"The Apple iPod iTunes Anti-Trust Litigation"

Filing 241

Expert Report of Michelle M. Burtis filed by Apple Inc.. (Attachments: # 1 Signature Page (Declarations/Stipulations), # 2 Exhibit A, # 3 Exhibit B, # 4 Exhibit C, # 5 Exhibit D-F)(Mittelstaedt, Robert) (Filed on 8/31/2009) Modified on 9/1/2009 (cv, COURT STAFF).

"The Apple iPod iTunes Anti-Trust Litigation"

Doc. 241 Att. 1

Case5:05-cv-00037-JW Document241-2

Filed08/31/09 Page1 of 76

8/31/09 Burtis Report: Exhibit A

Dockets.Justia.com

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page1 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page2 19

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

Robert A. Mittelstaedt #060359 Craig E. Stewart #129530 Michael Scott #255282 JONES DAY 555 California Street, 26th Floor San Francisco, CA 94104 Telephone: [phone] Facsimile: [phone] [e-mail] [e-mail] [e-mail] Attorneys for Defendant APPLE INC. UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF CALIFORNIA SAN JOSE DIVISION

THE APPLE iPOD iTUNES ANTITRUST LITIGATION

Case No. C 07-6507 JW (RS) EXPERT REPORT OF DR. MICHELLE M. BURTIS

This Document Relates to: Somers v. Apple Inc., No. C 07-6507 JW

I.

BACKGROUND AND EXPERIENCE. 1. I am a Vice President at Cornerstone Research, an economic and finance

consulting firm with six offices in the United States, including Washington, D.C. I have a Ph.D. in Economics from the University of Texas at Austin and have published in the field of economics. 2. My work in antitrust cases and class actions is summarized in my curriculum vitae

attached as Exhibit 1. In addition to testifying as an expert in other litigation matters, government regulatory proceedings and private arbitrations, I recently was the expert witness retained by
EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

Case5:05-cv-00037-JW Document241-2

Filed08/31/09 Page3 of 76

8/31/09 Burtis Report: Exhibit A

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page2 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page4 19

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

defendants in In Re Graphics Processing Units Antitrust Litigation where Judge Alsup denied class certification to the proposed class of indirect purchasers and certified a limited class of direct purchasers. I have taught undergraduate microeconomics at the University of Texas and graduate economics at George Mason University. 3. As an economist, I have substantial experience and familiarity with regression

analysis. In graduate school, one of my two fields of specialization was econometrics and my dissertation was based on econometric models of input substitution due to changes in relative prices. Since then, I have developed or analyzed numerous regression analyses for a wide variety of purposes including, for example, models developed to identify relevant product markets, models that analyzed possible reasons that wholesale gasoline prices increased and decreased at different rates, and models that have been offered to estimate damages in a variety of contexts. 4. II. My hourly rate is $540.

ASSIGNMENT. 5. Counsel for Apple asked me to review the reports and deposition transcript of Dr.

Gary L. French and to address issues associated with his proposed econometric analysis.1 Specifically, I have been asked to address whether his proposed econometric analysis is a valid methodology to demonstrate antitrust impact (or injury) and measure damages to proposed class members. 6. Exhibit 2. 7. I summarize my conclusions in Section III; provide a primer on regression A list of material that I considered in preparing this report is included as

analyses with examples in Section IV; and describe the basis and reasons for my conclusions in Sections IV and V. In particular, Section IV focuses on why Dr. French's regression model will not work, and Section V focuses on his inappropriate use of averages.

Affidavit of Gary L. French, Ph.D., February 23, 2009 ("French Report"); Reply Affidavit of Gary L. French, Ph.D., Regarding Class Certification, May, 19, 2009 ("French Rebuttal Report"); Deposition of Gary L. French, Ph.D., Somers v. Apple, April 3, 2009 ("French Dep."). -2EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

1

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page3 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page5 19

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.

