In Re: Electronic Books Antitrust Litigation

MEMORANDUM OF LAW in Support re:

S.D.N.Y.April 11, 2014

UNITED STATES DISTRICT COURT SOUTHERN DISTRICT OF NEW YORK THE STATE OF TEXAS, Plaintiffs, v. PENGUIN GROUP (USA), INC., et al., Defendants. x : : : : x : : : : : x 12-CV-03394 (DLC) IN RE ELECTRONIC BOOKS ANTITRUST LITIGATION x : : : x 11-MD-02293 (DLC) This Document Relates to: ALL ACTIONS x : : : x DEFENDANT APPLE INC.’S MEMORANDUM OF LAW IN SUPPORT OF ITS MOTION FOR RECONSIDERATION Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 1 of 19 i TABLE OF CONTENTS PRELIMINARY STATEMENT .................................................................................................... 1 STANDARD OF REVIEW ............................................................................................................ 2 BACKGROUND ............................................................................................................................ 2 ARGUMENT .................................................................................................................................. 3 A. The Court Should Reconsider Its Denial Of A Daubert Hearing. .................................. 3 B. The Court Should Reconsider Its Blanket Exclusion of Dr. Kalt’s Testimony and Near-Blanket Exclusion of Mr. Orszag’s Testimony. ............................................ 11 CONCLUSION ............................................................................................................................. 15 Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 2 of 19 ii TABLE OF AUTHORITIES Cases Arum v. Miller, 304 F. Supp. 2d 344 (E.D.N.Y. 2003) ........................................................................................ 2 City of Tuscaloosa v. Harcros Chems., Inc., 158 F.3d 548 (11th Cir. 1998) .................................................................................................... 4 Daubert v. Merrell Dow Pharm., 509 U.S. 579 (1993) .................................................................................................................... 3 Kumho Tire Co., Ltd. v. Carmichael, 526 U.S. 137 (1999) .................................................................................................................... 4 Lindsey v. Normet, 405 U.S. 56 (1972) .................................................................................................................... 11 Medisim Ltd. v. Bestmed LLC, 2012 WL 1450420 (S.D.N.Y. Apr. 23, 2012)............................................................................. 2 Munafo v. Metro. Trans. Auth., 381 F.3d 99 (2d Cir. 2004).................................................................................................... 2, 11 Padillas v. Stork-Gamco, Inc., 186 F.3d 412 (3d Cir. 1999).................................................................................................. 4, 10 Shrader v. CSX Transp., Inc., 70 F.3d 255 (2d Cir. 1995).......................................................................................................... 2 United States v. Apple Inc., 952 F. Supp. 2d 638 (S.D.N.Y. 2013) ......................................................................................... 5 United States v. Valencia, 600 F.3d 389 (5th Cir. 2010) .................................................................................................... 10 Other Authorities JEFFERY M. WOOLDRIDGE, ECONOMETRIC ANALYSIS OF CROSS SECTION AND PANEL DATA (2002) .................................................................................................................... 6 JEFFREY M. WOOLDRIDGE, INTRODUCTORY ECONOMETRICS: A MODERN APPROACH (5th ed. 2013) ......................................................................................... 6 Manual for Complex Litigation (Fourth) § 23.352 ......................................................................... 4 PETER KENNEDY, A GUIDE TO ECONOMETRICS (5th ed. 2003) ....................................................... 6 Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 3 of 19 1 PRELIMINARY STATEMENT Pursuant to Rule 6.3 of the Local Civil Rules of the United States District Court for the Southern District of New York, Apple respectfully moves for reconsideration of the Court’s March 28, 2014 orders (Dkts. 585, 586) excluding opinions offered by Apple’s experts, Dr. Joseph Kalt and Mr. Jonathan Orszag, and admitting opinions offered by Plaintiffs’ expert, Dr. Roger Noll.1 Apple believes that the Court should reconsider its rulings here for two specific purposes. First, the Court should reconsider its rejection of Apple’s request for a Daubert hearing.2 The Court concluded such a hearing was “unnecessary” (Dkt. 586 at 55), but a Daubert hearing would allow the experts to address critically important issues, which Apple believes were overlooked by the Court’s decision, and will demonstrate that Dr. Kalt and Mr. Orszag’s opinions are eminently relevant, reliable, and