"The Apple iPod iTunes Anti-Trust Litigation"

Filing 762

Administrative Motion to File Under Seal Apple's Reply Brief in Support of Motion for Summary Judgment 740 filed by Apple Inc.. (Attachments: # 1 Proposed Order, # 2 Declaration of David C. Kiernan in Support of Apple's Admin. Motion to Seal, # 3 Exhibit Exhibits 1-5 to David C. Kiernan Declaration, # 4 REDACTED Reply in Support of Apple's Motion for Summary Judgment, # 5 UNREDACTED Reply in Support of Apple's Motion for Summary Judgment)(Kiernan, David) (Filed on 1/31/2014)

1
2
3
4
5
6
7
8

Robert A. Mittelstaedt #60359
[e-mail]
Craig E. Stewart #129530
[e-mail]
David C. Kiernan #215335
[e-mail]
JONES DAY
555 California Street, 26th Floor
San Francisco, CA 94104
Telephone:
[phone]
Facsimile:
[phone]
Attorneys for Defendant
APPLE INC.

9

UNITED STATES DISTRICT COURT

10

NORTHERN DISTRICT OF CALIFORNIA

11

OAKLAND DIVISION

12
13
14

THE APPLE iPOD iTUNES ANTITRUST
LITIGATION

15

___________________________________

16

This Document Relates To:

17

ALL ACTIONS

18
19

Lead Case No. C 05-00037 YGR
[CLASS ACTION]
DEFENDANT’S REPLY
MEMORANDUM IN SUPPORT OF
MOTION FOR SUMMARY
JUDGMENT AND TO EXCLUDE
EXPERT TESTIMONY OF ROGER G.
NOLL
Date:
Time:
Courtroom:

To be set by Court
2:00 PM
5

20
21
22
23
24

***REDACTED/PUBLIC VERSION***

25
26
27
28
Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

TABLE OF CONTENTS

1
2

Page

3

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

4

ARGUMENT .................................................................................................................................... 1

5

I.

PLAINTIFFS MISSTATE THE SUMMARY JUDGMENT STANDARD. ............................. 1

6

II.

PLAINTIFFS HAVE NO EVIDENCE OF THE CRITICAL FACTS NECESSARY
TO SHOW ANTITRUST IMPACT. ....................................................................................... 2

7
8

III. PLAINTIFFS’ EXPERT TESTIMONY DOES NOT FILL THE VOID. ................................ 7

9

A.

Noll’s Model is Inconsistent with Real-World Facts. ...................................................... 7

10

B.

Noll Improperly Manipulates the iTunes 4.7 and 7.0 Variables. . ................................... 8

11

C.

Noll Fails to Account for Effect of Unchallenged Aspects of iTunes 7.0. ..................... 11

12

D.

Noll Fails to Account for Other Product Characteristics. .............................................. 11

13

E.

Noll’s Model Does Not Produce Statistically Significant Results. ................................ 13

14
15

IV. PLAINTIFFS HAVE NO EVIDENCE OF A PROPER RELEVANT PRODUCT
MARKET. .............................................................................................................................. 14

16

CONCLUSION ............................................................................................................................... 15

17
18
19
20
21
22
23
24
25
26
27
28
i

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

1

TABLE OF AUTHORITIES

2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

Page

CASES

Collins v. Assoc. Pathologists,
844 F.2d 473 (7th Cir. 1988)..................................................................................................... 2
Daubert v. Merrell Dow Pharms., Inc.,
509 U.S. 579 (1993) .................................................................................................................. 7
Freeland v. AT&T Corp.,
238 F.R.D. 130 (S.D. N.Y. 2006) ........................................................................................... 13
GE v. Joiner,
522 U.S. 136 (1997) ................................................................................................................ 11
HDC Med. v. Minntech Corp.,
474 F.3d 543 (8th Cir. 2007)..................................................................................................... 2
Ralph C. Wilson Indus., Inc. v. Am. Broad. Cos.,
598 F. Supp. 694 (N.D. Cal. 1984), aff’d, 794 F.2d 1359 (9th Cir. 1986)................................ 2
Raskin v. Wyatt Co.,
125 F.3d 55 (2d Cir. 1997) ........................................................................................................ 2
Thi-Hawaii, Inc. v. First Commerce Financial Corp.,
627 F.2d 991 (9th Cir. 1980)..................................................................................................... 2

