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

Filing 235

MOTION in Limine to Preclude Dr. Michelle Burtis from Offering at Trial Any Opinion on Competitive Effects. Document filed by Unites States of America.(Fairchild, Stephen)

UNITED STATES DISTRICT COURT
FOR THE SOUTHERN DISTRICT OF NEW YORK

__________________________________________
)
UNITED STATES OF AMERICA,
)
)
Plaintiff,
)
)
v.
)
)
APPLE, INC., et al.,
)
)
Defendants.
)
__________________________________________)

Civil Action No. 12-cv-2826 (DLC)

__________________________________________
)
)
)
)
Plaintiffs,
)
)
v.
)
)
PENGUIN GROUP (USA) INC. et al.,
)
)
Defendants.
)
__________________________________________)
THE STATE OF TEXAS;
THE STATE OF CONNECTICUT; et al.,

Civil Action No. 12-cv-03394 (DLC)

PLAINTIFFS’ MOTION IN LIMINE
TO PRECLUDE DR. MICHELLE BURTIS FROM OFFERING AT TRIAL
ANY OPINION ON COMPETITIVE EFFECTS

INTRODUCTION
Apple and Penguin’s joint economic expert, Dr. Michelle Burtis, intends to offer at trial
the opinion that the Apple Agency Agreements did not result in anticompetitive effects. Dr.
Burtis’s conclusion is based on her observations that, by April of 2012 (when the Department of
Justice filed its Complaint), e-book prices were lower on average (although not for Publisher
Defendants’ e-books), e-book output was higher, and e-reader devices were more advanced
technologically, than they were in April of 2010 (when the Apple Agency Agreements went into
effect). The problem is that in examining how the competitive landscape changed over those two
years, Dr. Burtis did not even attempt to isolate the impact caused by the Apple Agency
Agreements from any other trends or changes taking place in the industry. Indeed, in
contravention of basic economic effects analysis, Dr. Burtis made no attempt, in any way, to take
into account or control for any variables that may have affected price, output, or innovation
during her two-year window. Thus, the testimony Dr. Burtis intends to offer on competitive
effects is at best, unhelpful, and at worst, misleading.
The failure by Dr. Burtis to attempt to account for obvious alternative explanations that
affected price, output, and innovation means that her analysis failed to employ “the same level of
intellectual rigor that characterizes the practice of an expert in the relevant field,” Kumho Tire
Co. v. Carmichael, 526 U.S. 137, 152 (1999), and is so incomplete as to be inadmissible because
it is irrelevant, see Bickerstaff v. Vassar Coll., 196 F.3d 435, 449 (2d Cir 1999); see generally
Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579 (1993). Accordingly, Plaintiffs respectfully
submit that, pursuant to Federal Rules of Evidence 702 and 401, Dr. Burtis should be precluded

1

from offering any opinion at trial relating to competitive effects from the Apple Agency
Agreement.
BACKGROUND
By Dr. Burtis’s own admission, the average retail prices of Publisher Defendants’ ebooks increased as a result of the Apple Agency Agreements (Burtis Dep. (Attached hereto as
Ex. 1) 76:2-77:3) and remained higher two years after those agreements went into effect (Burtis
Report (Attached hereto as Ex. 2) ¶ 5a & graph 1; Burtis Dep. 81:25-82:8). Dr. Burtis
nonetheless opines that no anticompetitive effects resulted from the Apple Agency Agreements.
(Burtis Dep. 101:14-102:2.) In fact, Dr. Burtis concludes that the economic evidence is
“inconsistent with allegations that the agency agreements had an anticompetitive effect in the
alleged trade eBook market (after April 1, 2010).” (Burtis Report ¶ 23.) In support of that
conclusion, Dr. Burtis states:
Most significantly, the average retail price of eBooks was lower immediately prior to
when the DOJ filed its Complaint in April 2012 than when Apple opened the iBookstore
in April 2010; output has increased dramatically—122 million eBooks (including nearly
30 million paid eBooks) were downloaded from Apple’s iBookstore alone during the
same period; and innovation, in terms of the variety of eReader single- and multi-function
devices has continued apace. (Burtis Report ¶ 23.)
In other words, Dr. Burtis has concluded that there were no anticompetitive effects from
Defendants’ conduct because: (1) the average price of an e-book industry-wide was lower in
April 2012 than it was in April 2010; (2) there were a lot of e-books sold after the Apple Agency
Agreements went into effect; and (3) there has been a lot of device innovation since April 2010.

