Unites States of America v. Apple, Inc. et al
OPPOSITION BRIEF United States' Opposition to Apple Inc.'s Motion in Limine to Exclude Certain Expert Testimony of Professor Richard Gilbert. Document filed by Unites States of America.(Fairchild, Stephen)
UNITED STATES DISTRICT COURT
FOR THE SOUTHERN DISTRICT OF NEW YORK
UNITED STATES OF AMERICA,
APPLE, INC., et al.,
Civil Action No. 12-cv-2826 (DLC)
THE STATE OF TEXAS;
THE STATE OF CONNECTICUT; et al.,
PENGUIN GROUP (USA) INC. et al.,
Civil Action No. 12-cv-03394 (DLC)
UNITED STATES’ OPPOSITION TO APPLE INC.’S MOTION IN LIMINE TO
EXCLUDE CERTAIN EXPERT TESTIMONY OF PROFESSOR RICHARD GILBERT
Apple’s motion in limine is nothing more than a “tit for tat” attempt to deflect attention
away from the serious failings that pervade the analyses of its own experts, Dr. Murphy and
Dr. Burtis. Plaintiffs disclosed to Apple that they planned to move to exclude certain testimony
on three of Dr. Murphy’s opinions because, inter alia, they relied on no economic or empirical
analysis of any kind for the conclusion that Apple acted in its own interest in fixing the retail
price of e-books. Apple’s own motion in limine then argues that Professor Gilbert’s expert report
is essentially “120 pages . . . much of which merely recount the ‘evidence’ on which the DOJ’s
case rests.” 1 Similarly, Plaintiffs pointed out to Apple that Dr. Burtis’s decision to analyze
anticompetitive effects using a two-year window without controlling for any confounding factors
rendered her opinion unreliable under Federal Rules of Evidence 401 and 702. Apple’s own
motion in limine then argues that Professor Gilbert’s output analysis, which examines the effects
of the conspiracy using a shorter window of time immediately before and after the move to
agency, is “an unreliable empirical analysis.” 2
It is unclear how Apple can argue in the same motion that Professor Gilbert both
(1) conducted no economic analysis and (2) reached conclusions based on improper economic
analyses. But, putting that aside, even a cursory glance at Professor Gilbert’s testimony and
report makes clear how baseless Apple’s assertions are. Professor Gilbert undertook extensive
empirical work and economic analysis in this matter, which is reflected in both his report and
direct testimony. That empirical work includes, inter alia: an analysis of Publisher Defendants’
adherence to the price caps specified in the Apple Agency Agreements, 3 an analysis of the
increase in Publisher Defendants’ average e-book retail prices after switching to the agency
Apple Inc.’s Motion in Limine to Exclude Certain Expert Testimony of Professor Richard Gilbert (“Apple MIL.”)
Gilbert Direct ¶¶ 139-44 and Table 4.
model, 4 and an analysis of Publisher Defendants’ declining output growth and declining market
share after they increased their average retail prices. 5 To the extent Professor Gilbert’s report
and testimony refer to facts, those facts are included either because they were considered by
Professor Gilbert in forming his expert opinions, or because they help inform and confirm his
As for Professor Gilbert’s empirical work relating to output, that analysis is based on
generally accepted principles of economics, including that economists should attempt to isolate
the effects of what they are attempting to study, and should attempt to account for obvious
alternative explanations that could cast doubt on the validity of their conclusions. Accordingly,
the United States respectfully submits that there is no basis whatsoever to exclude any portion of
Professor Gilbert’s testimony.
PROFESSOR GILBERT’S TESTIMONY IS NOT AN IMPERMISSIBLE
Despite Apple’s broad statements, and given the fact that it is seeking to exclude some of
Professor Gilbert’s empirical work, Apple cannot legitimately assert that Professor Gilbert
“applies no economic analysis.” 6 Thus, Apple must acknowledge that its motion in limine
applies only to a subset of Professor Gilbert’s testimony. However, Apple has failed to specify
what portion of Professor Gilbert’s analysis it finds to be objectionable “factual narrative.” That
vagueness, by itself, suffices to justify the motion’s denial. See Nat’l Union Fire Ins. Co. v. L.E.
Myers Co. Grp., 937 F. Supp. 276, 287 (S.D.N.Y. 1996) (when a motion in limine lacks “the
necessary specificity with respect to the evidence to be excluded” the Court should “reserve
judgment on the motion until trial when admission of particular pieces of evidence is in an
Id. at ¶¶ 146-50 and Figures 3 and 4 and Table 5.
