`
`
`
`Karma M. Giulianelli (SBN 184175)
`BARTLIT BECK LLP
`1801 Wewatta St., Suite 1200
`Denver, CO 80202
`Telephone: (303) 592-3100
`Facsimile: (303) 592-3140
`karma.giulianelli@bartlitbeck.com
`
`Hae Sung Nam (pro hac vice)
`KAPLAN FOX & KILSHEIMER LLP
`850 Third Avenue
`New York, NY 10022
`Telephone: (212) 687-1980
`Facsimile: (212) 687-7715
`hnam@kaplanfox.com
`
`Interim Co-Lead Counsel for the Proposed Classes
`
`[Additional Counsel on Signature Page]
`
`
`
`
`
`
`
`
`UNITED STATES DISTRICT COURT
`NORTHERN DISTRICT OF CALIFORNIA
`SAN FRANCISCO DIVISION
`
`IN RE GOOGLE PLAY CONSUMER
`ANTITRUST LITIGATION
`RELATED ACTIONS:
`
`Epic Games Inc. v. Google LLC et al.,
`Case No. 3:20-cv-05671-JD
`
`In re Google Play Developer Antitrust
`Litigation, Case No. 3:20-cv-05792-JD
`
`State of Utah, et al., v. Google LLC, et al.,
`Case No. 3:21-cv-05227-JD
`Match Group, LLC, et al. v. Google LLC, et
`al., Case No. 3:22-cv-02746-JD
`
`
`No. 3:20-CV-05761-JD
`
`CONSUMER PLAINTIFFS’ REPLY
`IN SUPPORT OF THEIR MOTION
`FOR CLASS CERTIFICATION
`
`Hearing Date: August 4, 2022
`Hearing Time: 10:00 a.m.
`Courtroom: Courtroom 11, 19th Floor
`Judge: The Honorable James Donato
`
`
`CONSUMER PLAINTIFFS’ REPLY IN SUPPORT OF CLASS CERTIFICATION NO. 3:20-CV-05761-JD
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`
`
`Case 3:21-md-02981-JD Document 289 Filed 07/14/22 Page 2 of 22
`
`
`
`TABLE OF CONTENTS
`
`B.
`
`C.
`
`INTRODUCTION ...........................................................................................................................1
`ARGUMENT ...................................................................................................................................2
`I.
`Common Questions Predominate, Satisfying Rule 23(b)(3) ...............................................2
`Dr. Singer’s Model Shows All Developers Would Be Subject to
`A.
`Lower Take Rates Absent Google’s Monopolization ..............................................2
`Consumer Plaintiffs Will Prove Pass-Through with Classwide
`Evidence ...................................................................................................................4
`Dr. Singer’s Model Is Sound Classwide Proof of Pass-
`1.
`Through ....................................................................................................... 4
`App-By-App Analysis Is Not Required to Demonstrate
`Pass-Through .............................................................................................. 5
`Dr. Burtis’s Misguided Claim that There are Uninjured
`Class Members Does Not Undermine Class Certification .......................... 9
`Dr. Singer’s Play Points Model Provides Further, Independent
`Common Proof of Antitrust Impact .........................................................................9
`Class Members Will Not Be Worse Off In the But-For World .............................10
`D.
`Dr. Singer Provides a Common Method of Calculating Damages ........................11
`E.
`An Injunctive Relief Class Is Justified Under Rule 23(b)(2) .............................................12
`II.
`All Four of the Rule 23(a) Factors Are Met ......................................................................13
`III.
`CONCLUSION ..............................................................................................................................15
`
`2.
`
`3.