admissible, whereas Dr. Noll’s are not. For example, a Daubert hearing would establish that the Court overlooked key facts in relying on the adjusted-R2 statistic of Dr. Noll’s original model to conclude that his model “explains 90% of the variance in prices among e-book titles.” Id. at 8. Second, the Court’s conclusion that “Apple has not argued, nor has it established, that the remaining interstitial opinions [by Dr. Kalt] would aid the trier of fact” was mistaken. Id. at 38. Both Dr. Kalt and Mr. Orszag offered numerous opinions that were not challenged by Plaintiffs in their motions to exclude, nor addressed by the 1 Apple has filed a motion to stay all proceedings in these matters. Dkt. 603. This motion for reconsideration is not intended to derogate from that request in any way. 2 Apple therefore necessarily requests that the Court reconsider its class certification opinion (Dkt. 585) for purposes of interposing a Daubert hearing. The fact that Apple is requesting the Court to reconsider its orders and hold Daubert hearing(s) does not constitute waiver of its other objections to the orders, including but not limited to its objection described in its Rule 23(f) petition that the Court’s premature orders excluding Apple’s expert opinions and admitting Dr. Noll’s opinions for all purposes violate the rule against one-way intervention. Apple requests that any ruling rendered based on the Daubert hearing be issued at a time and posture of the litigation that is consistent with that rule. Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 4 of 19 2 Court in its opinion, and are reliable and relevant such that their exclusion is unwarranted. Apple respectfully requests that the Court reconsider its exclusion of those specific opinions. STANDARD OF REVIEW The Court may reconsider a ruling where it has overlooked matters that “might reasonably be expected to alter the conclusion reached by the court,” or where necessary to correct error and prevent manifest injustice. Shrader v. CSX Transp., Inc., 70 F.3d 255, 257 (2d Cir. 1995); see also, e.g., Munafo v. Metro. Trans. Auth., 381 F.3d 99, 105 (2d Cir. 2004) (“[D]istrict courts may alter or amend judgment to correct a clear error of law or to prevent manifest injustice.”) (citation and internal quotation marks omitted); Arum v. Miller, 304 F. Supp. 2d 344, 347 (E.D.N.Y. 2003) (a motion for reconsideration may be granted where the court has overlooked matters or controlling decisions which, if considered by the court, would have mandated a different result). Specifically, where the order to be reconsidered involves the exclusion of expert testimony, a court may grant a motion for reconsideration if the court finds that it overlooked certain aspects of the expert’s proffered testimony. See Medisim Ltd. v. Bestmed LLC, 2012 WL 1450420, at *2 (S.D.N.Y. Apr. 23, 2012) (granting reconsideration of its order excluding expert opinion for being based on flawed empirical testing where the court did not give “adequate consideration” to other independent bases for the expert opinion). BACKGROUND On October 11, 2013, Class Plaintiffs moved for class certification (Dkt. 423), a motion that they supported with expert testimony from Dr. Roger Noll (Dkt. 428). Apple opposed class certification (Dkt. 443), moved to exclude Dr. Noll’s opinions offered in support of class certification (Dkt. 445), and offered expert testimony by Dr. Joseph Kalt (Dkt. 481) and Mr. Jonathan Orszag (Dkt. 488). Class Plaintiffs, joined by Plaintiff States, subsequently moved to Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 5 of 19 3 exclude in full the opinions of Apple’s expert Dr. Kalt. Dkt. 557. Class Plaintiffs also moved to exclude in full the opinions of Mr. Orszag. Dkt. 558. Plaintiff States moved to exclude Mr. Orszag’s opinions about the “offsetting benefits” related to Apple’s conduct. 12-cv-3394, Dkt. 440. On March 28, 2014, the Court issued an order and opinion granting class certification in which it denied Apple’s motion to exclude Dr. Noll’s opinions. Dkt. 585 at 63-83. That opinion did not address Plaintiffs’ motions to exclude because the Court found that “Kalt’s and Orszag’s opinions would not prevent class certification even if they were admissible.” Id. at 3. The Court issued a separate order and opinion on March 28, 2014 in which it granted Plaintiffs’ motions to exclude, “with one exception.” Dkt. 586 at 9. The Court determined that Mr. Orszag “will be permitted to testify regarding his re-running of Noll’s regression analysis, assuming that there is an adequate evidentiary basis to support that analysis at trial.” Id. The Court also denied Apple’s request for a Daubert hearing on the motions to exclude (Dkt. 585 at 83-85; Dkt. 586 at 55-56), as well as Apple’s request to defer decision on the Daubert motions to the matters’ original, transferor courts (Dkt. 586 at 57-59). ARGUMENT A. The Court Should Reconsider Its Denial Of A Daubert Hearing. The