17
18
19
20
21
22
23
24
25
26
27
28
ii

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

1

INTRODUCTION

2

Plaintiffs’ opposition confirms the complete lack of evidence to support their claim.

3

Neither they nor their expert have any evidence to support critical steps in their purported causal

4

chain—e.g. that RealNetworks’ Harmony sales were significant, that iPod users were buying

5

RealNetworks music and using Harmony to play it on iPods, that anyone was forced to buy or not

6

buy an iPod because of iTunes 7.0, or that Apple took any such incremental change in demand for

7

iPods or competing devices into account in pricing iPods. Thus, their claim that iTunes 7.0

8

elevated iPod prices fails at every step. They lack even the most basic facts necessary to support

9

even the possibility of the impact they claim.

10

Plaintiffs’ reliance on their expert fares no better. They do not dispute that Noll has

11

gathered no real-world facts to support his theory. Nor do they answer Apple’s showing that Noll

12

has improperly manipulated his model to jury-rig the results. This is most evident in his

13

treatment of the critical variables for iTunes 4.7 and iTunes 7.0. Plaintiffs do not—and cannot—

14

reconcile that treatment with Noll’s own prior sworn testimony. Indeed, Noll does not attempt to

15

defend his critical decision to change the iTunes 7.0 variable from what he previously admitted

16

was the “right way to do it.” Fixing just this transparent manipulation eliminates the claimed

17

impact and damages. By itself, this is sufficient to exclude the model and grant summary

18

judgment. When the model’s other errors are considered, the case for exclusion is overwhelming.

19

As shown below, plaintiffs also fail to salvage Noll’s opinion on relevant market; they

20

only further demonstrate that he applied no relevant expertise. Summary judgment is thus proper

21

on this additional ground as well.
ARGUMENT

22
23
24

I.

PLAINTIFFS MISSTATE THE SUMMARY JUDGMENT STANDARD.
Contrary to plaintiffs’ assertion (Opp. at 9-10), the Ninth Circuit has no rule that summary

25

judgment should be “sparingly” granted in antitrust cases. While some cases (including the one

26

plaintiffs cite) have urged caution in cases involving issues of motive and intent, any such caution

27

is inappropriate outside that context. As the Ninth Circuit stated, “[t]he suggestion that summary

28

procedures are less appropriate in antitrust cases may be put aside, for the Federal Rules make
1

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

1

absolutely no distinction between antitrust and other cases.” Thi-Hawaii, Inc. v. First Commerce

2

Financial Corp., 627 F.2d 991, 994 n.2 (9th Cir. 1980) (quoting P. Areeda and D. Turner,

3

Antitrust Law ¶ 316b (1978)).

4

Indeed, if anything, “[g]iven the enormous expenditure of time and other resources

5

commonly necessary in antitrust cases, the complexity of the issues and general inexperience of

6

jurors in this area, as well as the need for uniformity and foreseeability in interpretation of the

7

antitrust laws, the court should, rather, be particularly alert to granting summary judgment in

8

appropriate cases.” Ralph C. Wilson Indus., Inc. v. Am. Broad. Cos., 598 F. Supp. 694, 699 (N.D.