2

I.

PRICE AND OUTPUT
During the course of her deposition, Dr. Burtis testified that, with respect to analyzing the

effects of the Apple Agency Agreements, prices and output prior to the agency agreements did
not constitute a “but-for world.” (See, e.g., Burtis Dep. 108:20-109:22.) Thus, Dr. Burtis
acknowledges that, even in the absence of the Apple Agency Agreements, prices, output and
innovation would have changed between April 2010 and April 2012. But neither Dr. Burtis nor
anyone on her staff conducted a regression analysis or undertook any other efforts to attempt to
control for those non-agency related changes.1 (Burtis Dep. 61:20-62:22; 69:15-21.) Dr. Burtis
fails to control even for the most obvious factors affecting e-book prices, such as the continuing
growth of low-priced, self-published e-books and the changing mix of titles that are offered in
digital formats as traditional publishers continue converting their print backlists to digital. Given
her failure to control for alternative explanations, Dr. Burtis cannot credibly conclude, as she
nonetheless has, that average retail e-book prices being lower in April 2012 than in April 2010 is
inconsistent with claims of anticompetitive effects. (Burtis Dep. 80:12-81:3.) Similarly, given
Dr. Burtis’s inability to quantify how much of the post-April 2010 increase in e-book output
(sales) was due to the Apple Agency Agreements (Burtis Dep. 161:19-162:16)—as opposed to,
for example, preexisting trends or the growing base of e-readers, tablets, smartphones, and other
devices on which consumers read e-books—Dr. Burtis cannot credibly conclude, as she
nonetheless has, that the growth in output is inconsistent with allegations of anticompetitive
effects (Burtis Dep. 164:22-165:8).
1

When directly questioned as to what the average retail price of e-books would have been in April 2012 absent the
Publisher Defendants entering into the Apple Agency Agreements, Dr. Burtis testified that she could not answer the
question. (Burtis Dep. 128:25-129:12.)

3

Dr. Burtis’s failure to control for non-agency effects is not merely a theoretical problem,
but rather one that has tangible impacts on her analysis.2 As Dr. Burtis acknowledges, between
2010 and 2012 there was “tremendous growth” in sales of self-published books (Burtis Dep.
162:17-24)—which typically cost less than books sold by Publisher Defendants (Burtis Dep.
212:7-12). Most of this tremendous sales growth occurred at Amazon, and Dr. Burtis admits that
at least a large portion of the growth was due to Amazon sweetening its royalty terms for selfpublished authors. (Burtis Dep. 189:3-192:9; Burtis Rebuttal Report (Attached hereto as Ex. 3)
¶ 19.) However, Amazon changed its royalty terms for self-published authors approximately a
week before the Apple Agency Agreements were ever signed. (Burtis Dep. 190:4-192:9.) Thus,
it is obvious that the tremendous growth in sales of self-published books occurred largely, if not
entirely, for reasons other than the Apple Agency Agreements. As such, it makes no sense for
Dr. Burtis to include Amazon self-publishers’ sales in determining the impact of the Apple
Agency Agreements. Yet Dr. Burtis did precisely that. (See, e.g., Burtis Dep. 214:5-19.) Even
more troubling, Dr. Burtis includes those sales despite acknowledging that doing so deflated the
April 2012 average retail price and inflated the April 2012 output level upon which she bases her
conclusions regarding competitive effects. (See Burtis Dep. 214:20-25.)
II.

INNOVATION
With respect to innovation, Dr. Burtis did not attempt any calculations. The sum total of

Dr. Burtis’s “analysis” is contained in three paragraphs of her initial report (and one chart) in
which she lists the devices that were released since April 2010 and states: “since the agency

2

What follows is only one example that illustrates the seriousness of Dr. Burtis’s failure to control for alternative
factors.