Court should reconsider its denial of Apple’s requests for Daubert hearings (Dkt. 585 at 83-85; Dkt. 586 at 55-56) and subsequently reconsider its exclusion of Apple’s expert testimony and admission of Dr. Noll’s testimony in light of such hearings.3 Courts have 3 Apple did not “belatedly request[]” a Daubert hearing (Dkt. 586 at 55); it requested a hearing on its motion to exclude Dr. Noll’s testimony at the time that motion was filed (Dkt. 445 at 2 n.1), and a hearing on Plaintiffs’ motions upon learning that Plaintiffs were resting their motion for summary judgment on the exclusion of Apple’s expert testimony (Dkt. 565 at 17 n.11). Moreover, holding such a hearing where appropriate is part of the district court judge’s “obligation” to ensure that expert testimony is relevant and reliable. Daubert v. Merrell Dow Pharm., 509 U.S. 579, 589-90 (1993); see also Kumho Tire Co., Ltd. v. Carmichael, 526 U.S. 137, 152 (1999). It is not an opportunity that a party can waive. See Padillas v. Stork-Gamco, Inc., 186 F.3d 412, 417-18 (3d Cir. 1999) (finding (Cont'd on next page) Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 6 of 19 4 “stressed the importance of in limine hearings . . . in making the reliability determination required under Rule 702 and Daubert.” Padillas v. Stork-Gamco, Inc., 186 F.3d 412, 417 (3d Cir. 1999). Daubert hearings are particularly appropriate in “complicated cases involving multiple expert witnesses” such as this one. City of Tuscaloosa v. Harcros Chems., Inc., 158 F.3d 548, 564 n.21 (11th Cir. 1998) (“Daubert hearings . . . are almost always fruitful uses of the court’s time and resources in complicated cases involving multiple expert witnesses . . . .”); see also Kumho Tire Co., Ltd. v. Carmichael, 526 U.S. 137, 152 (1999) (finding that “discretionary authority” is afforded to district court judges under Rule 702, in part, to allow them “to require appropriate proceedings in . . . more complex cases”). Moreover, the “need for [an] opportunity to be heard” “is compelling” when the expert testimony in question could be dispositive. Padillas, 186 F.3d at 417-18. The Court rejected Apple’s request to hold a Daubert hearing on its motion to exclude Dr. Noll’s opinions because “Apple has already had the opportunity to question Noll at his deposition.” Dkt. 585 at 83. But Daubert hearings are typically conducted in conjunction with Daubert challenges brought after the close of expert discovery, following expert depositions. See Manual for Complex Litigation (Fourth) § 23.352 (noting that “the best approach is to require that challenges to expert testimony be made during pretrial proceedings, either at the close of expert discovery [i.e., after expert depositions] or through in limine or other motion immediately prior to trial”). Moreover, Apple was given only six hours to depose Dr. Noll regarding his original Declaration, and its request to depose Dr. Noll for an additional four hours regarding his Reply Declaration was denied. Dkt. 502 at 1. (Cont'd from previous page) trial court abused its discretion by failing to hold in limine hearing even though proponent of testimony did not request a hearing). Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 7 of 19 5 Here, a Daubert hearing could potentially level the playing field4 and correct errors and matters overlooked in the Court’s order by elucidating the substantial flaws in Dr. Noll’s methodology and damages model, which were adopted by the Court’s opinion excluding testimony by Dr. Kalt and Mr. Orszag.5 Such a hearing would also demonstrate the reliability and relevance of Dr. Kalt’s and Mr. Orszag’s opinions.6 1. A Daubert hearing would establish the error in the finding that Dr. Noll’s model “explains 90% of the variance in prices among e-book titles,” because his original model had an adjusted-R2 statistic of 0.90. Dkt. 586 at 8. The Court relied on this finding to conclude Dr. Noll’s assumption that “if all other variables are equal, the relative effect of collusion will be the same” for each e-book transaction within one of Dr. Noll’s categories was “wholly proper.” Id. at 35. Further, the Court’s order frequently cites the 0.90 statistic as evidence of the reliability of Dr. Noll’s model and to dismiss Dr. Kalt’s opinions to the contrary. See id. at 23-24, 30; see also Dkt. 585 at 27, 38, 67, 74-75, 80, 83. As Apple and Dr. Kalt pointed out, the appropriate statistical measure of variance for Dr. Noll’s original model, the within-R2, was only 0.12, meaning that it captured only 12% of the relevant variance. See Dkt. 481 ¶ 134; Dkt. 445 at 14- 4 Apple believes that the Court’s decisions applied a double standard to the parties’ experts by accepting, without rigorous analysis, many of Plaintiffs’ and Dr. Noll’s representations as to the reliability of his model and, in turn, relying on these representations to reject many of Apple’s experts’ opinions. A Daubert hearing could rectify this error by providing the Court and the parties the opportunity to test and compare the reliability of the experts’ opinions. 