9

Cal. 1984), aff’d, 794 F.2d 1359 (9th Cir. 1986). 1

10

Nor, contrary to plaintiff’s suggestion, is “an expert’s report . . . a talisman against

11

summary judgment.” Raskin v. Wyatt Co., 125 F.3d 55, 66 (2d Cir. 1997). As Apple’s motion

12

shows, summary judgment has frequently been granted in antitrust cases where the court,

13

exercising its gatekeeper responsibility, finds the expert’s testimony unreliable, contradictory or

14

inconsistent with record facts.

15

II.

PLAINTIFFS HAVE NO EVIDENCE OF THE CRITICAL FACTS NECESSARY

16

TO SHOW ANTITRUST IMPACT.

17

Apple’s motion showed that plaintiffs cannot prove any of the critical links in their

18

improbable theory that disabling a single, insignificant source of music for iPods triggered a

19

cascading series of events that ultimately (or, under plaintiffs’ theory, immediately) resulted in

20

Apple overcharging for iPods. Plaintiffs’ opposition confirms this failure of proof.
First, plaintiffs effectively admit that they have no evidence RealNetworks was a

21
22

significant market participant at the time. They cite (Opp. at 11) a market analyst’s report that

23

RealNetworks had 2.4 million subscribers. But they offer no evidence that these subscribers

24

purchased permanent downloads on which Harmony worked, rather than using RealNetworks’

25

1

26
27
28

See also Collins v. Assoc. Pathologists, 844 F.2d 473, 475 (7th Cir. 1988) (“Supreme Court
made clear that, contrary to the emphasis of some prior precedent, the use of summary judgment
[in antitrust cases] is not only permitted but encouraged”); HDC Med. v. Minntech Corp., 474
F.3d 543, 546-47 (8th Cir. 2007) (“where there has been ample opportunity for discovery,
summary judgment is appropriate antitrust cases, just as in any other litigation”).
2

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

1

Rhapsody subscription service. See ECF No. 751-6 at 11 (describing Harmony as allowing users

2

to transfer downloads from the RealPlayer Music Store); see also “RealNetworks rekindles iPod

3

Tech Tussle (Apr. 26, 2005) (noting Harmony worked only on downloads, not subscription

4

service). 2 Nor do plaintiffs dispute that Real had only a 3% share of the digital downloads at the

5

time. Plaintiffs assert (Opp. at 4) that Apple’s share of digital music sales dropped when

6

Harmony was launched. But that was in 2004, when RealNetworks promoted Harmony for three

7

weeks with a half-price sale. RealNetworks’ sales in 2004 are irrelevant here, because the

8

disabling of Harmony in 2004 as a result of iTunes 4.7 was lawful and plaintiffs’ only remaining

9

challenge is to iTunes 7.0 issued in 2006. See ECF No. 627. 3
Second, plaintiffs offer no evidence that any of RealNetworks’ sales were to iPod users

10
11

(or potential users). They dismiss that information as irrelevant, asserting that effect on demand

12

is what is important. ECF No. 751-21 at Fact 3 (“Sep. St.”). 4 But that is exactly why the missing

13

evidence is critical—demand could not have been affected if iPod users (or potential buyers) were

14

not purchasing RealNetworks’ downloads in the first place.

15

Plaintiffs suggest that sales to iPod users can be inferred from customer complaints Apple

16

received “after Apple blocked Harmony through 7.0.” Opp. at 11. In fact, none of the complaints

17

plaintiffs cite says anything about Harmony. Nearly all of them were made before the

18

issued—and thus could not have been about any effect from the

19

Exhibit 35. 5 Instead, the customers were asking whether their DRM-protected music from stores

20
21
22
23
24
25
26
27
28

2

was

See ECF No. 751-12 at

Available at: http://news.cnet.com/2100-1027_3-5685286.html (Last accessed, Jan. 31, 2013).