4

agreements went into effect, eBook retailers have introduced many new and innovative eReader
devices and tablets at lower prices.” (Burtis Report ¶¶ 32-34 & ex. 3.) Dr. Burtis acknowledges
that she cannot draw a causal link between the Apple Agency Agreements and the introduction
of any devices identified in her report (Burtis Dep. 116:11-117:2), and that she did not undertake
any examination to determine which devices would not have been introduced but for the Apple
Agency Agreements (Burtis Dep. 117:22-118:6).3 In fact, falling prices and increased
functionality have long been a hallmark of the development of most technology.4
ARGUMENT
The problems with Dr. Burtis’s analysis are evident on their face: without controlling
for other variables, the simple fact that e-book prices were lower on average two years after the
conspiracy went into effect says nothing as to whether the conspiracy caused consumers to pay
higher prices than they would have otherwise. Similarly, the fact that more e-books were sold
post-agency than pre-agency is not probative of whether the Apple Agency Agreements
depressed output in the market. And Dr. Burtis does not even attempt to compare innovation
pre- and post- agency, but simply draws conclusions based on the fact that, post-agency, many
devices were introduced at cheap prices. For these reasons, Dr. Burtis’s competitive effects
testimony should be excluded.
For expert testimony to be admissible under Rule 702 of the Federal Rules of Evidence, it
must “rest[] on a reliable foundation and [be] relevant to the task at hand.” Daubert, 509 U.S. at
3

Indeed, Dr. Burtis indicated that it is difficult to separate the effects on innovation caused by the Apple Agency
Agreements from those caused by the introduction of the iPad. (Burtis Dep. 119:23-120:10.)
4

See, e.g., Gautam Gowrisankaran & Marc Rysman, “Dynamics of Consumer Demand for New Durable Goods,”
Working Paper 14737, National Bureau of Economic Research, Feb. 2009, available at
http://nber.org/papers/w14737, at 2.

5

597. A court has an obligation to act as a gatekeeper to ensure the “reliability and relevancy of
expert testimony.” Kumho Tire, 526 U.S. at 152. One factor that courts regularly examine in
determining whether expert testimony is sufficiently reliable to be considered by the trier of fact
is whether the expert has adequately accounted for obvious alternative explanations. See, e.g.,
Claar v. Burlington N. R.R., 29 F.3d 499, 502 (9th Cir. 1994). While it is true that, in general, an
expert’s “failure to include variables will affect the analysis’ probativeness, not its
admissibility,” the failure to account for major factors may render an analysis “so incomplete as
to be inadmissible as irrelevant” under Rule 401. Bickerstaff, 196 F.3d at 449.
Courts in this Circuit, including this Court, have ruled that the failure of an expert to
account for alternative explanations in an analysis may render the expert’s opinion inadmissible.
For example, in Raskin v. Wyatt Co., the Court upheld the exclusion of an economist’s report
analyzing data in an age discrimination case on the basis that the expert had failed in his analysis
to control for factors that could have impacted his conclusions. 125 F.3d 55, 67-68 (2d Cir.
1997). As Dr. Burtis did here, the expert in Raskin examined data over approximately two years
and reached conclusions by attributing all effects observed to the alleged conduct. Id. The
Second Circuit held that the expert’s arriving at his conclusions while making “no attempt to
account for other possible causes” made the exclusion of the report appropriate. Id.
Similarly, in In re Executive Telecard, Ltd. Securities Litigation, the court excluded a
damages analysis because the expert “fail[ed] adequately to distinguish between fraud related
and non-fraud related company specific influences on [the company’s] stock price.” 979 F.
Supp. 1021, 1025-26 (S.D.N.Y. 1997). In making its ruling, the court noted that the expert’s
failure “to isolate the influences of the information specific” to the company meant that the
6

expert was employing a “flawed analysis” that called into question the “magnitude and direction
of his value estimates” and rendered his opinion inadmissible under Fed. R. Evid. 702. Id. at
1026 (citing In re Oracle Sec. Litig., 829 F. Supp. 1176, 1181 (N.D. Cal. 1993)). And this Court,
in In re Wireless Telephone Services Antitrust Litigation, found an expert’s regression analysis
“methodologically unsound and therefore [inadmissible] pursuant to Rule 702” because the
expert failed to introduce any independent variables into his analysis to account for alternative
explanations that could have affected his analysis. 385 F. Supp. 2d 403, 427-28 (S.D.N.Y.
2005); see also Malletier v. Dooney & Bourke, Inc., 525 F. Supp. 2d 558, 667-68 (S.D.N.Y.
2007) (upholding findings of special master that, because expert testimony “did not take account
of any of these plausible alternative causes, the regression analysis must be excluded under Rule
702”).
During his deposition, the United States’ expert, Professor Richard Gilbert, explained the
fatal flaw in Dr. Burtis’s effects analysis:
we all know that many things are happening in the e-book industry. There are new
devices. It’s a maturing industry. There’s all kinds of things coming on -- new
publishers. And what you want to do is isolate those effects that are attributable
to the adoption of the agency contract. So the way to do it is -- using classical
event studies is to use narrow windows before and narrow windows after,
immediately after, in order to capture the effect of that event. Now, you can do it
with longer time frames, provided you can control for all these other effects. And
the longer those windows, the more difficult it’s going to be to control for those
effects. . . . So you can use longer windows. But to simply use long windows
without controls, as Dr. Burtis did, is clearly incorrect. (Gilbert Dep. 350:2-24.)
Notably, Professor Gilbert is not the only expert in this case who has explained that
failing to control for outside variables (either via a regression analysis or a short window)
renders an opinion unreliable. Penguin’s other expert, Professor Daniel Rubinfeld, has written

