Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 1 of 94 PageID #:21816 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 UNITED STATES DISTRICT COURT FOR THE NORTHERN DISTRICT OF ILLINOIS EASTERN DIVISION KLEEN PRODUCTS LLC, et al. individually and on behalf of all those similarly situated, Case No. 1:10-cv-05711 Hon. Harry D. Leinenweber Plaintiffs, CLASS ACTION v. PACKAGING CORPORATION OF AMERICA, et al., Defendants. EXPERT REPORT OF DENNIS W. CARLTON September 19, 2014 FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 2 of 94 PageID #:21817 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 I. INTRODUCTION. A. Qualifications. 1. My name is Dennis W. Carlton. I am the David McDaniel Keller Professor of Economics at the Booth School of Business of The University of Chicago. I received my A.B. in Applied Mathematics and Economics from Harvard University and my M.S. in Operations Research and Ph.D. in Economics from the Massachusetts Institute of Technology. I have served on the faculties of the Law School and the Department of Economics at The University of Chicago and the Department of Economics at the Massachusetts Institute of Technology. 2. I specialize in the economics of industrial organization. I am co-author of the book Modern Industrial Organization, a leading text in the field of industrial organization, and I also have published over 100 articles in academic journals and books. In addition, I serve as Co-Editor of the Journal of Law and Economics, a leading journal that publishes research applying economic analysis to industrial organization and legal matters; serve on the Editorial Board of Competition Policy International, a journal devoted to competition policy; and serve on the Advisory Board of the Journal of Competition Law and Economics. I have also served as an Associate Editor of the International Journal of Industrial Organization and Regional Science and Urban Studies, and on the Editorial Board of Intellectual Property Fraud Reporter. I have received several awards, including being designated as the 2014 Distinguished Fellow of the Industrial Organization Society. 3. In addition to my academic experience, I served as Deputy Assistant Attorney General for Economic Analysis, Antitrust Division, U.S. Department of Justice from October 2006 through January 2008. I also served as a Commissioner of the Antitrust Modernization Commission, created by Congress to evaluate U.S. antitrust laws. I have served as a consultant to the Department of Justice and Federal Trade Commission on the Horizontal Merger Guidelines, as a general consultant to the Department of Justice and Federal Trade FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 3 of 94 PageID #:21818 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 -2- Commission on antitrust matters, and as an advisor to the Bureau of the Census on the collection and interpretation of economic data. 4. I also am a Senior Managing Director of Compass Lexecon, a consulting firm that specializes in the application of economics to legal and regulatory issues and for which I served as President (of Lexecon) for several years. I have provided expert testimony before various U.S. state and federal courts, the U.S. Congress, a variety of state and federal regulatory agencies and foreign tribunals. My curriculum vitae, which includes a list of my testifying experience during the last four years, is attached as Exhibit 1 to this report. For my work in this matter, Compass Lexecon has billed my time at $1,350 per hour and neither its compensation nor mine depends on the outcome of this matter. A list of the materials I rely on in this report is attached as Exhibit 2. B. Assignment. 5. Plaintiffs claim that [b]eginning in or about 2004, the Containerboard Products industry was faced with decreasing profit margins, rising product demand, and a promising macroeconomic outlook. Nevertheless, Defendants began a coordinated across-the-board imposition of capacity restraints, leading to a subsequent restriction in the supply of Containerboard Products on the market. The goal of the conspiracy was to fix, raise, maintain and stabilize the price at which Containerboard Products were sold during the Class Period.1 Plaintiffs “seek certification of a class” of all customers that purchased from defendants during an alleged class period from February 15, 2004 through November 8, 2010.2 6. In support of their request for class certification, plaintiffs filed two expert reports: (1) Declaration of Michael J. Harris, Ph.D. in Support of Plaintiffs’ Motion for Class Certification (“Harris Report”); and (2) Report of Mark Joseph Dwyer, Ph.D. in Support of Plaintiffs’ Motion for 1. 2. Kleen Products LLC, et al. v. Packaging Corporation of America, et al., Consolidated and Amended Complaint For Violation of the Sherman Act and [Proposed] Amendments to Plaintiffs’ Consolidated and Amended Complaint (“Amended Complaint”), ¶ 6. See Plaintiffs’ Memorandum of Law in Support of Motion For Class Certification (“Plaintiffs’ Memorandum of Law”), at 1. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 4 of 94 PageID #:21819 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 -3- Class Certification (“Dwyer Report”). I have been asked by counsel for Georgia Pacific LLP (“GP”) to review and evaluate the Harris and Dwyer Reports. 7. Between them, Dr. Harris and Dr. Dwyer reach three major conclusions: (1) Dr. Harris concludes that common economic evidence shows that defendants successfully conspired.3 (2) Dr. Harris and Dr. Dwyer each conclude that they can show through common economic evidence that the alleged conspiracy “impacted” all or nearly all class members.4 (3) Dr. Dwyer concludes that “standard econometric methods . . . may be used to quantify aggregate overcharge damages to the class.”5 As I explain in this report, I find that the analyses presented by Drs. Harris and Dwyer contain numerous flaws and their conclusions are without merit. 8. The rest of my report is organized as follows. In Section II, I summarize my conclusions. In Section III, I provide background information on the containerboard industry. In Section IV, I explain that Dr. Harris fails to present common (or any) economic evidence that shows the existence of a conspiracy. In Section V, I explain that neither Dr. Harris nor Dr. Dwyer show that the alleged conspiracy had an impact on all or nearly all putative class members. In Section VI, I show that Dr. Dwyer’s statistical analysis purporting to measure aggregate overcharge damages to the purported class is flawed and unreliable. 3. 4. 5. See Harris Report, at 6 (“[t]he conduct only makes economic sense if the Defendants were engaged in collective action; and the empirical evidence, common to the class, indicates that the Defendants were able to successfully elevate prices above the level one would expect from non-coordinated behavior.”). See Harris Report, at 6 (“I have also concluded, based on the economic evidence common to the class, that the collusive conduct of the Defendants would have been broadly and generally successful, elevating prices above what they would have been without collusion, thus impacting all, or nearly all, class members.”). See also Dwyer Report, ¶ 11 (“all four of these measures of impact show that all or nearly all class members paid higher prices as a result of price increases announced and implemented by the defendants and alleged by the plaintiffs to be facilitated by the collusive restriction in supply.”). Dwyer Report, ¶ 12. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 5 of 94 PageID #:21820 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 -4- II. SUMMARY OF CONCLUSIONS. 9. Based on the analysis I present in this report, I have reached the following three major conclusions. (1) Plaintiffs fail to present common (or any) economic evidence that defendants engaged in a successful conspiracy. In particular, Dr. Harris fails to show, by common (or any) evidence, that defendants’ conduct was inconsistent with what is known in the economics literature as non-conspiratorial “oligopolistic interdependence.” Furthermore, Dr. Harris fails to show that any defendant’s actual conduct differed from what its conduct would have been absent the alleged conspiracy. Thus, there is no basis for Dr. Harris’s claim that economic evidence allows him to conclude that various actions taken by defendants are evidence of conspiracy. (2) Plaintiffs’ experts’ analyses fail to support their claim that common evidence can prove that the alleged conspiracy “impacted” all or nearly all class members. Dr. Harris presents no empirical evidence and instead provides only a theoretical analysis of impact that is completely at odds with the empirical evidence he presents. Dr. Dwyer presents an empirical analysis of impact, but his analysis contains numerous flaws – for example, he fails to isolate “antitrust impact” from non-conspiratorial changes in supply and demand conditions, and uses a model that does not assess impact on two-thirds of potential class members. Thus, Drs. Harris and Dwyer fail to show that plaintiffs can prove through the use of common evidence that the alleged conspiracy impacted all or nearly all putative class members. (3) Plaintiffs provide no economic analysis that reliably measures aggregate damages. Plaintiffs rely on an estimate of aggregate damages prepared by Dr. Dwyer that is based on a flawed and unreliable methodology. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 6 of 94 PageID #:21821 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 -5- III. BACKGROUND INFORMATION. 10. Plaintiffs in this case claim that several producers of “containerboard” entered into a conspiracy that had the effect of raising the price of containerboard and products made from containerboard beginning in 2004. Defendants in this matter are International Paper (“IP”); Cascades Canada, Inc. and Norampac Holdings U.S. Inc. (“Norampac”); Weyerhaeuser Co.; Georgia-Pacific LLC (“GP”); Temple-Inland Inc. (“Temple-Inland”); and Smurfit-Stone Container Corp (“SSCC”), now RockTenn CP, LLC. I understand that Packaging Corporation of America (“PCA”), which had been a defendant, has been dismissed from this case pursuant to a settlement reached with plaintiffs that has been approved by the Court. 11. Each of the defendants produces forms of paper known as “linerboard” and “corrugating medium,” also referred to as “medium.” Linerboard and medium are collectively referred to as “containerboard” or “rollstock.” Facilities that produce linerboard and/or medium are usually referred to as “containerboard mills.” Containerboard mills often operate more than one paper “machine” (e.g., mills may have separate machines for producing different grades of linerboard and/or medium). Several firms that are not defendants in this litigation also produce containerboard. 12. Linerboard and medium are combined to produce “corrugated sheets” using equipment referred to as a “corrugator.” Corrugated sheets are used to produce a wide variety of boxes and other products (e.g., display units) using additional equipment. Plants that use containerboard to produce and sell sheets are referred to as “sheet feeders.” Plants that use sheets to produce boxes and other products are referred to as “sheet plants.” Plants that use containerboard to produce sheets and use those sheets to produce boxes and other products are referred to as “box plants” (i.e., a box plant is an integrated sheet feeder and sheet plant). In this report, I use “corrugated products” to refer to sheets, boxes and other products made FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 7 of 94 PageID #:21822 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 -6- from sheets. Plaintiffs in this case refer to containerboard, sheets and all other corrugated products as “containerboard products.”6 13. Each of the defendants is “vertically integrated” to some degree – that is, each defendant produces containerboard and also corrugated products. GP, for example, uses roughly 70 percent of its containerboard production in its own box plants or sheet feeders.7 However, defendants also sell containerboard to other producers of corrugated products, (e.g., sheet feeders and box plants), including firms that produce both containerboard and corrugated products. IV. PLAINTIFFS FAIL TO PRESENT COMMON (OR ANY) ECONOMIC EVIDENCE THAT DEFENDANTS ENGAGED IN A SUCCESSFUL CONSPIRACY. A. Introduction. 14. Plaintiffs claim that defendants conspired to restrict supply in the containerboard industry, which allowed defendants to raise the price of containerboard products. In particular, “Plaintiffs’ central allegation is that the conspiracy was facilitated by agreements to restrict the supply of containerboard by cutting capacity, slowing back production, taking downtime, idling plants, and tightly restricting inventory, all of which set the stage for coordinated price increases of all Containerboard Products.”8 15. Dr. Harris claims to find common economic evidence that supports plaintiffs “central allegation.” Dr. Harris claims that: the economic evidence shows that the conduct of the Defendants was more consistent with collusion than with independent economic decision-making, with each Defendant exhibiting conduct that was contrary to its unilateral self-interest, acting independently. The conduct only makes economic sense if the Defendants were engaged in collective action.9 6. 7. 8. 9. See, for example, Harris Report, at 3, fn 1 (citing “Plaintiffs’ Complaint”). See, for example, Harris Report, at 10. Plaintiffs’ Memorandum of Law, at 1. Harris Report, at 6. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 8 of 94 PageID #:21823 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 -7- 16. As I discuss in this section of my report, Dr. Harris’s analysis is flawed and his claim is without merit. B.  First, Dr. Harris fails to show, by common (or any) economic evidence, that defendants’ conduct was inconsistent with what is known in the economics literature as nonconspiratorial “oligopolistic interdependence.”  Second, Dr. Harris fails to provide any economic analysis that even attempts to show that any specific actions taken by defendants would have been different absent the alleged conspiracy. That is, Dr. Harris fails to show that any defendant’s actual conduct differed from what its conduct would have been in a non-conspiratorial “but-for world” (i.e., the world absent the alleged conspiracy). Thus, there is no basis for Dr. Harris’s claim that various actions taken by defendants are evidence of conspiracy because those same actions are consistent with non-conspiracy. Defendants’ Conduct is Consistent with Non-Conspiratorial Oligopolistic Interdependence. 17. Dr. Harris states: “[t]he Defendants operate in what economists call an oligopolistic market. . . . In an oligopolistic market firms are interdependent. That is, each firm, even when acting unilaterally, develops ‘conjectures’ of what other firms will do in response to its actions.”10 Thus, in oligopolistic markets, firms do not act “independently” of their rivals. Instead, as Dr. Harris recognizes, firms in such markets take account of the likely actions and reactions of their rivals. For example, a firm considering whether to raise prices will take account of its rivals’ likely reactions to its price increase. Such oligopolistic conduct is common and not conspiratorial, and firms in such markets typically make efforts to gather information about their competitors from a variety of sources, such as customers, securities filings and thirdparty research firms. 18. Interdependent conduct in an oligopolistic industry may be illustrated by the following hypothetical example. Assume that there are two gasoline stations – owned independently by Bob and Steve – on opposite corners of the main intersection in a small town, and that no other gasoline stations are located within reasonable driving distance. Transaction 10. Harris Report, at 34 (footnote omitted). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 9 of 94 PageID #:21824 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 -8- prices charged by the stations are completely transparent – prices are posted on signs visible simultaneously by nearby drivers and by each station’s competitor. 19. If both stations have equal capacity, and each can supply the entire amount of gasoline demanded in the area at a competitive (i.e., marginal cost) price, then each will have an incentive to try to sell all its capacity, leaving the other with no sales. To do this, each station would appear to have an incentive to lower its price and undercut its competitor, thereby increasing its market share. However, if Bob actually does reduce price, Steve likely will react immediately by lowering his price in order to protect his own market share. If this continues, then a competitive (marginal cost) price will result, as both stations attempt to make sales at the expense of the other. 20. As a result of experience trying to win a greater share of customers, or of Bob’s and Steve’s individual, common sense predictions of the outcome of such an attempt, each station owner will realize that if he cuts price by, say, five cents, his competitor will as well. Thus, a price cutter will not gain market share, but instead will reduce the price at which he sells. In other words, Bob and Steve each will realize that price competition between them can reduce profits. The fact that each firm’s price decision affects the other firm means that it is in the individual interest of each firm to refrain from undercutting the other’s price down to the competitive level.11 Although each firm may achieve a short-term increase in profits by undercutting a price that is above the competitive level, it will recognize that such a gain will be fleeting since it then will be in its competitor’s self-interest to cut price as well. A firm’s recognition that a rival may respond to its conduct is common and not conspiratorial. 21. For similar reasons, Bob and Steve each may have an incentive to raise price. For example, if Bob raises price, Steve may have an incentive to match the price increase. That is, Steve would realize that if he did not match the price increase, he would quickly gain share. 11. See Dennis W. Carlton, Robert H. Gertner and Andrew M. Rosenfield (1997), “Communication Among Competitors: Game Theory and Antitrust,” 5 George Mason Law Review 423. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 10 of 94 PageID #:21825 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 -9- As a result, Bob is likely to quickly revoke his increase and return to the lower price, and Steve’s market share gain would be short-lived. If, instead, Steve matched the price increase, he would not gain share, but he would benefit by charging a higher price for an extended period of time. Again, this type of conduct is common and not conspiratorial. 22. Although my gas station example describes an industry with only two firms and a simple pricing structure (e.g., all customers pay the same easily observed price), nonconspiratorial oligopolistic interdependence is possible in a wide variety of industries. As Dr. Harris recognizes, there are many real-world industries in which firms have an economic incentive to consider rivals’ likely responses to their own pricing and other actions. For example, Coke and Pepsi can be expected to monitor each other’s pricing and other actions; similarly, General Motors, Ford, Toyota and Honda can be expected to monitor each other’s pricing and other actions. Evidence of oligopolistic interdependence in such industries is not economic evidence of conspiracy in those industries. 23. Thus, evidence that defendants took account of likely actions and reactions by other defendants is not economic evidence of conspiracy because it is consistent with the behavior of non-conspiratorial firms in an oligopoly industry. C. Dr. Harris’s Claim that Various Actions Taken by Defendants are Economic Evidence of Conspiracy is Flawed and Without Merit. 24. Dr. Harris reviews defendants’ conduct during the alleged conspiracy period and concludes that various actions taken by each defendant were not consistent with their “unilateral self-interests.” In particular, Dr. Harris evaluates defendants’ conduct with respect to:  Mill capacity cuts;  Operating rates, inventory levels and the use of trades;  Taking “downtime” or “slowbacks”;  “Coordinated pricing”;  Use of industry consultants; FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 11 of 94 PageID #:21826 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 10 -  Monitoring of rivals; and  Communication with rivals. For each of these categories, however, Dr. Harris provides no economic analysis that shows that defendants’ actual conduct differed from what its conduct would have been absent the alleged conspiracy. Thus, Dr. Harris provides no common (or any) economic evidence to support his claim that defendants’ conduct is evidence of conspiracy. 1. The Mill Closures in this Case are Not Economic Evidence of Conspiracy. 25. According to Dr. Harris, “[t]he capacity cuts by Defendants in this case were undertaken during periods of strong growth in box demand.”12 Dr. Harris concludes that “[t]he actions undertaken by the Defendants to close capacity do not appear to be consistent with their unilateral self-interests. Rather, it is only rational in the context of coordinated collective action.”13 Dr. Harris’s analysis is flawed and his conclusion is without merit. 26. 12. 13. 14. Dr. Harris’s analysis is flawed for two major reasons.  First, non-defendants, as well as defendants, permanently shut mills (or a machine at a mill) during the alleged conspiracy period, showing that there were non-conspiratorial reasons for a containerboard manufacturer to reduce its capacity.  Second, Dr. Harris appears to be arguing that defendants, which account for a large share of industry capacity, reduced industry capacity during a period of increasing demand by closing mills, thereby creating supply constraints that allowed defendants to fix prices.14 But Dr. Harris analyzes only industry-wide containerboard mill closures and ignores additions to industry containerboard capacity. Despite the mill closures during the alleged conspiracy that Dr. Harris discusses, industry capacity was relatively stable between 2003, the year prior to the start of the alleged conspiracy, and 2008, the last year of “increasing demand.” This shows that additions to industry capacity by Harris Report, at 37. Despite Dr. Harris’s claim, a substantial proportion of the total reduction in capacity from mill closures that he discusses occurred after the start of the Great Recession (e.g., mills closed in 2009 and 2010 had a combined capacity of 2,666 thousand tons; see Harris Report, Exhibit 3). Harris Report, at 39. See, for example, Harris Report, at 3 (“Plaintiffs allege that beginning in 2004 the Defendants, during a period of stable or increasing demand, imposed coordinated supply constraints, including the shutting of capacity and coordinated price increases in order to fix, raise, maintain and stabilize the prices of Containerboard Products above competitive levels they otherwise would have been during the Class period.”). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 12 of 94 PageID #:21827 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 11 - defendants and non-defendants offset mill closures during this period. Contrary to Dr. Harris’s apparent claim, industry capacity did not fall during this period of increasing demand. a. Dr. Harris’s Analysis is Flawed Because he Ignores that Firms Had NonConspiratorial Reasons to Close Mills. 27. Even in a period of growing demand, firms may have non-conspiratorial reasons to close a plant – for example, a firm may close a relatively high-cost plant and replace it with a lower-cost plant. Indeed, during the alleged conspiracy period, PCA closed only one plant – in Tomahawk, Wisconsin – but restarted a second plant to offset the reduction in capacity and reduce transportation costs: “PCA restarted PM#1 at Filer City, MI in Feb. 2005. Once fully operational, the company will close PM#3 at Tomahawk, WI. The objective of the shift is to reduce transportation costs. Both machines have same capacity so there will be no change in company's semichemical medium capacity.”15 28. Dr. Harris’s Exhibit 3 shows that non-defendants, as well as defendants, permanently closed mills during the alleged conspiracy period. For example, non-defendants accounted for about 23 percent of the capacity reductions reported in Dr. Harris’s Exhibit 3 in the years 2004 – 2008. These firms must have had non-conspiratorial reasons for cutting capacity because they are not participants in the alleged conspiracy. At his deposition, Dr. Harris testified that the closing of a mill by a non-defendant would have only a negligible effect on market price,16 and so concluded that: “If we're looking at an individual firm and assuming they're not part of the conspiracy, then we wouldn't expect them to reduce capacity in some 15. 16. AFPA00001929. See also Harris Report, Exhibit 3 (showing Tomahawk plant as only mill closure by PCA since 2000). See Deposition of Michael J. Harris, July 22, 2014 (“Harris dep.”), at 134: 4 – 9 (“on the margin it could perhaps affect the market price, but it's going to be negligible. By definition, they’re a small part of the market, and what you want are the larger firms cutting capacities where it makes a meaningful reduction in supply.”). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 13 of 94 PageID #:21828 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 12 - hopes of impacting market price. They're likely doing that for a business reason or business strategy.”17 29. However, Dr. Harris applies the opposite reasoning to mill closures by a defendant. For example, Dr. Harris claims that the closure of a mill by a single defendant would have a “minimal” effect on market price,18 and so closures by defendants are “only rational in the context of coordinated collective action.”19 That is, Dr. Harris claims that: (1) a mill closure by a non-defendant would have a “negligible” effect on market price, and so concludes that non-defendants likely are closing the mill “for a business reason or business strategy”; yet (2) a mill closure by a defendant would have a “minimal” effect on market price, and so concludes that a mill closure by a defendant is “only rational in the context of coordinated collective action.” That is, Dr. Harris draws opposite conclusions – one for defendants and one for non-defendants – from the same conduct (i.e., a mill closure). There is no valid economic basis for Dr. Harris drawing different conclusions from the same conduct.20 30. Dr. Harris offers no other basis to distinguish mill closures by non-defendants from mill closures by defendants. Indeed, Dr. Harris testified that a defendant also could have a non-conspiratorial “business reason or strategy” to cut capacity: 17. 18. 19. 