3

Indeed, many of the “facts” outlined in plaintiffs’ opposition relate to events before iTunes 4.7
and thus are irrelevant. See, e.g., ECF No. 751-21 at Fact 45.
4

Plaintiffs’ Responsive Separate Statement includes purported “additional material facts” (Nos.
42-48). Apple is not filing a separate responsive document to those purported facts, as the
Court’s Standing Order does not appear to contemplate or authorize one. Plaintiffs’ asserted
facts, however, are immaterial (e.g., Facts 42, 45) or unsupported (Facts 43, 44, 45, 46, 47, 48)
for the reasons stated in Apple’s moving and reply papers.
5

Of the 19 complaints plaintiffs cite, only four were made after September 12, 2006, when the
was released for new iPod nanos. Two of those complaints were from customers who had
purchased music from the iTS and wanted to play that music on a non-iPod. That inquiry has
nothing to do with the
or Harmony. The other two were from consumers who wanted to
(continued)
3

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

1

such as Napster, Yahoo, or Walmart would play on their iPod. These inquiries reflect only that

2

these stores’ DRM was incompatible with Apple’s FairPlay. Plaintiffs do not claim this had

3

anything to do with iTunes 7.0. It was simply the result of Apple using FairPlay (rather than

4

another DRM) which Judge Ware ruled was lawful. ECF No. 303, pp. 2, 10; ECF No. 274, p. 10.

5

Nor can Harmony sales to iPod users be inferred from the reaction of Apple’s executives

6

to Harmony. Plaintiffs cite documents from 2004 when RealNetworks first broke FairPlay’s

7

code. Opp. at 11. Again, however, Apple’s issuance of iTunes 4.7 in 2004 was lawful, and the

8

issue now is whether Harmony sales in 2006 were sufficient to affect iPod prices. None of the

9

documents plaintiffs cite shed any light on that question.

10

Third, even if iPod owners (or potential owners) were buying from RealNetworks,

11

plaintiffs still do not point to any consumer who became locked in or locked out as a result. They

12

cite testimony from two of the named plaintiffs that they bought an iPod because of their iTS

13

purchases. Opp. at 5. But neither plaintiff testified that this had anything to do with Harmony or

14

any updates to iTunes. Neither said he or she purchased from RealNetworks or would have done

15

so absent the

16

whose decision to buy an iPod was influenced by the disabling of Harmony.

17

, and was locked in or out as a result. Nor do plaintiffs identify anyone else

Plaintiffs suggest they should be excused from offering this pivotal evidence because they

18

“have no way of identifying Harmony users, or determining exactly how many customers were

19

locked in or locked out.” Opp. at 11. This argument is flawed at every turn. First, litigants

20

cannot get to trial—particularly in a suit of this size—simply by throwing up their hands and

21

proclaiming that the necessary evidence cannot be found. Second, plaintiffs have not shown that

22

the evidence is unavailable. This case has been pending for over eight years, and iTunes 7.0 was

23

released over seven years ago. Plaintiffs have had ample opportunity to do whatever

24

investigation was necessary—including consumer surveys, questionnaires and third-party

25

discovery from RealNetworks or others—to establish the factual basis for their claim, if one

26

existed. The point is not simply that plaintiffs do not know “exactly” how much music

27
28

play music from Napster or BearShare.com—again, an issue unrelated to the
4

or Harmony.

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
8

22
23

In short, after years of litigation with every opportunity to develop the facts supporting

24

their claim if any existed, plaintiffs offer only argument, not facts. This failure of proof requires

25

that summary judgment be granted.

26
27

8

28
6

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

1
2

III.

PLAINTIFFS’ EXPERT TESTIMONY DOES NOT FILL THE VOID.
Noll’s regression does not cure this failure of proof. That regression is unsupported by

3

any real-world facts, produces results contrary to the known facts, and has been improperly

4

manipulated to show impact where there is none.