SUMMARY OF CONCLUSIONS. 8. Dr. French fails to support his opinion that the alleged impact on indirect

purchasers and the amount of damages caused by the alleged antitrust violation can be determined and measured through regression models. He has not developed any actual model or provided sufficient information about his proposed models to demonstrate or even suggest that they would work. He has collected little if any data. He has not identified or quantified any specific variables. He has written no equations. He has proposed no solutions to the numerous obstacles that his proposed methodology creates. 9. Dr. French has not shown that a regression model would work here, and my

conclusion based on my analysis to date is that his proposed regression models would not work. The purported baseline period before the alleged violation is too short; most of the iPod models at issue were not even sold during that period; the differences among the models are too great; and pricing in this nascent, dynamic industry was affected by too many complex factors that cannot be measured. Dr. French proposes using, for example, the 2001 price of the original white iPod to determine what the prices of all the different iPods, including the nano, shuffle, touch, photo and classic models, would have been today but for the alleged violation. These newer models are so much smaller in size and weight with so much more capacity and so many new and different features that they bear little resemblance to the original iPod models. The current iPod shuffle, for example, is one-twentieth the size of the original iPod with nearly the same capacity.2 10. Dr. French has not proposed a reliable method to determine the dividing line

between the baseline, or "before," period and the period when the conduct at issue allegedly had any effect on pricing. Nor has he proposed any method to distinguish the effects of Apple's business decisions that are not challenged in this lawsuit (e.g., the launch of the iTunes Music Store (iTS) or the launch of progressively smaller, multi-colored, lower priced, higher capacity iPod models) from any effects of the alleged unlawful conduct (e.g., use of proprietary DRM). 11.
2

Dr. French exclusively addresses iPod prices. He ignores the prices of iTS music

See iPod shuffle technical specifications at http://www.apple.com/ipodshuffle/specs.html; first generation iPod technical specifications at http://web.archive.org/web/20011217064651/ www.apple.com/ipod/specs.html (Exhibit 3). -3EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page4 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page6 19

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

even though, under plaintiff's theory, if iPod prices were higher because of the alleged violation than they would have been in the "but-for" world, iTS music prices may have been lower. Individuals with large iTS music purchases relative to their iPod purchases therefore would not have paid any net overcharge under plaintiff's theory. By ignoring iTS music prices, Dr. French fails to propose any methodology to determine whether any consumer paid a net overcharge. 12. Dr. French's proposed method would use average prices, so that the results (if he

could overcome all other obstacles) would be at best a putative measure of an average overcharge. His methodology will not determine whether or not an individual proposed class member has been impacted or the particular amount of any such impact. IV. BACKGROUND ON ECONOMETRIC ANALYSIS AND LIMITS TO ITS USE TO DETERMINE IMPACT AND OVERCHARGE AMOUNTS. 13. In this section, I describe a simple regression model to provide background for my

conclusions regarding Dr. French's methodology. Assume a product is sold during two relevant periods of time ­ a period before some alleged anti-competitive conduct and a period during which the alleged conduct may have impacted prices. Assume in addition that the price of the product depends solely on demand for the product which, in turn, is driven solely by advertising in popular magazines. The greater the amount of advertising, the higher the demand for the product, and as a result, the higher the price.3 14. A regression equation may be prepared, or "specified" and "estimated," as

econometricians use those terms, to determine the relationship between the price of the product and the amount of magazine advertising. The price and the amount of advertising in popular magazines are shown in Example 1. This example shows that in the "before" period, price increased due to magazine advertising and that, in the "during" period, price continued to increase even though magazine advertising leveled off and remained constant. This example and related graphs are obviously highly simplified. The example relates to a single product, sold by one firm at a single price to all consumers in a given time period. Realworld pricing, even for a single product, can be substantially more complex, varying over customers, geography and other dimensions, dependent on more variables than a single one, and not measured so precisely that each price point lines up one right after another in a straight line, over time. -4EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW 3

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page5 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page7 19

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 15. If advertising in popular magazines was the only variable that determined price, an

average overcharge could be obtained from a regression analysis of price and the amount of advertising in popular magazines by using a "dummy variable" in the regression equation that is given a different value depending on time period. This is shown in Example 2.4 The dummy variable is given a value of zero when it is "off," i.e., before the misconduct occurred, and a value of one when it is "on," i.e., during the misconduct period. The regression will produce a number or "coefficient" for the dummy variable, reflecting the number of units by which the price is higher (or lower) when the dummy variable is on. In this way, a basic regression model would provide a measure of the average overcharge, controlling for the relationship between advertising and price. The regression would generate a single "measured overcharge," applicable to the The graph is drawn so that actual prices and "but for" prices are identical until the change in magazine advertising occurs. An actual regression, estimated even with the overly simplified data hypothesized in the example, would generate some differences in the actual and "but for" price lines. Also, in more realistic scenarios, price variability would exist due to the presence of factors in addition to advertising in popular magazines. (To simplify this example, the units of price and advertising are shown on a single scale.) -5EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW 4