20. Harris dep., at 137: 2 – 7. See Harris Report, at 37. Harris Report, at 39. At his deposition, Dr. Harris testified: “And in fact you look at the size composition of the defendants relative to all the other firms – and there are a number of documents that show this – it’s rather markedly – the defendants are much, much larger than all of the remaining nondefendants’ companies” (Harris dep., at 137: 14 – 19). However, Dr. Harris also testified: “Q. Is every defendant larger than every nondefendant? A. I don’t recall what the presentation shows, so I can’t say for sure. Q. You don’t know? A. I don’t know” (Harris dep., at 137: 20 – 25). In fact, I understand that one non-defendant (Visy) is similar in capacity to Norampac. In any event, Dr. Harris concluded that no single firm – defendant or non-defendant – could increase market price materially by closing a mill, yet drew different conclusions about defendants that closed mills as opposed to nondefendants that closed mills. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 14 of 94 PageID #:21829 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 13 - Q. A. You said that a nondefendant might have closed a mill because of a business reason or strategy. Could that also be true for a defendant? Yes, there could be a reason for a mill closure.21 Dr. Harris also testified: Q. A. 31. If mills were closed to increase efficiency, reduce costs over the long run and increase profits, would your structure, conduct, performance analysis characterize those closures as consistent with cartelization? Well, the way you've defined it, the mill was closed for pure business reasons, and if there's a justified business reason, it wouldn't be part of the cartel; it would be a normal business decision.22 Despite his admission that mill closures are not necessarily evidence of conspiracy, Dr. Harris did no analysis to determine which mill closures would have occurred even in the absence of the alleged conspiracy. Q. A. Okay. In a world absent an alleged conspiracy, which of the mill closures appearing on Exhibit 3 would have occurred? I don't know. . . . . So have you made any analysis of, in the absence of an alleged conspiracy, which of the mill closures appearing on Exhibit 3 would have occurred? No, I have not.23 32. Dr. Harris also ignores that while six non-defendants – Groveton Paperboard, A. Q. Longview Fiber, Menasha, Bay State Paper, Banner Fibreboard and Catalyst – closed mills in the years 2004 to 2008, two defendants – GP and Temple-Inland, did not (and PCA’s mill closure was completely offset by an addition to its capacity).24 That is, during the period that Dr. Harris claims that mill closures are evidence of conspiracy (i.e., a period during which demand is increasing), many non-defendants closed capacity and two defendants did not. 21. 22. 23. 24. Harris dep., at 138: 6 – 12. Harris dep., at 295: 8 – 17. Harris dep., at 290: 12 – 291: 1. See Harris Report, Exhibit 3. Additionally, Norampac did not close any mills in the United States, the geographic market for containerboard defined by Dr. Harris (see Harris dep., at 268: 24 – 269: 3). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 15 of 94 PageID #:21830 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 14 - 33. Because there are non-conspiratorial reasons for any firm to close mills during 2004 – 2008, and not all defendants closed mills, mill closures by defendants during this period are not economic evidence of conspiracy.25 b. Dr. Harris’s Analysis is Flawed Because He Ignores Additions to Industry Capacity. 34. Dr. Harris claims that “during the alleged class period, capacity and inventories in this industry were reduced and operating rates increased. This was done during periods of constant or increasing demand. The Defendants during this time sought to reduce supply.”26 Dr. Harris then turns to a discussion of mill closures. That is, Dr. Harris appears to be claiming that defendants conspired to reduce industry containerboard capacity in order to reduce industry containerboard supply because, all else equal, a reduction in the supply of a product results in higher prices for that product. Dr. Harris’s apparent claim that industry capacity fell while demand was increasing is wrong. Furthermore, even if industry capacity had fallen during this period (which it did not), a reduction in industry capacity can be offset by an increase in the industry operating rate, so that a reduction in industry capacity does not necessarily result in a reduction in industry supply. As I discuss later in this report, industry output in every year between 2004 and 2008 was higher than in 2003, so there is no basis for plaintiffs’ claim that defendants “reduced supply” during this period. Industry output did not fall below the 2003 level until 2009, in the midst of the Great Recession. 35. To evaluate a claim that firms conspired to reduce industry capacity, changes in net capacity (i.e., additions to capacity minus reductions in capacity) – not just reductions in capacity – are relevant, because reductions in industry capacity can be offset, or more than 25. 26. Industry capacity fell between 2008 and 2010, but industry demand also fell substantially during that period. See Harris dep., at 111: 7 – 11 (“. . . I think I note in a number of areas in the declaration that, clearly, with the onset of the recession, 2008 through 2010, there was a rather dramatic decline in demand for corrugated products.”). Harris Report, at 35. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 16 of 94 PageID #:21831 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 15 - offset, by additions to industry capacity. Dr. Harris, however, only discusses industry-wide mill closures, and ignores additions to industry capacity. In particular, in his Exhibit 3, Dr. Harris presents a list of all containerboard mill closures – both by defendants and non-defendants – since 2000.27 Dr. Harris reports that “[s]ince 2004, 31 mills have been closed with a combined capacity of 5,967 thousand tons.”28 36. The 31 mill closures that Dr. Harris discussed took place during the years 2004 – 2010, but this period was not one of “constant or increasing demand.” In particular, demand fell substantially after the start of the Great Recession in about 2008. Thus, I evaluate Dr. Harris’s claim during the period 2004 – 2008, not 2004 – 2010. According to Dr. Harris’s Exhibit 3, industry mill closures in the years 2004 – 2008 were 3,445 thousand short tons. But according to Dr. Harris’s Table 1, industry capacity was relatively stable during this period, and industry capacity in 2008 was 0.6 percent higher than in 2003, the last full year before the start of the alleged conspiracy (i.e., industry capacity increased from 35,906 thousand metric tons in 2003 to 36,139 thousand metric tons in 2008, an increase of 233 thousand metric tons, or about 256 thousand short tons).29 Thus, Dr. Harris’s apparent claim that industry capacity fell during a period of increasing demand is wrong.30 27. 28. 29. It appears that the units in Dr. Harris’s Exhibit 3 are in thousands of short tons instead of metric tons: one short ton is 2,000 pounds; a metric ton is 1,000 kilograms, or 2,204 pounds. That is, a metric ton is equal to about 1.1 short tons. In contrast, the data in Dr. Harris’s Table 1 appear to be in thousands of metric tons. Harris Report, at 35. There is an error in Dr. Harris’s calculation. Based on Exhibit 3, the 31 mills closed since 2004 had a combined capacity of 6,111 thousand tons. Dr. Harris appears to have inadvertently excluded the capacity of the Catalyst mill in Elk Falls Mill, BC from his calculation. See Harris Report, Exhibit 3, pg. 2. Some of the closures listed in Dr. Harris’s Exhibit 3 are of a single machine, not an entire mill (e.g., I understand that IP closing “Plymouth, NC #1” in 2006 refers to a single machine, not an entire mill). In comparison, U.S. capacity of other paper products fell during this period. For example, according to information compiled by the American Forest & Paper Association (“AFPA”), U.S. containerboard capacity increased from 36,466 thousand tons in 2003 to 37,007 thousand tons in 2008, an increase of 1.5 percent. In contrast, U.S. capacity for other paper products (i.e., newsprint, paper for printing and writing, paper for packaging and industrial converting, and tissue) fell from 47,106 thousand tons in 2003 to 43,329 thousand tons in 2008, a decline of 8.0 percent. If newsprint is excluded, U.S. capacity for the other paper products fell by 4.2 percent during the same period. See AFPA00001314, FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 17 of 94 PageID #:21832 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 16 - 37. Industry capacity did not fall between 2003 and 2008, in part, because defendants added capacity during those years. Examples of mill capacity added by defendants during this period include:  In the second quarter of 2004, IP added 235 thousand tons of liner capacity in Savannah, Georgia.31  In the first quarter of 2005, PCA added 65 thousand tons of medium capacity in Filer City, Michigan.32  In the second quarter of 2007, SSCC added 170 thousand tons of capacity (that can produce both liner and medium) in Jacksonville, Florida.33  In the third quarter of 2007, IP added 500 thousand tons of liner capacity in Pensacola, Florida.34 38. Dr. Harris did no analysis to determine what the level of industry capacity, or industry production, would have been in the absence of the alleged conspiracy Q. A. Do you have an opinion what defendants' overall capacity of containerboard would have been between 2004 and 2010 in the absence of a cartel? No, I do not have that. Do you have an opinion what defendants' overall production of containerboard would have been between 2004 and 2010 in the absence of a cartel? No, I've done no such empirical work. Do you have an opinion what defendants' overall production of corrugated products would have been between 2004 and 2010 in the absence of a cartel? No, I've done no such empirical work.35 39. Because industry participants added capacity, as well as closed capacity during A. Q. A. Q. the alleged conspiracy period, industry capacity did not, as Dr. Harris appears to claim, decline during a period of increasing demand. Mill closures during a period of increasing demand by (...continued) at 16 and 17. The AFPA containerboard capacity figures do not match the data in Dr. Harris’s Table 1 because the information in Table 1 includes Canadian mills (and is expressed in metric tons instead of short tons). 30. As I have discussed, industry capacity fell between 2008 and 2010, but industry demand also fell substantially during that period. 31. See AFPA00002792, at 23. 32. See AFPA00002792, at 25. As I have discussed, the addition of capacity at Filer City offset the shutdown of a mill in Tomahawk, Wisconsin. 33. See AFPA00002856, at 24. 34. See AFPA00002824, at 23. 35. Harris dep., at 309: 5 – 21. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 18 of 94 PageID #:21833 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 17 - defendants and non-defendants that are offset by additions to industry capacity during the same period are not economic evidence of conspiracy.36 2. High Operating Rates, Low Inventory Levels and Trading with Rivals are Not Economic Evidence of Conspiracy. 40. Dr. Harris discusses the defendants’ operating rates, inventory levels and use of “trades” and concludes: The target and achievement of high operating rates and low inventory levels to the detriment of customer service and competitive advantage is not the behavior of a firm pursuing its unilateral self interest but rather only rational for a firm coordinating its actions with other firms. Similarly, the use of trades provided a vehicle to communicate between the Defendants and intertwines the business operations of the Defendants. Finally, the cooperative behavior observed and the reliance on trades as an outright substitution for mill production is not the behavior of competitive firms pursuing their independent self-interest.37 Dr. Harris’s analysis of defendants’ operating rates, inventory levels and use of trades is flawed and his conclusion that defendants’ conduct is “not the behavior of competitive firms” is without merit. 36. 37. In his report, Dr. Harris suggests that other regions can be used as a benchmark to evaluate changes in North American containerboard capacity. In particular, Dr. Harris states: “Notably, over the period 2002 to 2010 every region, with the exception of North America, has shown substantial increases in aggregate containerboard capacity” (Harris Report, at 12). However, Dr. Harris provides no basis to conclude that differences in containerboard capacity growth rates between North America and other regions in the world can only be explained by the alleged conspiracy. See Harris dep., at 308:13 – 309:9 (“Q. Your Table 1 also prints RISI data showing that capacity increased in Europe, Asia, Latin America, Africa and the Middle East between 2002 and 2010; correct? A. Correct. Q. What analysis have you done to determine why containerboard capacity increased in each of those regions? A. I have done no such analysis. Q. What analysis have you made of any macroeconomic factors in each of the regions to see whether they applied similarly in the containerboard industry in those regions as in the United States? A. I have done no such analysis or comparison between North America and non-North American regions. Q. Do you have an opinion what defendants’ overall capacity of containerboard would have been between 2004 and 2010 in the absence of a cartel? A. No, I do not have that.”). Moreover, the growth in foreign capacity (shown in Dr. Harris’s Table 1) likely reduced the total demand for U.S. containerboard by reducing foreign demand. According to Dr. Harris, “[t]he volume of [U.S.] containerboard exports is, relatively speaking, much greater than imports . . . In December 2010, U.S. mills exported approximately 369 thousand short tons of containerboard representing about 13 percent of all containerboard produced” (Harris Report, at 17). Harris Report, at 43. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 19 of 94 PageID #:21834 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 18 - 41. Dr. Harris’s analysis is flawed because he ignores that each of the actions he analyzes can reduce a firm’s costs. Every firm has a non-conspiratorial incentive to reduce its overall costs, and so will choose a combination of operating rate, inventory level and amount of trading that it expects will maximize its profits, all else equal. Because Dr. Harris ignores the effect of defendants’ actions on their overall costs, he has no basis to claim that such actions are “not the behavior of competitive firms pursuing their independent self-interest.” 42. High operating rates can preserve a firm’s ability to produce output but invest less in capacity, thus allowing a firm to reduce its capital costs (e.g., by closing a high-cost plant). That is, high operating rates can allow a firm to produce a given amount of output with less capacity (or more output with the same capacity).38 Whether increasing operating rates minimizes costs depends on the relative costs of operating at high rates vs. the costs of investing in additional capacity. Dr. Harris presents no analysis of this issue.39 43. Similarly, holding inventory generates “carrying costs” for that inventory, and, all else equal, firms generally have an incentive to reduce inventory carrying costs. At his deposition, Dr. Harris testified that “all firms” are interested in “prudent inventory management”: “I don't disagree that prudent inventory management isn't important. All firms do that. If you look in the business literature, there's fairly lengthy discussions of – of optimizing costs and lowering inventories.”40 Dr. Harris presents no analysis of the effect of changes in inventory 38. 39. 40. As I have discussed, total industry capacity was relatively constant from 2003 through 2008. Industry operating rates in 2004 – 2008 were higher than the 2003 level, and industry output in 2004 – 2008 was higher than in 2003. Industry output did not fall below the 2003 level until 2009. See Harris Report, Tables 1 and 2 (industry output equals industry capacity – from Table 1 – times industry operating rate – from Table 2). If a firm reduces its capacity and increases its operating rate, its total output may increase or decrease. If a firm does not reduce its capacity (e.g., GP) and increases its operating rate, its total output will necessarily increase. Dr. Harris fails to explain how an individual firm that did not reduce its capacity and increased its operating rate would be a participant in the alleged conspiracy. Harris dep., at 195: 3 – 7. Dr. Harris also testified that there is a “tradeoff” between lowering inventory costs and providing customer service: “But you do that both from a cost basis and the tradeoff of customer service, and what’s clear from the documents is the defendants clearly don’t care about customer service, and it doesn’t appear as though FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 20 of 94 PageID #:21835 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 19 - levels on defendants’ costs.41 Instead, Dr. Harris claims only that he has “seen no reference to the carrying costs [for inventory] in any of the documents” he reviewed.42 But whether or not any defendant document that Dr. Harris reviewed contains a discussion of inventory carrying costs is irrelevant – Dr. Harris’s failure to consider the benefits of inventory reductions is an economic error that invalidates his analysis. 44. All else equal, a firm has an economic interest in reducing its overall costs, and cost reductions give a firm a competitive advantage relative to firms that do not reduce their costs. Thus, there is no basis for Dr. Harris’s claim that the “achievement of high operating rates and low inventory levels” produces a “detriment of . . . competitive advantage.” Even if high operating rates and/or low inventory levels reduced the level of customer service that a firm could provide (as Dr. Harris claims, but does not analyze), some customers could prefer a combination of a lower level of customer service/lower prices to a combination of a higher level of customer service/higher prices. Dr. Harris presents no analysis of the effect of high operating rates and/or low inventory levels on customer service, and thus no basis to evaluate any tradeoff between inventory levels and customer service. 45. Dr. Harris also discounts the cost savings that firms can gain by engaging in trades with rivals. For example, a firm with a mill in geographic area A and box plants in geographic area B can save freight costs if it finds a trade partner with a mill in geographic area (...continued) they’re lowering inventories just from a cost perspective” (Harris dep., at 195: 8 – 13). Dr. Harris provides no economic analysis that shows that defendants “don’t care about customer service.” Indeed, Dr. Harris also testified: “I didn’t say they didn’t care about customer service. Clearly, all firms have to care about customer service.” (Harris dep., at 195: 24 – 196: 1). Dr. Harris also testified that with respect to his statement that defendants “don’t care about customer service”: “Well, I didn’t mean it as a general proposition; I meant it with respect to inventories and some of their inventory decisions.” (Harris dep., at 196: 12 – 14). 41. See, for example, Harris dep., at 193: 20 – 23 (“Q. Do you know what the defendants’ carrying costs for inventory are? A. No, but I’ve seen no reference to the carrying costs in any of the documents.”). 42. Harris dep., at 193: 22 – 23 (emphases added). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 21 of 94 PageID #:21836 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 20 - B and box plants in geographic area A. Each firm has a non-conspiratorial interest in reducing its freight costs. 46. Dr. Harris recognizes that trades can be used to reduce freight costs: “For an integrated firm trading allows it to minimize freight costs from its mill to its converting facilities and to obtain incremental supply when it is incapable of supplying the entire need of its converting facilities.”43 At his deposition, Dr. Harris agreed that there are non-conspiratorial reasons for engaging in trades: Q. A. Q. A. Okay. Can there be legitimate business reasons for firms to engage in trades? Yes. I believe I discussed that in the declaration. And trades could be observed in a competitive market; correct? Correct.44 47. Dr. Harris also ignores that defendants engaged in trades with rivals before and after the alleged conspiracy period. If defendants engaged in trades in periods during which the plaintiffs do not allege that defendants conspired, there clearly are non-conspiratorial reasons for trading, and thus evidence of engaging in trades, by itself, cannot be evidence of conspiracy. Furthermore, defendants engaged in trades with non-defendants during the period of alleged conspiracy (as well as before and after the alleged conspiracy period). Trades with nondefendants also show that there are non-conspiratorial reasons for engaging in trades. 48. Dr. Harris claims that “[t]he trade in containerboard provides an instrument to punish firms that might stray from a collusive agreement and thus supports collusive outcomes.”45 But Dr. Harris presents no analysis that shows that trades were ever used as “an instrument to punish firms.” At his deposition, Dr. Harris testified that he had done no analysis of whether any defendant was harmed by the use of trades: Q. A. 43. 44. 45. Okay. Do you provide any conclusion anywhere in your declaration as to whether any defendant came out net ahead or net behind at any time as a consequence of trades? Not on that specific point, no. Harris Report, at 15 – 16. Harris dep., at 393: 25 – 394: 6. Harris Report, at 31. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 22 of 94 PageID #:21837 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 21 - Q. A. Q. A. Okay. Did you do any analysis on that point? I did not. Okay. Did you do any analysis as to whether any defendants received a net financial benefit from trades? I did not. And did you do any analysis as to whether any defendant suffered a net financial detriment to trades? I did not.46 49. Dr. Harris also claims that “trades and swaps also provide a source of pricing A. Q. information among and across the Defendants which facilitates collusive behavior.”47 That is, Dr. Harris appears to be suggesting that trades are a source of pricing information that would not otherwise be available, and so “facilitate” a conspiracy. But elsewhere in his report Dr. Harris states that defendants “utilize the [Pulp & Paper Week] index when trading with each other and their downstream corrugating facilities.”48 Because the Pulp & Paper Week index is publicly available, trades and swaps based on that index do not provide any additional pricing information to trading partners. 50. Dr. Harris’s analysis also is flawed because he conducts no analysis of what defendants’ operating rates, inventory levels or trade levels would have been in the absence of the alleged conspiracy. That is, Dr. Harris has no basis for claiming that operating rates, inventory levels or trade levels would have been different absent the alleged conspiracy. For example, with respect to operating rates, Dr. Harris testified: 46. 47. 48. Harris dep., at 395: 4 – 19. Dr. Harris also provides no basis to conclude that trades were used as “side payments.” See Harris dep., at 396: 8 – 23 (“Q. Okay. Are you offering any opinion as to what the price terms were in the product trades between defendants? [Objection omitted] A. No, other than I highlight the fact that trades at market prices can be an indicator of collusion if used as side payments, but I think we've established that I have no evidence that those were actually side payments. Q. Okay. And, actually, you don’t offer any opinion that the trades here were in fact used as side payments; correct? A. Correct.”). GP’s trades “netted out” (i.e., GP provided and received the same number of containerboard tons trades from its trade partners). See Deposition of Mollie Hilliard, December 12, 2013 (“Hilliard dep.”), 44: 23. Hilliard was formerly “trades manager for containerboard” for GP (see Hilliard dep., 17: 8 – 9). Harris Report, at 31. Harris Report, at 19. See also Hilliard dep., at 67: 3 – 5 (“Q. Okay. And trades were typically priced off of indices reported in Pulp & Paper Week? A. Yes.”). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 23 of 94 PageID #:21838 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 22 - Q. A. And you have not made any analysis of a “but for” operating rate across a specific defendant's mills; have you? That's correct.49 With respect to inventory levels, Dr. Harris testified: Q. A. Okay. Did you do any study of equilibrium or “but for” inventory levels? I did not.50 With respect to trade levels, Dr. Harris testified: Q. A. Q. A. Q. A. Okay. Now, did you do any analysis of the use of trades before the class period compared to during the class period? I did not. Did you do any analysis of the use of trades after the class period compared to the class period? I did not. Do you do any analysis of trades between defendants and nondefendants? I did not. Did you do any analysis of trades between nondefendants? I did not.51 51. Because Dr. Harris ignores that high operating rates, low inventory levels and A. Q. trades can reduce a firm’s overall costs, he has no basis to conclude that taking such actions is “not the behavior of competitive firms pursuing their independent self-interest” and that those actions are economic evidence of conspiracy. 3. Evidence of “Downtime” or “Slowback” is Not Economic Evidence of Conspiracy. 52. Dr. Harris states: In addition to permanent shutdowns of capacity, it is also possible to slow or temporarily stop the productive capabilities of a mill. There are a number of terms for this, including downtime, market downtime, economic downtime, slowback, etc. but the effect is the same—the removal of productive capacity and thus the ability to supply the market, causing prices to rise or stabilize.52 49. 50. 51. 52. Harris dep., at 349: 1 – 4. Harris dep., at 193: 5 – 7. Harris dep., at 396: 24 – 397: 12. Harris Report, at 43. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 24 of 94 PageID #:21839 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 23 - Dr. Harris discusses examples of downtime and slowback and concludes that “Defendants knowingly employed downtime as a means to either increase or support prices and did so in a manner inconsistent with their unilateral self-interests.”53 53. Taking downtime can reduce a firm’s costs, all else equal. For example, Dr. Harris states: “Downtime might be a rational response to slowing market conditions. That is, if a firm anticipates a significant and prolonged downturn in orders, it might make economic sense to take downtime rather than incur the ongoing costs of an underutilized mill.”54 Nonetheless, Dr. Harris concludes that “an examination of GP’s downtime shows that it is not fully justified by economic conditions.”55 In his report, Dr. Harris does not explain what he means by “not fully justified by economic conditions.”56 54. Dr. Harris also provides no analysis that shows that defendants’ use of downtime and slowbacks was different during the alleged conspiracy period than in non-conspiracy periods. Similarly, Dr. Harris fails to provide any analysis that shows that defendants’ use of downtime and slowbacks differed from that of non-defendants during the alleged conspiracy period. 