50% over Amazon’s
$9.99, it did not want to be embarrassed by what it considered
unrealistically high prices.

Id. at 658-59.

Apple realized that if it moved to an agency model with the
Publishers, Apple would be at a competitive disadvantage so long
as Amazon remained on the wholesale model and could price New
Releases and NYT Bestsellers at $9.99, or even lower, to compete
with Apple.

Since it was inevitable that the Publishers would

raise e-book prices when given the opportunity -- indeed, Apple
expected the Publishers to raise the prices to the tier caps -e-books priced at $9.99 by Amazon would doom the iBookstore.
Id. at 659.
To ensure that the iBookstore would be competitive at
higher prices, Apple concluded that it needed to eliminate all
retail price competition.

Thus, the final component of its plan

was to require the Publishers to move all of their e-retailers
(“e-tailers”) to the agency model.
Things moved quickly.

Id.

The week following Apple’s first

meetings with the Publishers, Cue met with key executives from
Simon & Schuster, Macmillan, and Random House.

Id. at 659-60.

On January 4 and 5, Cue wrote six essentially identical emails
9

to the Publishers, describing the key components of Apple’s
proposed agency model.

Id. at 661-62.

It was as apparent to

the Publishers as it was to Apple that Apple’s proposal would
only allow the Publishers to raise the consumer prices for
e-book versions of their key titles above Amazon’s $9.99 price
point to the proposed caps if they moved Amazon and their other
e-tailers to an agency arrangement.

Id.

On January 11, Apple sent its proposed eBook Agency
Distribution Agreement (“Draft Agreement”) to each of the
Publishers.

The Draft Agreement included a “Most Favored

Nation” provision (“MFN”) that guaranteed that the e-books in
Apple’s e-bookstore would be sold for the lowest retail price
available in the market.

The MFN eliminated any risk that Apple

would ever have to compete on price when selling e-books, while
as a practical matter forcing the Publishers to adopt the agency
model across the board with their e-tailers.

Id. at 662-63.

The final agency agreements (the “Agreements”) included an MFN.
Id. at 666.
In the two intervening weeks before the Launch, Apple and
the Publishers engaged in intensive negotiations.

Id. at 664.

The Publisher Defendants fought hardest to raise the price caps.
They and Apple knew that these negotiations were really about
setting the new industry prices for e-books.

Id. at 667.

debate over price caps essentially ended on January 16.
10

The
Id. at

668.

Apple decreased the hardcover list price triggers for the

$12.99 and $14.99 e-book caps, but carved out NYT Bestsellers
for special treatment.

Id. at 669.

Except for small exceptions

which were immaterial to Apple, this pricing proposal was the
one finally adopted in the Agreements.

Id.

As of January 16, the Launch was just eleven days away and
Cue did not have a single agreement executed.

Id. at 670.

By

January 26, the day before the Launch, Apple had executed its
fifth Agreement.

Id.

The only Publisher to decline to sign the

Agreement was Random House.

Id. at 677.

In separate

conversations on January 20 and over the next few days, the
Publisher Defendants all told Amazon that they wanted to change
to an agency distribution model with Amazon.

Id. at 672.

Thus, in less than two months, Apple had signed agency
contracts with five of the six Publishers, and those Publisher
Defendants had agreed with each other and Apple to solve the
“Amazon issue” and eliminate retail price competition for ebooks.

The Publisher Defendants would move as one, first to

force Amazon to relinquish control of pricing, and then, when
the iBookstore went live, to raise the retail prices of e-book
versions of New Releases and NYT Bestsellers to the caps set by
Apple.

This would not have happened without Apple’s ingenuity

and persistence.

Id. at 677-78.

11

On January 27, Jobs launched the iPad, introducing both the
iPad’s e-reader capability and the iBookstore.

Id. at 678.

On

January 28, John Sargent, CEO of Macmillan, met with Amazon and
advised that Amazon had just two options: either (1) move to an
agency arrangement, or (2) not receive Macmillan’s Kindle
versions of New Releases for seven months.

Seven months was no

random period -- it was the number of months for which titles
were designated New Release titles under the Agreement and
restrained by the Apple price caps and MFN.
roughly twenty minutes.

The meeting lasted

Amazon let Macmillan know in blunt

terms that it was unhappy.

Id. at 679.

Macmillan had anticipated that Amazon might retaliate
against it by removing the “buy buttons” on the Amazon site that
allow customers to purchase books from Amazon’s online store or
from the Kindle.

The evening of Thursday, January 28, Amazon

removed the buy buttons for both print and Kindle versions of
Macmillan titles.

Id.

Over the weekend, it became obvious to Amazon that its
strategy had failed.

Amazon knew that its battle was not just

with Macmillan but with five of the Big Six.

Amazon announced

on its website on Sunday, January 31, that it would “capitulate
and accept” Macmillan’s agency terms.

Id. at 680-81.

With help

from Apple, Macmillan negotiated an agency agreement with
Amazon, which was signed that Friday, February 5.
12

Id. at 681.

In light of the Publisher Defendants’ overlapping threats to
remove content from Amazon’s platform if it did not move to
agency in early April, when the iPad became available, Amazon
moved quickly to execute agency agreements with the remaining
Publisher Defendants.

Id.

By the end of March 2010, Amazon had completed agency
agreements with Macmillan, HarperCollins, Hachette, and Simon &
Schuster.

Because of circumstances that were unique to Penguin

and its reseller contract, its agency agreement with Amazon was
the last to be executed.

Penguin signed its agency contract

with Amazon on June 2, 2010, but before that date, Penguin had
refused to allow Amazon to sell Penguin’s new e-books.

Id. at

682.
III. Prices After Agency
Just as Apple expected, after the iBookstore opened in
April 2010, the price caps in the Agreements became the new
retail prices for the Publisher Defendants’ e-books.

In the

five months that followed, the Publisher Defendants collectively
priced 85.7% of their New Release titles sold through Amazon and
92.1% of their New Release titles sold through Apple within 1%
of the price caps.

This was also true for 99.4% of the NYT

Bestseller titles on Apple’s iBookstore, and 96.8% of NYT
Bestsellers sold through Amazon.

The increases at Amazon within

roughly two weeks of moving to agency amounted to an average per
13

unit e-book retail price increase of 14.2% for their New
Releases, 42.7% for their NYT Bestsellers, and 18.6% across all
of the Publisher Defendants’ e-books.

Id.

The following chart, prepared by one of Apple’s experts for
the liability trial, illustrates this sudden and uniform price
increase.

While the average retail prices for Random House’s

e-books hovered steadily around $8, for four of the Publisher
Defendants, the price increases occurred at the opening of the
iBookstore; Penguin’s price increases awaited the execution of
its agency agreement with Amazon and followed within a few
weeks.

The bottom flat line represents the average prices of

non-major publishers.

Id.
The Publisher Defendants raised more than the prices of
just New Release e-books.

The prices of some of their New

14

Release hardcover books were also raised in order to move the ebook version into a correspondingly higher price tier.

And, all

of the Publisher Defendants raised the prices of their backlist
e-books, 7 which were not governed by the Agreements’ price tier
regimen.

Id. at 683.

The following two charts, one prepared by the DOJ and
States’ expert and another by an expert for Apple, compare the
price increases for the Publisher Defendants’ New Releases with
the price increases for their backlist books.

Despite drawing

from different time periods, their conclusions are very similar.
The Publisher Defendants used the change to an agency method for
distributing their e-books as an opportunity to raise the prices
for their e-books across the board.

Id.

E–Book Average Price Increases at Amazon by Publisher
Defendants Following the Move to Agency

Backlist books are books that have been on the market for more
than one year. Frontlist books are those that have been on the
market for less than one year.
7

15

E–Book Average Price Increases at Amazon by Publisher
Defendants Following the Move to Agency

Id. at 683-84.
If there were any doubt about the impact of the Apple
Agreements on e-book prices, at least in so far as the market
for trade e-books is concerned, the experience of Random House
confirms each of the observations just made about the prices and
sales of the five Publisher Defendants.

Random House adopted

the agency model in early 2011, and promptly raised the prices
of its e-books and experienced a concomitant decline in e-book
sales.
IV.

Id. at 685.

The Instant Litigation
Beginning on August 9, 2011, a number of putative class

actions were filed alleging that Apple and the Publisher
Defendants conspired to fix prices and consequently injured
those who purchased e-books from the Publisher Defendants.
e.g., 11 Civ. 3892 (N.D. Cal.); 11 Civ. 5576 (S.D.N.Y.).

See,

Actions filed outside the Southern District of New York were

16

transferred here by the United States Judicial Panel on
Multidistrict Litigation.
As noted above, the DOJ and States have also brought suits
against Apple and the Publisher Defendants alleging price-fixing
in violation of the antitrust laws.

On April 11, 2012, the DOJ

Action was filed, seeking declaratory and injunctive relief
against Apple and the Publisher Defendants.

United States v.

Apple Inc., et al., 12 Civ. 2826 (S.D.N.Y.).

The DOJ quickly

agreed to settle with Hachette, HarperCollins, and Simon &
Schuster, and the first of the Publisher Defendants began
terminating their Agreements on May 21, 2012.

Final judgment

was entered as to these three defendants on September 6, 2012.
Penguin agreed to settle with the DOJ on December 18, 2012, and
Macmillan executed a settlement agreement on February 8, 2013.
Also on April 11, 2012, the States’ Action was filed
against Apple, Macmillan, and Penguin, bringing claims parens
patriae for injunctive relief and damages.

The States’ Action

was transferred to this Court by the United States Judicial
Panel on Multidistrict Litigation.

State of Texas, et al. v.

Penguin Grp. (USA), Inc., et al., 12 Civ. 3394 (S.D.N.Y.).
In a separately filed action, all of the states of the
Union except for Minnesota, as well as the District of Columbia
and five U.S. territories and possessions, settled their claims
against Hachette, HarperCollins, and Simon & Schuster by a
17

settlement agreement executed June 11, 2012 and approved
February 8, 2013.

State of Texas, et al. v. Hachette Book Grp.,

Inc., et al., 12 Civ. 6625 (S.D.N.Y.). 8

The litigating States

and class plaintiffs entered into settlement agreements with the
remaining Publisher Defendants, Macmillan and Penguin, on April
25 and May 20, 2013, respectively.

The States’ and class

settlements were approved on December 6, 2013.
As noted above, the liability trial was held in the DOJ
Action and the States’ Action from June 3 to 20, 2013.

The July

10 Liability Opinion found that Apple had committed a per se
violation of the Sherman Act, and that the DOJ and States had
also carried their burden to show a violation under the rule of
reason test.
On October 11, 2013, plaintiffs Anthony Petru, Thomas
Friedman, and Shane S. Davis moved for class certification,
pursuant to Rules 23(a) and (b)(3), Fed. R. Civ. P, for those in
twenty-three states and U.S. territories who purchased an e-book
published by one of the Publisher Defendants after the agency
model was adopted but before May 21, 2012, the date the first
Publisher Defendants began terminating their Agreements pursuant

A class of Minnesota e-book consumers subsequently settled with
Hachette, HarperCollins, and Simon & Schuster on June 20, 2013,
through the class action.
8

18

to their settlements with government authorities.