`
`
`
`
`
`i
`CONSUMER PLAINTIFFS’ REPLY IN SUPPORT OF CLASS CERTIFICATION NO. 3:20-CV-05761-JD
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`
`
`Case 3:21-md-02981-JD Document 289 Filed 07/14/22 Page 3 of 22
`
`
`
`TABLE OF AUTHORITIES
`
`Cases
`Allied Orthopedic Appliances, Inc. v. Tyco Healthcare Grp. L.P.,
`247 F.R.D. 156 (C.D. Cal. 2007) .............................................................................................. 11
`B.K. by next friend Tinsley v. Snyder,
`922 F.3d 957 (9th Cir. 2019) .................................................................................................... 12
`Berni & Barilla S.p.A.,
`964 F.3d 141 (2d Cir. 2020) ..................................................................................................... 13
`Boeing Co. v. Van Gemert,
`444 U.S. 472 (1980) .................................................................................................................. 14
`Buchanan v. Tata Consultancy Servs., Ltd.,
`No. 15-CV-01696-YGR, 2017 WL 6611653 (N.D. Cal. Dec. 27, 2017) ................................. 13
`Cummings v. Connell,
`316 F.3d 886 (9th Cir. 2003) ..................................................................................................... 15
`In re Apple iPhone Antitrust Litig.,
`No. 11-cv-6714-YGR, 2022 WL 1284104 (N.D. Cal. Mar. 29, 2022) ...................................... 8
`In re Digital Music Antitrust Litig.,
`321 F.R.D. 64 (S.D.N.Y. 2017) .................................................................................................. 6
`In re Electronic Books Antitrust Litig.,
`Case No. 11-md-02293 (DLC) (S.D.N.Y.) ............................................................................... 14
`In re Flash Memory Antitrust Litig.,
`No. C 07-0086 SBA, 2010 WL 2332081 (N.D. Cal. June 9, 2010) ........................................... 3
`In re FPI/Agretech Sec. Litig.,
`105 F.3d 469 (9th Cir. 1997) .................................................................................................... 14
`In re Lithium Ion Batteries Antitrust Litig.,
`No. 13-MD-2420 YGR, 2018 WL 1156797 (N.D. Cal. Mar. 5, 2018) ...................................... 8
`In re Live Concert Antitrust Litig.,
`247 F.R.D. 98 (C.D. Cal. 2007) ................................................................................................ 11
`In re Online DVD Rental Antitrust Litig.,
`No. M 09-2029 PJH, 2010 WL 5396064 (N.D. Cal. Dec. 23, 2010) ....................................... 15
`In re Optical Disk Drive Antitrust Litig.,
`303 F.R.D. 311 (N.D. Cal. 2014) ............................................................................................ 5, 8
`In re Optical Disk Drive Antitrust Litig.,
`No. 3:10-md-2143 RS, 2016 WL 467444 (N.D. Cal., Feb. 8, 2016) ........................................ 10
`In re Packaged Seafood Prod. Antitrust Litig.,
`332 F.R.D. 308 (S.D. Cal. 2019), aff’d on reh’g en banc sub nom. Olean,
`31 F.4th 651 ................................................................................................................................ 8
`In re Pre-Filled Propane Tank Antitrust Litig.,
`No. 14-02567-MD-W-GAF, 2021 WL 5632089 (W.D. Mo. Nov. 9, 2021) .............................. 6
`In re Suboxone Antitrust Litig.,
`421 F. Supp. 3d 12 (E.D. Pa. 2019) .......................................................................................... 13
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`ii
`CONSUMER PLAINTIFFS’ REPLY IN SUPPORT OF CLASS CERTIFICATION NO. 3:20-CV-05761-JD
`
`
`
`Case 3:21-md-02981-JD Document 289 Filed 07/14/22 Page 4 of 22
`
`
`
`Kamakahi v. Am. Society for Reproductive Med.,
`305 F.R.D. 164 (N.D. Cal. 2015) .............................................................................................. 11
`MacRae v. HCR Manor Care Services, LLC,
`Case No. SA CV 14-00715-DOC (RNBx), 2018 WL 8064088 (C.D. Cal. Dec. 10,
`2018) ......................................................................................................................................... 13
`Mueller v. Puritan’s Pride, Inc.,
`No. 3:16-cv-06717-JD, 2021 WL 5494254 (N.D. Cal. Nov. 23, 2021) ................................... 12
`Olean Wholesale Grocery Coop., Inc. v. Bumble Bee Foods, LLC,
`31 F.4th 651 (2022) (en banc) ........................................................................................... passim
`Paul, Johnson, Alston & Hunt v. Graulty,