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page6 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page8 19

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 -6EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

entire period of alleged misconduct, even though (as shown in Example 2) the actual overcharge, the difference between the actual price and the price but for the alleged conduct, would vary over time. 16. However, there is no way to know from the regression model whether the

coefficient of the dummy variable (the "measured overcharge") is due to the alleged wrongful conduct or whether it reflects some other new factor. Rather, the regression result shows only that the price was, on average, higher during the period when the dummy variable is on. 17. This simple example is based on the assumption that no other variable affects

price. Relaxing this assumption shows its importance. If the firm being modeled here supplemented its advertising in popular magazines with advertising on the internet at the same time as the alleged unlawful conduct is alleged to have taken place, then attributing the increase in price to unlawful conduct would be fallacious. This is shown in Example 3. It shows that there is no change in the relationship between price and total advertising (in both magazines and through the internet) from the "before" period to the "during" period and thus no overcharge.

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page7 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page9 19

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 18. If the model ignored internet advertising, the effect of the internet advertising on

price would be attributed to the alleged unlawful conduct, represented by the dummy variable. If data were available to measure the actual amount of internet advertising, the internet advertising could be added as another variable in the regression equation. If such data were not available, another dummy variable could be used to measure the period when internet advertising began, like the one that measures the effect of the alleged conduct. However, given that internet advertising and the alleged conduct occurred during the same time periods, the regression results would attribute the same number or coefficient to both dummy variables. In other words, the regression would have no way to separate the effects of internet advertising and the misconduct. 19. These examples are highly simplistic in assuming that price is determined by only

one variable. Example 4 shows price and advertising data, where price is affected by more factors than advertising. In this example, the variation (or "dispersion") in price may be related to various factors, such as the particular customer to whom the product is sold, the terms of the sale, or whether the product is sold alone or bundled with some other product. Price variation may also be due to the product being offered by different suppliers that charge varying prices. In addition, price variation may occur if there are variations in the product. A regression analysis that does not account for the important reasons for variation in prices cannot produce accurate or -7EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page8 of of 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page10 19

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 21. reliable estimates. 20. As will be described below, the models proposed by Dr. French to determine and

measure impact suffer from a number of flaws illustrated by this simplified example. V. WHY DR. FRENCH'S PROPOSED MODELS WILL NOT WORK. A. Dr. French Has Provided No Actual Empirical Analysis To Support His Conclusion That His Proposed Econometric Models Can Be Appropriately Estimated. Dr. French says that he intends to use a "temporal competitive benchmark," i.e.,

"a time period outside of the Class Period when the conduct at issue did not occur." French Report, ¶ 66. The theory underlying this approach is to ascertain how relevant factors determine prices in a period unaffected by the alleged conduct (the "before" period), and then use the same relationship of relevant factors to price to determine what the prices would have been during the misconduct period but for the alleged misconduct (the "during" period). If the "but for" prices in the during period are lower than the actual prices, the alleged violation is said to have impacted or caused injury during the class period. -8EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page9 of of 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page11 19

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

22.

To obtain a robust and accurate measure of any alleged overcharge, a regression

model must be able to account for all of the important factors relevant to determining iPod prices. Otherwise, it has no probative value. At his deposition, Dr. French agreed that: "[i]f you don't account for the other factors you can't draw a conclusion [regarding the price impact of the violation]." French Dep. 158:4­158:12. The effect of omitted factors or variables is well known in the econometrics literature.5 23. Dr. French has not demonstrated that he can identify or account for necessary

variables. He has not obtained iPod prices for direct or indirect purchases. He has not identified specific variables affecting iPod prices or obtained the data relevant to those variables. He simply refers to the need to identify "other relevant variables." Although he recognizes the need to control for various iPod features (French Dep. 31:7-18), he has not attempted to catalog them. For example, at deposition he was unsure what a USB connection was and whether he would include a variable for it. French Dep. 76:23 ­ 77:6. A USB port, which Apple first offered on iPods in April 2003, made it significantly easier for PC owners to use iPods and thus would be expected to increase demand for iPods.6 24. Nor has he determined whether the value of those features can be quantified for