Specifically,

plaintiffs seek to certify a class consisting of
[a]ll persons in the Non-Litigating Jurisdictions who
purchased ebooks between April 1, 2010 and May 21,
2012, published by Hachette Book Group, Inc.
(“Hachette”), HarperCollins Publishers L.L.C.
(“HarperCollins”), Holtzbrinck Publishers, LLC d/b/a
Macmillan (“Macmillan”), Penguin Group (USA) Inc.
(“Penguin”), or Simon & Schuster, Inc. (“Simon &
Schuster”) directly from that publisher (including any
of its imprints) after the adoption of the agency
model by that publisher. The “Non-Litigating
Jurisdictions” are American Samoa, California,
Florida, Georgia, Guam, Hawaii, Kentucky, Maine,
Minnesota, Mississippi, Montana, Nevada, New
Hampshire, New Jersey, North Carolina, Northern
Mariana Islands, Oklahoma, Oregon, Rhode Island, South
Carolina, U.S. Virgin Islands, Washington, and
Wyoming. Excluded from the Class are Defendants,
their employees, co-conspirators, officers, directors,
legal representatives, heirs, successors, and wholly
or partly owned subsidiaries of affiliated companies,
as well as the Honorable Denise L. Cote and persons
described in 28 U.S.C. § 455(b)(4)-(5). 9
Plaintiffs’ motion for class certification was fully submitted
on January 21, 2014.
With their reply, plaintiffs filed Noll’s reply declaration
on December 18, 2013. 10

On December 27, this Court granted Apple

By Order of December 20, 2011, the Court waived any interest it
might have in the putative class action and ordered that any
motion for class certification would define the class so as to
exclude individuals described in 28 U.S.C. § 455(b)(4)-(5).
9

On September 30, 2013, plaintiffs requested permission for
Noll to file a rebuttal report. The Court granted this request
on October 3, also ordering that “[t]o the extent the
plaintiffs’ rebuttal report includes new opinions that could not
have been anticipated by Apple’s expert(s), Apple may file a
sur-reply report.”
10

19

permission to file a sur-reply brief and sur-reply expert
reports addressing “new opinions that could not have been
anticipated by Apple’s experts.”

The only “new opinion[] that

could not have been anticipated” was Noll’s decision to use a
supercomputer to re-run his regression study using individual
transaction records, rather than four-week average prices.

That

regression study is described below.
On January 21, 2014, Apple filed its sur-reply memorandum
of law in opposition to class certification and the sur-reply
declarations of Kalt and Orszag, all of which went well beyond
the scope of Noll’s addition of a finer-grained analysis.

On

January 27, class plaintiffs requested that the Court strike
Apple’s sur-reply memorandum of law in opposition to plaintiffs’
motion for class certification and class plaintiffs and the
States jointly requested that the Court strike the sur-reply
declarations of Kalt and Orszag.

Apple responded on January 31.

Plaintiffs’ request to strike is granted as to Apple’s surreply memorandum except for those passages that properly address
Noll’s new analysis. 11

Plaintiffs’ request is also granted as to

any new analyses by Kalt that are not directly responsive to

The motion to strike is denied as to the first two paragraphs
of the memorandum and the four paragraphs beginning with the
first full paragraph on page 4.
11

20

Noll’s use of individual transaction data. 12

Kalt and Orszag

also devote a great deal of space to addressing Noll’s
criticisms of their rebuttal declarations.

These criticisms

could have been anticipated, and thus Kalt’s and Orszag’s
responses are beyond the scope of the December 27 Order.

Yet,

in the interests of fully exploring the bases for Apple’s
experts’ opinions, the Court declines to strike these responses.
Accordingly, plaintiffs’ request is denied as to the remainder
of Kalt’s sur-reply and as to Orszag’s sur-reply.
A.

Noll’s Damages Model

Plaintiffs offer Dr. Roger G. Noll’s declarations in
support of their motion for class certification.

Noll is a

Professor Emeritus of Economics at Stanford University and a
Senior Fellow in the Stanford Institute for Economic Research,
where he has served as the Director of the Program in Regulatory
Policy since 1984.

He holds a B.S. in mathematics from the

California Institute of Technology and a Ph.D. in economics from
Harvard University.

Noll’s primary field of research is

industrial organization, which includes antitrust economics, and
he has taught courses in the fields of antitrust and regulation
to undergraduate and graduate students for almost fifty years.

The motion to strike is granted as to the following portions
of Kalt’s sur-reply: paragraphs 22-29, 39-43, 58-59, 71, 76, 81,
86, and 88-91; and figures 2A-2F, 3, 8A-8B, 11A-11B, and 18.
12

21

He has published more than 300 scholarly articles, books, and
reviews, many of which concern antitrust or the information
technology sector.
He sits on the board of editors of a number of economics
journals, including the International Journal of the Economics
of Business, Journal of Risk and Uncertainty, and the Economics
of Governance.

He has served on various committees of the

National Research Council, and he is a member of the Board of
Advisors of the American Antitrust Institute.

His awards

include, most recently, the Alfred E. Kahn Distinguished Career
Award, given by the American Antitrust Institute in 2012, and a
Distinguished Member Award given by the American Economic
Association’s Transportation and Public Utilities Group in 2013.
Plaintiffs asked Noll to determine whether anticompetitive
harm arising from the conspiracy can be demonstrated for all
class members, and whether the method for calculating damages to
individual consumers is common to class members.

Noll also

calculated damages for both the class and the States.

After

considering the Liability Opinion, Noll determined that one
element of anticompetitive harm is the transfer of wealth from
consumers to sellers as a result of prices that are elevated due
to anticompetitive conduct.

Noll calculated that the

anticompetitive conduct by the defendants caused prices to be
higher for e-books that account for 99.8% of e-book sales by the
22

Publisher Defendants.

Through this analysis, he concluded that

the requirement to show class-wide anticompetitive harm had been
satisfied.
To calculate damages, Noll adopted the widely-used “before
and after” approach.

Noll and his team used prices for titles

from the Publisher Defendants before they began selling e-books
under their agency agreements, 13 as well as prices throughout
this period for titles from other publishers that had not
adopted the agency model, to calculate “competitive benchmark”
prices.

These benchmark prices -- which include Random House

sales through mid-January 2011, when Random House adopted the
agency model -- are presumed to be free of the effects of
collusion. 14
Noll controlled for a host of other factors that might
influence an e-book’s price.

To do so, Noll built a hedonic

Noll’s “competitive benchmark” includes sales of e-books from
Hachette, HarperCollins, Macmillan, and Simon & Schuster between
June 8, 2008 and April 1, 2010, and sales of e-books from
Penguin between June 8, 2008 and May 26, 2010. It is not clear
from Noll’s declaration whether Penguin sales between May 26 and
May 31 are considered pre- or post-agency.
13

Noll’s use of these benchmarks creates a conservative model
that likely underestimates damages. After all, the prices of
titles from other publishers may well have risen because of the
price-fixing of their major competitors. Thus, Noll’s
“competitive benchmark” may well include prices that were
inflated by collusion.
14

23

pricing model to separate out these effects. 15

Noll’s model

considers the following characteristics as independent variables
for each e-book title: whether the title was “frontlist” or
“backlist” when purchased (i.e., whether it was published more
than a year earlier); whether it was a “new release,” as
determined by Amazon (i.e., available for 90 days or fewer);
whether it was a NYT Bestseller; which of several genres it
belonged to; whether the title had a hardcover print edition;
whether it had a paperback edition; and which publisher offered
the title.

The model also includes, in its analysis of each

transaction, a variable for monthly personal consumption
expenditures on nondurable goods, to account for changing
amounts of disposable income; 16 a variable measuring how long the
agency model had been in place, to account for unrelated trends
affecting e-book prices; an indicator variable specific to each
title, to account for pricing effects specific to a particular

A hedonic pricing model -- “hedonic” from the Greek meaning
pleasure, as the method relates to consumers’ desires -measures the effect of various product attributes on price. See
Freeland v. AT&T Corp., 238 F.R.D. 130, 149 n.15 (S.D.N.Y.
2006).
15

This variable incorporates data from the Bureau of Economic
Analysis of the U.S. Department of Commerce.
16

24

e-book title; and a variable that reflects whether the sale was
made under the agency model. 17
To calculate the effects of these variables on an e-book’s
price, Noll ran a multiple regression analysis 18 on transaction
records for more than 149 million sales of 1.3 million different
titles.

Noll’s data was comprised of transaction records

compiled by Amazon, Barnes & Noble, Apple, Sony, Kobo, Google,
and Books-A-Million for each e-book (except textbooks) 19 sold
between June 8, 2008 and April 8, 2012 (five weeks before the
end of the damages period, May 21, 2012).

To be included in the

data set, an e-book title had to be purchased at least once
In addition, Noll includes a variable to account for the
effects of Amazon’s removal of the “buy button” from listings
for all Macmillan titles, including e-books, between January 29
and February 6, 2010.
17

“Multiple regression analysis is a statistical tool used to
understand the relationship between or among two or more
variables. [It] involves a variable to be explained -- called
the dependent variable -- and additional explanatory variables
that are thought to produce or be associated with changes in the
dependent variable.” Federal Judicial Center, Reference Manual
on Scientific Evidence, 305 (3d ed. 2011); see also LavinMcEleney v. Marist Coll., 239 F.3d 476, 482 (2d Cir. 2001)
(noting, in sex discrimination case, that “[i]t is undisputed
that multiple regression analysis . . . is a scientifically
valid statistical technique for identifying discrimination”);
Bickerstaff v. Vassar Coll., 196 F.3d 435, 448 (2d Cir. 1999)
(recognizing that “multiple regression analysis [is] a commonly
accepted method of statistical analysis for examining the effect
of independent variables on a dependent variable”).
18

Textbooks fall outside the relevant market definition: trade
e-books in the United States. Liability Opinion, 952 F. Supp.
2d at 694 n.60.
19

25

after the first Agreements went into effect on April 1, 2010.
Of the 720 possible combinations of the variables listed above, 20
all sales of the Publisher Defendants’ e-books were captured by
502 of these combinations.
After controlling for the factors listed above, Noll’s
model calculated the effect, if any, of Apple’s anticompetitive
conduct for each of these 502 combinations of the above
independent variables.
these 502 categories.

Each e-book falls into exactly one of
For instance, Noll’s model calculates

that a customer who purchased an e-book version of a Penguin
hardcover book of fiction on the NYT Bestseller list during the
agency period paid an overcharge of approximately 29.4%.
Damages calculations for each transaction are
straightforward: damages for a given sale are equal to the price
paid multiplied by the overcharge for that title’s category.

In

the example above, purchasing the Penguin NYT Bestseller for
$14.99 would result in damages of $14.99 x 29.4% = $4.41.
Subtracting these damages from the actual price reveals the butfor price -- that is, the price a consumer would have paid but
While there are 720 possible combinations of these variables
for the Publisher Defendants’ e-books, there are 1008 possible
combinations for any given e-book: seven possible publisher
values (one for each of the Big Six, and one for other), six
genre categories, two possible values for the hardcover edition
variable, two possible values for the paperback edition
variable, and then three possible values for the title’s age
(new release, other frontlist, or backlist).
20

26

for the price fixing.

In the example of the Penguin NYT

Bestseller, the but-for price is $10.58, that is, $14.99 less
$4.41.
In the first model Noll submitted to this Court, he ran his
regression analysis using the average price for each title over
a four-week period, rather than the actual transaction price, to
reduce computational complexity.

Thus, a unit of observation

was an average sale price of an e-book title for a four-week
period through a specific retailer.

After Apple’s experts

criticized him on this point, Noll employed a supercomputer to
re-run his regression using both a one-week average sale price
and the actual prices for each sales transaction.
The model estimated by Noll’s regression analysis has an
adjusted R2 of 0.90 -- that is, it explains 90% of the variance
in prices among e-book titles. 21

Using the individual sales

transaction data, Noll calculates the total damages to
consumers, both putative class members and in the States, to be
just over $280 million.

The fraction of e-book sales for which

the model finds no damages is 0.2%.

This is the adjusted R2 of Dr. Noll’s initial model, not his
later model that used individual transaction prices rather than
four-week averages. The parties have not suggested that the
adjusted R2 of the later model is substantially different,
although Apple’s experts have re-run both of Noll’s regressions
and calculated related R2 statistics for both models.
21

27

DISCUSSION
I.

Plaintiffs’ Motion for Class Certification
Apple opposes plaintiffs’ motion for class certification on

narrow grounds.

It chiefly argues that Noll’s damages model

cannot reliably determine each class member’s damages, and
consequently plaintiffs cannot meet the commonality and
predominance requirements.
A.

Legal Standard

A party seeking certification of a class must
“affirmatively demonstrate” compliance with each of the
requirements of Rule 23.