7

extensively on the problem of “spurious correlation” between two variables, and the fact that “an
attempt should be made to identify all known or hypothesized and measurable major explanatory
variables.”5 And in fact, in previous work for Apple, Dr. Burtis herself criticized an opposing
expert for doing exactly what she did here—failing to take into account variables that could have
impacted prices. See Somers v. Apple, Inc., 258 F.R.D. 354, 360-61 (N.D. Cal. 2009)
(“Ultimately, [Dr. Burtis’s] testimony was that Dr. French had failed to propose a model that
could adequately account for the morass of variables that make up the iPod pricing dynamic.”).
While Plaintiffs do not dispute Dr. Burtis’s qualifications as an expert, her analysis here
does not meet the standard for admissibility under Daubert and subsequent cases. Dr. Burtis
could have chosen a short window for analyzing the effects of the Apple Agency Agreement—
thereby minimizing the likelihood that other factors would influence her results. But Dr. Burtis
chose not to. Instead, Dr. Burtis decided to employ a two-year window. Though Plaintiffs do
not suggest a long window is necessarily inappropriate for effects analysis, by choosing such a
window Dr. Burtis introduced the likelihood that her data would contain a significant amount of
“noise.” At that point, in order for her analysis to be reliable, it was incumbent on Dr. Burtis to
attempt to control for that noise. Most, if not all, of the alternative factors that, due to the
lengthy window, likely affected e-book pricing and output were quantifiable and could therefore
have been analyzed in a regression analysis. See Malletier, 525 F. Supp. 2d at 668. Dr. Burtis’s

5

Daniel L. Rubinfeld & Peter O. Steiner, Quantitative Methods in Antitrust Litigation, 46 LAW & CONTEMP. PROBS.
69, 90 (Fall 1983); see also Daniel L. Rubinfeld, Reference Guide on Multiple Regression, in Reference Manual on
Scientific Evidence 179, 184 (Fed. Judicial Ctr. 3d ed. 2011), available at
http://www fjc.gov/public/pdf nsf/lookup/sciman03.pdf/$file/sciman03.pdf.

8

failure to take into account any of the plausible alternatives makes her opinion unreliable and,
therefore, inadmissible. 6
CONCLUSION
For the foregoing reasons, Plaintiffs respectfully request that Dr. Burtis be precluded at
trial from offering any opinions on competitive effects.

Dated: April 26, 2013

Respectfully submitted,

Mark W. Ryan
Lawrence E. Buterman
Katharine S. Mitchell-Tombras
Attorneys for the United States
United States Department of Justice
Antitrust Division
450 Fifth Street, NW, Suite 4000
Washington, DC 20530
[phone]
[e-mail]
On Behalf of the United States of America

6

Dr. Burtis’s rebuttal report also contains criticisms of the reports of Plaintiffs’ experts. To be clear, Plaintiffs are
not seeking to preclude Dr. Burtis from offering those criticisms—only from offering any opinions on competitive
effects.

9

Gabriel Gervey
Eric Lipman
David Ashton
Assistant Attorneys General
Office of the Attorney General of Texas
P.O. Box 12548
Austin, TX 78711
[phone]
[e-mail]

W. Joseph Nielsen
Gary M. Becker
Assistant Attorneys General
Office of the Attorney General of Connecticut
55 Elm Street
Hartford, CT 06106
[phone]
[e-mail]
On Behalf of the Plaintiff States

10