regression purposes, or if the data necessary to quantify them is available. For example, at deposition, he admitted that the "coolness" factor, i.e., the perception that the iPod is a "cool" product," was not something that could be captured in a regression analysis. French Dep. 125:1 ­ 125:7.7 And he has not demonstrated that his regression could capture and quantify other features For example: "the omission of relevant variables can bias the results. If, for example, costs were high during those periods of alleged wrongful behavior because of the influence of variables not included in the regression model, or if demand grew more inelastic during that period in ways not captured by the included demand-side variables, then a dummy variable reflecting the likely effect of wrongful behavior might have a large positive coefficient for reasons unrelated to the existence of the alleged conspiracy." Daniel L. Rubinfeld, Quantitative Methods in Antitrust, in 1 Issues in Competition Law and Policy 723 (ABA Section of Antitrust Law 2008), p. 726. "Apple Admits USB Dominance Too Big to Ignore; Experts See it as Right Move," The MacObserver, February 24, 2005 (Exhibit 4).
7 6 5

"[The iPod] has gone from gizmo to life-changing cultural icon. . . . [It] has smacked right into the sweet spot where a consumer product becomes something much, much more: a pet, a status symbol and an indispensable part of one's life." Newsweek, July 26, 2004 (Exhibit 5); -9EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page10 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page12 19

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

like iPod's design, the availability of multiple product colors, the touch screen interface, video capabilities, battery life, weight, size, operating system, and other product characteristics. These features, and others, vary among iPod models, and also differ from the features in competing players. 25. Dr. French appears to consider the use of dummy (or "indicator" as he calls them)

variables to be a panacea for accounting for many of these factors. French Rebuttal Report, ¶ 17. He may use a dummy variable to indicate, for example, the time period when a particular version of iPod shuffles was sold (i.e., to reflect that price data are for an iPod shuffle). But, as illustrated in the earlier examples, turning on a dummy variable during the period that an iPod shuffle is sold simply results in a measure of average price differences between an iPod shuffle and other iPod models, showing for example that iPod shuffles are, on average, less expensive than the baseline iPod. It does nothing to show how much the shuffle could have cost but for the alleged violation. 26. Consider the difference between an "explanatory variable" such as the amount of

advertising expenditures, and a dummy variable that simply indicates whether or not there was advertising. In the former case, the number (or coefficient) for the explanatory variable for advertising expenditures in the regression example would indicate the amount that price changes when advertising changes. For example, if the coefficient were 10 (using a log-linear specification as described by Dr. French), the interpretation of the coefficient is that, for every one percent change in advertising expenditures, the price of the product increases 10 percent. The regression coefficient indicates something about the relationship between the variables and provides information about the change in one variable (advertising) and the consequent change in another variable (price).
(continued...)

"No other product has the incredible, loyal devotion that the iPod inspires. . . . It revolutionized and popularized music players . . . ." CNet.com (http://www.cnet.com/1990-11136_1-63122461.html) (Exhibit 6); "Apple Computer's iPod portable digital music player has become the gold standard for digital music fans. No other player has been able to match its combination of elegant design, small size and high capacity." Wall Street Journal, May 1, 2003 (Exhibit 7). - 10 EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page11 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page13 19

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

27.

Contrast this with the information derived from a dummy variable in a regression.

Instead of using the amount of advertising, a dummy variable is included in the equation to indicate whether or not there is advertising. The coefficient provides no information about how price changes when advertising changes. It shows only that price was different by some average amount in the period with advertising compared to the period without advertising.8 28. There is also a technical limit to the number of dummy variables that can be used.

If Dr. French proposes to use dummy variables for different retailers, various product features, characteristics, and capabilities (French Rebuttal Report, ¶ 17), it is likely that as a group they will be too overlapping in time for the model to be able to isolate any effect of the alleged misconduct (also proposed by Dr. French to be measured with a dummy variable). In econometrics jargon, they will be perfectly correlated with the misconduct dummy variable. Thus, the results will not be valid. B. Dr. French Has Not Proposed A Methodology For Determining When The Price Effect Of The Alleged Tie Began. 29. Dr. French has not determined when any alleged impact on iPod prices began and

has not proposed any specific methodology to determine any such date. French Dep. 42:9­ 42:15. This is a critical date for his proposed approach because it defines the "temporal competitive benchmark," or the "before" period. It also defines the time period for the dummy variable that he claims can be used to measure the price effect, or overcharge, due to the alleged violation. 30. If the period of impact of the alleged misconduct is not accurately defined, the