5

Predictably, plaintiffs argue that regression models are “standard and widely accepted”

6

and that Apple’s objections must be treated as going to weight rather than admissibility. Opp. at

7

16. But the courts have rejected this head-in-the-sand approach to expert testimony. Instead,

8

exercising their “gatekeeping” role (Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579, 597

9

(1993)), they have frequently excluded expert regression models that fail to meet basic standards

10

of reliability. See Mot. at 18 n.5 (citing cases). Regression models must be consistent with the

11

real-world facts and with accepted econometric principles. Noll’s regression fails that standard.

12

A.

13

As shown above, plaintiffs have no valid answer to the point that Noll’s regression results

14

are inconsistent with the indisputable evidence that RealNetworks was not a significant source of

15

music downloads for iPod owners, that the postulated lock-out or lock-in effect did not occur, or

16

that Apple was not aware of, and did not, consider any such lock-out or lock-in from the

17

setting its prices. Nor can plaintiffs reconcile Noll’s results with the uncontradicted facts

18

regarding how Apple sets its prices.

19

Noll’s Model is Inconsistent with Real-World Facts.

in

Likewise groundless is plaintiffs’ effort to defend the immediate and constant overcharge

20

postulated by Noll’s model. They assert that it makes sense the overcharge amount would be

21

immediate and unchanging because the alleged “lock-out” effect would occur immediately and

22

remain constant throughout the period, while the alleged “lock-in” effect would not kick in at all

23

during the damage period. Opp. at 15. As discussed, however, there is no factual basis for the

24

posited immediate “lock-out” effect because there is no evidence that Apple knew about or

25

considered any effect of the

26

customers.

27
28

on elasticity of demand resulting from supposedly locked out

As for the lock-in effect, plaintiffs quote Noll’s rebuttal report as saying that “for new
iPod purchasers” the lock-in effect “would not be an important factor affecting iPod prices.” Id.
7

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

1

At his deposition, however, Noll admitted that, by “new iPod purchasers,” he meant only

2

consumers who had not purchased an iPod before September 2006. Ex. 11 at 51:9-15. For those

3

consumers, his view is that, on average, they would not be purchasing a new player for another 18

4

to 24 months—which, according to him, means any increased demand for iPods from those

5

consumers would not be felt until then. Id. at 51:16-52:1. He admitted, however, that (under his

6

theory) the lock-in effect would occur earlier for consumers who already owned iPods. As to

7

those consumers, he said he would expect they would be buying new iPods earlier and the lock-in

8

effect as to them would therefore impact demand during the damages period. ECF No. 751-15 at

9

Exhibit 50, 50:19-52:11. And he admits that (again, under his theory) that lock-in effect—

10

beginning within the damage period—would increase over time. Ex. 8 at 7. These admissions—

11

lock-in for some consumers, lock-out for others, with differing times of impact for each—are

12

irreconcilable with the single, unchanging overcharge amount predicted by his damages model.

13

B.

Noll Improperly Manipulates the iTunes 4.7 and 7.0 Variables.

14

Contradicting a basic econometric principle that he previously endorsed Noll changed his

15

latest regressions to turn off the iTunes 4.7 variable as of September 2006 for iPod models on

16

which iTunes 7.0 was implemented and turn on the iTunes 7.0 variable only for those same

17

models. The right approach would have been to leave on both variables for all iPod models. As

18

shown in Apple’s motion, correcting for this manipulation results in no impact or damages.
iTunes 4.7 variable: Plaintiffs’ sole argument for turning off the iTunes 4.7 variable is

19
20

that the

21

This argument fails for two reasons. First, turning off the iTunes 4.7 variable is improper because

22

it assumes that, in the actual world, iTunes 4.7 had no possible continuing effect. But Noll

23

testified that iTunes 4.7’s blocking of Harmony in 2004 could continue to affect demand for

24

iPods even after iTunes 7.0 was released and replaced iTunes 4.7. Ex. 10 at 68:2-70:16 (“Q: And

25

could that consumer expectation continue even after 7.0 is issued? A: Exactly, it could.”). 9 Thus,

26
27
28

in iTunes 7.0 supposedly “replaced” iTunes 4.7’s DRM technology. Opp. at 16.