`886 F.2d 268 (9th Cir. 1989) .................................................................................................... 14
`Sandoval v. M1 Auto Collisions Ctrs.,
`309 F.R.D. 549 (N.D. Cal. 2015) .............................................................................................. 15
`Walters v. Reno,
`145 F.3d 1032 (9th Cir. 1998) .................................................................................................. 12
`Statutes
`15 U.S.C. § 15c(b)(2) .................................................................................................................... 15
`Other Authority
`Newberg on Class Actions § 3:75 ................................................................................................. 15
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`iii
`CONSUMER PLAINTIFFS’ REPLY IN SUPPORT OF CLASS CERTIFICATION NO. 3:20-CV-05761-JD
`
`
`
`Case 3:21-md-02981-JD Document 289 Filed 07/14/22 Page 5 of 22
`
`
`
`INTRODUCTION
`In its opposition, Google derides consumers for spending the “bulk of their motion … at-
`tacking Google’s business practices.” Dkt. 300 (“Opp.”) at 1. But Google’s monopolistic conduct
`demonstrates the enormous amount of common evidence of wrongdoing that would predominate
`over any individualized issues at trial. The conduct at issue includes predation, an array of exclu-
`sionary contracts, and numerous bribes and threats aimed at potential competitors. A trial on
`Google’s anticompetitive schemes will require the presentation of voluminous common evidence,
`which will plainly predominate over any individualized issues.
`Committing the very transgression Google falsely accuses consumers of, Google’s oppo-
`sition focuses on arguments about the merits of the case, including its criticisms of Dr. Singer’s
`substantive analysis, that are inappropriate at the class certification stage. Those arguments also
`fail on the merits for the same reason they fail in the Daubert context.
`The main thrust of Google’s opposition—that pass-through is “not susceptible to general-
`ized class-wide proof”—is foreclosed by the Ninth Circuit’s recent decision in Olean Wholesale
`Grocery Cooperative, Inc. v. Bumble Bee Foods, LLC, 31 F.4th 651 (2022) (en banc). Google
`argues that
` (based on its own economist’s analysis), and therefore
`must be proven with individualized evidence. Opp. at 15. Google’s economist is free to take this
`position—one that Plaintiffs will prove is fundamentally flawed—but it cannot outweigh Dr.
`Singer’s reliable methodology assessing classwide damages. Instead, Google merely argues that
`“plaintiffs’ evidence relating to the common question … [is] unpersuasive and unlikely to suc-
`ceed.” Olean, 31 F.4th at 667. Plaintiffs can prove their case with common evidence, and certifi-
`cation is warranted. It will be up to the jury to decide whether the evidence is persuasive. Id.
`Google is also wrong that its monopolization has benefitted some consumers. Google ar-
`gues that competition would harm consumers by forcing it to
` or to reduce
`security. Speculation that Google might begin
` for the first time
`in its history—causing some class members to pay more under competitive conditions—is no rea-
`son to deny class certification. There is no reason or economic rationale to suggest competition
`would lead Google to make its product worse, and acceptance of its argument would allow any
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`1
`CONSUMER PLAINTIFFS’ REPLY IN SUPPORT OF CLASS CERTIFICATION NO. 3:20-CV-05761-JD
`
`
`
`Case 3:21-md-02981-JD Document 289 Filed 07/14/22 Page 6 of 22
`
`
`
`I.
`
`monopolist to escape liability through unilateral speculation about what it would be willing to do
`if it lost monopoly power. In any event, this, too, is a question for the merits.
`ARGUMENT
`Common Questions Predominate, Satisfying Rule 23(b)(3)
`Ignoring the overwhelming predominance of undisputedly common issues in this monop-
`olization case, Google focuses just on antitrust injury, claiming that “Plaintiffs have no common
`proof” of pass-through, that Google would “lower service fees at all,” or that consumers “would
`have been injured under their Play Points theory.” Opp. at 8. But Google’s argument ignores Dr.
`Singer’s expert analysis demonstrating common impact through well-accepted economic models
`and is a rehash of its Daubert motion, which fails for multiple reasons. See Dkt. 298 (“Daubert
`Opp.”). Dr. Singer’s modeling reliably demonstrates common impact. Id.