Comcast Corp. v. Behrend, --- U.S.

---, ---, 133 S. Ct. 1426, 1432 (2013) (citation omitted).
Thus, plaintiffs will be able to sue as representatives of a
class only if
(1)

the class is so numerous that joinder of all members
is impracticable,

(2)

there are questions of law or fact common to the
class,

(3)

the claims or defenses of the representative parties
are typical of the claims or defenses of the class,
and

(4)

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

Fed. R. Civ. P. 23(a); see Police & Fire Ret. Sys. of the City
of Detroit v. IndyMac MBS, Inc., 721 F.3d 95, 104 n.10 (2d Cir.
2013).

28

If the Rule 23(a) criteria are satisfied, an action may be
maintained as a class action only if it also qualifies under at
least one of the categories provided in Rule 23(b).

Fed. R.

Civ. P. 23(b); In re U.S. Foodservice Inc. Pricing Litig., 729
F.3d 108, 117 (2d Cir. 2013).
class under Rule 23(b)(3).

Plaintiffs seek to certify a

Rule 23(b)(3) permits certification

“if the questions of law or fact common to class members
predominate over any questions affecting only individual
members, and . . . a class litigation is superior to other
available methods for fairly and efficiently adjudicating the
controversy.”

Fed. R. Civ. P. 23(b)(3); In re U.S. Foodservice,

729 F.3d at 117 (citation omitted).

Among other factors, courts

are to consider:
(A)

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

(B)

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

(C)

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

(D)

the likely difficulties in managing a class action.

Fed. R. Civ. P. 23(b)(3); In re U.S. Foodservice, 729 F.3d at
130 n.15 (citation omitted).
“To certify a class, a district court must make a
definitive assessment of Rule 23 requirements, notwithstanding

29

their overlap with merits issues, must resolve material factual
disputes relevant to each Rule 23 requirement, and must find
that each requirement is established by at least a preponderance
of the evidence.”

In re U.S. Foodservice, 729 F.3d at 117

(citation omitted).

In other words, the district judge must

“receive enough evidence, by affidavits, documents, or
testimony, to be satisfied that each Rule 23 requirement has
been met.”

Teamsters Local 445 Freight Div. Pension Fund v.

Bombardier, Inc., 546 F.3d 196, 204 (2d Cir. 2008) (citation
omitted).
B.

Rule 23(a) Requirements
1.

Numerosity

Rule 23(a) requires a finding that the putative class
members are so numerous as to make joinder of each
“impracticable.”

Fed. R. Civ. P. 23(a)(1).

Numerosity is

presumed when a class consists of forty or more members.

See

Consol. Rail Corp. v. Town of Hyde Park, 47 F.3d 473, 483 (2d
Cir. 1995).
Apple does not contest numerosity and does not dispute
plaintiffs’ claim that the putative class consists of millions
of consumers.
2.

Such a class meets the numerosity requirement.
Commonality

Commonality is established where “plaintiffs’ grievances
share a common question of law or of fact.”
30

Shariar v. Smith &

Wollensky Rest. Grp., Inc., 659 F.3d 234, 252 (2d Cir. 2011)
(citation omitted).

As the Supreme Court taught in Dukes,

[w]hat matters to class certification is not the
raising of common “questions” -- even in droves -but, rather the capacity of a classwide proceeding to
generate common answers apt to drive the resolution of
the litigation. Dissimilarities within the proposed
class are what have the potential to impede the
generation of common answers.
Wal–Mart Stores, Inc. v. Dukes, ––– U.S. –––, –––, 131 S. Ct.
2541, 2551 (2011) (citation omitted).
There are a host of common issues that will generate common
answers in this litigation.

They include the collateral

estoppel effect of the Liability Opinion on issues to be
litigated in the damages trial, and the applicability of Noll’s
damages model.
Apple contends that the plaintiffs cannot establish through
common proof an injury to each individual plaintiff and his or
her damages.

For this same reason it argues that the common

issues here do not predominate over individual issues.

For the

reasons described below in connection with the discussion of
plaintiffs’ showing of predominance, Apple’s argument is
rejected.
Because “the predominance criterion is far more demanding”
than the commonality requirement, when plaintiffs move for
certification of a class pursuant to Rule 23(b)(3), “Rule
23(a)(2)’s ‘commonality’ requirement is subsumed under, or
31

superseded by, the more stringent Rule 23(b)(3) requirement” of
predominance.

Amchem Prods., Inc. v. Windsor, 521 U.S. 591,

609, 624 (1997).

As the plaintiffs have satisfied Rule

23(b)(3)’s predominance requirement, Rule 23(a)(2)’s commonality
requirement is met as well. 22
3.

Typicality

The typicality requirement is met when “each [class]
member’s claim arises from the same course of events, and each
class member makes similar legal arguments to prove the
defendant’s liability.”

Brown v. Kelly, 609 F.3d 467, 475 (2d

Cir. 2010) (citation omitted).

Commonality and typicality “tend

to merge into one another, so that similar considerations
animate analysis of both.”

Id. (citation omitted).

Here, each putative class member’s claim arises from the
same conduct: Apple’s conspiracy with the Publisher Defendants
to fix e-book prices, which caused the prices of e-books to
rise.

They share the same measurement of their damages, using

Noll’s model.

And each class member would make similar legal

arguments as to liability, including the extent to which
collateral estoppel applies, barring Apple from disputing this
In a footnote, Apple refers to a September 27, 2013 letter it
provided to the Court. To the extent Apple has made an argument
in the text of its opposition to the motion for certification,
it has been considered. The Court declines Apple’s invitation
to review as well its September 27 letter for any arguments it
might include that may be relevant to this motion practice.
22

32

Court’s earlier finding of liability.
establish typicality.

This is sufficient to

Apple does not dispute the existence of

typicality here.
4.

Adequacy

To determine the adequacy of representation, courts
determine whether: “1) plaintiff’s interests are antagonistic to
the interest of other members of the class and 2) plaintiff’s
attorneys are qualified, experienced and able to conduct the
litigation.”

In re Flag Telecom Holdings, Ltd. Sec. Litig., 574

F.3d 29, 35 (2d Cir. 2009) (citation omitted).

The adequacy

inquiry serves to “uncover[] conflicts of interest between named
parties and the class they seek to represent.”
Amchem, 521 U.S. at 625).

Id. (quoting

Not every potential conflict will

preclude a finding of adequacy, however.

Id.

“The conflict

that will prevent a plaintiff from meeting the Rule 23(a)(4)
prerequisite must be fundamental, and speculative conflict
should be disregarded at the class certification stage.”

In re

Visa Check/MasterMoney Antitrust Litig., 280 F.3d 124, 145 (2d
Cir. 2001) (citation omitted), overruled on other grounds by In
re Initial Public Offerings Sec. Litig., 471 F.3d 24 (2d Cir.
2006) (“In re IPO”).
The named plaintiffs are adequate representatives of the
class and class counsel are qualified, experienced, and able to
conduct this litigation.

Apple’s only argument against a
33

finding of adequacy is that plaintiffs are not “seeking damages
after May 21, 2012” and are consequently “waiving the rights of
absent class members to seek such damages.”

Here, declining to

seek damages for transactions after May 21, 2012 -- the date the
first of the Publisher Defendants began terminating their agency
agreements pursuant to their settlements with the DOJ and States
-- is entirely reasonable and justified.

Any effort to extend

the class period beyond May 21, 2012 would substantially
complicate plaintiffs’ damages analysis.

Apple cites no case

where a similar limitation on a demand for damages was found to
defeat class certification.
C.

Rule 23(b)(3)’s Requirements

Plaintiffs having satisfied each of the elements of Rule
23(a), the next inquiry concerns the requirements of Rule
23(b)(3).

After a discussion of the predominance requirement,

the issues of superiority and ascertainability will be
addressed.
1.

Predominance

Predominance is established where the legal or factual
issues that can be resolved through generalized proof are “more
substantial than the issues subject only to individualized
proof.”
omitted).

In re U.S. Foodservice, 729 F.3d at 118 (citation
In the predominance analysis, “resolved” issues are

to be treated just the same as “contested” issues and are
34

weighed in analyzing the extent to which common issues will
predominate over individual ones.

In re Nassau Cnty. Strip

Search Cases, 461 F.3d 219, 228 (2d Cir. 2006).

The Second

Circuit has emphasized that “Rule 23(b)(3) requires that common
questions predominate, not that the action include only common
questions.”

Brown, 609 F.3d at 484.

“As long as a sufficient

constellation of common issues binds class members together,”
individualized issues will not “automatically foreclose class
certification under Rule 23(b)(3).”

Id. at 483.

The essential

inquiry for predominance is whether the proposed class is
“sufficiently cohesive to warrant adjudication by
representation.”

Amgen Inc. v. Conn. Ret. Plans & Trust Funds,

--- U.S. ---, ---, 133 S. Ct. 1184, 1196 (2013) (citation
omitted).
Predominance is readily shown “in certain cases alleging
. . . violations of the antitrust laws.”
625.

Amchem, 521 U.S. at

For “where plaintiffs were allegedly aggrieved by a single

policy of the defendant[], and there is a strong commonality of
the violation and the harm, this is precisely the type of
situation for which the class action device is suited.”

Brown,

609 F.3d at 484 (citation omitted).
This is just such a case.

Apple conspired with the five

Publisher Defendants to fix national e-book prices.

Working

together, the e-book prices of the Publisher Defendants rose
35

precipitously and with one exception, simultaneously, 23 after
their adoption of Apple’s Agreements.
To prove a violation of Section 1 of the Sherman Act,
plaintiffs must show “a combination or some form of concerted
action between at least two legally distinct economic entities
that constituted an unreasonable restraint of trade either per
se or under the rule of reason.”

PepsiCo, Inc. v. Coca-Cola

Co., 315 F.3d 101, 109 (2d Cir. 2002) (citation omitted).

Apple

does not dispute that findings regarding any “unreasonable
restraint of trade” will be based on class-wide proof and
arguments. 24
To bring a private damages action pursuant to Section 4 of
the Clayton Act, plaintiffs must also establish “antitrust
standing.”

Gatt Commc’ns, Inc. v. PMC Assoc., LLC, 711 F.3d 68,

75 (2d Cir. 2013) (citation omitted).

To do so, they must prove

that they suffered an “antitrust injury” -- an injury “of the
type that the antitrust statute was intended to forestall” -and that they are “suitable plaintiff[s] to pursue the alleged

As noted above, Penguin did not sign its agency contract with
Amazon until June 2, 2010. Liability Opinion, 952 F. Supp. 2d
at 682.
23

Class plaintiffs’ motion for partial summary judgment on
collateral estoppel grounds is still pending before the Court.
But how and where collateral estoppel applies is a class-wide
issue, just as relitigation of any particular issue covered by
the Liability Opinion would be based on class-wide evidence.
24

36

antitrust violations and thus [are] efficient enforcer[s] of the
antitrust laws.”

Id. at 76 (citation omitted).

Apple does not

dispute that a class member who suffered an overcharge on the
Publisher Defendants’ e-books has standing to bring a claim
here.
Instead, Apple challenges a finding of predominance on two
other grounds.

First, Apple argues that the putative class

members are dissimilar, because each e-book transaction is
“unique,” the prices of many e-books fell after the adoption of
the agency model, many iBookstore customers purchased only one
or two e-books, and many injured class members received
offsetting benefits.

Second, Apple argues that Noll’s damages

model cannot reliably establish damages for each class member,
and that the use of that model to determine damages would
constitute an improper “trial by formula.”

None of these

arguments alters the conclusion that the common issues in this
litigation will predominate over any individual ones.
a.

Common Injury
i.

Every E-Book Is Unique.

In presenting its first argument, Apple offers three
reasons why the sale of every e-book is unique.

According to

Apple, class certification is improper here because “each of the
150 million e-book purchases that took place during the putative
class period has its own unique history and must be evaluated
37

separately.”