55. Dr. Harris’s analysis also is flawed because he conducts no analysis of what the defendants’ use of downtime and slowbacks would have been in the absence of the alleged conspiracy. That is, Dr. Harris has no basis for claiming that defendants’ use of downtime and slowbacks would have been different absent the alleged conspiracy. For example, Dr. Harris has done no analysis of but-for levels of downtime in the years 2008 to 2010: 53. 54. 55. 56. Harris Report, at 47. Harris Report, at 43. Harris Report, at 44. Dr. Harris’s claim ignores that different mills have different costs, and that the cost of a machine can vary with its rate of use. See, for example, Deposition of Travis Ballard, August 14, 2013 (“Ballard dep.”), at 21: 18 – 22 (“Sweet spot is typically the point in the cost curves where your machine is making product at the lowest cost. So that, you know, if you’re trying to get to the lowest cost place, that’s what we typically call the sweet spot.”). Also, as a mill increases its output, key inputs (e.g., wood or old corrugated containers) must typically be sourced from greater distances, which tends to increase the cost of producing containerboard. See Ballard dep., at 20: 19 – 24. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 25 of 94 PageID #:21840 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 24 - Q. Q. A. Q. A. Absent the alleged conspiracy, is it your testimony the defendants would have taken no downtime during 2008 to 2010? No, I think during the recession there was a rather sharp decrease in demand, and it’s likely that they would have. How much? I have no idea. You have not analyzed that question? I have not.57 56. Because he ignores the effect of downtime and slowbacks on a firm’s costs, Dr. A. Harris has no basis to conclude that taking downtime or slowbacks is economic evidence of conspiracy. 4. The “Coordinated Pricing” Cited by Dr. Harris is Not Economic Evidence of Conspiracy. 57. Dr. Harris claims that the use of price increase announcements in the containerboard industry is a form of “coordinated pricing” and claims that the “[t]he price increase announcements themselves act to facilitate coordinated behavior. The timing, amounts, and economic conditions in which the price increases were implemented are consistent with collusive behavior.”58 I note, however, that Dr. Harris does not claim in his report that defendants’ pricing practices during the alleged conspiracy period were inconsistent with non-conspiratorial pricing. As I showed in my gasoline station example, non-conspiratorial price matching and interdependent pricing often occurs in oligopolistic industries and thus, by itself, such pricing patterns are not evidence of conspiracy. 58. Dr. Harris states that “[f]rom 2004 to 2010 the Defendants were successful in full implementation of price increases on nine occasions.”59 However, nowhere in this section of his report does Dr. Harris discuss the defendants’ six other price increase announcements. That is, despite Dr. Harris’s claim that price increase announcements “facilitate coordinated behavior,” 57. 58. 59. Harris dep., at 312: 1 – 11. Harris Report, at 47, 49. Harris Report, at 47. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 26 of 94 PageID #:21841 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 25 - Dr. Harris does not explain why 40 percent of the 15 price increase announcements during the alleged class period were “unsuccessful.” 59. Dr. Harris uses a Pulp & Paper Week (“PPW”) index – specifically, the index for “42 lb. unbleached kraft linerboard for delivery east of the Rocky Mountains” -- to determine whether a price increase announcement was “successful” – a price increase was “successful” only if the PPW index increased in the month “following” the price increase announcement.60 During the alleged conspiracy period, 40 percent of announced price increases did not result in an increase in the PPW index – indeed, in one case, the PPW index fell. As a result, there was almost no relationship between the size of an announced price increase and the change in linerboard price as measured by the PPW index. 60. A “correlation coefficient” (or “correlation”) is a standard measure of the extent to which two series move together. The correlation coefficient ranges in value from -1.0 (which indicates that two series always move in opposite directions) to 1.0 (which indicates that two series always move in the same direction); a correlation coefficient of 0.0 indicates no correlation between two series (that is, changes in one series provide no information about changes in the other series). Given that 40 percent of announced price increases failed, it is not surprising that the correlation coefficient between the size of the announced price increase and the net change in the PPW index during the alleged conspiracy period is 0.05, and not statistically significantly different from zero.61 Thus, despite Dr. Harris’s claim that “price increase announcements themselves act to facilitate coordinated behavior,” announced price 60. 61. See Dwyer Report, ¶ 23 (for definition of index) and Harris Report, Exhibit 4. For the purpose of my analysis, I will adopt Dr. Harris’s terminology that a price increase was “successful” if the PPW index increased and, consistent with this terminology, I will label a price increase as “unsuccessful” if the PPW index did not increase. I calculate the correlation between Target Increase and PPW Increase. For Target Increase, I use the column headed “Target Increase ($/ton)” reported by Dr. Harris for all price announcements except 1, 4, 9 and 10. For those four price increase announcements, Dr. Harris reports a range; I use the midpoint of that range in my analysis. For PPW Increase, I use the column headed “Net PPW Increase ($/ton).” FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 27 of 94 PageID #:21842 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 26 - increases alone provide no statistically significant information about the likely change in the PPW index. 61. Dr. Harris also claims that “[t]he quick succession of price increases [during the alleged class period] was not justified on a cost or demand basis.”62 Dr. Harris presents no economic evidence in support of his claim. For example, Dr. Harris does not discuss changes in costs during the alleged conspiracy period, even though Dr. Dwyer’s models of defendant pricing, which Dr. Harris reviews and endorses, includes variables to control for costs.63 In fact, costs in the containerboard industry increased substantially during the alleged conspiracy period. For example, Dr. Dwyer uses as one measure of cost the price of U.S. pulp (“pulp”),64 which reflects the opportunity cost of a key input in the production of containerboard.65 During the alleged class period, the price of pulp increased 59 percent.66 62. Furthermore, pulp prices increased in the period between the “Prior Month” and “Following Month” reported in Dr. Harris’s Exhibit 4 for each of the nine “successful” price increases. Pulp prices also sometimes increased substantially between “successful” price increases. For example, in the month “following” Price Increase Announcement 6 (May 2006), pulp prices were at $665 per ton. The next two price increases were “unsuccessful” despite a substantial increase in pulp prices. In September 2007 – the month “following” the next “successful” price increase (Price Increase Announcement 9) – pulp prices had increased to $810 per ton. That is, despite Dr. Harris’s claim that defendants’ price increases were “not justified” on a cost (or demand) basis, the price of pulp (which reflects the opportunity cost of a 62. 63. 64. 65. 66. Harris Report, at 47. See Harris dep., 312: 20 – 23 (“Q. You reviewed [Dr. Dwyer’s] work? A. Yes. Q. You endorsed his work in full? A. Yes.”). In his damages analysis, Dr. Dwyer measures U.S. pulp prices with a series called “SBSK [Southern Bleached Softwood Kraft] US sources delivered to US East.” See, for example, Haizheng Li and Jifeng Luo (2008), “Industry Consolidation and Price in the US Linerboard Industry,” Journal of Forest Economics 14, 101 (“For linerboard producers, the major cost is pulpwood, which represents 40-50% of the total cost.”). As I discuss later in this report, I find that pulp prices are highly correlated with the PPW containerboard index used by Dr. Harris to determine the success of an announced price increase. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 28 of 94 PageID #:21843 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 27 - key containerboard input) increased substantially during the alleged conspiracy period, as well as before and after each “successful” price increase. See Table 1. Table 1 Changes in U.S. Pulp Prices Between "Prior" and "Following" Months in Dr. Harris's Analysis Price Increase Before  Month/Year After  Month/Year Successful/ Unsuccessful Pulp Before ($/Metric Ton) Pulp After ($/Metric Ton) Change 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Jan‐04 May‐04 Feb‐05 Sep‐05 Dec‐05 Feb‐06 Dec‐06 Mar‐07 Jul‐07 Feb‐08 Jun‐08 Sep‐08 Dec‐09 Mar‐10 Jul‐10 Apr‐04 Jul‐04 May‐05 Nov‐05 Feb‐06 May‐06 Feb‐07 Jun‐07 Sep‐07 Apr‐08 Aug‐08 Nov‐08 Feb‐10 May‐10 Sep‐10 S S U S S S U U S U S U S S U 560 620 650 600 620 630 730 760 790 840 840 830 780 860 975 620 650 630 620 630 665 750 770 810 840 845 745 830 955 945 60 30 ‐20 20 10 35 20 10 20 0 5 ‐85 50 95 ‐30 Source: Dr. Harris's Exhibit 4; Dr. Dwyer's backup materials. 63. Despite Dr. Harris’s claim that “[t]he quick succession of price increases [during the alleged class period] was not justified on a cost or demand basis,” Dr. Harris testified at his deposition that some of the price increases were justified by changes in supply and demand conditions: Q. A. Q. A. 64. Okay. Now, in the “but for” world, is it your opinion that there would have been no price increases during this period? No, and I don't think Dr. Dwyer's model indicates that. Well, I'm not talking about Dr. Dwyer's model; I'm talking about your opinion. You described these price increase analysis, you described price increases. Is it your opinion that in the “but for” world there would have been no price increases over the class period? I mean I – that opinion has to be based on Dr. Dwyer's empirical model which controls for supply and demand factors, and his model doesn't indicate that the conspiratorial overcharge is in every single one of these price increases. There's some portion of these that was a result of supply and demand factors.67 Dr. Harris provides no basis to justify his claim that defendants’ price increases during the alleged conspiracy period were unjustified by changes in economic conditions, 67. Harris dep., at 216:20 – 217:13. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 29 of 94 PageID #:21844 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 28 - including costs. Thus, Dr. Harris has no basis for his claim that defendants’ price increases are economic evidence of a conspiracy. 5. The Use of Industry Consultants is Not Economic Evidence of Conspiracy. 65. Dr. Harris claims that “advice by industry consultants provided a common focal point for the Defendants – a focal point that facilitated coordinated behavior. The consultants thereby provided to Defendants a means to collude.”68 Although Dr. Harris claims that industry consultants provided a “means to collude,” nowhere does Dr. Harris conclude that defendants in fact used industry consultants as a method of conspiring. 66. Dr. Harris ignores that the use of industry consultants is common in a wide variety of industries not subject to conspiracy and therefore cannot by itself indicate the existence of a price-fixing conspiracy. Furthermore, Dr. Harris provides no evidence that defendants would not have used industry consultants absent the alleged conspiracy. For example, Dr. Harris provides no evidence that defendants did not use industry consultants during non-conspiratorial periods, or that non-defendants did not use industry consultants. Thus, there is no basis for Dr. Harris to conclude that the use of industry consultants is economic evidence of conspiracy. 6. Monitoring Rivals is Not Economic Evidence of Conspiracy. 67. Dr. Harris claims that “Defendants expended considerable effort to monitor the actions and operational status of ostensibly competing firms in this industry.”69 Because Dr. Harris refers to “ostensibly” competing firms, he appears to be suggesting that competing firms would not expend effort monitoring rivals.70 68. 69. 70. Harris Report, at 55. Harris Report, at 55. “Ostensibly” is defined as “apparently or purportedly, but perhaps not actually” (https://www.google.com/search?q=definition+ostensibly&gws_rd=ssl). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 30 of 94 PageID #:21845 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 29 - 68. Dr. Harris provides no basis in economics to claim that the observation of rivals’ actions and forming “conjectures” about those actions is inconsistent with competition and therefore is evidence of conspiracy, or that defendants would not have “monitored” each others’ actions absent the alleged conspiracy. In general, rivals in oligopolistic industries can be expected to monitor rivals’ actions (as I have discussed, Coke and Pepsi can be expected to monitor each other’s pricing and other actions; GM, Ford, Toyota and Honda can be expected to monitor each other’s pricing and other actions). Similarly, the gas station owners in my example monitor each other’s prices but do not conspire. Thus, there is no basis for Dr. Harris to conclude that monitoring the actions of rivals is economic evidence of conspiracy. 7. The Examples of “Communication” Cited by Dr. Harris are Not Economic Evidence of Conspiracy. 