`A.
`Dr. Singer’s Model Shows All Developers Would Be Subject to Lower Take
`Rates Absent Google’s Monopolization
`As explained in Plaintiffs’ opening brief, Dr. Singer’s models of the App Distribution Mar-
`ket and In-App Aftermarket demonstrate that Google’s 30% headline take rate is supra-competi-
`tive, and that Google would be forced to lower prices in the but-for world. Dkt. 280 (“Mot.”) at
`10-11. Aside from questioning an input, Google does not challenge the reliability of that model in
`its Daubert briefing. Dkt. 282 (“Daubert Br.”) at 13-14. Instead, Google appears to argue that “not
`all developers would be subject to lower service fees in the but-for world” because it would keep
`its 30% take rate in the face of competition, competing with other stores only through a “targeted
`approach” of offering discounts to some developers. Opp. at 16-17. Most importantly, whether
`Google’s contention has merit can be resolved on a classwide basis and, therefore, cannot defeat
`class certification. Second, Google’s argument has no support in economics, the law, or the facts.
`First, it is false that Google and its competitors have employed a “targeted approach” to
`pricing. Id. at 17. Google has used its uniform 30% take rate for the vast majority of developers,
`with negotiated programs formulaically reducing its rate from the headline rate representing
`
` of U.S. developers. Ex. 31 (Singer Reply) ¶ 8. Likewise, Google’s recent
`
`
`1 Exhibit citations in this reply refer to the consecutively-numbered exhibits of the Giulianelli
`Declaration filed in support of class certification, Dkt. 280-1, and the Giulianelli Declaration filed
`in support of this reply.
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`2
`CONSUMER PLAINTIFFS’ REPLY IN SUPPORT OF CLASS CERTIFICATION NO. 3:20-CV-05761-JD
`
`
`
`Case 3:21-md-02981-JD Document 289 Filed 07/14/22 Page 7 of 22
`
`
`
`pricing reductions from 30% to 15% for the first $1 million in developer revenue per year and for
`subscription products apply uniformly to all developers. Ex. 68 (https://android-developers.goog-
`leblog.com/2021/03/boosting-dev-success.html); Ex. 69
`(https://android-developers.google-
`blog.com/2021/10/evolving-business-model.html). Even for the
`—
`which are anticompetitive and would be absent in the but-for world—Google’s headline take rate
`.” Ex. 40 (GOOG-PLAY-004588725); Ex. 2 (Singer Rpt.) ¶¶ 110-14.2 A key
`was “
`Google executive confirmed that Google
`
`
`” Ex. 70 (Rosenberg Dep.) 123:22-124:23. A uniform approach
` are the exception. Likewise, contrary to Google’s
`is Google’s norm, and
`suggestion, entrants have already attempted to compete with cheaper take rates for all developers.
`International stores offer 20% (One Store) and 25% (Aptoide). Ex. 2 (Singer Rpt.) ¶¶ 196-98.
`
`
`. Id. ¶¶ 104-09.
`Second, Google ignores that a competitive but-for world would foster more general com-
`petition than the current world constrained by Google’s monopoly power. Google’s anticompeti-
`tive conduct affects “all market participants, creating an inference of class-wide impact” that
`cannot be undone due to “different bargaining power.” Olean, 31 F.4th at 671, 677; see also Mot.
`at 21-22 (citing further cases). In an open market, it is unrealistic to expect that Google would
`individually negotiate with tens of thousands of developers. Ex. 3 (Singer Reply) ¶ 10. Without
`Google’s contractual and technical restrictions, all apps would have the ability to freely steer to
`cheaper alternatives. And Google recognized that
`, the result
`would
`
` and Google would
` Ex. 71 (GOOG-PLAY-000542516.R)
`at -532. In other words, Google’s current ability to price discriminate by
`
` reflects its market power. Ex. 2 (Singer Rpt.) ¶ 148. In short, Google would be
`
`
`2 This case is nothing like In re Flash Memory Antitrust Litig., No. C 07-0086 SBA, 2010 WL
`2332081 (N.D. Cal. June 9, 2010), where three direct purchasers who controlled 82% of the market
`made “individualized negotiations and contracts with Defendants.” Id. at *9.