5 For example, Dr. Noll’s methodology provides no reliable or explanatory bases for calculating damages on titles whose prices stayed at or below their pre-agency levels. Dr. Kalt’s treatment of Penguin and Simon & Schuster prices in this regard is consistent with the Court’s findings regarding defendants’ most-favored-nation provisions (see United States v. Apple Inc., 952 F. Supp. 2d 638, 701 (S.D.N.Y. 2013)) and record evidence relied upon by the Court in its order (see Dkt. 586 at 15). 6 The Court rejected Dr. Kalt’s analysis demonstrating the lack of a stable price structure for ebooks, concluding that it is flawed because it fails to “control for systematic factors affecting prices.” Dkt. 586 at 29-30. Dr. Kalt’s figures and testimony clearly state that his analyses examine price changes after “Accounting for All Noll Explanatory Factors.” Dkt. 538 at Figs. 2A-2F; see also id. ¶¶ 23-34, 41-45, 81-86, 90-91, Figs. 4-6, 8B, 9-10, 18- 24. In the same vein, the Court errs in its understanding of the prices employed by Dr. Kalt in analyzing the correlations of prices over time. Dkt. 586 at 31. A Daubert hearing would provide for proper consideration of these and other matters described below. Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 8 of 19 6 15.7 Dr. Noll did not dispute the applicability of the within-R2 measure in his Reply Declaration. Moreover, when he revised and purportedly improved his damages model using supposed transaction-level data,8 he made the conscious decision to not report the within-R2 or adjusted-R2 of the revised model. The statistical software package used by Dr. Noll does report the within- R2, but does not report the adjusted-R2 statistic in its standard output for the estimation method employed by Dr. Noll. Dr. Noll had to instruct the software package to calculate the adjusted-R2 with respect to his first model. Dkt. 481 ¶ 134(a). After Dr. Kalt demonstrated the inappropriateness of relying on the adjusted-R2, Dr. Noll did not even instruct the software to calculate the adjusted-R2 of his revised model, and all R2 measures that his software did provide (including the within-R2) were deleted from the revised model’s output produced by Plaintiffs. Dkt. 538 ¶ 51 n. 68. The Court seemingly dismissed this omission as irrelevant (Dkt. 586 at 8 n.5), but Dr. Noll’s silence on the R2 of his revised model in the face of criticism by Apple and Dr. Kalt is telling. The reason for Dr. Noll’s silence on this important point would be illuminated at a Daubert hearing. As Dr. Noll himself admitted, the within-R2 measure is “about the extent to 7 As Dr. Kalt explained in his declaration, and as could be further elaborated during a Daubert hearing, it is inappropriate under professional economic standards to rely on the adjusted-R2 to assess how much variance is explained by Dr. Noll’s fixed-effects model. Dkt. 481 ¶¶ 133-34. The adjusted-R2 of a fixed-effects model, such as Dr. Noll’s model, is driven by its use of dummy variables. Dr. Noll uses over 1 million such variables in his revised model, and thus its adjusted-R2 does not provide meaningful insight into the model’s explanatory ability. See PETER KENNEDY, A GUIDE TO ECONOMETRICS 254 (5th ed. 2003) (“Care must be taken in evaluating models containing dummy variables … since these dummy variables could play a major role in generating a high R2, hiding the fact that the independent variables have little explanatory power.”). It is for this reason that the econometrics textbook cited by the Court (Dkt. 585 at 69 n.34) relies on the within-R2 to measure how well fixed-effects models explain variations in the data. See JEFFREY M. WOOLDRIDGE, INTRODUCTORY ECONOMETRICS: A MODERN APPROACH 487 (5th ed. 2013); see also JEFFERY M. WOOLDRIDGE, ECONOMETRIC ANALYSIS OF CROSS SECTION AND PANEL DATA 269 (2002) (explaining that fixed effects models use the “within estimator” to measure the effects of their explanatory variables, such as collusion). 8 The “transaction-level” data used by Dr. Noll are still aggregated into weekly groupings and thus Dr. Noll’s revised model continues to lack important transaction-specific information necessary for understanding and explaining e-book pricing. Dkt. 538 ¶ 52. Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 9 of 19 7 which predicting an average in a category predicts the individual observations in a category.” Dkt. 481-1 at 154:8-10. Here, assessing how reliably “predicting an average in a category [i.e., the average overcharge in each of Dr. Noll’s 502 categories] predicts the individual observations in a category” is central to determining whether Dr. Noll’s assumption of common impact within his categories is valid. According to Dr. Noll, a within-R2 of 0.12 (12%) for his initial model, which utilized data averaged into month-long periods, “tell[s] me . . . that within a month-long period, there’s lots of variation in price around the [category’s] average price; and, of course, you’re not explaining it.” Id. at 186:22-25. He predicted that a transactions model “would be better than that,” i.e., produce a within-R2 of higher than 0.12. Id. at 187:2-7. He was wrong. When it came time to test that prediction with his revised “transactions” model, Dr. Noll eschewed even reporting the relevant results, and they were removed from the revised model’s output files. Dkt. 538 ¶ 51 n.68. Dr. Kalt ran Dr. Noll’s software and found that the within-R2 of Dr. Noll’s revised model was even lower, at 0.05 or 5%. Id. ¶ 51. Far from improving the model’s ability to estimate average overcharges in his categories that reliably reflect the individual e-book price movements within those categories, the revised model’s use of “transaction-level” data does an even worse job at this critical task. The Court places great weight on the value of Dr. Noll’s methodology of grouping e- book titles into hundreds of categories. See, e.g., Dkt. 586 at 22, 32-34. The fact that Dr. Noll’s revised model explains only 5% of the variation in the data within Dr. Noll’s categories undermines the central assumption in Dr. Noll’s model, erroneously adopted as appropriate by the Court’s decision, that the average overcharge in each of these categories can be applied to each individual transaction in a category. Rather, it demonstrates that the average overcharge effect attributed to collusion by Dr. Noll’s model cannot reliably measure the effect (if any) Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 10 of 19 8 experienced by each individual transaction in that category. The Court should therefore reconsider its denial of a Daubert hearing and receive testimony from Dr. Kalt to clarify this vital error. 2. The Court’s exclusion of Dr. Kalt’s opinions regarding the false positives generated by Dr. Noll’s model appears to be based on a misunderstanding of Dr. Kalt’s analysis. The Court’s decision asserts that “Kalt’s analysis compares average but-for prices to actual prices, and so only confirms that titles were sold at more than one price over the period.” Id. at 35. This is incorrect. Dr. Kalt engages in numerous analyses aimed at assessing the degree to which Dr. Noll’s model is plagued by potential false positives, and none of them involves averaging of prices. These analyses demonstrate how Dr. Noll has assumed, without testing, his central conclusion that in the but-for world all transaction prices in a category differ from actual prices by the same percentage amount. For example, Dr. Kalt compares “actual transaction prices to the but-for prices predicted by the model,” where such but-for prices are constructed in accord with the mathematical formula used by Dr. Noll’s revised model for each specific title, at each specific retailer, at a given moment in time (i.e., not for average but-for prices, as the Court appears to have misunderstood was the case). Dkt. 538 ¶ 32.9 Doing such a transaction-by-transaction comparison, Dr. Kalt finds “no reliable demonstration of injury for approximately 11.9 to 16.5 million transactions.” Id; see also id. at Fig. 5. Dr. Kalt also compares the prices paid by current and former named class representatives to the but-for prices for each transaction calculated in accordance with Dr. Noll’s revised model. Seventeen of the twenty-one individuals for whom 9 The Court recognized that but-for prices for individual transactions can be calculated using Dr. Noll’s model. See Dkt. 586 at 8. Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 11 of 19 9 data were available paid actual prices on one or more transactions that were less than the model’s calculated but-for price for that transaction. Id. ¶ 33; see id. at Fig. 6. Rather than examine these analyses, the Court’s decision adopted a “simple example” proffered by Dr. Noll that purports to show that “Kalt misidentifies transactions as being unaffected by price collusion.” Dkt. 586 at 36. But this example by Dr. Noll is nothing more than a hypothetical that assumes that the same overcharge percentage should apply to each of thousands of titles within a given category. Indeed, Dr. Kalt provided a mathematical demonstration—unaddressed and unrebutted by Dr. Noll or Plaintiffs—that Dr. Noll has assumed, rather than proven, precisely this proposition. See Dkt. 481 at App. A. Dr. Kalt’s “false positives” analysis tests the proposition and demonstrates the fallacy in Dr. Noll’s application of a common, average overcharge to measure the effect experienced by individual transactions within a category. He finds that, for millions of transactions, the price that Dr. Noll’s model predicts that an individual would have paid in a given transaction “but-for” the conspiracy is higher than the price actually paid. See Dkt. 538 ¶¶ 30-32. In these cases, however, Dr. Noll’s model attributes individual harm by extrapolating the positive overcharge the model applies to all transactions falling within that given category. Dr. Noll’s criticism of Dr. Kalt’s analysis, which the Court’s order adopted, is misleading and baseless, and Dr. Kalt should be permitted to explain as much in a Daubert hearing. 