regression model will be inaccurate. It may find an overcharge when none exists. It may overstate or understate the amount of any overcharge. 31. To resolve this critical issue, Dr. French proposes to make a "judgment call" as to

when a sufficient number of iTS consumers became "locked-in" to using the iPod so that Apple The limited ability of dummy variables to provide explanatory power is well known in econometrics. Peter Kennedy, A Guide to Econometrics, 254 (5th ed. 2003) ("Care must be taken in evaluating models containing dummy variables designed to capture structural shifts or seasonal factors, since these dummies could play a major role in generating a high R2, hiding the fact that the independent variables have little explanatory power."). - 11 EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW 8

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page12 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page14 19

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

could allegedly price iPods at higher levels. French Dep. 43:15­51:5, 52:9­59:8, 82:9-8, 80:9­ 85:9. He apparently believes that this date can be determined by analyzing purchases of iTS music although he has not explained exactly what he would be looking for. His contention is obviously flawed. First, purchasing from iTS does not mean that the purchaser is "locked-in" to iTS or iPods. A consumer, for example, may have intended to buy an iPod regardless of whether he bought or planned to buy iTS music. Second, Dr. French admits that he does not know how many iTS songs it would take to be "locked-in" or how much "lock­in" would be needed to affect price. French Dep. 57:18­58:2. He has not proposed a methodology for dealing with these issues. Third, the only way to determine if an individual was "locked-in" would be to make an individual inquiry. Without such inquiry, Dr. French would simply be assuming the very thing that he is supposedly testing for. C. 32. The "During" Or "Class Period" Contains Different Products. Dr. French's proposed "temporal competitive benchmark" method depends on an

assumption that the relationship between prices and their determining factors (other than the alleged misconduct) is known and remains the same. For example, if price is assumed to depend on cost and demand changes, it is assumed that price reacts similarly to cost changes, or demand changes, in both the before and during periods. Without this stability, the model is mis-specified and any estimate of the impact of the alleged violation is invalid. 33. This critical assumption is violated in this case. Most obviously, the products in

the before period are different from the products in the during period. Although Dr. French has not determined the beginning of the alleged overcharge period, he surmised that: "I'd be amazed if it's more than a couple of months after December `03." French Dep. 177:7 ­ 177:8. The large majority of iPod products were not introduced until after this time. As shown in Exhibit 8, only the first three generations of the original iPod and the original iPod mini were sold in the "before" period (assuming he were able to demonstrate that the "before" period began "a couple of months" after December 2003). The rest of the 42 iPod models--including the iPod photo, video, touch, nano and shuffle--were all sold only in the "during" period.9 The characteristics, technical
9

See Exhibit 9: "Apple Introduces iPod Photo," Apple press release, October 26, 2004; "Apple - 12 EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page13 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page15 19

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

capabilities and features of these models changed dramatically. The original iPods were substantially larger, heavier, and had less capacity than the models that followed. 10 Products introduced later were more advanced with more features and technical capabilities. The iPod touch, for example, was introduced with a 3.5 inch widescreen display and "multi-touch interface," allowing users to "pinch" the screen to make images larger or smaller or "flick" the screen to change the image. The iPod touch automatically senses when its position changes and rotates the image to a landscape position. It allows users to play music, watch videos and full­ length feature movies, send and receive e-mail, store and view photographs, and access the internet with wi-fi capability. It is 8 millimeters thick, less than half the thickness of the original iPod.11 To take another example, as innovative as the original iPod was, the iPod nano was hailed in its own right in 2006 as "irrevocably alter[ing] the landscape for portable audio players." It took "clean design aesthetics to a new level [and] brought us the first high-capacity (4GB) flashbased player--and one priced within reach of the masses, no less."12 34. Dr. French proposes to use the prices of the early generations of iPods, with their

much more limited capabilities and features, to determine what the prices of the more advanced models should have been absent the alleged conduct in this case. But he has not considered or demonstrated that the factors that determined prices in the before period (whenever that period may have ended) had similar effects on prices in the during period (whenever that period may have started). For example, the popularity and demand for MP3 players in general grew tremendously in the during period whereas the benchmark period was quite early in the development of these products.

(continued...)