Id. at ¶¶ 68-70 and Figure 2, ¶ 235 and Figure 9.
Apple MIL. at 5.
appropriate factual context.”); see also Baxter Diagnostics, Inc. v. Novatek Med., Inc., 94 Civ.
5520, 1998 WL 665138, at *3 (S.D.N.Y. Sept. 25, 1998) (denying motions in limine to exclude
“all ‘evidence of Baxter’s financial condition’” and “evidence on its punitive damages claim”
because the motions in limine “lack ‘the necessary specificity’”).
Apple’s lack of specificity is particularly problematic here because, of course, Professor
Gilbert references facts at various points in his opinion. That is what he is required to do under
the Federal Rules of Civil Procedure. According to Rule 26, a testifying expert must include in a
written report, inter alia, “the facts . . . considered by the witness” in the forming of “all the
opinions the witness will express.” Fed. R. Civ. P. 26(a)(2)(B)(i), (ii). Indeed, many of the facts
that Apple complains Professor Gilbert includes in his report—the bulk of which are not even in
dispute 7—fall within that category. 8
But beyond that, “[i]t is well settled that an expert is free to offer testimony to provide a
background for the case.” U.S. Info. Sys., Inc. v. Int’l Bhd. of Elec. Workers Local Union No. 3,
AFL-CIO, 313 F. Supp. 2d 213, 237 (S.D.N.Y. 2004) (citing United States v. Mulder, 273 F.3d
91, 102 (2d Cir. 2001); United States v. Daly, 842 F.2d 1380, 1388 (2d Cir. 1988)). And that is
what Professor Gilbert also has done. In order to provide a proper context for his empirical
analysis, Professor Gilbert has provided the Court with (1) background facts that describe
conditions in the e-books market prior to Apple’s entry, and (2) a discussion of the negotiation
and implementation of the Apple Agency Agreements.
And are, in fact, found in Dr. Murphy’s direct testimony, too. Compare Murphy Direct ¶ 47 (referencing Amazon’s
“large share” of e-book sales), with Apple MIL. at 5 (criticizing Professor Gilbert for stating that Amazon “had the
largest share” of e-book sales) and compare Murphy Direct ¶ 59 (referencing Amazon’s “$9.99” price for “many
new and popular e-books” and its discouraging effect that pricing had on its competitors) with Apple MIL. at 5
(criticizing Professor Gilbert for stating that Amazon “priced its ebooks very competitively”).
For example, in order to assess the effects on price and output that resulted from Defendants’ conspiracy, it was
important for Professor Gilbert to understand the state of the market as it existed before the conspiracy came into
play. To the extent Professor Gilbert relied on those background facts (which include that Amazon had low prices
for e-books and large market share), Professor Gilbert was correct to reference them in his testimony and report.
Simply put, the facts included in Professor Gilbert’s report are not intended to be
substitutes for economic analyses, nor could they reasonably be considered as such. Rather, they
both inform Professor Gilbert’s analysis and provide confirmation that his expert opinion
comports with the real world. 9 “The exercise of quantifying damages must be supported by an
in-depth qualitative analysis of the industry, which should help provide the justification for the
methodology and specification chosen. To carry weight, any econometric results will need to be
plausible given the known facts about the industry.” 10
Professor Gilbert uses the factual information “to describe the setting in which the parties
were operating. From there he applies economic principles,” including analyzing the economic
incentives created by the Apple Agency Agreements and conducting an empirical analysis, to
determine whether Apple’s agency model harmed consumers. 11 See U.S. Info. Sys., 313 F. Supp.
2d at 237. Professor Gilbert’s “ultimate conclusions are based on the application of standard
economic methodology to the facts at hand.” Id. Accordingly, it was permissible for Professor
Gilbert to testify about those facts on which he relied in his analysis. To the extent Apple
disagrees with Professor Gilbert’s understanding of any factual matter, his disclosure of his
understanding allowed Apple to question Professor Gilbert in his deposition and, of course, will
allow it to do so again at trial.
For example, Professor Gilbert’s economic analysis found that each Publisher Defendant lost profit in the short
term by adopting the agency model. Gilbert Direct ¶¶ 66-83. That is to say, it was against each publisher’s
unilateral self-interest. Thus, Professor Gilbert concluded that each Publisher Defendant would not have moved to
agency unless it expected that its competitors were going to do the same. Professor Gilbert confirmed his
conclusion by looking at the factual evidence, which showed that both Apple and Publisher Defendants expected
that Publisher Defendants would move Amazon and all other retailers to the agency model.