9

Contrary to plaintiffs’ footnote 17, Noll’s testimony is clear that iTunes 4.7 could continue to
have an effect after iTunes 7.0. The supposedly contrary testimony cited by plaintiffs is directed
at the continuing effects of iTunes 7.0. He said that iTunes 7.0 might have had a “similar story”
(continued)
8

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

1

the proper treatment of iTunes 4.7 was to leave it on for all models.

2

Second, by turning off the iTunes 4.7 variable, Noll’s regressions cannot begin to answer

3

the critical question of what the price would have been but for iTunes 7.0. See Mot. at 15-16; Ex.

4

4 ¶¶ 88-97; Ex. 14 ¶ 13. To answer that question, the regression must compare the prices after

5

iTunes 7.0 replaced iTunes 4.7 to the prices before iTunes 7.0 replaced iTunes 4.7—i.e., a world

6

in which iTunes 4.7 existed. But turning off the iTunes 4.7 variable forced the regressions to

7

compare prices after iTunes 7.0 replaced iTunes 4.7 to a period before iTunes 4.7 was released

8

(the three-month period from July to October 2004)—i.e., a world in which iTunes 4.7 did not

9

exist. This causes the iTunes 7.0 variable to estimate the cumulative impact of iTunes 4.7 and

10

iTunes 7.0. Ex. 4 ¶¶ 88-97. 10

11
11

12
13

In his “supplemental rebuttal” report, Noll tries to defend using the wrong but-for world

14

by asserting that he has seen no evidence that iTunes 4.7 would have continued to work on iPods

15

if iTunes 7.0 had not been implemented—i.e., no evidence that iTunes 4.7 would have continued

16

to exist in the but-for world. ECF No. 751-5 at 14. But plaintiffs’ liability case is premised on

17

the notion that Harmony would have continued to work in the but-for world. For that to be the

18

case, iTunes 4.7 would have had to continue to work on iPods because Harmony, as relaunched in

19

to iTunes 4.7 in causing consumers to stay away from Harmony even after iTunes 7.0 was
replaced or overcome. “But, in fact, 7.0 was never undone, so we can’t test that hypothesis.” Ex.
10 at 68:14-20. This testimony has nothing to do with testing for the continuing effect of iTunes
4.7.

20
21
22
23
24
25
26
27

10

Noll has admitted that leaving on the iTunes 4.7 variable would cause the iTunes 7.0 variable to
estimate only the incremental impact of iTunes 7.0 on price. He explained that where variable A
(which represents a given technology) is left on after variable B (which represents new
technology that replaces variable A) is turned on, variable B “measure[s] the incremental” impact
on price “as a result from moving from” one technology (variable A) to the new technology
(variable B). The effect then of turning off variable A is to cause variable B to measure the
cumulative effect of both technologies. See ECF No. 685 at 10-11.
11

As Apple’s experts and Noll explain, to determine the incremental effect of iTunes 7.0, one
need only subtract the coefficient of iTunes 4.7 from the coefficient of iTunes 7.0. See Ex. 4 at
41; see also ECF No. 685 at 11.

28
9

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

1

Apple marketed these features to attract customers (e.g., battery life, display size and screen

2

resolution),

3

They then confirmed the

4

importance of the features by showing that adding them significantly improves the explanatory

5

power of the regressions while not affecting the precision of the coefficient on the iTunes 7.0

6

variable. Ex. 3 at 56 & Exhibit 15 thereto, Ex. 4 at 48-49 & Exhibit 13 thereto, Exhibit 14 at 10-

7

11 & Exhibit JT-5 thereto. Plaintiffs do not answer any of these points.