Although each transaction has “its own unique

history,” it need only be “evaluated separately” if it is not
susceptible to class-wide proof.

Here, Noll’s multivariate

regression analysis has disentangled the effect of collusion on
e-book prices, generating a model that explains 90% of the
variance among titles’ pricing.

Apple has not identified any

variable that Noll should have but did not include in his
regression model. 25

An abstract argument that each transaction

is “unique” or that some number of class members made very few
purchases through the iBookstore does not cast doubt on Noll’s
model.
Moreover, as the Liability Opinion explains, the Publisher
Defendants set the prices of individual e-book titles by
For the first time in its reply brief in support of its motion
to exclude Noll’s opinions, Apple contends that a class should
not be certified because Noll “disregards factors that would
affect the price for a given transaction, including authors’
reputations, reviews, events such as a movie release of the
title, ‘buzz’ and ‘word-of-mouth’ effects, and advertising and
other marketing effects.” Apple could have raised this argument
in its initial brief in opposition to class certification, but
chose not to. Accordingly, the argument is waived. See Conn.
Bar Ass’n v. United States, 620 F.3d 81, 91 n.13 (2d Cir. 2010)
(“Issues raised for the first time in a reply brief are
generally deemed waived.”).
25

Even if the argument were not waived, it would be easily
dismissed. As Noll has explained, his regression does capture
effects specific to an individual title, as it includes an
“indicator variable” for every e-book title. Noll’s indicator
variables are not perfect, as they must average these effects
over time, but they allow his model to account for the
idiosyncratic pricing influences on each e-book title.
38

reference to the categories to which they assigned them.
E-books were not purchased by a consumer haggling over a price
in a bookstall on a street corner.

For example, Apple’s pricing

caps, which became the sales price for most of the Publisher
Defendants’ books, set the e-book price at $12.99 for a NYT
Bestseller whose hardcover was priced between $25.01 and $27.50,
and at $14.99 for NYT Bestsellers with a hardcover list price
between $27.51 and $30.

Liability Opinion, 952 F. Supp. 2d at

669.
In a variation on the same argument, Apple contends that
the existence of any injury will depend on the specific titles
purchased by a class member and the timing of those purchases
since “over 50% of the Publisher Defendants’ e-book titles’
prices decreased or stayed the same after the switch to agency.”
Plaintiffs hotly contest the reliability of the analysis
undertaken by Kalt which produced the over 50% figure, but it
will be assumed for purposes of addressing this portion of
Apple’s argument that the figure is accurate. 26
In any event, Apple has identified no reason to believe
that a decrease in the price of an e-book over the course of the
class period is inconsistent with inflation due to the price-

The reliability and admissibility of Kalt’s work, including
this figure, is discussed in the Opinion issued today addressing
the motions to strike Kalt’s and Orszag’s expert opinions.
26

39

fixing scheme.

A number of independent factors, like the

release of a paperback edition, tend to cause a price decrease.
Those independent factors did not cease to exist during the
period of Apple’s conspiracy with the Publisher Defendants.
Noll’s model accounts for such independent variables while
measuring the impact of the scheme.

Each e-book title and its

history has been accounted for through the use of a titlespecific indicator variable and through Noll’s calculation of
overcharges specific to each of 502 categories.
ii.

Offsets

Apple argues that the dissimilarities among class members,
which are evident when one examines each of their purchases of
e-books, will prevent the plaintiffs from demonstrating common
injury or damages.

It contends that the examination of every

e-book purchase by each class member -- including purchases of
e-books issued by publishers other than the Publisher Defendants
-- is necessary, and that such an examination will show that
many class members benefitted by Apple’s entry into the e-book
market, that such benefits warrant offsets, and that
individualized inquiry is necessary to determine the magnitude
of these offsets.

Thus, according to Apple, even if members of

the class are entitled to damages due to Apple’s participation
in a price-fixing conspiracy and the inflated prices that
consumers paid because of that conspiracy, Apple is entitled to
40

reduce that recovery by applying offsets to the damages.

Apple

identifies four benefits: (1) the prices of some e-books
declined; (2) free and self-published e-books were more
available; (3) some consumers purchased e-books from the
iBookstore that they would not have purchased from another
e-tailer; and (4) the agency model ended the practice of
“windowing” e-books, allowing earlier access to certain
e-books. 27
Apple’s offset argument is rejected on several grounds.
First, Apple is not entitled to reduce the amount of any damages
that it owes because of any benefits that it claims consumers
received when Apple entered the e-book market by selling e-books
through its iBookstore.

In simplest terms, Apple was not

accused of nor found liable for violating the antitrust laws by
the act of opening the iBookstore.

What was illegal was joining

and facilitating a conspiracy to raise e-book prices, a
conspiracy that was effected through Apple’s Agreements and
their MFNs.

At trial, the DOJ and States showed both that

Apple’s price-fixing conspiracy was a per se violation of the
antitrust laws and that they were entitled to judgment under the
rule of reason test.

At that trial, Apple failed to show that

Apple’s expert, Orszag, provides lengthy opinions regarding
another proposed benefit: cheaper e-reader prices. Apple has
elected not to raise this as a reason to deny class
certification.
27

41

“the execution of the Agreements had any pro-competitive
effects.”

Liability Opinion, 952 F. Supp. 2d at 694.

As the

Liability Opinion found, “[t]he pro-competitive effects to which
Apple has pointed, including its launch of the iBookstore, the
technical novelties of the iPad, and the evolution of digital
publishing more generally, are phenomena that are independent of
the Agreements and therefore do not demonstrate any procompetitive effects flowing from the Agreements.”

Id. 28

Even if Apple could challenge these findings, plaintiffs
strenuously contest these “benefits” that Apple now identifies.
As described below, the plaintiffs contest in some instances
that the phenomena to which Apple points existed, or existed to
the extent it asserts.

In other instances, they contest that

the phenomena are properly associated with the collusive
behavior that triggered this action.

Apple has proffered little

reliable evidence in support of any pro-competitive benefits
flowing from its illegal conduct.

But, even if Apple were

entitled to litigate these issues at trial and if it also had
sufficient evidence to suggest that these purported benefits
existed and could be attributed to the price-fixing conspiracy,
the existence or non-existence of each of these offsetting

Whether or not collateral estoppel applies, Apple does not
dispute that these issues are subject to class-wide evidence and
argument in the class action.
28

42

benefits would constitute a common issue for the class to
litigate.
There is another reason, however, that the existence of
these supposed pro-competitive effects do not provide an
impediment to certification of a class.

Apple appears to

contend that common issues will not predominate over individual
ones because it will be necessary to determine the extent to
which each class member benefited from these four phenomena, to
calculate the dollar value of that benefit, and to subtract that
amount from any damages due the class member from the antitrust
violation.

In this case, however, the damages will be

calculated by subtracting the but-for prices of e-books from the
prices paid by class members. 29

Apple will not be entitled, as a

matter of law, to reduce the amount it owes by applying its
proposed offsets.
The proper measure of damages in a suit concerning a pricefixing conspiracy is “the difference between the prices actually
paid and the prices that would have been paid absent the
conspiracy.”

New York v. Hendrickson Bros., Inc., 840 F.2d

1065, 1077 (2d Cir. 1988); see also New York v. Julius Nasso
Concrete Corp., 202 F.3d 82, 88 (2d Cir. 2000) (noting damages

This is equivalent to multiplying the relative overcharge by
the actual purchase prices, as a but-for price is equal to the
difference between the actual price and the overcharge.
29

43

in antitrust suit alleging bid-rigging conspiracy would
“ordinarily be the difference between the price actually paid by
[plaintiff] . . . and the price it would have paid absent the
conspiracy”).

This Court has not found, and the parties have

not cited, any authority for a different measure of damages in
the context of a price-fixing conspiracy where the injured party
paid an inflated price for a good because of the defendants’
illegal conduct.

Cf. IIA Phillip E. Areeda, et al., Antitrust

Law ¶ 395 (3d ed. 2007) (explaining damages in overcharge cases
are determined either by lost profits or by overcharge, which is
“the difference between the price actually paid and the price
that would have been paid ‘but for’ the unlawful conduct
multiplied by the quantity purchased”).
Moreover, antitrust jurisprudence not only limits the proof
of damages in price-fixing cases to this formula, but it
expressly refuses to impose extraordinary burdens on a plaintiff
to construct the but-for price.

Where the but-for price is

uncertain, “the plaintiff’s burden of proving damages is, to an
extent, lightened,” Hendrickson Bros., 840 F.2d at 1077, for
“the wrongdoer shall bear the risk of the uncertainty which his
own wrong has created.”

Id. at 1078 (quoting Bigelow v. RKO

Radio Pictures, Inc., 327 U.S. 251, 265 (1946)).

Indeed, the

Supreme Court has long taught that “damage issues in [antitrust]
cases are rarely susceptible of the kind of concrete, detailed
44

proof of injury which is available in other contexts.”

J.

Truett Payne Co., Inc. v. Chrysler Motors Corp., 451 U.S. 557,
565 (1981) (citation omitted); accord Comcast, --- U.S. at ---,
133 S. Ct. at 1433 (noting, in antitrust case, that damages
“[c]alculations need not be exact”) (citation omitted); see also
Julius Nasso Concrete, 202 F.3d at 88 (quoting J. Truett
language).

For this reason, “it does not come with very good

grace for the wrongdoer to insist upon specific and certain
proof of the injury which it has itself inflicted.”

J. Truett,

451 U.S. at 566-67 (citation omitted).
As significantly, an antitrust defendant may not alter this
well-settled measurement of damages by speculatively raising
potential offsets, even when those offsets are directly related
to the goods at issue.

In Hanover Shoe, Inc. v. United Shoe

Machinery Corp., 392 U.S. 481 (1968), a monopolist refused to
sell its shoe machinery, insisting that the plaintiff rent it.
Id. at 483-84.

As calculated in the lower courts, the damages

were measured by the difference between what the plaintiff paid
the monopolist in rental fees and what it would have paid to
purchase the machines.

Id. at 484.

In affirming that

straightforward measurement of damages, the Court rejected the
monopolist’s argument that it was entitled to show that the
plaintiff had passed on the overcharge to its customers and had
suffered no actual loss.

Id. at 488.
45

Particularly relevant here, the Court noted “the nearly
insuperable difficulty of demonstrating that [a business] could
not or would not have raised [its] prices absent the overcharge
or maintained the higher price had the overcharge been
discontinued.”

Id. at 493.

Because proving (or disproving)

these features of the but-for world “would normally prove
insurmountable,” and because antitrust defendants may
“frequently seek to establish [the] applicability” of offsets
based on such arguments if allowed, the Court taught that they
should be barred.

Id.

Otherwise, “[t]reble-damage actions

would often require additional long and complicated proceedings
involving massive evidence and complicated theories.”

Id.;

accord Illinois Brick Co. v. Illinois, 431 U.S. 720, 737 (1977)
(rejecting suit by plaintiff consumer alleging overcharge of
retailer was passed-on on the basis of Hanover Shoe, noting any
other result would “add whole new dimensions of complexity to
treble-damages suits and seriously undermine their
effectiveness”); cf. In re Coordinated Pretrial Proceedings in
Petroleum Prods. Antitrust Litig., 691 F.2d 1335, 1341 (9th Cir.
1982) (holding Illinois Brick bars an “umbrella” theory of
damages that would seek recovery for overcharge by price-fixer’s
non-conspiring competitors on the grounds that price-fixing
permitted competitors to raise prices, in part because such
claims are “unacceptably speculative and complex”); Mid-West
46

Paper Prods. Co. v. Continental Grp., Inc., 596 F.2d 573, 585
(3d Cir. 1979) (rejecting umbrella theory, as “[a]part from its
speculative nature, any attempt to determine the effect of
defendants’ overcharges upon their competitors’ prices would
transform this antitrust litigation into the sort of economic
proceeding that the Illinois Brick Court was desirous of
avoiding if at all possible”).
Applying these principles, the jury’s task in assessing
damages will be confined to multiplying the relative overcharge
by the actual purchase prices of Publisher Defendants’ e-books
during the class period.