Peter Davis & Eliana Garcés, QUANTITATIVE TECHNIQUES FOR COMPETITION AND ANTITRUST ANALYSIS 353
Professor Gilbert concluded that defendants’ actions caused “consumers of e-books [to] have paid substantially
higher prices,” Gilbert Direct ¶ 17, but Professor Gilbert, like any good economist, was careful to make sure that his
economic reasoning was both informed by, and consistent with, the underlying evidence: documents, testimony,
and even Defendants’ expert reports.
II. PROFESSOR GILBERT’S OUTPUT ANALYSIS IS RELIABLE
As a threshold matter, Apple erroneously contends that under Brooke Group Ltd. v.
Brown & Williamson Tobacco Corp., 509 U.S. 209, 233 (1993), the United States must, even
when “some prices rise,” offer “proof of an output restriction . . . to sustain a claim of
anticompetitive effects.” 12 Properly understood, the Court’s actual holding was that a plaintiff’s
failure to show a reduction in output “was not dispositive.” Conwood Co. v. U.S. Tobacco Co.,
290 F.3d 768, 789 (6th Cir. 2002) (emphasis added). The Court in Brooke Group was merely
observing an economic truism; namely that when prices for normal goods such as e-books
increase, all other things being equal, the output of the firms that increased their price will
decrease. 13 Here, Professor Gilbert’s empirical work properly focused primarily on the price
effects arising from the Apple Agency Agreements. Specifically, Professor Gilbert found, as
shown below, that the shift to agency caused the average prices of Publisher Defendants’ ebooks to increase significantly. 14
Apple MIL. at 7.
That hardly means that an antitrust plaintiff must prove that decrease. To the contrary, if a plaintiff can show a
price increase, a decrease in output is assumed. See General Leaseways, Inc. v. Nat’l Truck Leasing Ass’n, 744 F.2d
588, 594-95 (7th Cir. 1984). Nor is the fact that output overall continued to increase damaging to Professor
Gilbert’s conclusions. See Fleischman v. Albany Med. Ctr., 728 F. Supp. 2d 130, 155 (N.D.N.Y. 2010) (expert
opinion that information exchanges softened wage competition for nurses in area hospitals not contradicted by the
fact that overall compensation continued to rise).
Gilbert Direct ¶¶ 146-53.
Gilbert Direct Figure 3: The average per unit e-book prices at Amazon of each First Wave
Agency Publisher increased significantly when it switched to agency
Professor Gilbert also found that, as standard economics predicts, this price increase
coincided with a reduction in Publisher Defendants’ unit sales and an increase in non-defendant
publishers’ unit sales, resulting in an immediate net decrease in total sales across both sets of
titles. This is sufficient evidence of anticompetitive effects arising from the conspiracy.
Apple’s motion does not directly contest Professor Gilbert’s finding that prices increased.
Instead, it only critiques Professor Gilbert’s findings regarding the net output decrease that
accompanied the price increases, which is a minor point in his analysis. To the extent relevant,
Apple’s attacks are misleading and factually incorrect. Though nominally expressed as three or
four separate flaws, the essence of each is that Apple disagrees with the window Professor
Gilbert used to analyze the effects of Apple’s conspiracy with Publisher Defendants. 15
Professor Gilbert’s decision to select a short time window was well-considered and is
consistent with generally accepted economic principles. Far from being “litigation-driven,”
Professor Gilbert’s use of a short window allowed him to isolate the direct effects of the
conspiracy, all of which occurred within a few days, while decreasing the “noise” from other
factors that may have affected e-book prices and sales, including the rapid maturation of the
industry and any life-cycle trend. Thus, Apple’s attack that Professor Gilbert fails to account for
the sales of a particular title declining over time is incorrect. 16
Apple also misleadingly contends that Professor Gilbert’s analysis regarding output “is
based on 0.4% of the sales of trade e-books during the period April 2010 through March 2012,”
to suggest that Professor Gilbert did not show harm to the whole market. 17 This assertion is
irrelevant. Professor Gilbert did not analyze output over Apple’s arbitrarily selected time
period. 18 For the window in which Professor Gilbert actually did analyze output, he looked at
the sales of all trade e-books sold at the beginning and end of his before-and-after study.
Notably, and notwithstanding Apple’s contention to the contrary, this does not amount to an attack on Professor
Gilbert’s methodology. Professor Gilbert used a “before-and-after” approach, which is generally accepted in
economics and validated by courts if the expert accounts for differences in market conditions across the compared
periods. See, e.g., Sun Microsystems Inc. v. Hynix Semiconductor Inc., 608 F. Supp. 2d 1166, 1208 (N.D. Cal. 2009)
(citing Pacific Coast Agricultural Export Ass’n v. Sunkist Growers, Inc., 526 F.2d 1196 (9th Cir. 1975)).