8

Plaintiffs’ multicollinearity argument is likewise flawed. As Noll recognized,

9

multicollinearity among a set of explanatory variables may affect the precision of the estimates

10

for some variables but not others. Ex. 8 at 31; see also ECF No. 754-8 at Exhibit 3 (American

11

Bar Assoc., Proving Antitrust Damages (2d. Ed. 2010) at 151 (hereafter “ABA”). And

12

multicollinearity may present a problem only when the variables in question are highly collinear

13

with the variable of interest, here the iTunes 7.0 variable: “[i]f the precision of the coefficient

14

estimate on the variable of primary interest [here iTunes 7.0 variable] is not substantially affected

15

by the multicollinearity, there is no reason to drop any variables that have a strong economic

16

rationale for inclusion…. dropping variables that are truly important could lead to omitted

17

variable bias…. Therefore, in general the best course of action in that case would be to retain

18

the collinear variables.” ABA 151 (emphasis added).

19

Here, although Noll says he belatedly ran tests that show that the omitted variables are

20

collinear with explanatory variables in the model, none of those tests show that they are highly

21

collinear with the iTunes 7.0 variable as opposed to other variables in the model. See ECF No.

22

751-5 at 10-12. By contrast, applying a standard test, Apple’s experts show that the omitted

23

variables are not highly collinear with the iTunes 7.0 variable and have no impact on the precision

24

of the estimated coefficient for that variable. Ex. 14 at 12-13 & Exhibit JT-6 thereto. 13

25

13

26
27
28

Plaintiffs assert that “Professor Murphy admitted to running no tests to determine whether there
was a multicollinearity problem.” Opp. at 19. This is contradicted by the very pages they cite.
When asked if he tested whether the additional variables are collinear with the existing variables,
Murphy answered, “Yes, I did,” and then described those tests. ECF No. 751-15 at Exhibit 53,
294:2-15.
12

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

1

Finally, if Noll were correct that adding the omitted features would cause severe

2

multicollinearity with the iTunes 7.0 variable, that is an added reason to exclude his regressions

3

as unreliable. See ABA 152 (“With severe multicollinearity between the variable of interest and

4

other economically important explanatory variables, the coefficient on the variable of interest

5

cannot be estimated precisely, and the regression model may therefore be unreliable for

6

estimating damages.”). As Noll has recognized, the standard procedure for handling

7

multicollinearity is to drop one or both of the variables that are highly collinear with each other.

8

ECF No 751-5 at 10. But when the “variable of interest is affected by the multicollinearity, . . .

9

[g]enerally no valid basis exists for choosing to retain the variable of interest [here iTunes 7.0] as

10

opposed to retaining the variable(s) with which it is collinear.” ABA 151. And “[d]ropping

11

explanatory variables to reduce the multicollinearity without a sound economic reason does not

12

fix the problem”; indeed, it “‘amounts to elevating ignorance to arrogance.’” ABA 152 & n.2

13

(quoting Jacques Dreze, Comment, Nonspecialist Teaching of Econometrics: A Personal

14

Comment and Personalistic Lament, 2 Econometric Reviews 291, 296 (1983)); see also Freeland

15

v. AT&T Corp., 238 F.R.D. 130 (S.D. N.Y. 2006) (rejecting multicollinearity argument and

16

excluding regressions for omitting variables).

17

E.

Noll’s Model Does Not Produce Statistically Significant Results.

18

Plaintiffs’ attempt to defend Noll’s calculation of his standard errors (and thus the claimed

19

statistical significance of his results) consists of summarizing the arguments in plaintiffs’ recent

20

motion to exclude Dr. Murphy’s and Topel’s opinions on this issue. ECF No. 737-4. Apple has

21

opposed that motion (ECF No. 754-6) and respectfully refers the Court to that brief. As shown

22

there, plaintiffs’ arguments are refuted by their own newly (and belatedly) offered expert on this

23

issue. Contrary to plaintiffs’ assertion that correlation of standard errors can never be a problem

24

when the regression uses the entire population of data rather than a sample, their new expert

25

admits that his graduate textbook in econometrics says the opposite, as does the ABA’s treatise

26

on econometrics in antitrust cases. The new expert’s textbook similarly endorses the use of the

27

“ex post” clustering that he and plaintiffs claim was improperly used by Apple’s experts. And

28

plaintiffs’ argument that clustering in these circumstances can bias the results by overestimating
13

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

1

the standard errors is contradicted by the very authorities on which Noll and the new expert rely.