Introduces iPod shuffle," Apple press release, January 11, 2005; "Apple Introduces iPod nano," Apple press release, September 7, 2005; "Apple Unveils iPod touch," Apple press release, September 5, 2007.
10 11

http://web.archive.org/web/20011217064651/www.apple.com/ipod/specs.html (Exhibit 3).

See, for example, "Apple Unveils iPod touch," Apple press release, September 5, 2007 (Exhibit 9). 2006 PC World Innovations Awards, ABC News (Exhibit 10). - 13 EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

12

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page14 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page16 19

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
13

35.

Dr. French's only proposal is to include "dummy' variables in the regression

analysis to account for the differences in product features. I have addressed the problems with this approach above. In addition, this proposed solution begs the important question. For a proposed model to reliably estimate the effect of alleged misconduct, the relationship between prices and the variables that determine the prices must be the same in the two periods. But there is no reason to assume that the relationship between the type of screen on an iPod and its price in 2001 is the same as the relationship in 2008, particularly when the type of screen was different. Therefore, this assumption underlying the proposed model cannot be valid. 36. The number of products and the number of prices of early iPods are also limited in

number, which makes it more difficult to estimate a regression model where the premise of the model is a comparison of prices in two periods.13 37. Also, Dr. French has suggested that he could prepare separate models for different

iPods. However, any individual regression for iPods introduced after late 2003 or early 2004 (or whatever date Dr. French may pick) would by definition not have a "before" period. In other words, there is no "temporal competitive benchmark." French Report, ¶ 67. Thus, this proposed method fails. D. Dr. French Has Not Proposed A Methodology To Isolate The Effect Of The Alleged Tie From Other Confounding Events Not At Issue. 38. A regression analysis does not work when it cannot differentiate the effect of the

alleged misconduct (e.g., Apple's use of proprietary DRM) from other confounding events not alleged to be unlawful (e.g., other factors affecting the demand for iPods). For example, to the extent that iTS created demand for iPods independent of Apple's use of proprietary DRM, that effect would be confounded with any effect of the alleged violation. Dr. French has proposed no methodology to separate the pro-competitive and beneficial effect of iTS from the aspect of iTS alleged to be unlawful. French Dep. 78:5 ­ 78:14.

Dr. French's own analysis shows the limited number of products that were available in a "before" period defined to end in late 2003 or early 2004. See French Report, Exhibits 1a and 2. - 14 EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page15 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page17 19

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
14

39.

The problem of confounding events or factors would also exist because the

changing features of different iPod models may be confounded with the allegedly wrongful conduct. For example, a regression of the price of an iPod touch cannot differentiate any effect of the alleged violation from the effect of the touch screen, because both factors exist (i.e., the dummy variable is turned "on") for the entire period of iPod touch sales. E. Dr. French's Methodology Does Not Consider The Overall Or Net Injury To Individual Proposed Class Members. 40. Plaintiff's theory is that Apple's use of proprietary DRM for its music store

increased the demand and therefore prices for iPods. It would follow from plaintiff's theory that some consumers would have elected not to purchase iTS music for the same reason and, to that extent, the demand for and price of iTS music would have decreased due to the use of DRM. Accordingly, to determine whether any consumer paid a net overcharge would require an analysis of both the prices of iPod products and the price of iTS music. For example, depending on the amount of any iPod "overcharge" and the amount of any iTS music "undercharge," whether an individual paid a net overcharge would turn on the number of iPods and music files purchased by that individual. Individuals with sufficiently large purchases of music relative to iPods would not have paid a net overcharge, even under plaintiff's theory and even if they could establish an iPod overcharge. Dr. French does not address this issue at all, or propose any method to do so.14

In addition to the econometric methods discussed in this report, Dr. French also mentions that direct purchasers' overcharge could be determined using a "yardstick" approach or a "margin analysis." However, in each case he simply describes the approach in a single sentence with no further analysis or discussion (French Report, ¶ 65). A "yardstick" method would require Dr. French to identify products comparable to iPod products (but without the alleged wrongful conduct) and, based on a comparison of the prices of those products to the prices of iPod products, determine the impact of the alleged conduct. Dr. French has not identified "yardstick" products that could be used in such an exercise and, given the unique features of iPod products, this method is likely infeasible. Similarly, any "margin analysis" would require some benchmark of comparison. Dr. French has not identified any benchmark margin or provided information that would lead to its discovery. - 15 EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page16 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page18 19

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

VI.