Apple MIL. at 9-10. Moreover, the principal purpose of this particular analysis was to test whether agency
increased the output of those who adopted agency relative to those who did not. When agency contracts and MFNs
are procompetitive, they generally keep prices low or increase output or shares of the adopting firms. Here, the
answer is clear: Not only did the average prices of those who adopted agency rise, but their output also fell relative
to those who did not adopt agency—a result Apple has not contested. Gilbert Direct ¶¶ 67-70, 128, 234-38.
Apple MIL. at 9 (emphasis added).
Apple’s motion includes a citation to “[Burtis trial affidavit]” for this point. Id. at 9. Dr. Burtis’s affidavit makes
no reference whatsoever to the point being cited.
In contrast, Apple’s expert Dr. Burtis “offered no rationale for her choice of such long
time periods, other than it happened to be all the data she received.” 19 Dr. Burtis’s failure to be
deliberate in selecting an appropriate window to analyze the conspiracy renders her analysis
unreliable, because any observed effect may be driven by factors other than the move to
agency. 20 These factors include the upward trend in e-book growth as more consumers adopt ereading, a changing mix of titles, and even the peak book-buying holiday season. For this
reason, so-called “long horizon” studies like the one conducted by Dr. Burtis “require extreme
caution.” 21 Dr. Burtis failed to show that caution here.
Finally, when Apple attacks Professor Gilbert’s opinion regarding industry-wide growth
in output, its argument actually applies with force to Dr. Burtis’s opinions. In her initial report,
Dr. Burtis stated that “[s]ales of eBooks increased dramatically during the post-agency period—a
result that is inconsistent with allegations that the agency agreements had an anticompetitive
effect.” 22 But as Professor Gilbert points out in his testimony:
nowhere did Dr. Burtis show or even argue that the increase in sales of e-books is
in any way attributable to defendants’ adoption of the Apple Agency Agreements.
In particular, Dr. Burtis did not isolate any increases in e-reading caused by the
launch of the iPad (as opposed to the iBookstore) or the continuation of the preexisting trend, and instead implicitly attributes those increases to the adoption of
the Apple Agency Agreements. 23
Gilbert Direct ¶ 175.
Id. at ¶¶ 175-77. Apple’s apparent awareness that external factors affect price and output over time makes Dr.
Burtis’ failure to even attempt to control for such factors through regression or some other method all the more
inexplicable. See Plaintiffs’ Motion in Limine re: Dr. Burtis at 4.
S.P. Kothari & Jerold B. Warner, Econometrics of Event Studies, in HANDBOOK OF CORPORATE FINANCE:
Vol. 1, at 8 (B. Espen Eckbo, ed., 2007).
EMPIRICAL CORPORATE FINANCE,
Burtis Initial Report ¶ 27.
Gilbert Direct ¶ 228.
That is, Dr. Burtis “offer[ed] no opinion that the alleged relevant market would have expanded at
a [lesser] rate but-for the Apple agency agreements.” 24 Thus, far from “admitting” that one
could not isolate the relationship between the adoption of agency and the industry’s growth rate,
Professor Gilbert was criticizing Dr. Burtis’s analysis because it did not account for the general
growth trend. Apparently, Apple now agrees with Professor Gilbert’s criticism.
Professor Gilbert’s only conclusion related to the industry growth rate was that
Dr. Burtis’s opinion was not only irrelevant, but also inaccurate. Specifically, in an analysis
using Dr. Burtis’s data, Professor Gilbert found that the number of paid e-book purchases of all
titles at all e-retailers grew more slowly in the year following the switch to agency than during
the year prior to agency. 25 By failing to account for industry sales growth over time, Dr. Burtis
underestimates the true competitive harm caused by the conspiracy and falsely attributes the
natural growth in e-book sales to the agency agreements. Apple itself concedes that output was
growing “at very high rates” both before and after the adoption of the Apple Agency
Agreements, so it is inexcusable that in Dr. Burtis’s analysis this fact was ignored. 26
For the foregoing reasons, the United States respectfully requests that Apple’s motion in
limine be denied.
Apple MIL. at 10.
Gilbert Direct ¶ 233.
Apple MIL. at 9.
Dated: May 3, 2013
Mark W. Ryan
Lawrence E. Buterman
David Z. Gringer
United States Department of Justice
450 Fifth Street, NW, Suite 4000
Washington, DC 20530