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`3
`CONSUMER PLAINTIFFS’ REPLY IN SUPPORT OF CLASS CERTIFICATION NO. 3:20-CV-05761-JD
`
`
`
`Case 3:21-md-02981-JD Document 289 Filed 07/14/22 Page 8 of 22
`
`
`
`forced to compete on price for all developers if it could not rely on its anticompetitive conduct to
`stymy competitors.
`Third, that some developers may negotiate even lower prices in the but-for world does not
`defeat common impact. Dr. Singer’s model accounts for those deals by extending the discount
`achieved in the actual world, albeit from the lower but-for headline rate. Ex. 3 (Singer Reply) ¶¶ 7-
`12. And Dr. Singer reliably shows that Google’s take rate would be lower for all apps in the but-
`for world. For example, as noted above, Google has a policy favoring uniformity and it uses form
`Developer Distribution Agreements for all developers. Ex. 2 (Singer Rpt.) ¶ 119 & n.276.
`B.
`Consumer Plaintiffs Will Prove Pass-Through with Classwide Evidence
`1. Dr. Singer’s Model Is Sound Classwide Proof of Pass-Through
`Dr. Singer provides reliable proof of pass-through, “a class-wide question[,] in one stroke.”
`Olean, 31 F.4th at 666. Google attacks this pass-through model, claiming that it relies on theoret-
`ical assumptions, that it fails to employ regressions, and that it cannot identify uninjured class
`members. These arguments ignore the rigorous work Dr. Singer performed to validate his work.
`Google claims that Dr. Singer did not run any regressions to calculate pass-through, and
`instead relied on “a theory of universal pass-through.” Opp. at 14-15.3 But Dr. Singer did not
`simply rely on a theory. He ran regressions on real-world market shares and prices using each of
`Google’s 35 app categories as a separate demand system to determine the demand curve faced by
`each developer and to test the fit of the logit model to the data. Ex. 2 (Singer Rpt.) ¶¶ 235-38 &
`Table 7. Google has not challenged those regressions. Only after the logit model proved to fit the
`data, Dr. Singer calculated pass-through for each app using the standard logit pass-through for-
`mula, derived mathematically in peer-reviewed literature, based on each app’s share of the cate-
`gory in which it competed. Id. ¶ 239. The result is a model that calculates pass-through on an app-
`by-app level using established economic methods. Id. ¶¶ 235-39; Ex. 72 (Singer Dep.) 130:12-25.
`
`
`3 Google notes that Dr. Singer did not use his usual approach of “regressing retail price changes
`on wholesale price changes,” Opp. at 14, but omits that the same deposition answer confirms that
`“that’s just not available here.” Ex. 72 (Singer Dep.) 134:24-135:8. Google’s supra-competitive
`prices infect the entire period for which data is available, and no wholesale pricing exists given
`the two-sided platform structure of the market. See id. 138:16-142:13. Dr. Singer therefore turned
`to the well-accepted economic models in his report. Ex. 2 (Singer Rpt.) ¶168.
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`4
`CONSUMER PLAINTIFFS’ REPLY IN SUPPORT OF CLASS CERTIFICATION NO. 3:20-CV-05761-JD
`
`
`
`Case 3:21-md-02981-JD Document 289 Filed 07/14/22 Page 9 of 22
`
`
`
`Nor did Dr. Singer simply assume universal pass-through—the model calculates a pass-
`through rate only if the logit demand explains the variation in the market share data within a cate-
`gory, which Dr. Singer validated. Unlike in In re Optical Disk Drive Antitrust Litigation, 303
`F.R.D. 311 (N.D. Cal. 2014), Dr. Singer ran regressions to confirm the applicability of the model;
`he did not merely “assume[] the very proposition” he was trying to prove. Id. at 321.
`Contrary to Google’s suggestions, Dr. Singer also used real-world data. First, Dr. Singer
`confirmed the applicability of the logit model using Google’s voluminous transaction data. Dr.