Unless they are part of this

calculation, none of Apple’s proposed offsets are relevant to
that damages calculation, and therefore, exploration of these
offsets will not affect any evaluation of whether common issues
will predominate over individual ones.
Apple cites two principal cases in support of its argument
that these four purported phenomena create offsets against any
calculated damages.

Neither supports Apple’s argument.

first is Hanover Shoe, which is quoted at length above.
Hanover Shoe does not aid Apple.

The
But,

The plaintiffs have already

adopted the analogous calculation of damages: the difference
between what plaintiff(s) paid to the defendant and what
plaintiff(s) would have paid, but for the illegal conduct.

In

Hanover Shoe, this was equal to the (actual) rental price less
47

the (but-for) purchase price; here, it is equal to the actual
price paid for an e-book less the but-for price for that title.
Indeed, the teachings of Hanover Shoe militate strongly against
acceptance of Apple’s proposed offsets.
The other case Apple cites is the Ninth Circuit’s decision
in Los Angeles Memorial Coliseum Commission v. National Football
League, 791 F.2d 1356 (9th Cir. 1993).
Apple.

It is no more helpful to

Although the calculation of damages it endorsed was made

more complex by the fact that the antitrust conspiracy had
denied the plaintiffs an opportunity to make a profit, the
calculation was entirely consistent with that which the
plaintiffs proffer here.
L.A. Memorial concerned a suit by a Los Angeles football
stadium and the Raiders football team against the National
Football League (“NFL”).

The Raiders were prevented from moving

from Oakland to Los Angeles by an NFL rule requiring unanimous
consent of NFL teams.

Id. at 1364.

The NFL rule was held to be

illegal as applied to the Raiders’ move, and the trial court
ordered that the Raiders be granted damages in the form of lost
profits for the years the Raiders were prevented from moving.
Id. at 1372.

The Ninth Circuit held that, because the NFL was

entitled to charge a new expansion team owner for the expansion
opportunity, the (but-for) costs of a transaction (the move)
must include that expansion charge and be netted against the
48

(but-for) lost profits expected from that transaction.
1373.

Id. at

The Ninth Circuit opined that a plaintiff’s antitrust

damages “are to be calculated by comparison of profits, prices
and values as affected by the conspiracy, with what they would
have been in its absence under freely competitive conditions.”
Id. at 1367 (citing Bigelow, 327 U.S. at 264).
Like the Supreme Court in Hanover Shoe, the Ninth Circuit
went on to reject offsets requested by the defendant.

The Ninth

Circuit noted that the alleged “benefits,” which included
increased television revenue supposedly due to the challenged
rule and benefits received from the Oakland community, were
independent of the illegal conduct and would have existed
regardless.

It consequently affirmed the exclusion of

defendant’s proposed offsets.

Id. at 1374.

The only other cases Apple cites in its offset argument are
two district court opinions concerning securities fraud and an
antitrust decision from this Court concerning an illegal tying
agreement. 30

Each discusses a traditional measurement of damages

that is directly related to the transactions at issue.

In Freeland v. AT&T Corp., 238 F.R.D. 130, 150 (S.D.N.Y.
2006), this Court adopted the “package” approach in antitrust
tying cases. Because a tying theory is based on the link
between the tying and tied products, the Court held that damages
should be set at the net amount plaintiffs were overcharged (or
undercharged) for the tied and tying products. Id.
30

49

Applying these principles to each of the four offsets that
Apple requests, it is readily apparent that none will create
individualized issues that preclude class certification.

Some

of these offsets do not directly relate to the transactions at
issue here; consequently, any imagined benefits are too remote
to be considered in damages calculations.

Others concern the

relevant transactions, but are based purely on conjecture about
speculative happenings in the but-for world.

Pursuant to the

sound policy taught by Hanover Shoe, such offsets must be
rejected.
First, Apple argues that some plaintiffs were not injured
because the prices of some e-books declined during the agency
period. 31

To the extent that Apple is referring to e-books not

published by one of the Publisher Defendants, this phenomenon is
too remote from the e-book transactions at issue here to be
relevant to the damages calculation.

To the extent that Apple

is referring to the Publisher Defendants’ e-books, plaintiffs
claim no damages unless they can demonstrate that the conspiracy
created a higher price for an e-book than would otherwise have
existed.

Noll’s model identifies the few instances in which a

purchaser of a title was not damaged, amounting to 0.2% of
transactions.

Of course, where an e-book’s price was inflated

The plaintiffs vigorously dispute that Apple’s experts have
accurately described the path of e-book prices.
31

50

due to Apple’s conduct but declined for an independent reason,
like the introduction of a paperback edition, the decline is
irrelevant to the proper calculation of damages and has been
accounted for in the regression analysis.
Second, Apple argues that the advent of the iBookstore
motivated Amazon to change its business terms to promote free
e-books, largely from self-publishing authors.

Yet, as noted in

the Liability Opinion, the “launch of the iBookstore” was
“independent of the Agreements and therefore do[es] not
demonstrate any pro-competitive effect[] flowing from the
Agreements.”

Liability Opinion, 952 F. Supp. 2d at 694.

Regardless, because the self-published and free e-books were not
the Publisher Defendants’ e-books, they may not be incorporated
into the damages calculation.

Simply put, they are not relevant

transactions when calculating damages.

Moreover, for the very

policy reasons identified in Hanover Shoe, such an offset would
invite speculation and add unnecessary complexity to the trial.
Third, Apple speculates that some of the consumers who
purchased e-books from the iBookstore may not have chosen to
purchase these same e-books from any other e-tailer.

If Apple

is able to confirm that that occurred, it contends that these
consumers should not be permitted to recover any damages even
though the price they paid for the e-books that Apple sold to
them was inflated by Apple’s illegal conduct.
51

Because Apple

cannot imagine any less onerous way to identify such consumers
than examination of each and every iBookstore customer, it
contends that individualized inquiry will defeat certification.
As noted above, the launch of the iBookstore was already
held to be independent of Apple’s illegal conduct, and thus as a
factual matter plaintiffs have shown that Apple is unlikely to
be able to establish any basis for this offset even if Apple
were entitled to relitigate the issue.

Furthermore, Apple and

its expert, Kalt, cannot explain why someone who purchased an ebook on the iBookstore would be unwilling to purchase the same
e-book for the same price (or less) via, for instance, a Kindle
app on the iPad.
But regardless of these impediments, any argument
concerning which iBookstore customers would have elected not to
purchase which e-books in the but-for world is too speculative
to support an offset.

Any such argument would be shot-through

with conjecture, and per Hanover Shoe, such an offset must be
rejected.

In addition, as an antitrust violator, Apple may not

escape its obligation to compensate iBookstore customers whom it
overcharged by arguing that some of them were too foolish or
stubborn to have used the Kindle app.

Cf. United States v. E.I.

du Pont de Nemours & Co., 366 U.S. 316, 366 n.12 (1961) (“In
general the object of the remedies under the anti-trust laws is
to . . . deprive the wrongdoers of the fruits of their unlawful
52

conduct . . . .” (citation omitted)); Berkey Photo, Inc. v.
Eastman Kodak Co., 603 F.2d 263, 296 (2d Cir. 1979) (“An
unlawful monopolist must be deprived of the fruits of its
wrongful conduct, and one of the forbidden fruits is an
excessive price. . . .

So long as a monopolist enjoys the

flower of evil at the expense of its customers, those victims
must have a remedy.” (citation omitted)).
Finally, Apple alleges that the Agreements ended the
practice of “windowing” e-books, which saved certain consumers
from having to wait to purchase an e-book or being forced to
purchase a more expensive hardcover edition.

Factually, the

Liability Opinion observed that “there is no reason to find that
windowing would have become widespread, long-lasting, or
effective,” and pointed to substantial evidence to the contrary.
Liability Opinion, 952 F. Supp. 2d at 702.

Legally, again,

Hanover Shoe bars an offset argument rooted in rank speculation
about the but-for world.

Because of Apple’s conduct, there is

no way to know whether a given title would or would not have
been “windowed” and whether a given purchaser of its e-book
would or would not have waited through the “windowing” period or
purchased a hardcover edition.

Furthermore, Apple may not avoid

liability for overcharging a customer by speculating that that
customer may have been better off paying the overcharge than
being unable to purchase the good at all.
53

Because these proposed offsets are inappropriate, they do
not inject a need for “individualized inquiry for each of the
millions of class members,” as Apple argues.

In fact, these

disputes concerning the existence and relevance of procompetitive benefits are themselves susceptible to class-wide
argument and proof.

Plaintiffs have therefore established that

common issues concerning liability and damages overwhelmingly
predominate over any individualized issues.

Cf. Shahriar, 659

F.3d at 253 (affirming finding of predominance despite “some
individualized damage issues”).
b.

Noll’s Model Reliably Estimates Class Member
Damages.

Apple also objects to class certification on the ground
that Noll’s damages model is unreliable for the reasons raised
in support of Apple’s Daubert motion.
below and is denied.

That motion is addressed

As explained below, this Court has

conducted a “rigorous analysis” of Noll’s model and finds it
capable of reliably estimating class members’ damages.

As a

consequence, the application of this damages model will not be
an impediment to a finding that common issues predominate over
individualized ones.
Apple makes a few arguments in addition to those raised in
its Daubert motion regarding the extent to which individual
issues may exist.

They are addressed here.

54

i.

Noll’s Model Is Reliable.

Apple raises only two arguments in its opposition to class
certification regarding Noll’s model that it does not raise in
its motion to exclude Noll’s opinions. 32

First, Apple charges

that Noll’s model is unreliable because he “fails to account for
key factors that would have increased e-book prices in the
absence of a conspiracy.”

Apple speculates that Amazon’s

pricing strategy may have changed with the launch of the iPad,
forcing Amazon to raise e-book prices.

But the record evidence

strongly indicates that Amazon was not planning to raise retail
prices; indeed, the Liability Opinion noted that, if anything,
the Publisher Defendants feared Amazon would force them to lower
wholesale prices.

See, e.g., Liability Opinion, 952 F. Supp. 2d

at 649 (noting Publishers “were concerned that, should Amazon
continue to dominate the sale of e-books to consumers, it would
start to demand even lower wholesale prices for e-books”).
Because the Publisher Defendants judged that Amazon would
continue to resist their pleas that it raise e-book prices, they
illegally joined with Apple to wrest control of retail pricing
from Amazon and raise those prices themselves.

Apple’s charge that Noll’s model impermissibly aggregates data
by using four-week average prices for e-books rather than
transaction prices has been mooted, as Noll subsequently offered
a damages model based on individual transaction prices.
32

55

In any event, Apple’s speculation that Amazon would have
unilaterally raised retail prices for e-books does not undermine
the reliability of Noll’s damages calculation.

As already

explained, such speculation is not a component of a damages
calculation in a price-fixing case and Noll’s regression model
is not unreliable because it fails to account for such
speculation.
Apple also wonders whether Barnes & Noble may have exited
the e-books market but for the conspiracy’s success in raising
e-book prices.

Apple complains that Noll does not account for

the fact that Barnes & Noble consumers might not have purchased
e-books at all in that case.
better.

This line of speculation fares no

Apple does not explain why an e-book purchaser would

not obtain e-books elsewhere if Barnes & Noble exited the market
or how one would reliably determine that today for any Barnes &
Noble customer in the period between 2010 to 2012.

In any

event, the law does not require Noll to engage in such
speculation when constructing a damages model.
Apple next notes that Noll has admitted he has not studied
Amazon’s pricing algorithm, as it is proprietary and was not
produced in the litigation.

Apple charges that, given this

admission, Noll is misleading when he states that “his
methodology is ‘consistent with the structure of Amazon’s
pricing formula.’”

Noll’s statement is not misleading.
56

Noll

did review and rely on the summary of Amazon’s pricing rules
that was produced in discovery.