69. Dr. Harris claims that “[t]here are numerous instances in which the Defendants have communicated competitive information directly with each other and/or shared non-public information.”71 Each of the three examples of “direct communication” cited by Dr. Harris could occur in a non-conspiratorial oligopolistic industry and thus none of the examples is evidence of conspiracy. 70. The first example cited by Dr. Harris involves the move of an individual employee from IP to SSCC, allegedly taking with him IP information. Even if this claim is correct, there is no basis for Dr. Harris to conclude that this individual’s actions could not take place in an oligopolistic industry and thus are not evidence of conspiracy. 71. The second example cited by Dr. Harris appears to relate to discussions between defendant employees regarding “trade negotiations.” As I have discussed, trades between defendants can reduce costs, thereby promoting competition. As I also have discussed, Dr. Harris provides no support for his claim that trades between defendants provided them 71. Harris Report, at 56. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 31 of 94 PageID #:21846 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 30 - information that was not publicly available. There is no basis for Dr. Harris to conclude that discussions regarding trade negotiations could not take place in an oligopolistic industry and thus are evidence of conspiracy. 72. The third example cited by Dr. Harris relates to an e-mail from SSCC to GP regarding pricing. The product at issue (cup stock) is not a containerboard product. Instead, “cup stock” is a raw material used to manufacture Dixie® cups.72 That is, the e-mail cited by Dr. Harris reflects a request from GP’s Dixie® business for a price quote from a potential supplier. 73. Dr. Harris provides no basis to conclude that any of his three examples of “communication” would not have occurred absent the alleged conspiracy. Thus, there is no basis for Dr. Harris to conclude that his examples of “communication” are economic evidence of conspiracy. V. PLAINTIFFS’ EXPERTS’ ANALYSES FAIL TO SUPPORT THEIR CLAIM THAT THE ALLEGED CONSPIRACY “IMPACTED” ALL OR NEARLY ALL CLASS MEMBERS. 74. Plaintiffs rely on the reports of both Dr. Harris and Dr. Dwyer to support their claim that “all or nearly all class members would have been harmed” by the alleged conspiracy.73 Dr. Harris provides no empirical analysis in support of his impact opinion, and so appears to have taken the position that impact can be assessed on theoretical grounds alone.74 Dr. Dwyer, instead, bases his conclusion on an empirical analysis. As I discuss in this section 72. 73. 74. At his deposition, Dr. Harris testified that he did not know what cup stock is, or whether it is a product at issue in this case. See Harris dep., at 402: 22 – 403: 4 (“Q. What is cup stock? A. I don’t know. Q. So you don’t know whether it’s a product that’s at issue in this case? A. Well, I’m somewhat perplexed why an E-mail would be produced for a product not in this case, but sitting here, I don’t know if cup stock specifically is in the case.”). Plaintiffs’ Memorandum of Law, at 54 (“Both of Class Plaintiffs’ experts offer consistent opinions on common impact—that all or nearly all class members would have been harmed by Defendants collusive conduct, as alleged by the plaintiffs.”). “Impact” is a legal, not economics concept. For the purpose of my analysis, I assume that a customer is “impacted” by the alleged conspiracy if the customer paid more than it would have in the absence of the conspiracy. That is, I understand that a customer is “impacted” if it paid more for the relevant products it purchased than it would have paid if the alleged conspiracy had not occurred. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 32 of 94 PageID #:21847 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 31 - of my report, the analyses of both experts are flawed and neither provides any common means of showing that the alleged conspiracy impacted all or nearly all putative class members. A. Dr. Harris’s Theoretical Analysis Fails to Show that the Alleged Conspiracy Impacted All or Nearly All Putative Class Members. 75. Dr. Harris’s impact analysis consists of two steps. First, Dr. Harris analyzes the structure of the industry and concludes that “[a]ny attempt at collusion would have been broadly and generally successful given the economic characteristics of the industry.”75 Second, Dr. Harris draws a further conclusion: “I have also concluded, based on the economic evidence I have described, that the conduct of the Defendants would have impacted all, or nearly all, class members.”76 1. Dr. Harris’s Claim that the Alleged Conspiracy “Would Have Been Broadly and Generally Successful” is Flawed and Without Merit. 76. Dr. Harris provides no common (or any) economic evidence demonstrating that the alleged conspiracy would have been successful in reducing supply and thereby elevating prices relative to the prices that would have prevailed absent – or “but for” – the alleged conspiracy (henceforth, I refer to such prices as “but-for” prices).77 That is, Dr. Harris fails to show that the supply of containerboard was actually reduced below the but-for level, or that the prices of those products were therefore increased above the but-for price, by the alleged conspiracy. 75. 76. 77. Harris Report, at 59. Harris Report, at 59. Dr. Harris also states that “[t]he empirical analysis on common impact performed by Dr. Dwyer is consistent with my economic findings” (Harris Report, at 59). However, Dr. Harris does not rely on Dr. Dwyer’s impact analysis, but reaches his conclusion on impact on theoretical grounds alone. I assume that the term “broadly and generally successful” as used by Dr. Harris means the raising of price by a significant amount above the but-for level. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 33 of 94 PageID #:21848 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 32 - a. Economic Theory Alone Cannot Show that an Alleged Conspiracy Would be Successful, and the Empirical Evidence Does Not Support Plaintiffs’ Claim that the Alleged Conspiracy Succeeded in Reducing Supply. 77. As a general matter, although firms are jointly better off if they all agree to raise price, any individual firm is better off if, having convinced other firms to raise their prices, it undercuts the agreed-upon price, provided it can hide its price-cutting from the other cartel members. Such price cutting is often referred to as “cheating” on the agreement. As the Nobel Prize-winning economist George Stigler explained: “It is a well-established proposition that if any member of the agreement can secretly violate it, he will gain larger profits than by conforming to it. . . . If price is above marginal cost, marginal revenue will be only slightly less than price (and hence above marginal cost) for price cuts by this one seller.”78 Generally, every firm that is a party to a cartel agreement has an individual incentive to undercut the agreed-upon price. This conflict between the incentives of individual firms and the incentives of the group acting together makes it difficult to maintain cartels. 78. According to Dr. Harris, the conspiracy alleged by plaintiffs is one in which “beginning in 2004 the Defendants, during a period of stable or increasing demand, imposed coordinated supply constraints, including the shutting of capacity and coordinated price increases in order to fix, raise, maintain and stabilize the prices of Containerboard Products above competitive levels they otherwise would have been during the Class period.”79 However, for allegedly conspiratorial “supply constraints” to lead to higher prices, those constraints must, in fact, reduce supply compared to the level of supply in the absence of the alleged conspiracy. 79. Dr. Harris fails to show that an alleged conspiracy to shut capacity (or reduce supply through downtime and slowbacks) would in fact reduce supply for two reasons: (1) firms in the industry – including defendants and non-defendants – could add capacity, offsetting capacity cuts by members of the alleged conspiracy; and (2) firms in the industry – including 78. 79. George J. Stigler (1964), “A Theory of Oligopoly,” Journal of Political Economy, 72(1), 46. Harris Report, at 3. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 34 of 94 PageID #:21849 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 33 - defendants and non-defendants – could use their assets more intensively to produce the same or more output with reduced capacity. For both reasons, an alleged agreement to restrict capacity need not result in a reduction in the amount supplied. Thus, whether the conspiracy alleged by plaintiffs “would have been broadly and generally successful” cannot be evaluated on the basis of theory alone, but instead requires an empirical analysis. 80. As I have discussed, the evidence in Dr. Harris’s own report shows that despite the reductions in capacity described by Dr. Harris, capacity additions during the alleged class period largely or completely offset those reductions, so that total capacity in the industry was higher in 2008 (around the start of the Great Recession) than in 2003 (before the beginning of the alleged conspiracy). Furthermore, because of increases in operating rates after 2003, industry production in every year in the period 2004 – 2008 was higher than it had been in 2003. b. Dr. Harris’s “Structure-Conduct-Performance” Analysis Does Not Show that the Alleged Conspiracy “Would Have Been Broadly and Generally Successful.” 81. Dr. Harris uses the “structure-conduct-performance paradigm” to “evaluate the containerboard industry.”80 According to Dr. Harris, the structural characteristics of the industry: (1) “are consistent with and would facilitate successful collusion among the Defendants”; (2) “certain of these structural characteristics – presence of trade associations, frequency of interactions and trades among the firms – provide to Defendants ample opportunity to collude”; and (3) the industry’s “structural characteristics also provide substantial support for the conclusion that any collusion by the Defendants would ultimately be broadly and generally 80. Harris Report, at 4. At his deposition, Dr. Harris suggested that “containerboard products” is the relevant product market for his analysis (see Harris dep., at 248: 14 – 249: 13), and that the United States is the relevant geographic market (Harris dep., at 270: 14 – 18). Defining a relevant market requires determining which products are substitutes (i.e., to determine the product dimension of the market) and the areas from which those products can be obtained (i.e., to determine the geographic dimension of the market). Dr. Harris presents no economic analysis to evaluate the extent of the relevant market, in either the product or geographic dimension. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 35 of 94 PageID #:21850 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 34 - successful.”81 As I explain in this section of my report, a review of the containerboard industry’s structural characteristics, by itself, does not show that the alleged conspiracy would have been successful. 82. Dr. Harris’s structure-conduct-performance analysis is based on his review of eight industry characteristics: (1) market share/concentration; (2) barriers to entry; (3) demand elasticities; (4) product homogeneity; (5) contract provisions; (6) frequency of interactions; (7) market transparency; and (8) trade associations. From his review of those characteristics, Dr. Harris concludes that, if defendants had conspired, that conspiracy “would have been broadly and generally successful.”82 83. There is no theoretical basis for Dr. Harris’s claim that, if defendants had conspired, as alleged by the Plaintiffs, the conspiracy would have been successful. As a matter of economic theory, it is not possible to determine – based on the industry “characteristics” that Dr. Harris has examined – that the conspiracy alleged would have been successful. Instead, determining whether an alleged conspiracy was successful requires an empirical analysis and cannot be based solely on the presence of certain industry characteristics. 84. According to Dr. Harris, plaintiffs claim that each of the 15 attempted price increases during the alleged class period were conspiratorial: Q. A. Q. A. Q. A. 81. 82. 83. Doctor, this [referring to Dr. Harris’s Exhibit 4] lists 15 price increase announcements by the defendants during the class period; correct? Yes. Which of these were part of the conspiracy, as you understand it? All of them. Every single one? Yes.83 Harris Report, at 33. Harris Report, at 6. As I discussed in the previous section of my report, Dr. Harris’s conclusion that the available evidence supports plaintiffs’ conspiracy claim is flawed and without merit. Harris dep., at 86: 11-19. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 36 of 94 PageID #:21851 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 35 - As I have discussed, six of the 15 price increase announcements were “unsuccessful.” At his deposition, Dr. Harris testified that he did not know why six price increases were “unsuccessful,” but that those outcomes could reflect cartel “instability”: Q. A. Okay. So your testimony, as I understand it, is where the price increases succeeded, that was conspiratorial; when they failed, it wasn't? I really don't have a reason for why they failed. I just proffered up an explanation . . . that economists know that cartels are unstable, both for their duration and within periods in which they're colluding. And one of the reasons that you could have price announcements not stick is some instability in the cartel.84 Furthermore, Dr. Harris testified that he did not know whether industry prices returned to the but-for (i.e., non-conspiratorial) level in the periods around “unsuccessful” price increases: Q. A. Q. A. Q. A. Q. A. Q. A. Q. A. 84. 