2

Far from refuting Apple’s showing that Noll’s results are not statistically significant, plaintiffs

3

have only confirmed as much, further demonstrating that his model should be excluded.

4

IV.

5

PLAINTIFFS HAVE NO PROOF OF A PROPER PRODUCT MARKET.
Plaintiffs do not dispute that, without Noll’s testimony, they lack sufficient evidence to

6

withstand summary judgment on the issue of the relevant product market—another key element

7

of their claim. The question then is whether Noll has the relevant expertise and has done the

8

requisite analysis to make his opinion on this issue admissible. He has not.

9

Plaintiffs admit that Noll did not estimate cross-elasticity of demand or employ the

10

hypothetical monopolist test. They claim that the data to conduct these accepted economic

11

methods was “unavailable.” Opp. at 23. But neither they nor Noll cite any evidence in support,

12

nor offer any explanation why they could not have obtained data “on percentage changes in price

13

and quantity from competing firms.” Id. at 23:4-5. Noll did not even attempt such analysis.

14

But even if Noll had adequately demonstrated that no quantitative analysis was possible,

15

his testimony is still inadmissible because he did not engage in any other reliable expert analysis.

16

As evidence of Noll’s supposed expert analysis, plaintiffs point to a list of materials Noll

17

purportedly considered. Opp. at 24. This argument is flawed for at least two reasons.

18

First, plaintiffs do not show that Noll employed any relevant expertise in reviewing these

19

materials. They do not claim that he has worked in the music or music player industry or

20

otherwise possesses specialized knowledge about such industries or about what participants in

21

these industries perceive about competition. His discussion about what competitors or consumers

22

thought, or about supposed product similarities or dissimilarities, is thus not informed by any

23

particular expertise beyond the ability to search the internet and read a report. Plaintiffs assert he

24

has “45 years of experience in analyzing the economics of the communications industry” (Opp. at

25

24:7), but they do not explain how that qualifies him to opine on whether CDs compete with

26

digital music or what devices consumers consider as substitutes to digital players. See Mot. § IV.

27
28

Second, reflecting that he was simply reading documents without any relevant expertise,
Noll’s analysis is fundamentally flawed, incomplete and misleading.
14

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR

1
2
3
4
5
6
7
8
9
10
11

Noll’s purported reliance on other sources is equally misguided. For example, plaintiffs’

12

claim that Noll relied on industry analysts, citing to page 28 of his report. But the cited articles

13

assert, contrary to Noll’s opinion, that smartphones are substitutes for iPods. Noll asserts this did

14

not actually happen until after the damage period. But he relied for that assertion on his belief

15

that smartphones could not replace iPods until smartphones became capable of downloading

16

music over a wireless network—ignoring that iPods have never had the capability.

17

In short, this portion of Noll’s report is not expert testimony at all. It is simply his own

18

untrained, subjective and deeply flawed view of an issue as to which he failed to conduct the

19

required investigation or analysis. It should be excluded and summary judgment granted on the

20

additional ground that plaintiffs lack evidence to establish a proper relevant product market.
CONCLUSION

21
22
23

For the reasons above, Apple is entitled to summary judgment.
Dated: January 31, 2014

24

Respectfully submitted,
By: /s/ Robert A. Mittelstaedt
Robert A. Mittelstaedt

25
26

Counsel for Defendant APPLE INC.

27
28

SFI-851314v6

15

Reply ISO Mot. for Sum. Jgmt. and
To Exclude -- C 05-00037 YGR