In any event, this reference to

Amazon’s pricing formula occurred in the context of a discussion
of how Noll chose to express his results.

It did not concern

the underlying process used to reach those results.

Noll chose

to express his results, specifically the coefficients of his
variables (e.g., a title’s publisher and its age), as percentage
price effects, because that is “more consistent with the
structure of Amazon’s pricing formula.”

Noll’s choice of the

form in which to express the results of his regression model is
purely superficial; his choice simply made his equation easier
to read.

Noll has not misleadingly claimed special insight into

Amazon’s proprietary algorithm, and there is no reason to reject
his analysis as unreliable on this account.
For these reasons, and for the reasons set out below in the
discussion of Apple’s motion to exclude Noll’s opinions, the
Court finds that Noll’s model can reliably estimate class
members’ damages.
ii.

Remaining Arguments Regarding
Individualized Issues

Apple raises three additional objections to class
certification that appear to address the requirement that
plaintiffs show that common issues will predominate over
individualized ones.

First, Apple argues that class

57

certification is inappropriate unless plaintiffs can prove “that
all class members were in fact injured by the alleged
conspiracy.”

With the exception of dicta quoted from In re Rail

Freight Fuel Surcharge Antitrust Litig., 725 F.3d 244, 252 (D.C.
Cir. 2013), none of the cases Apple cites supports the
proposition that certification is inappropriate where some
putative class members may not have suffered injury.

Rail

Freight made the statement on which Apple relies as it rejected
an admittedly defective damages model.

The plaintiff’s expert

in Rail Freight conceded that the model yielded false positives.
Id. at 253.

There is no such concession here. 33

As for the general principle, it is widely recognized that
a class will often include persons who have not been
injured by the defendant’s conduct; indeed, this is
almost inevitable because at the outset of the case
many of the members of the class may be unknown, or if
they are known still the facts bearing on their claims
may be unknown. Such a possibility or indeed
inevitability does not preclude class certification.
Apple argues that many class members may not have been
injured, based on Kalt’s opinions that most e-book prices did
not increase as a result of price fixing and that Noll’s model
produces millions of false positives. As discussed in today’s
Opinion deciding the States’ and class plaintiffs’ motions to
exclude Apple’s experts’ opinions, both of these analyses are
fundamentally flawed and inadmissible. Kalt’s finding that most
e-book prices did not increase depends on his misclassification
of prices set by Penguin under the agency model as “pre-agency”
prices. And Kalt’s false positives analysis turns on his
comparison of actual transaction prices to average but-for
prices: his finding of “false positives” simply reflects the
fact that actual prices for a given e-book were both above and
below that e-book’s average price.
33

58

Kohen v. Pac. Inv. Mgmt. Co. LLC, 571 F.3d 672, 677 (7th Cir.
2009) (Posner, J.).
Next, Apple argues that Noll’s damages model -- because it
is a “formula to determine, and then distribute, damages” -violates the Supreme Court’s admonition in Dukes against “Trial
by Formula.”

Apple is mistaken.

The Supreme Court has never

suggested that widely used tools of economic analysis like
regression models must be banned from trials because they rely
on a “formula.”
The Supreme Court’s reference to a “Trial by Formula” in
Dukes was to a plan to try a sample set of class members’ claims
of sex discrimination and then multiply the average backpay
award to determine the class-wide recovery without further
individualized proceedings.
2561.

--- U.S. at ---, 131 S. Ct. at

Such a “novel” process would have robbed Wal-Mart of its

right to litigate its defenses to individual claims, as
liability for all but the sample set would have never been
tried.

Id.

Here, Apple has had a full opportunity to present

defenses to liability, and it will also have an opportunity to
challenge plaintiffs’ evidence in support of a damages award for
the entire class.

The plaintiffs’ proposal reflects a classic

method for arriving at a class-wide damages figure in an

59

antitrust lawsuit and is not the novel trial by formula rejected
in Dukes.
Finally, citing to the Second Circuit’s decision in
McLaughlin v. American Tobacco Co., 522 F.3d 215 (2d Cir.),
abrogated on other grounds by Bridge v. Phoenix & Indem. Co.,
553 U.S. 639 (2008), Apple argues that Noll’s model is “nothing
more than an improper ‘fluid recovery’ mechanism” designed to
“mask[] the prevalence of individual issues.”

Noll’s

individualized damages calculations are the very opposite of the
gross estimate of damages that would be used in a “fluid
recovery” procedure.
The plaintiffs in McLaughlin sought to recover damages for
cigarette smokers who had been injured by the manufacturers’
implicit representation that light cigarettes were healthier
than full-flavored cigarettes.

Id. at 220.

In decertifying

this class, the Second Circuit rejected a “fluid recovery”
procedure whereby the court would “[r]oughly estimat[e] the
gross damages to the class as a whole and only subsequently
allow[] for the processing of individual claims.”

Id. at 231.

This “rough estimate” was calculated according to “an initial
estimate of the percentage of class members who . . . have valid
claims” and an estimate of the average loss per plaintiff.

The

district court had conceded that the evidence supporting either
estimate “appears to be quite weak.”
60

Id. at 231-32.

The court

of appeals explained that “such an aggregate determination is
likely to result in an astronomical figure that does not
accurately reflect the number of plaintiffs actually injured by
defendants and that bears little or no relationship to the
amount of economic harm actually caused by defendants.”

Id. at

231.
In this case, by contrast, Noll’s damages model produces
individualized damage estimates based on individual transaction
records.

Apple will not be made to pay some “rough estimate” of

damages; rather, it will pay to each injured claimant a
conservatively calculated amount equal to damages due that
claimant.

Consequently, plaintiffs have established that common

issues in this litigation will predominate over any
individualized ones.
2.

Superiority and Ascertainability

Finally, plaintiffs must establish that a class action
would be “superior to other available methods for fairly and
efficiently adjudicating the controversy.”
23(b)(3).

Fed. R. Civ. P.

Where individual class members’ possible recoveries

are so small that no other practical method of adjudication
exists, superiority is often satisfied.

See Amchem, 521 U.S. at

617.
Here, where millions of putative class members each
suffered minor injury -- an average overcharge of less than $7,
61

in the case of iBookstore customers -- a class action is the
superior method for adjudicating these claims.

Apple argues

that a class action is not a superior method for adjudicating
this controversy because it violates its due process rights.
This argument is an apparent reference to Apple’s argument
concerning a “Trial by Formula,” and is rejected.
Plaintiffs’ proposed class must also meet the related
requirement of ascertainability.

“[C]lass members must be

ascertainable at some point in the case, but not necessarily
prior to class certification.”
(citation omitted).

In re IPO, 471 F.3d at 45

“To be ascertainable, the class must be

readily identifiable, such that the court can determine who is
in the class and, thus, bound by the ruling.”

Charrons v.

Pinnacle Grp. N.Y. LLC, 269 F.R.D. 221, 229 (S.D.N.Y. 2010).
Apple argues that class members are not ascertainable,
citing Carrera v. Bayer Corp., 727 F.3d 300 (3d Cir. 2013).

In

Carrera, the Third Circuit decertified a class of consumers of
an over-the-counter diet supplement on the ground that such a
class could not be reliably ascertained.

Id. at 308.

In doing

so, the court observed that the method of determining whether
someone is in the class must be “administratively feasible” and
not require “individualized fact-finding or mini-trials.”
at 307-08 (citation omitted).

The Carrera court found it

62

Id.

unlikely that records existed establishing who had purchased the
alleged falsely advertised diet supplement.

Id. at 308.

By contrast, here, the parties have in their possession
detailed transaction records, and using those records the class
is already distributing proceeds from settlements in this
action.

Amazon, Barnes & Noble, Apple, and Kobo have each

“identified, in its internal system, all individual customers
who purchased qualifying E-books.”

Due to the existence of

digital transaction records, therefore, the class’s members in
this lawsuit are readily ascertainable.
Plaintiffs have more than met their burden to establish
each of the requirements of Rule 23(a), as well as the
predominance, superiority, and ascertainability requirements of
Rule 23(b)(3).

Consequently, their motion for class

certification is granted.
II.

Apple’s Motion to Exclude the Opinions of Dr. Noll
In opposing class certification, Apple moved to strike

Noll’s expert reports.

The decision certifying a class rests in

part on the rejection of that motion, and it is to that motion
to strike that this Opinion now turns.
A.

Legal Standard

Federal Rule of Evidence 702 grants an expert witness
testimonial latitude unavailable to other witnesses, provided
that (1) “the testimony is based on sufficient facts or data,”
63

(2) “the testimony is the product of reliable principles and
methods,” and (3) “the expert has reliably applied the
principles and methods to the facts of the case.”
702.

Fed. R. Evid.

“[T]he proponent of expert testimony has the burden of

establishing by a preponderance of the evidence that the
admissibility requirements of Rule 702 are satisfied.”
States v. Williams, 506 F.3d 151, 160 (2d Cir. 2007).

United
The

district court performs the role of “gatekeeper” -- ensuring
that the proponent has made the necessary showing and that the
expert’s testimony “both rests on a reliable foundation and is
relevant to the task at hand.”

Daubert v. Merrell Dow Pharm.,

Inc., 509 U.S. 579, 597 (1993).
In order to be admissible, “[a]n expert opinion requires
some explanation as to how the expert came to his conclusion and
what methodologies or evidence substantiate that conclusion.”
Riegel v. Medtronic, Inc., 451 F.3d 104, 127 (2d Cir. 2006),
aff’d on other grounds, 552 U.S. 312 (2008).

An explanation is

necessary because “when an expert opinion is based on data, a
methodology, or studies that are simply inadequate to support
the conclusions reached, Daubert and Rule 702 mandate the
exclusion of that unreliable opinion testimony.”

Ruggiero v.

Warner–Lambert Co., 424 F.3d 249, 255 (2d Cir. 2005) (citation
omitted).

64

While a district court has “broad latitude” in deciding
both “how to determine reliability” and in reaching “its
ultimate reliability determination,” it may not abandon this
“gatekeeping function.”
omitted).

Williams, 506 F.3d at 160–61 (citation

“[N]othing in either Daubert or the Federal Rules of

Evidence requires a district court to admit opinion evidence
that is connected to existing data only by the ipse dixit of the
expert.”

Kumho Tire Co. v. Carmichael, 526 U.S. 137, 157 (1999)

(citation omitted).
“[A] trial judge should exclude expert testimony if it is
speculative or conjectural or based on assumptions that are so
unrealistic and contradictory as to suggest bad faith.”

Zerega

Ave. Realty Corp. v. Hornbeck Offshore Transp., LLC, 571 F.3d
206, 213–14 (2d Cir. 2009) (citation omitted).

“Other

contentions that the assumptions are unfounded go to the weight,
not the admissibility, of the testimony.”
omitted).

Id. (citation

“A minor flaw in an expert’s reasoning or a slight

modification of an otherwise reliable method” does not itself
require exclusion; exclusion is only warranted “if the flaw is
large enough that the expert lacks good grounds for his or her
conclusions.”

Amorgianos v. Nat’l R.R. Passenger Corp., 303

F.3d 256, 267 (2d Cir. 2002) (citation omitted).

This is

because “our adversary system provides the necessary tools for
challenging reliable, albeit debatable, expert testimony.”
65

Id.

“Vigorous cross-examination, presentation of contrary evidence,
and careful instruction on the burden of proof are the
traditional and appropriate means of attacking shaky but
admissible evidence.”

Id. (quoting Daubert, 509 U.S. at 596).

The Daubert inquiry is “flexible” and “gives the district
court the discretion needed to ensure that the courtroom door
remains closed to junk science . . . .”

Id.

And it is

“critical that an expert’s analysis be reliable at every step,”
for “any step that renders the analysis unreliable” renders the
expert’s testimony inadmissible.
B.

Id. (citation omitted).

Noll’s Model Is Reliable

Noll’s opinions easily meet the standards of Rule 702 and
Daubert.

Noll is eminently qualified and nationally respected

in the field of antitrust economics.