`Singer ran regressions to confirm that real world price changes explained changes in market shares
`for the applications. Those regressions allowed Dr. Singer to confirm that the demand for the ap-
`plications within the various Google “app categories” is interdependent—that is, an app’s share
`within a given category declines with increases in the price of that app. Ex. 2 (Singer Rpt.) ¶¶ 235-
`38. Second, Dr. Singer’s analysis of developers’ reaction to taxes in the real world confirmed cost
`pass-through. Id. ¶ 244. Third, Dr. Singer examined (and rejected) the data underlying the flawed
`analysis of Dr. Burtis. Id. ¶¶ 242-44; Ex. 3 (Singer Reply) ¶¶ 112-15.
`In short, none of Google’s criticisms of Dr. Singer’s pass-through model undermine its
`rigor or preclude its use as classwide proof of antitrust impact.
`2. App-By-App Analysis Is Not Required to Demonstrate Pass-
`Through
`Dr. Singer’s reliable proof of impact—which as explained above, considers the pass-
`through rate of each app individually through the application of a well-accepted economic model—
`means that no further individualized app-by-app analysis is required to prove Plaintiffs’ claims.
`Google’s argument to the contrary relies on the flawed analysis of its economist, Dr. Burtis. She
`concluded that when Google applied a limited “service fee” (or take rate) reduction under the
`market conditions that exist today, pass-through was “rare.” Opp. at 9. Google thus argues that
`“pass-through must be proven for each app.” Id. (emphasis in original). But the Olean court re-
`jected arguments like these: “[A] district court cannot decline certification merely because it con-
`siders plaintiffs’ evidence relating to the common question to be unpersuasive and unlikely to
`succeed in carrying the plaintiffs’ burden of proof on that issue.” 31 F.4th at 667. Google’s expert
`unsurprisingly disagrees on the merits, but that is no barrier to class certification. Opp. at 9.
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`5
`CONSUMER PLAINTIFFS’ REPLY IN SUPPORT OF CLASS CERTIFICATION NO. 3:20-CV-05761-JD
`
`
`
`Case 3:21-md-02981-JD Document 289 Filed 07/14/22 Page 10 of 22
`
`
`
`In any event, Google’s merits arguments are wrong. Google’s claim that Dr. Singer has
`“no answer to real-world evidence showing that prices did not change with service fees,” Opp. at
`16, ignores the many fatal flaws in Dr. Burtis’s and the developers’ expert analyses. The one-
`month and six-month time horizons used by Dr. Burtis are insufficient to show the long-run effects
`of lower prices in the but-for world. Ex. 3 (Singer Reply) ¶¶ 103. Google’s anticompetitive conduct
`infected the initial pricing of applications, which economic evidence suggests would linger due to
`sticky prices. Id. ¶¶ 114-15. Both experts analyzed pass-through after a 2018 price reduction for
`second-year subscriptions, but ignored that
`
`
`. Id. ¶ 122 & n.248 (citing Ex. 73 (Scalise Dep.) at 269:9-21). Dr. Burtis’s model suffers
`from numerous other flaws, including failing to use appropriate controls and using artificially
`small product-groups. Ex. 3 (Singer Reply) ¶¶ 102-33.
`Google cannot paper over those flaws by claiming only its expert uses “actual data.” As
`discussed, Dr. Singer employed regression analysis using real-world pricing data to confirm the
`applicability of his economic modeling of pass-through.4 Google’s mere assertion that pass-
`through is “
`” is not supported by the factual record. For that reason, the cases Google
`cites do not undermine the pertinency of Dr. Singer’s model. In re Pre-Filled Propane Tank Anti-
`trust Litig., No. 14-02567-MD-W-GAF, 2021 WL 5632089, at *12 (W.D. Mo. Nov. 9, 2021) (ex-
`cluding pass-through analysis limited to 3 retailers and less than 10% of total sales); In re Digital
`Music Antitrust Litig., 321 F.R.D. 64, 94 (S.D.N.Y. 2017) (excluding expert opinion for ignoring
`20% of the market and for fatal flaw that “renders his inference from the regression invalid”).