85. 86. Doctor, you've mentioned several times today periods of instability or periods that the cartel or the conspiracy may have failed temporarily. Do you recall that? Yes. Okay. During those periods, is it possible that the prices were reverted to the competitive level or fell to the competitive level? No. If you look generally at the prices, they seem to be cumulative. There are decreases, but it doesn't appear as though the prices went back to pre-class period levels, no. Well, when I say the competitive level, I don't mean the pre-class level; what I mean is the level that would have prevailed in the “but for” world. Did you do any analysis to determine what the “but for” world price would have been? I did not. Dr. Dwyer did and found that the prices were elevated over the entire class period. But during the periods where the cartel or collusion or conspiracy, as you put it, might have been unstable or failing, did either you or Dr. Dwyer do any analysis to see whether the price at those times was elevated above the “but for” world price? I don't know if Dr. Dwyer parsed out specific periods like that, but they are certainly included in his model. Okay. Because they're included in the aggregate numbers of his model, but do you know if there's any specific analysis of whether the prices during those periods were at or above or below the “but for” price? You'd have to ask Dr. Dwyer. I don't know if he conducted that sort of analysis.85 You had no analysis on that. As I said before, I did not develop an empirical model.86 Harris dep., at 92: 16 – 93:4. Dr. Harris also testified that during period of cartel instability, defendants deviate “from the collusive objective.” See Harris dep., at 368: 10 – 14 (“Q. During periods of cartel instability, are defendants acting as a single firm? A. No. By definition, there’s deviation from the collusive objective.”). Dr. Dwyer presented no such analysis in his report. Harris dep., at 242: 18 – 244: 10. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 37 of 94 PageID #:21852 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 36 - 85. Defendants announced price increases that were “successful” in 2004, 2005, 2006, 2007, 2008 and 2010. Defendants also announced price increases that were “unsuccessful” in 2005, 2007, 2008 and 2010. Thus, in four years – 2005, 2007, 2008 and 2010 – defendants announced both “successful” and “unsuccessful” price increases. Dr. Harris presents no evidence that the eight factors that he discusses changed materially during any calendar year of the alleged conspiracy. In light of Dr Harris’s admission that there were six “unsuccessful” price increases, yet no change in underlying economic factors, it follows that any one, or any combination of those factors, cannot predict whether a particular alleged conspiratorial price increase in a particular year would be “successful.” In short, Dr. Harris’s own analysis shows that even if one assumes the existence of the conspiracy alleged by plaintiffs, 40 percent of the attempted price increases were “unsuccessful.” There is thus no basis for Dr. Harris to claim that the existence of the eight industry factors he analyzes imply that if defendants had conspired, as alleged by plaintiffs, they would have been successful. 86. These results – the same industry characteristics are associated with “successful” and “unsuccessful” price increases – confirm that economic theory alone cannot replace empirical evidence in determining whether an alleged conspiracy, if entered into, would be successful.87 87. To the extent that Dr. Harris claims that the alleged conspiracy was successful because some of the price increases were “successful,” such a definition of “success” would imply that even conspiracies that had a non-material effect on prices could be considered successful. For example, suppose that firms in an industry with the same relevant characteristics as the containerboard industry attempted 100 price increases over a 20year period, and 99 of them were “unsuccessful.” Using “one or more attempt to increase prices is successful” as a benchmark, the industry with a one percent success rate would be considered an example of a successful price conspiracy, even if the one “successful” price increase did not have a material effect on average prices over the 20-year period. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 38 of 94 PageID #:21853 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 37 - 2. Dr. Harris’s Claim that a “Broadly and Generally Successful” Conspiracy Would Have Impacted All or Nearly All Putative Class Members is Flawed and Without Merit. 87. Despite his failure to conduct an empirical analysis, Dr. Harris concludes not only that “the collusive conduct of the Defendants would have been broadly and generally successful,” he also concludes, as a matter of economic theory, that the alleged conspiracy would have impacted “all, or nearly all, class members.”88 Dr. Harris provides no basis for this conclusion. 88. Even if the alleged conspiracy had been successful in the sense that it had the effect of increasing the average price paid by all putative class members, it does not follow that all or nearly class members would have been impacted by the alleged conspiracy.89 As I have discussed, Dr. Harris has presented no analysis that shows whether prices were above the butfor level during periods of what he claims is cartel instability. Thus, Dr. Harris has no basis to conclude that the alleged conspiracy (even if had been successful in raising prices during portions of the alleged conspiracy period) had an effect on putative class members that purchased only during periods of “cartel instability.” As I show later in this report, there were thousands of putative class members that purchased containerboard or corrugated products only during such periods. 89. Most of the putative class members in this case purchased corrugated products (e.g., cardboard boxes), not containerboard. For example, in Dr. Dwyer’s analysis of impact (which I will discuss in detail later in this report), he “assesses” 78,224 putative class members, and 77,134 of those putative class members bought corrugated products, not containerboard.90 In order for a customer that did not purchase containerboard but did purchase corrugated 88. 89. 90. Harris Report, at 6. In contrast, Dr. Dwyer claims to analyze customer-specific prices. I discuss the flaws in Dr. Dwyer’s analysis later in this section of my report. See Dwyer Report, Exhibits 6 and 7. In an alternative analysis, Dr. Dwyer “assesses” about 94,000 putative class members. See Dwyer Report, Exhibit 8. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 39 of 94 PageID #:21854 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 38 - products to be impacted by the alleged conspiracy, the price of corrugated products must increase.91 90. In general, the extent to which a price increase for an input (in this case, containerboard) is reflected in the price of a final product (in this case, a corrugated product) can depend on a variety of factors.92 For example, the extent of “pass through” can depend on the extent to which substitutes are available for the final product. Some purchasers of corrugated products (e.g., produce customers) may have substitutes for that product (e.g., plastic containers), while other purchasers may not. Dr. Harris presents no analysis of the extent to which substitute products are available to different putative class members that purchased corrugated products. 91. The extent of “pass through” also will, in general, depend on economic conditions in a market, including the degree of competition in that market. Corrugated products like boxes typically are shipped relatively short distances, and so the geographic market for corrugated products may be relatively narrow.93 Thus, the extent to which an increase in the price of containerboard will affect the price of corrugated products can differ across the country. Dr. Harris, however, has conducted no analysis of the extent of the geographic market for corrugated products, and has not analyzed the extent of “pass through” of containerboard prices to corrugated product prices by region.94 92. 91. 92. 93. 94. For the foregoing reasons, Dr. Harris’s analysis of impact is flawed. See Harris dep., at 87: 20 – 23 (“as a practical matter what it means is once you've restricted the supply of containerboard, linerboard and medium, you have, by definition, restricted the supply of boxes.”). See also Harris dep., at 90: 20 – 23 (“The effect of the conspiracy was manifested in the price of those [corrugated] products, but the conduct that I evaluate is for rolled stock.”). See Harris dep., at 62: 23 – 63: 13. See Harris dep., at 272: 22 – 24 (“I know generally that customers for a box plant are within 150 miles of the plant, due primarily to significant freight costs.”). See Harris dep., at 279: 24 – 280: 4 (“Q. What work did you do to determine whether the structure, conduct, performance of corrugated producers was similar or different by region for purposes of your analysis? A. I didn’t break it up by region. Viewed it as one market.”). Dr. Harris agreed at his deposition that “the Fibre Box Association tracks industry data by 47 distinct market regions” (Harris dep., at 273: 9 – 12). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 40 of 94 PageID #:21855 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 39 - B. Dr. Dwyer’s Empirical Analysis Fails to Show that the Alleged Conspiracy Impacted All or Nearly All Putative Class Members. 93. Unlike Dr. Harris, Dr. Dwyer presents an empirical impact analysis, from which he concludes that “all or nearly all class members paid higher prices as a result of price increases announced and implemented by the defendants and alleged by the plaintiffs to be facilitated by the collusive restriction in supply.”95 Dr. Dwyer’s analysis, however, fails to show that the alleged conspiracy affected all or nearly all putative class members. 1. Dr. Dwyer’s Claim that a Published Market Price Index Provides Evidence of Class-Wide Impact is Wrong. 94. Dr. Dwyer claims that his review of an index of containerboard prices supports his view that the alleged conspiracy had “class-wide impact.” 96 Dr. Dwyer’s analysis consists of two steps. First, Dr. Dwyer presents a probability analysis which he claims shows that announced price increases during the alleged conspiracy are statistically related to increases in the same PPW index that Dr. Harris uses to determine which price announcements were “successful” (i.e., “42lb. unbleached kraft linerboard for delivery east of the Rocky Mountains”).97 Second, Dr. Dwyer appears to claim that transaction prices move together with the PPW index 95. 96. 97. Dwyer Report, ¶ 11. Dr. Dwyer’s “impact” (and damage) analyses are based on transactions data from each defendant. For the purposes of my analyses in this report, I use the datasets Dr. Dwyer created from each defendant’s transactions data, but I do not necessarily endorse how he created those data from the “raw” transactions data. See Dwyer Report, ¶ 30 (“In summary, the prices for linerboard and medium reflected in the PPW Index increased for 9 of the 15 PIIEs during the class period. The timing, direction, and magnitude of PPW Index movements supports the conclusion of class-wide impact for purchasers of linerboard and medium. Further, given that much evidence ties corrugated prices to the PPW Index, the results of this examination of the PPW index provides indirect evidence that the defendants' price increase announcements had a class-wide impact on class-members purchase prices for these corrugated Containerboard Products as well. These results are sufficiently significant and robust to support my opinion that all or nearly all class members were impacted by the defendants' collusive behavior as alleged by the plaintiffs.”). See Dwyer Report, ¶ 29. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 41 of 94 PageID #:21856 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 40 - and so changes in the PPW index are indicative of changes in transaction prices.98 Dr. Dwyer’s analysis is flawed and provides no basis for a claim that the alleged conspiracy “impacted” all or nearly all putative class members – the first step of Dr. Dwyer’s analysis is irrelevant and provides no information about “impact” on individual putative class members, and the second step is wrong as an empirical matter. 95. The first step in Dr. Dwyer’s analysis consists of his claim that “[t]he probability that the 9 PPW Index price increases that we observe in the defendants' implementation months of Exhibit 3 are a matter of chance is less than one-hundredth of one percent (0.0054%).”99 But Dr. Dwyer’s calculation shows only that in months during which there was a price increase announcement, the PPW index sometimes increased (i.e., 60 percent of the time), and in months during which there was no price increase announcement, the PPW index did not increase. In other words, Dr. Dwyer finds that the PPW index was likely to increase only when sellers announced a price increase (but not every time an increase was announced), and that the PPW index did not increase when sellers did not announce a price increase. This is a completely unsurprising result that provides no information about whether the transaction prices paid by any particular putative class member increased when the PPW index increased. 