DR. FRENCH'S PROPOSED USE OF AGGREGATED AND AVERAGED DATA WILL NOT SHOW IMPACT FOR ANY INDIVIDUAL INDIRECT PURCHASER. 41. At his deposition, Dr. French stated that he intends to calculate only an average

overcharge and not to determine whether individual sale transactions included any overcharge. French Dep. 20:25-21:23, 64:11-15. He further stated that, even if transactional data is available, he will aggregate that data into monthly average data before running his regression. French Dep. 18:8-20, 21:11-12. 42. This use of average monthly prices hides pricing variation that Dr. French's

analysis needs to explain. As stated in ABA Section of Antitrust Law, Econometrics (2005): [s]ometimes the prices used by economists are averages of a number of different prices charged to different customers or for somewhat different products. Using such averages can lead to serious analytical problems. For example, averages can hide substantial variation across individual cases, which may be key to determining whether there is common impact.15 In other words, because this methodology assumes that the conduct in question "has the same effect on every purchaser and focuses on an average effect," it does not "test whether there is impact on all members of a proposed class." Rather, "it assumes the very proposition that is being tested." Id. at 222. 43. In his Reply Affidavit (¶ 22), Dr. French confirms that his proposed regression

models will yield only an "average overcharge or average pass-through." He states that he believes that he will be able to obtain monthly average prices on a retailer-by-retailer basis, rather than an average price across all retailers. He has not actually obtained those data, however. Nor does he state whether the data are available for all of the relevant retailers or for the entire relevant period. 44. Even assuming that such data are available, however, these monthly averages will

mask significant price differences in any given month arising from such things as sales or promotions in which the iPod is effectively discounted by being bundled with other merchandise or gift cards. Exhibit 1 to Dr. French's Reply Affidavit (which Dr. French offers to show

15

ABA Section of Antitrust Law, Econometrics (2005), p. 220. - 16 EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page17 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page19 19

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

supposed price uniformity) illustrates these differences. In just a single month, four of the listed retailers offered sale prices that differed from that retailer's regular price for that month, ranging from $15 to $70. To take another example, on one day in November 2007, Amazon.com offered an iPod nano (4GB) for $30 or almost 20% lower than Apple's store, and raised it two days later by $15, which was $23 above CompUSA's price that day. See Declaration of Michael Scott (filed 4/20/09), Exs. 15-16. Moreover, Dr. French admits that he omitted from that exhibit "[p]rices of refurbished models and bundled products." French Rebuttal Report, Exhibit 1, fn. c. Retailers such as Best Buy, WalMart and Target have offered iPods bundled with gift cards worth as much as $50 or valuable merchandise such as exercise machines, Sony stereo headphones, FM transmitters and multi-device USB AC/DC chargers. See Scott Declaration, Exs. 23-27. 45. In addition to using only monthly averages, Dr. French is vague about the degree

to which he intends to aggregate data across iPod models. Over the years, Apple has sold the iPod, iPod photo, iPod classic, iPod mini, iPod nano, iPod shuffle and the iPod touch as well as special edition iPods such as the U2 iPods. And most of these broad model categories include a number of different generations, with at least 42 different specific models sold since the iPod was first introduced. These models vary widely in features and price. For example, in 2003 Apple sold at a retail price of $399 a 15GB iPod with a black and white screen and no video capabilities.16 Compare that to a 16GB iPod touch that is now sold for three-quarters the price, $299, with all of the new features noted above (¶ 33) plus the availability of thousands of video games and software applications.17 And compare those to an iPod shuffle now sold for $79 that weighs less than half an ounce and has no display at all, but instead a "voice" that indicates the song and identity of the performer.18 There is no reason to believe that, if any overcharge existed, it would be uniform across all of these models.

16 17 18

"Apple Introduces New iPods," Apple press release, April 28, 2003 (Exhibit 9). "Apple Introduces New iPod Touch," Apple press release, September 9, 2008 (Exhibit 9).

"Apple Announces Incredible New iPod Shuffle," Apple press release, March 11, 2009 (Exhibit 9). http://www.apple.com/ipodshuffle/specs.html (Exhibit 3) - 17 EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page18 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page20 19

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

46.