`The remainder of the individualized factors Google says require further app-by-app analy-
`sis are simply recast versions of its Daubert arguments. For the same reason those arguments fail
`at Daubert, they fail to raise individualized issues that defeat class certification.
`Marginal Costs. Dr. Singer’s model does not ignore marginal costs. The peer-reviewed
`literature on which Dr. Singer relied for his standard pass-through formula demonstrated that the
`
`
`4 Dr. Singer did not analyze pass-through of Google’s take rate changes against real world data
`because, as he testified, those changes were too
` to allow robust testing. Ex. 72 (Singer
`Dep.) 138:16-141:12. Dr. Burtis’s analysis of the same data is flawed for that reason.
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`6
`CONSUMER PLAINTIFFS’ REPLY IN SUPPORT OF CLASS CERTIFICATION NO. 3:20-CV-05761-JD
`
`
`
`Case 3:21-md-02981-JD Document 289 Filed 07/14/22 Page 11 of 22
`
`
`
`formula is derived mathematically from a profit-maximation model in which marginal cost is a
`key component. Ex. 3 (Singer Reply) ¶ 71. Based on that derivation, Dr. Singer calculates pass-
`through based on the change in each developer’s marginal costs from a reduced take rate. Id. ¶ 72.
`To argue otherwise, Google relies on the same misconstrued deposition testimony and faulty eco-
`nomics it did in its Daubert motion. See Daubert Opp. at 7-8.5 Dr. Burtis opines that pass-through
`is “less likely” when a developer’s marginal costs are zero. Ex. 5 (Burtis Rpt.) ¶¶ 142-43; Ex. 3
`(Singer Reply) ¶¶ 21-24. Google has pointed to zero evidence that any app developers’ marginal
`costs are actually zero. The best it can do is to note that “replication costs” are sometimes zero,
`Opp. at 10-11, but as the full deposition answer Google only partially cited notes, replication costs
`are merely one type of marginal costs. Ex. 72 (Singer Dep.) 97:19-98:19 (noting it “doesn’t cost
`any more to replicate that sword, but that doesn’t mean there aren’t any marginal costs incurred in
`the transaction.”). As Dr. Singer demonstrated, developers face numerous other marginal costs like
`Google’s take rate, taxes, customer service, and server hosting. Ex. 3 (Singer Reply) ¶¶ 21-25.
`Given this unrebutted evidence, analysis of levels of developers’ marginal costs is unnecessary.
`Focal Point Pricing. Nor will proof of pass-through require app-by-app inquiry of focal
`point pricing as Google contends. Google misleadingly suggests that Dr. Singer conceded that
`focal point pricing is an “important consideration” in the but-for world solely by quoting a response
`to a question asking whether focal point pricing “explain[s] any developers’ pricing in the actual
`world.” Ex. 72 (Singer Dep.) 202:2-7 (emphasis added). As Dr. Singer explained, developers’
`behavior in the face of Google’s restraints are not necessarily predictive of the but-for world. Id.
`at 197:19-207:6; Ex. 3 (Singer Reply) ¶¶ 26-31. Nonetheless, Google’s transaction data records
` transactions at prices that do not end in “99.” Ex. 3 (Singer Reply) ¶¶ 26-
`31. As of February 2022, Google reduced the minimum app price from $0.99 to $.05 worldwide.
`Id. ¶ 29 & n.58. Even in the actual world, developers depart from focal point pricing, and Google
`itself has eased restrictions which effectively required focal point pricing at the $0.99 price point.
`And Dr. Singer’s model does in fact allow for developers to retain focal-point pricing with price
`
`
`5 Google’s reference to a “standard economic model” that Dr. Singer purportedly chose not to use,
`is an economic model that cannot calculate pass-through because it does not provide a method for
`determining a developer’s demand curve. Ex. 72 (Singer Dep.) 106:4-107:22; Daubert Opp. at 7.