3. The Court’s order erroneously discarded Mr. Orszag’s well-supported argument that the change in business environment would have required Amazon to raise e-book prices, based on “the overwhelming evidence at trial . . . that no one expected Amazon to raise e-book prices.” Dkt. 586 at 44 n.22. This finding ignores record evidence that Amazon had modified its $9.99 pricing strategy in the past and raised its e-book prices (see Orszag Decl. ¶ 68), and would Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 12 of 19 10 have lost large amounts of money if had not further changed its pricing strategy during the years following the introduction of the iPad. See Dkt. 488 at 22-42.10 Nonetheless, there is ample evidence in the record substantiating Mr. Orszag’s opinions and a Daubert hearing would allow Mr. Orszag to explain that he undertook a detailed analysis of financial data, internal Amazon documents, testimony from Amazon executives, industry analyst reports, and economic literature to reach his opinions. Economists can—and often do—rely on tools outside of econometric modeling to offer reliable opinions. See, e.g., United States v. Valencia, 600 F.3d 389, 427 (5th Cir. 2010) (“regression analysis is not a mandatory feature in all applications of economics or statistics”). The Court’s conclusion that Mr. Orszag engages in a “chain of shaky inferences, without any modeling or analysis” (Dkt. 586 at 44) ignores Mr. Orszag’s use of such tools. These are only a few illustrative examples of errors adopted and matters overlooked by the Court’s rulings that could be addressed through a Daubert hearing and subsequent reconsideration of the Court’s orders.11 Apple should be afforded the “opportunity to be heard on [such] critical issues” given their potential to significantly impact the outcome of these matters (Padillas, 186 F.3d at 417), and respectfully requests that the Court reconsider its denial of a Daubert hearing. 10 Although the Court faults Mr. Orszag for lacking “evidence to support a contention that Amazon subsidized one side of its business with another line of business” (Dkt. 586 at 43), the sufficiency of the evidence relied on by Mr. Orszag must be assessed in light of the fact that Apple sought discovery directly relevant to this subject, as well as other Amazon pricing strategy information highly probative of pricing in the but-for world, but was barred from obtaining it. See, e.g., Dkt. 256 at 11:9-11; Dkt. 309-5 at 24:20-27:10; Dkt. 405 at 22:4-21. 11 Apple has numerous objections to the Court’s orders granting class certification, denying Apple’s motion to exclude Dr. Noll’s opinions, and excluding Apple’s experts’ opinions. Apple is in no way waiving these objections or any of its rights with respect to challenging these errors to the extent permitted by law, including on appeal, and is preserving all rights consistent with its Rule 23(f) petition. Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 13 of 19 11 B. The Court Should Reconsider Its Blanket Exclusion of Dr. Kalt’s Testimony and Near-Blanket Exclusion of Mr. Orszag’s Testimony. In response to Plaintiffs’ Daubert motions, the Court has excluded Dr. Kalt’s opinions in their entirety for all purposes, and has excluded all of Mr. Orszag’s opinions except his opinion that “Noll’s regression analysis improperly included publishers outside the Big Six in the control group, and used data from too long a period.” Dkt. 586 at 39. However, the Court’s order only specifically addressed a subset of each expert’s opinions. Significant opinions of Dr. Kalt and Mr. Orszag were not addressed by the Court’s order. By overlooking these opinions, the Court’s blanket exclusions stretch well beyond Plaintiffs’ specific challenges, are overbroad, manifestly unjust, and should be reconsidered. Cf. Munafo, 381 F.3d 99, 105 (2d Cir. 2004) (“[D]istrict courts may alter or amend judgment to correct a clear error of law or to prevent manifest injustice.”) (citation and internal quotation marks omitted). Such wholesale (and unjustified) exclusion of Apple’s expert evidence denies Apple its due process right to present “every available defense” and deprives it of its right to a fair trial. See Lindsey v. Normet, 405 U.S. 56, 66 (1972) (citation and internal quotation mark omitted). 1. Dr. Kalt’s Opinions After excluding what it deemed Dr. Kalt’s “major opinions,” in a single sentence, the Court also excluded Dr. Kalt’s remaining opinions on the basis that “Apple has not argued, nor has it established, that the remaining interstitial opinions would aid the trier of fact.” Dkt. 586 at 38. That conclusion mischaracterizes and diminishes the significance of Dr. Kalt’s remaining opinions. The remaining opinions unaddressed by the Court’s order are foundational rather than interstitial. They are pillars of Dr. Kalt’s central critique of the defects in Dr. Noll’s model infecting both Dr. Noll’s assumptions and his methodology. Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 14 of 19 12 The Court’s rationale for excluding the remaining opinions is error. Plaintiffs challenged only five “subsidiary” opinions, the exclusion of which do not substantively affect Dr. Kalt’s remaining opinions. The remaining opinions relate to the excluded opinions insofar as they also support Dr. Kalt’s overarching criticism of Dr. Noll’s methodology but do not depend upon the excluded opinions. Plaintiffs have not argued—and the Court has not found—otherwise. For example, neither Plaintiffs nor the Court addressed Dr. Kalt’s detailed analysis, which mathematically demonstrates that the results-oriented engineering behind Dr. Noll’s damages formula assumes—rather than finds—common impact. Dkt. 481 ¶¶ 108-12, App. A; Dkt. 538 ¶¶ 16-21, Fig. 1. Similarly, neither Plaintiffs nor the Court addressed Dr. Kalt’s analysis of the implausible assumptions demanded by Dr. Noll’s methodology when it is applied to individual transactions for e-book titles whose prices stayed at or below their pre-agency levels. Dr. Kalt shows how, to find damages on these titles, Dr. Noll’s calculations require one to assume that, immediately upon agency, the prices of those titles would have inexplicably fallen in the precise amount that Dr. Noll calculates as the price-increasing effect that the conspiracy turned out to have on those titles. Dkt. 481 ¶¶ 130-31, Figs. 28A-28C, 29A-29C; Dkt. 538 ¶¶ 77-80, 82-85, Figs. 19-20. Both of these opinions are independent of the challenged “subsidiary” opinions, and where neither of them is addressed by the Court’s order, their exclusion is unjust and incorrect.12 Although Apple bears the burden at trial of establishing the reliability of Dr. Kalt’s opinions, a blanket exclusion of independent opinions that have nowhere been addressed effectively deprives Apple of an opportunity to do so. Dr. Kalt’s remaining opinions are highly relevant to the core issues in this case and embrace topics that fall within the expertise of an 12 The Court’s rulings regarding Dr. Kalt’s testimony also support the usefulness of a Daubert hearing in this case. A live Daubert hearing would allow the Court to more fully explore the relationship of Dr. Kalt’s opinions to one another. Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 15 of 19 13 economic expert. An ordinary fact-finder would not be able to deduce these facts, and therefore Dr. Kalt’s remaining opinions would aid the trier of fact and should be reinstated. The Court should reinstate the following opinions of Dr. Kalt: • His criticism of the methodological flaws inherent in Dr. Noll’s damages calculation formula (both the one originally described and endorsed by Dr. Noll and the one actually implemented in his work papers). In particular, Dr. Kalt demonstrates that Dr. Noll’s methodology is rigged by its very structure to find that within each and every one of the 500+ groups where his model reports average overcharges every transaction in the same category suffered the same, common percentage overcharge amount. Dr. Kalt illustrates this in painstaking detail in his Mathematical Appendix, and demonstrates how Dr. Noll’s formula is unreliable and unsound because it estimates an average overcharge for all observations in a single grouping and then forces a common impact of the unlawful conduct across all observations in the grouping. In his Reply Declaration, Dr. Noll did not rebut or even respond to this demonstration. These opinions are central to Dr. Kalt’s core criticism of Dr. Noll’s methodology, and neither Plaintiffs nor the Court have offered any basis for excluding them. See Dkt. 481 ¶¶ 108-12, App. A; Dkt. 538 at ¶¶ 16-21, Fig. 1. • His criticism leveled at the inability of Dr. Noll’s use of averages to provide evidence as to the fact of injury and damage of individual class members. Dkt. 481 ¶ 86; Dkt. 538 ¶ 6. • His analysis demonstrating why a claim that publishers’ response to agency was to raise all titles’ prices is not economically sound, particularly in a market in which any one title is differentiated from another. Dkt. 481 ¶¶ 83-88. • His criticism of Dr. Noll’s assertion that the entire distribution of prices increased with the adoption of the agency model. Dkt. 538 ¶¶ 60-63, Fig. 12. • His analysis demonstrating that numerous titles of Hachette, Macmillan, and HarperCollins stayed the same or went down upon the advent of agency marketing (see Dkt. 481 ¶¶ 89-90, Figs. 10A-10F, 11A-11C, 12A-12F, 13A-13F, 18A-18K, 19A-19C, 20A-20C; Dkt. 538 Figs. 15B-15D), and that this is inconsistent with Prof. Noll’s modeling and its but-for prices (see Dkt. 538 ¶¶ 82-85, Figs. 19-22). • His analysis demonstrating the implausible implications of Dr. Noll’s methodology when applied to actual individual transactions. Dr. Kalt shows how Dr. Noll’s methodology provides no reliable or explanatory bases for calculating damages on titles whose prices stayed at or below their pre-agency levels, or even on titles whose prices increased. See Dkt. 481 ¶¶ 128-31, Figs. 28A-28C, 29A-29C, 30A-30C, 31A- 31C; Dkt. 538 ¶¶ 77-80, 82-85, Figs. 19-22. Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 16 of 19 14 • His criticism of the inability of Dr. Noll’s damages model reliably to isolate the effects of the unlawful conduct from legitimate movements in price. See Dkt. 481 ¶ 137, Fig. 33. • His critique of the method by which Dr. Noll’s revised model aggregates e-book prices into weekly groupings, which eradicates important transaction-specific information necessary for understanding and explaining e-book pricing. See Dkt. 538 ¶¶ 52-54. • His analysis showing that the appropriate statistical measure of how reliably Dr. Noll’s analysis has measured individual damages is properly the within R-squared (as calculated by Dr. Kalt in re-running Dr. Noll’s regression models), and that under such a measure, Dr. Noll’s model poorly accounts for factors that determine and move e-book prices. See Dkt. 481 ¶¶ 132-34 & nn.138, 141; Dkt. 538 ¶ 51. • His analysis arguing that the modal price provides a better representation than the mean of the distribution of prices actually paid by class members, and thus is more illustrative of the common experience of class members. See Dkt. 538 ¶¶ 72-75. • His analysis of the pertinent economics of the e-book publishing industry. See Dkt. 481 ¶¶ 25-58; Figs. 1-8. • His opinion that Dr. Noll’s aggregate damages are unreliable and based on ad hoc reasoning. See Dkt. 538 ¶¶ 77-80, 82-85. 2. Mr. Orszag’s Opinions With respect to Mr. Orszag’s opinions, the Court appears to adopt Plaintiffs’ characterization of Mr. Orszag’s opinions as neatly falling into one of four categories. Dkt. 586 at 39. However, as shown below, that is incorrect. Even after excising the Court’s excluded opinions, numerous opinions remain that do not fall within the excluded categories. Specifically, the Court’s order did not address: • Mr. Orszag’s economic analysis of the relevant industry background and of retailers’ profit margins and pricing strategies before and after the switch to agency to demonstrate that Dr. Noll’s model is incomplete and leads to unrealistic and incorrect conclusions that retailers would lose money on e-books absent Apple’s conduct. See Dkt. 488 ¶¶ 12-22, 50-58, 60, 84, 87, 126, nn.215, 220, Figs. III-1, III-2, VI-1, Tbl. VI-1; Dkt. 541 ¶ 75. • Relatedly, the Court does not address Mr. Orszag’s criticism of the incurable contradiction between Dr. Noll’s assumption that Amazon would have continued to emphasize low e-book prices in the but-for world (which is a key driver of his Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 17 of 19 15 damages model) at the same time it was also incurring substantial losses on Kindle sales, and his claim that Amazon would have been profitable in the but-for world. See Dkt. 488 ¶ 87. • Mr. Orszag’s criticism directed at the impropriety of including almost $20 million in Dr. Noll’s damages calculation for consumers whose location is “unidentified” and thus cannot be confirmed to be U.S. residents. See id. ¶ 124, 57 n.227. • Mr. Orszag’s contrasting of the method employed by Dr. Noll to those employed by Plaintiffs’ earlier experts to highlight the fact that a higher damages figure is achieved by Dr. Noll’s questionable modeling choices. See id. ¶¶ 23-26. These opinions are central to Mr. Orszag’s core criticism that Dr. Noll’s damage analysis is flawed and overstates wildly any damages resulting from Apple’s allegedly unlawful conduct. By only allowing Mr. Orszag to testify regarding the narrow issue of whether Dr. Noll’s “regression analysis improperly included publishers outside the Big Six in the control group, and used data from too long a period,” the Court’s order is overbroad and unjustifiably eviscerates Mr. Orszag’s opinions. The Court should therefore reinstate these opinions. CONCLUSION For the reasons stated herein, Apple respectfully requests that the Court reconsider its March 28, 2014 order (Dkt. 586) excluding opinions offered by Apple’s experts, Dr. Joseph Kalt and Mr. Jonathan Orszag, and its March 28, 2014 order (Dkt. 585) admitting the opinions offered by Plaintiffs’ expert, Dr. Roger Noll. Dated: April 11, 2014 Respectfully submitted, /s/ Theodore J. Boutrous, Jr. Theodore J. Boutrous, Jr. Daniel G. Swanson Gibson, Dunn & Crutcher, LLP 333 South Grand Avenue Los Angeles, CA 90071 (213) 229-7000 tboutrous@gibsondunn.com Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 18 of 19 16 Cynthia Richman Gibson, Dunn & Crutcher, LLP 1050 Connecticut Avenue, N.W. Washington, DC 20036 (202) 955-8500 On behalf of Defendant Apple Inc. Case 1:12-cv-03394-DLC-MHD Document 489 Filed 04/11/14 Page 19 of 19