While Dr. French stated at deposition that he would "be surprised if it was very

different from model to model" (French Dep. 37:3­37:4), Dr. French suggests that he could prepare separate regressions, with varying specifications, for different iPods. French Dep. 35:5­ 37:10, 63:22­64:9. Under this scenario, there could be as many as 42 different specifications of the regression model for direct purchasers and an equal number for indirect purchasers.19 Without doing this, he would simply be relying on an assumption that any overcharge would be common to all iPods. In addition, Dr. French's methodology would not work because, as described above, there would be no "before" period for the vast majority of these products and therefore no way for the regression to be estimated. 47. A similar problem exists for regressions run for separate resellers. Not only would

separate regression models for dozens or hundreds of different resellers be inconsistent with the required commonality for class treatment, but for those resellers that purchased only during the class period there would be no "before" period and therefore the proposed models could not be used to estimate any alleged overcharge. 48. Dr. French suggests that he would combine data for various iPod models to create

a "before" period. In addition to the defects discussed above, this methodology would not determine whether individual purchasers of a particular iPod model were impacted by the alleged violation. In such a specification the estimated overcharge would be, as Dr. French describes, an average across all the included iPod models, retailers, and purchasers, direct or indirect.

French Report, Exhibit 2. Dr. French is not clear about how he would determine what constitutes an iPod "model" for his proposed separate regressions (e.g., whether he would consider a 20GB iPod photo to be a distinct model from a 60GB iPod photo). - 18 EXPERT REPORT OF DR. MICHELLE M. BURTIS Case No. C 07-6507 JW

19

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page19 ofof 76 Case5:07-cv-06507-JW Document74 Filed08/31/09 Page21 19

CCase5:07-cv-06507-JW Document241-2 ase5:05-cv-00037-JW Document74-1

Filed06/17/09 Page1 of 55 Filed08/31/09 Page22 of 76

CCase5:07-cv-06507-JW Document241-2 ase5:05-cv-00037-JW Document74-1

Filed06/17/09 Page2 of 55 Filed08/31/09 Page23 of 76

CCase5:07-cv-06507-JW Document241-2 ase5:05-cv-00037-JW Document74-1

Filed06/17/09 Page3 of 55 Filed08/31/09 Page24 of 76

CCase5:07-cv-06507-JW Document241-2 ase5:05-cv-00037-JW Document74-1

Filed06/17/09 Page4 of 55 Filed08/31/09 Page25 of 76

CCase5:07-cv-06507-JW Document241-2 ase5:05-cv-00037-JW Document74-1

Filed06/17/09 Page5 of 55 Filed08/31/09 Page26 of 76

CCase5:07-cv-06507-JW Document241-2 ase5:05-cv-00037-JW Document74-1

Filed06/17/09 Page6 of 55 Filed08/31/09 Page27 of 76

CCase5:07-cv-06507-JW Document241-2 ase5:05-cv-00037-JW Document74-1

Filed06/17/09 Page7 of 55 Filed08/31/09 Page28 of 76

CCase5:07-cv-06507-JW Document241-2 ase5:05-cv-00037-JW Document74-1

Filed06/17/09 Page8 of 55 Filed08/31/09 Page29 of 76

CCase5:07-cv-06507-JW Document241-2 ase5:05-cv-00037-JW Document74-1

Filed06/17/09 Page9 of 55 Filed08/31/09 Page30 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page10 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page31 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page11 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page32 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page12 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page33 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page13 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page34 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page14 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page35 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page15 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page36 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page16 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page37 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page17 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page38 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page18 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page39 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page19 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page40 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page20 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page41 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page21 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page42 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page22 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page43 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page23 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page44 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page24 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page45 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page25 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page46 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page26 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page47 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page27 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page48 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page28 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page49 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page29 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page50 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page30 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page51 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page31 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page52 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page32 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page53 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page33 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page54 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page34 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page55 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page35 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page56 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page36 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page57 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page37 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page58 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page38 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page59 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page39 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page60 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page40 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page61 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page41 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page62 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page42 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page63 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page43 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page64 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page44 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page65 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page45 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page66 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page46 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page67 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page47 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page68 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page48 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page69 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page49 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page70 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page50 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page71 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page51 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page72 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page52 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page73 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page53 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page74 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page54 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page75 of 76

Case5:05-cv-00037-JW Document241-2 Filed06/17/09 Page55 of 55 Case5:07-cv-06507-JW Document74-1 Filed08/31/09 Page76 of 76