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`7
`CONSUMER PLAINTIFFS’ REPLY IN SUPPORT OF CLASS CERTIFICATION NO. 3:20-CV-05761-JD
`
`
`
`Case 3:21-md-02981-JD Document 289 Filed 07/14/22 Page 12 of 22
`
`
`
`points ending in 9. Ex. 3 (Singer Reply) ¶ 31. Dr. Singer’s consideration of these factors is suffi-
`cient for his model to support certification. See In re Packaged Seafood Prod. Antitrust Litig., 332
`F.R.D. 308, 343 (S.D. Cal. 2019) (finding that “[t]he Court is not persuaded” by defendants’ argu-
`ment that expert “ignored loss-leader and focal point pricing” because the expert “in fact discusses
`both in his report”), aff’d on reh’g en banc sub nom. Olean, 31 F.4th 651.6 Dr. Singer’s model
`reliably considered the role focal point pricing would play in the but-for world.
`Competitive Conditions. Google ignores that Dr. Singer’s logit model measures, and con-
`trols for, the competitive conditions faced by developers in each of Google’s 35 app categories.
`Dr. Singer determined that the logit model he employed explained real-world variation in prices
`and shares within an app category. Ex. 2 (Singer Rpt.) ¶ 238; Ex. 3 (Singer Reply) ¶ 70. In other
`words, the logit model verifiably described the competitive conditions faced by developers.7 It
`yields the economically intuitive result that a highly concentrated app category where there is less
`competition will have a low pass-through rate, and an unconcentrated app category where there is
`more competition will have a high pass-through rate. Id. ¶ 73. Dr. Singer’s modeling already ac-
`counts for competitive conditions, and no further individualized analysis is required.
`“Other Idiosyncratic Factors.” Google’s argument that Dr. Singer failed to consider
`“other idiosyncratic factors” is another Daubert argument that mischaracterizes Dr. Singer’s work.
`Dr. Singer’s models are agnostic to how developers choose to use the portion of savings not passed
`through to consumers. Ex. 2 (Singer Rpt.) ¶ 266. Speculation that some developers may spend the
`remainder of their savings on marketing, improvements to their apps, or in
`
`neither undermines his pass-through calculation nor requires individualized analysis. Daubert
`
`6 Google’s cases all involved significantly different evidence of, and theories supporting, focal
`point pricing. In re Lithium Ion Batteries Antitrust Litig., No. 13-MD-2420 YGR, 2018 WL
`1156797, at *3 (N.D. Cal. Mar. 5, 2018) (finding that theory of pass-through based on lower qual-
`ity to compensate for battery costs in laptops was unsupported by record); In re Optical Disk Drive
`Antitrust Litig., 303 F.R.D. at 324-25 (finding that pass-through of cost that is “relatively small
`portion of the cost” of products priced at $100 increments defeated class certification); In re Apple
`iPhone Antitrust Litig., No. 11-cv-6714-YGR, 2022 WL 1284104, at *8 (N.D. Cal. Mar. 29, 2022)
`(finding expert should have considered focal point pricing where record included “overwhelming
`evidence” of focal point pricing in but-for world, including expert’s own admission).
`7 While Google is correct that Dr. Singer did not model “which apps in each category are comple-
`ments and which are substitutes,” Opp. at 12-13, he explained in the same answer that “it’s not
`necessary to get the implied pass-through rate.” Ex. 72 (Singer Dep.) 159:21-160:1; see Dkt. 298
`(Daubert Opp.) at 13 (further discussing suitability of categories for logit model).
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`8
`CONSUMER PLAINTIFFS’ REPLY IN SUPPORT OF CLASS CERTIFICATION NO. 3:20-CV-05761-JD
`
`
`
`Case 3:21-md-02981-JD Document 289 Filed 07/14/22 Page 13 of 22
`
`
`
`Opp. at 9. In short, Google’s “other idiosyncratic factors” are irrelevant to the determination of
`whether Plaintiffs have provided a reliable methodology for calculating antitrust injury.
`3. Dr. Burtis’s Misguided Claim that There are Uninjured Class
`Members Does Not Undermine Class Certification
`Google’s argument that class members who purchased from only one app may not be in-
`jured if that app’s developer did not pass through Google’s supracompetitive costs is just a restate-
`ment of Dr. Burtis’s error-riddled analysis that pass-through is “a rare exception.” Opp. at 1. Dr.
`Singer’s analysis proves otherwise that