Here, Noll applies the

“before and after” method, which is a well accepted method of
measuring antitrust damages.

See, e.g., In re Scrap Metal

Antitrust Litig., 527 F.3d 517, 529 (6th Cir. 2008) (the “before
and after” method “is broadly accepted for proving antitrust
damages”); Eleven Line, Inc. v. N. Tex. State Soccer Ass’n,
Inc., 213 F.3d 198, 207 (5th Cir. 2000) (noting “the two most
common methods of quantifying antitrust damages are the ‘before
and after’ and ‘yardstick’ measures”); In re NASDAQ MarketMakers Antitrust Litig., 169 F.R.D. 493, 521 (S.D.N.Y. 1996)
(identifying “before and after” and “yardstick” methods as
66

accepted methodologies for measuring damages in antitrust cases)
(quoting ABA Antitrust Section, Antitrust Law Developments, 66973 (3d ed. 1992)); American Antitrust Institute, Private
Enforcement of Antitrust Law in the United States: A Handbook,
229 (2012).
Noll has constructed a sophisticated hedonic pricing model
that separately estimates damages for 502 categories.

Noll’s

fixed-effects model not only accounts for all identifiable (and
reliably estimable) structural influences on pricing, it also
accounts for the idiosyncratic effects peculiar to a given
title.
To generate this model, Noll performed a multiple
regression analysis on 1.3 million e-book transactions,
including nearly every e-book sold before and during the pricefixing conspiracy.

The resulting model is remarkably able: it

explains 90% of the variance among e-book titles’ prices.
Noll’s methods are uniformly reliable, and his model will surely
aid the trier of fact.

Indeed, if Noll’s opinions did not pass

muster under Rule 702, it is difficult to imagine an opinion
that could.
Apple moves to exclude the opinions Noll offers in support
of class certification on very narrow grounds.
that follow, Apple’s motion is denied.

67

For the reasons

Apple does not dispute that Noll is well qualified to serve
as an expert with regard to each of the opinions that he is
offering here.

Nor does Apple take issue with the use of either

a “before and after” method or a multiple regression analysis to
calculate damages.

Taking each of Apple’s arguments singly or

considering them as a whole, Apple has not identified any good
reason to exclude Noll’s opinion.
Apple makes four arguments:
1.

Noll’s opinions fail to meet professional standards;

2.

Noll did not conduct a “factual study” of the but-for
world and, at his deposition, did not have an opinion
regarding “important specific features of th[e] butfor world”;

3.

Noll’s model does not reliably fit the data;

4.

Noll’s methodology “assumes, rather than proves,
common impact” and consequently “forces a positive
overcharge estimate.”

These arguments are considered in the order raised by
Apple.

Apple’s arguments suggest, at most, lines of cross-

examination that may be appropriate for trial.

Many of these

arguments, however, are at odds with fundamental tenets of
econometrics and antitrust law, or rest on phrases pulled from
court decisions and taken out of context.

Indeed, one of

Apple’s chief arguments -- that Noll assumes, rather than
proves, common injury -- misconstrues Noll’s use of multiple
regression analysis.

Similarly, although Apple cites Comcast,

--- U.S. at ---, 133 S. Ct. at 1433, no fewer than ten times in
68

its briefing of its motion to exclude Noll’s opinions, it seems
to have overlooked the Court’s recognition that, in an antitrust
case, damages “[c]alculations need not be exact.”

Id. at ---,

1433; accord J. Truett, 451 U.S. at 565-66; Julius Nasso
Concrete, 202 F.3d at 88.
C.

Noll’s opinions do not fail to meet professional
standards.

Apple claims that the Noll opinion on damages fails to meet
professional standards because (a) Noll did not consider
“whether his regression coefficients were statistically
significant”; (b) at his deposition Noll could not recall the
meaning of particular lines of code used to implement his
regression analysis; and (c) Noll did not consider the analysis
of Dr. Abraham Wickelgren.
Apple’s first argument reflects a fundamental
misunderstanding of econometrics.

While the statistical

significance of the forecasting model is relevant, the
statistical significance of each and every coefficient is not. 34

In particular, Noll noted that multicollinearity among the
variables could be one reason useful variables appear not to be
statistically significant. Multicollinearity occurs when
several variables are correlated with each other -- that is,
they overlap in the sense that they each capture some part of
the same effect -- such that it is impossible from the data set
to objectively determine the influence of one variable as
opposed to the influence of these others. See Freeland, 238
F.R.D. at 147 & n.13. In that case, a model might estimate a
smaller coefficient (influence) for one variable and a greater
coefficient for other correlated variables; while the overall
34

69

Apple has offered no expert testimony to support its critique,
and in its reply Apple has abandoned this argument.
Apple’s second complaint, that at his deposition Noll was
able to explain “very little of the coding” used to run his
regression, is premised on a false assumption.

Noll’s role as

the plaintiffs’ expert was to identify the variables that would
be included in the regression analysis.
code.

It was not to write the

Therefore, Noll’s inability to identify for Apple’s

counsel particular variable names in certain lines of code is of
no significance.
Finally, Apple charges that Noll’s failure to consider a
damages analysis performed by Dr. Wickelgren exhibits a “studied
ignorance of the work of fellow economists.”
distorts the record.

This unfair charge

Apple has not shown that Noll is

unfamiliar with or failed to consider any published work in the
field of antitrust economics or economics in general that would
be relevant to his expert opinion in this case.

Apple’s charge

relates instead to a non-public study done by another expert
retained by plaintiffs’ counsel in this case.

Dr. Wickelgren

effect of these variables may be strongly statistically
significant, each coefficient may not appear statistically
significant on its own. Such an effect is noted in a resource
upon which Apple’s expert, Kalt, relies. See Jeffrey M.
Wooldridge, Introductory Econometrics: A Modern Approach, 148-49
(5th ed. 2013) (setting out an example where jointly significant
variables appear singly insignificant due to multicollinearity).
70

entered a declaration to support a finding that the amounts paid
in settlement by the Publisher Defendants were reasonable.

He

performed a far less sophisticated analysis than Noll has,
simply comparing the average price of a title in the four weeks
after the introduction of agency with the average price in the
four weeks before, and then weighting those differences
according to unit sales, to estimate an average overcharge for
the settling publishers for bestsellers, other frontlist titles,
and backlist titles.

Apple identifies no reason to believe that

review of Wickelgren’s analysis would have informed Noll’s far
more complex work.
D.

Study of the but-for world

Apple next argues that Noll’s opinions should be excluded
because he did not conduct an independent analysis of the ebooks market and failed to consider the following important
specific features of the but-for world: whether the publishers
would have entered into agency contracts; the size of Amazon’s
market share; whether Apple would have distributed e-books;
whether Barnes & Noble would have exited the e-books market;
Amazon’s pricing algorithm; and the impact of evidence
“suggesting” that “many” e-books’ buyers bought only one or two

71

e-books during the class period.

Apple’s argument has little

merit. 35
It is unsurprising that Noll did not develop an opinion
about many of the issues that Apple contends are important
features of a but-for world.
not exist.

After all, the but-for world does

Noll was given the task of calculating what the

prices for a given e-book would have been had Apple not
conspired with the Publisher Defendants to engage in illegal
price-fixing.

To do so, he performed a sophisticated

multivariate analysis to construct a pricing model.
While Apple argues that “Dr. Noll’s damages calculations
cannot stand without the false assumption that pricing dynamics
in the but-for world were the same as in the pre-agency world,”
Noll makes no such assumption.

To generate his model, Noll

constructed a competitive benchmark against which to compare
collusive prices -- benchmark data that included e-books sold by
In its reply, Apple shifts the focus of its attack on Noll’s
opinion and contends that he failed to consider that there would
have been higher e-reader prices in the but-for world. Because
Apple elected to raise this argument for the first time in its
reply -- thus preventing plaintiffs from responding -- Apple has
waived it. See Conn. Bar Ass’n, 620 F.3d at 91 n.13 (“Issues
raised for the first time in a reply brief are generally deemed
waived.”).
35

Were it not waived, this argument would fail. For the reasons
explained in the discussion above concerning Apple’s proposed
offsets, damages for e-book overcharges will not be reduced to
account for speculative benefits in purchases of e-readers, a
wholly distinct product sold in a distinct market.
72

the Publisher Defendants in the pre-agency period and by other
publishers during the pre- and post-agency periods -- and
included variables in his regression to capture unrelated
differences between the pre- and post-agency periods.

Noll used

a well-established mode of analysis to compare collusive pricing
to competitive pricing and thereby to capture pricing that would
have prevailed in the but-for world.
None of the features which Apple has identified in this
line of attack on Noll’s methodology suggest that his choice of
variables or his methodology is wanting.

Most of these features

of the but-for world lost all relevance once the conspirators
raised the prices for e-books.

Only one of them -- the

frequency with which a consumer may have purchased e-books -has any relevance to a damages calculation.
E.

Apple argues Noll’s model does not reliably fit the
data.

Apple next contends that Noll’s opinions should be rejected
because his model “does not reliably fit the data.”

Apple

submits three bases for such a finding: (a) Noll’s model does a
poor job of explaining variation in a given title’s prices;
(b) his model generates “millions of false positives”; and
(c) his model does not conform to the plaintiffs’ theory that
more e-book titles would have been priced at $9.99 but for the
conspiracy.

Each argument is considered in turn.

73

1.

Variation in pricing of a given title

Apple argues that Noll’s model should be excluded because
it cannot explain much of the variation in the prices of a given
e-book title.

This argument is a straw man.

The purpose of Noll’s model is not to explain the variation
in the pricing of an individual title; the purpose is to
estimate the effects of collusion on e-book prices.

To do so,

it effectively compares prices of e-books affected by collusion
to prices of e-books that are not affected.

Thus, the relevant

question is not whether the model can explain variances within
the pricing data for a given e-book, but rather whether the
model can reliably explain the differences between collusive ebook prices and competitive prices so that it can disentangle
the effects of collusion. 36

Apple does not dispute that the Noll

model has an adjusted R2 of 0.90, which means that it explains
90% of the variation in the prices among e-book titles.
Apple argues that a pricing model that cannot explain
variations within the pricing of a given e-book cannot reliably
explain the differences between collusively and competitively

While some comparisons between competitive and collusive
transaction prices concern the same e-book title sold pre- and
post-agency, and thus the effect of collusion implicates
“within” variance rather than “between” variance, this is only a
subset of such comparisons.
36

74

priced e-books so as to isolate the effect of collusion.

This

is not true.
The two cases Apple cites for this mistaken proposition do
not even concern evaluation of the reliability of an expert’s
study.

In Amorgianos, the Second Circuit held that it was not

an abuse of discretion for a district court to exclude testimony
that short-term exposure to a certain chemical could cause
certain symptoms where the expert relied on articles reporting
on long-term exposure to a variety of chemicals causing symptoms
different from plaintiff’s.

303 F.3d at 270.

Here, far from

relying on tangentially relevant studies performed by others,
Noll has performed his own study of the very transactions at
issue.

And in In re Rezulin Products Liability Litigation,

Judge Kaplan excluded expert testimony where the experts
“offered no evidence” that injury like plaintiffs’ was
“possible” and indeed “nothing in the challenged expert reports
gives any indication” of the possibility of such injury.
Supp. 2d 398, 427 (S.D.N.Y. 2005).

369 F.

Here, Noll has generated a

model that explains 90% of the variance in pricing among
e-books.
Finally, in making this critique of Noll’s model, Apple
relies extensively on an analysis by Kalt purporting to measure
the variation in price for individual titles.

This work by Kalt

is subject to a separate motion to strike because it is error75

ridden, relies on irregular techniques, and manipulates data.
Even if that work could survive a motion to strike, it would
provide at most an avenue for cross-examination of Noll at
trial.

It does not provide a basis to exclude Noll’s work.
2.

False positives

Apple next argues that Noll’s model produces “millions of
false positives.”

To reach this result, Apple’s expert, Kalt,

compared the model’s predicted average but-for price against
actual transaction prices, and counted as a “false positive” any
transaction where the actual price was below the average but-for
price.

There are several problems with Kalt’s analysis.

A

discussion of two of them will suffice.
Although Noll’s model was not intended to be used in this
way, the model can be made to produce an estimated average butfor price of an e-book in a given four-week period.

Noll’s goal

was to calculate the percentage elevation in prices due to price
collusion for each e-book.

By focusing on the level of prices,

rather than the changes in prices attributable to collusion,
however, Kalt misidentifies transactions as being unaffected by
price collusion.
Noll provides a simple example to demonstrate the error.
Assume the standard price for an e-book is $20 before collusion
(the “competitive price”) and $24 after, with the $4 increase a
result of price-fixing.

Assume also that in both periods one76

third of customers pay the standard price; one-third receive a
25% discount; and one-third receive a 50% discount.

Thus,

consumers would pay the following:
Competitive

Collusion

Overcharge

No discount:

$20

$24

$4

25% discount:

$15

$18

$3

50% discount:

$10

$12

$2

As the model’s predicted price is the average price, the model
predicts a competitive price of $15 and a collusive price of
$18.
Kalt’s test, which simply compares the actual transaction
price (i.e., the actual collusive price) to the predicted
(average) competitive price, would flag as a “false positive”
the transaction where consumers received a 50% discount, since
the actual price they paid ($12) is less than the average
competitive price ($15).

But as this example presupposes, these

consumers were damaged, by $2 -- in the absence of price fixing,
they would have paid $2 less than $12.

Although Noll’s model

would properly estimate damages of $4, $3, or $2 in this case
(depending on the discount received), Kalt would mistakenly find
one-third of these results to be “false positives.”

Apple’s

arguments concerning “false positives” provide no basis to
challenge the reliability of Noll’s regression model.

77

3.

$9.99 price point

Apple argues that Noll’s model fails to fit the data in a
third way.

Although plaintiffs have argued that an “industry

standard $9.99 price point” existed, Apple asserts that Noll
testified in his deposition that he did not know or care how
many predicted but-for prices in his model are at the $9.99
price point.
In fact, Noll’s deposition testimony was more nuanced than
Apple describes.

At his deposition, Noll explained that it may

be relevant whether the but-for prices of certain books are
close to $9.99, but that the percentage of but-for prices equal
to “exactly” $9.99 is not a “valid measure of [the] reliability”
of his model.

Noll also pointed out that $9.99 -- which Apple

here terms the “industry standard . . . price point” -- was the
price Amazon charged for certain e-books, not all e-books.

See

Liability Opinion, 952 F. Supp. 2d at 649 ($9.99 was the price
for certain New Releases and NYT Bestsellers).

Consequently, a

perfect damages model would calculate but-for prices of $9.99
for fewer than 100% of e-books.
Apple may cross-examine Noll at trial regarding the extent
to which his model predicts prices at or near $9.99.

The fact

that Noll had not done this calculation at the time of his

78

deposition is not a ground for striking his report or rejecting
his model. 37
F.

Common impact

Finally, Apple charges that Noll’s pricing model assumes
common impact.

“Inherent in Dr. Noll’s methodology is the

assumption that all proposed class members that purchased titles
within a certain category were either injured or not, regardless
of whether an individual purchaser actually paid more for a
specific title as compared to the but-for price of that title on
the same day.”

Similarly, Apple contends that Noll’s “averaged

and aggregated percentage overcharges . . . are nothing more
than a fictional composite.”

In a related argument, Apple

asserts that Noll’s 502 e-book categories are suspect because
they are constructed from “highly aggregated” genre categories.
Because the genre categories are over-inclusive, Apple argues,
they undermine the reliability of any calculation of an
overcharge for the book category.
Addressing this last argument, it should be noted that Noll
largely adopted the genre categories from the New York Times
Within a stricken portion of Kalt's sur-reply, Apple has
offered a study performed by Kalt of the percentage of Noll’s
but-for prices that fall within 50 cents of $9.99. It does not
appear, at first blush, that that study makes any attempt to
isolate the extent to which the predicted prices of New Releases
and NYT Bestsellers fall close to $9.99. In any event, the
plaintiffs have not had an opportunity to respond to this
calculation and it will not be further considered here.
37

79

bestseller lists, and his genre distinctions helped to generate
720 possible categories for the Publisher Defendants’ e-books.
The categories were so narrowly drawn that no e-book was sold in
218 of those categories.

Apple has proffered no credible

evidence that Noll’s genre categories are too “aggregated” to
reliably measure damages.
Apple’s argument, at root, is that no expert analysis could
reliably isolate the effects of collusion because every book is
different.

This ignores the factual record -- establishing that

pricing grids were used both before and during Apple’s pricefixing, based on the variables Noll considers -- and the fact
that Noll’s model succeeds in explaining 90% of the variance
among titles’ prices.
As reflected in the Liability Opinion, after the iBookstore
opened in April 2010, the price caps in Apple’s Agreements
became the new retail prices for many of the Publisher
Defendants’ e-books.

Liability Opinion, 952 F. Supp. 2d at 682.

In the five months following the adoption of the agency model,
the Publisher Defendants collectively priced 85.7% of their New
Release titles sold through Amazon and 92.1% of their New
Release titles sold through Apple to a price that was within 1%
of the price caps.

Id.

This was also true for 99.4% of the NYT

Bestseller titles on Apple’s iBookstore, and 96.8% of NYT
Bestsellers sold through Amazon.
80

Id.

The increases at Amazon

within roughly two weeks of moving to agency amounted to an
average per unit e-book retail price increase of 14.2% for their
New Releases, 42.7% for their NYT Bestsellers, and 18.6% across
all of the Publisher Defendants’ e-books.

Id.

Noll’s hedonic pricing model disaggregates the major
quantifiable factors that might influence e-book pricing -publisher, genre, bestseller status, age, and the availability
of a hardcover or paperback edition -- and then controls for
these factors to compare competitive prices against collusive
prices.

Noll’s regression analysis assumes that, if all other

variables are equal, the relative effect of collusion will be
the same.

So long as Noll has captured the salient

characteristics for pricing purposes, this assumption is wholly
proper. 38
To attack Noll’s model, Kalt misleadingly quotes from a
report on econometrics by the American Bar Association’s Section
of Antitrust Law, warning that
[t]he reduced-form pricing equation assumes that a
conspiracy has the same effect on every purchaser and
focuses on average effect, which may hide variation
across class members. If one is attempting to test
whether there is an impact on all members of a
proposed class, however, that assumption is not valid,
as it assumes the very proposition that is being
tested.

Apple was free to conduct its own regression study with other
variables, but has elected not to do so.
38

81

ABA Section of Antitrust Law, Econometrics: Legal, Practical,
and Technical Issues, 222 (2005) (Kalt’s emphasis).
itself quotes the second sentence in its brief.

Apple

But,

astonishingly, Apple and its expert neglect to mention that in
this passage the ABA report is discussing a single dummy
variable to determine the full effect of an antitrust
conspiracy.

The ABA report goes on to recommend the very

approach taken by Noll:
As a result, somewhat more complex models that do not
make such an assumption must be used to test classwide impact.
One approach is to divide the proposed class into
categories and use a model that allows the value of
the dummy variable to be different for different
categories. This would be appropriate if members of
the proposed class can be grouped using some
observable structural characteristic that is believed
to affect the price . . . .
Id.

Noll has done just this, dividing the e-books purchased by

the putative class into 502 categories and calculating the
effect of collusion on each category.
Thus, Apple has not undermined Noll’s model’s assumption
that similarly situated e-books -- that is, e-books for which
the values of all explanatory variables identified by Noll’s
model are identical -- suffered the same relative overcharge.
Noll’s model identifies the variables that should account for
most of the differences in e-book pricing, including

82

idiosyncratic factors affecting each title, and it proves able
to explain 90% of the pricing variation among e-books.
G.

Apple’s Request for a Hearing

Apple requests that, before deciding the motion to exclude
Noll’s opinions, the Court hold a hearing “with live testimony
by Dr. Noll.”

But Apple has already had the opportunity to

question Noll at his deposition, and Apple’s motion to exclude
is based wholly on written materials, not disputed issues of
fact best resolved through live testimony.

Apple has not

explained why it believes a hearing would aid the Court in
deciding this motion.

Because it is manifestly clear from the

papers that Noll’s opinions are admissible, a hearing is
unnecessary.
A district court is granted “the same kind of latitude in
deciding how to test an expert’s reliability, and to decide
whether or when special briefing or other proceedings are needed
to investigate reliability, as it enjoys when it decides whether
that expert’s relevant testimony is reliable.”
526 U.S. at 152.

Kumho Tire Co.,

“While the gatekeeping function requires the

district court to ascertain the reliability of [an expert’s]
methodology, it does not necessarily require that a separate
hearing be held in order to do so.”

United States v. Williams,

506 F.3d 151, 161 (2d Cir. 2007); accord In re United States
Foodservice Inc. Pricing Litig., 729 F.3d 108, 129-30 (2d Cir.
83

2013); United States v. Jawara, 474 F.3d 565, 582-83 (9th Cir.
2006); Nelson v. Tenn. Gas Pipeline Co., 243 F.3d 244, 248-49
(6th Cir. 2001); Oddi v. Ford Motor Co., 234 F.3d 136, 155 (3rd
Cir. 2000).
not required.

When a hearing would be a mere “formality,” it is
Williams, 506 F.3d at 161.

The cases Apple cites are not to the contrary. 39

Apple

quotes from the Second Circuit’s decision in Borawick v. Shay,
68 F.3d 597 (2d Cir. 1995), but fails to note the context.

The

Borawick decision addressed the utility of a hearing when
addressing the admissibility of lay witness testimony in a
highly unusual circumstance.

In Borawick the Second Circuit

held that the admissibility of a lay witness’s “hypnoticallyenhanced testimony” should be determined on a case-by-case basis
and noted, after listing seven fact-intensive factors courts
should consider in such cases, that “a pretrial evidentiary
hearing is highly desirable to enable the parties to present
expert evidence and to test credibility through crossexamination.”

Id. at 608.

Apple did not cite any cases in its request for a hearing on
its motion to exclude Noll’s opinions, which was fully submitted
on January 21, 2014. It first presented argument in favor of a
hearing weeks later -- on February 21 -- in its opposition to
class plaintiffs’ motion for summary judgment. In that brief,
Apple requested a hearing on plaintiffs’ motions to exclude
Apple’s experts, which had also been fully submitted weeks
earlier, on February 4.
39

84

Apple twice cites to the district court’s opinion in Colon
ex rel. Molina v. BIC USA, Inc., 199 F. Supp. 2d 53 (S.D.N.Y.
2001), but Colon excluded expert testimony without a hearing,
noting that “[t]he party proferring [expert] testimony is not
entitled” to an evidentiary hearing; that “[n]othing in Daubert,
or any other Supreme Court or Second Circuit case, mandates that
the district court hold a Daubert hearing before ruling on the
admissibility of expert testimony, even where such ruling is
dispositive of a summary judgment motion”; and that “the fact
that the evidentiary record is well-developed in this case makes
a Daubert hearing that much less necessary.”

Id. at 70-71.

When, as here, it is overwhelming clear that an expert’s
opinions meet the standards of Daubert and Rule 702, a hearing
would be an empty “formality” and is not required.
506 F.3d at 161.

Williams,

Thus Apple’s request for a hearing is denied.

CONCLUSION
Plaintiffs’ October 11, 2013 motion for class certification
is granted; Apple’s November 15, 2013 motion to exclude Noll’s
opinions is denied; and plaintiffs’ January 27, 2014 request to
strike portions of Apple’s sur-reply memorandum of law in
opposition to plaintiffs’ motion for class certification and the

85

sur-reply declarations of Kalt and Orszag is granted in part.
SO ORDERED:
Dated:

New York, New York
March 28, 2014
__________________________________
DENISE COTE
United States District Judge

86