96. Dr. Dwyer’s apparent claim that movements in individual transaction prices closely track changes in the PPW index is an empirical question that can be examined directly. According to RISI, the publisher of the PPW indices, those indices are not intended to reflect the prices paid by all customers. A RISI online publication explains that the indices it reports are based on prices that:  “are normalized to represent prices paid by small to mid-size buyers (i.e. buyers of <5000 tons/month”; 98. 99. See Dwyer Report, ¶ 23 (“One reason for the prominence of the PPW Index is its pervasive role as a reference price for contracts. This index serves as a reference price for contracts not just for linerboard, but for semichemical medium and generally, for other Containerboard Products as well.”). Dwyer Report, ¶ 29. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 42 of 94 PageID #:21857 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 41 -  exclude transactions that are contractually linked to an index not published by RISI; and  exclude certain spot transactions (i.e., “orders filled over a period of less than a month without the expectation of future business”).100 Thus, for example, prices paid by large buyers are not reflected in the PPW index, and the PPW index is not “reflective of class member transaction prices for Containerboard products”101 for all putative class members. 97. If Dr. Dwyer’s claim were correct, transaction prices should change only when the PPW index changes. I now illustrate that Dr. Dwyer’s claim is contradicted by the empirical evidence. One implication of Dr. Dwyer’s claim is that if the PPW index is unchanged, then the prices paid by individual customers would be unchanged. To see that this implication is contradicted by the empirical evidence, consider Figure 1. Figure 1 shows the PPW index used by Drs. Harris and Dwyer to identify “successful” price increases and the actual “mode” transaction prices calculated by Dr. Dwyer for 42 lb. linerboard paid by each of eight major purchasers of that product from GP during the period April 2010 to August 2012. Note that during this period, the PPW index does not change (and is represented by the horizontal black line in Figure 1).102 Figure 1 shows that transaction prices paid by individual putative class members clearly are not always constant over this period during which the PPW index did not change. Dr. Dwyer’s claim that the PPW index can be used to infer “impact” on all or nearly all individual putative class members that purchase containerboard is therefore wrong.   100. See http://www.risiinfo.com/Marketing/OBM-MAR/RISI_ctb_price_plan_final.pdf. 101. Dwyer Report, ¶ 26. 102. The data underlying Figure 1 are derived from Dr. Dwyer’s backup information. Each customer in Figure 1 represents a customer/plant/product/zip code combination (where product is 42 lb. linerboard). I base Figure 1 on GP’s 10 largest 42 lb. containerboard customer/plant/product/zip code combinations in terms of sales over the entire time period for which data are available. One of those customer/plant/product/zip code combinations did not make purchases in the period April 2010 – August 2012, and so is not included in the figure; another combination reflects GP internal purchases of containerboard, which Dr. Dwyer appears to have inadvertently left in his analysis; I exclude GP internal purchases from Figure 1. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 43 of 94 PageID #:21858 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 42 - Figure 1 98. Furthermore, Dr. Dwyer claims that: given that much evidence ties corrugated prices to the PPW Index, the results of this examination of the PPW index provides indirect evidence that the defendants' price increase announcements had a class-wide impact on class-members purchase prices for these corrugated Containerboard Products as well. These results are sufficiently significant and robust to support my opinion that all or nearly all class members were impacted by the defendants' collusive behavior as alleged by the plaintiffs.103 That is, Dr. Dwyer claims that movements in the PPW index also provide information about changes in transaction prices to individual customers for corrugated products. 99. To evaluate Dr. Dwyer’s claim about corrugated products, I examine prices paid for a corrugated product by selected putative class members during the same April 2010 to August 2012 (i.e., a period during which the PPW index is unchanged). If Dr. Dwyer’s claim were correct, transaction prices for corrugated products to individual customers should not change during this period. Figure 2 shows that the “mode” price calculated by Dr. Dwyer for a 103. Dwyer Report, ¶ 30. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 44 of 94 PageID #:21859 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 43 - specific corrugated product by each of 10 major purchasers of that product from GP is clearly not always constant during a period in which the PPW index is unchanged.104 Dr. Dwyer’s claim that the PPW index can be used to infer “impact” on all or nearly all individual putative class members that purchase corrugated products is therefore wrong.   Figure 2 100. I conclude that Dr. Dwyer’s claim that his analysis of movements in the PPW index over time supports his claim that the alleged conspiracy had “class-wide impact” is wrong 104. The data underlying Figure 2 are derived from Dr. Dwyer’s backup information. The corrugated product in the analysis (NEW 25# S/L TOMATO BODY (V2) 62315) is the largest selling GP corrugated product during the alleged conspiracy period that was purchased by at least 10 putative class members. Each customer in Figure 2 represents a customer/plant/product/zip code combination (where product is (NEW 25# S/L TOMATO BODY (V2) 62315). I base Figure 2 on GP’s 10 largest customer/plant/product/zip code combinations in terms of sales of this product over the entire time period for which data are available. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 45 of 94 PageID #:21860 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 44 - because it assumes – contrary to the evidence – that prices paid by all or nearly all individual putative class member move in tandem with the PPW index. 2. Dr. Dwyer’s Analysis of Transaction Prices Does Not Measure the Impact of the Alleged Conspiracy. 101.  Dr. Dwyer’s impact analysis is based on comparisons of prices paid by individual  customers before and after a series of what he refers to as “Price Increase Implementation  Events” or “PIIEs.”105  In effect, Dr. Dwyer compares the price paid by a customer for a particular  product at a particular location before a “successful” price increase to the prices paid by the  same customer for the same product in the same location after that “successful” price increase  at up to three different points in time.    102.  In particular, Dr. Dwyer compares a “pre­event price” to prices at up to three  different months after the PIIE: “The pre-event price is the monthly net price in the closest month up to (and including) the month prior to the PIIE as listed in Exhibit 5. The comparisons then are to the monthly net prices for that same product within 3 months, 6 months, and 9 months after the PIIE.”106  If the customer’s price is higher at some point in the  period after the PIIE than before the PIIE, Dr. Dwyer classifies that customer’s purchase of that  product at that location as “impacted.”107  For example, suppose a customer paid a pre­event  price of $10.00 for a particular product at a particular location before PIIE 1.  If that same  customer bought the same product at the same location three, six and nine months after PIIE 1,  and paid $9.50, $10.01 and $9.50 for the product, Dr. Dwyer would find “impact” because the  105. Because some “successful” price increases occurred within a few months of each other, Dr. Dwyer combines the nine “successful” price increases during the alleged conspiracy period into five PIIEs. 106. Dwyer Report, ¶ 37. 107. See Dwyer Report, ¶¶ 32 – 38. See also Deposition of Mark J. Dwyer, August 12, 2014 (“Dwyer dep.”), at 255: 11 – 17 (“Q. But your analysis for common impact is essentially that if any customer has a one-cent higher price in the post period than in the pre period, you would consider them someone you would count within your common impact group; is that correct? A. Yes.”). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 46 of 94 PageID #:21861 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 45 - price six months after the PIIE ($10.01) was higher than the pre­event price ($10.00), even  though the price three and ninth months after the PIIE was lower than the pre­event price.    103.  Using Dr. Dwyer’s methodology, a customer can be impacted even though the  average price paid by the customer over the entire nine months after the PIIE can be lower than  the pre­event price.  In my prior example, if the customer purchased one unit in each month, its  average price in the three months after the PIIE is $9.67 (i.e., the average of $9.50, $10.01 and  $9.50).  But Dr. Dwyer would nonetheless classify that customer as “impacted” because one of  the prices after the PIIE was greater than the pre­event price.  Dr. Dwyer classifies all of a  customer’s purchases as impacted if any one of the products it purchased in any location is  “impacted” after one or more PIIEs.108    104. At his deposition, Dr. Dwyer agreed that antitrust “impact” refers to the effect of the alleged conspiracy on the prices paid by customers: Q. A. Q. A. And is it correct that the impact that you are looking at here or that you are attempting to analyze here is impact caused by some antitrust wrong? That's the allegation I'm taking as given, yes. That you understand the impact needs to be caused by the antitrust wrongdoing as opposed to some change in price that comes from something else; correct? That's correct.109 However, in his impact analysis, Dr. Dwyer makes no attempt to measure the effect of the alleged conspiracy on prices compared to what otherwise would have occurred (i.e., Dr. Dwyer does not compare actual prices to but-for prices). Instead, as I have discussed, Dr. Dwyer’s impact analysis is based only on comparisons of prices at different points in time, without controlling for any changes in economic conditions during the relevant period – such as costs – that could affect prices. At his deposition, Dr. Dwyer confirmed that he made no attempt to control for other factors that could explain price changes over time: 108. Dr. Dwyer also presents alternative analyses that define “impact” in somewhat different ways, but none of his analyses are based on comparing actual prices to but-for prices. See Dwyer Report, ¶¶ 39 – 50. 109. Dwyer dep., at 29: 22 – 30: 7. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 47 of 94 PageID #:21862 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 46 - Q. A. Q. A. Q. A. Q. A. Q. A. Q. A. In your common-impact analysis, as you've explained it, if a customer paid a dollar for a product in the pre-price increase implementation of that period and paid [$1.01] in the post-increase implementation period, you would count that as someone who has suffered an impact for purposes of your analysis; correct? That's correct. You would consider that person to have suffered an impact regardless of whether costs went up 5 cents; is that correct? Sure, yes. You would consider that an impact even if there was some change in demand that caused an increase in price; correct? I'm focused on these particular events where the defendants are announcing price increases, so I'm not looking behind their announcements to see what the motivation for it was in any impact analysis or price increase analysis. You're not adjusting for inflation, are you? No. So that in your regression model you included five variables for inflation; correct? That's correct. But in your price model for common impact, if some portion of any price increase is attributable to inflation generally, you're not factoring that out; correct? That's correct.110 Dr. Dwyer also testified that he had not attempted to measure the price that any customer would have paid but for the conspiracy: Q. A. Q. A. . . . . Is it correct that you have not at this point attempted to calculate the “but for” price for an individual customer within the class period? That's correct. And that your aggregate model does not estimate the “but for” price for an individual customer; is that also correct? Yes.111 Thus, Dr. Dwyer’s impact analysis fails to distinguish between changes in prices caused by nonconspiratorial changes in economic conditions with changes in price (if any) caused by the alleged conspiracy.112 For this reason, Dr. Dwyer’s impact analysis cannot and does not measure the impact of the alleged conspiracy. 105. Dr. Dwyer’s impact analysis also does not measure the impact of the alleged conspiracy because it largely consists of the comparison of allegedly conspiratorial prices with 110. Dwyer dep., at 252: 21 – 254: 3. 111. Dwyer dep., at 75: 9 – 17. 112. In contrast, in his damages analysis, Dr. Dwyer states: “In comparing prices between a benchmark period and the class period, I employ standard econometric methods to adjust for non-conspiratorial factors that may distinguish prices in these two periods” (Dwyer Report, ¶ 57). I discuss the flaws in Dr. Dwyer’s damages analysis in the next section of this report. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 48 of 94 PageID #:21863 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 47 - other allegedly conspiratorial prices. For example, in his analysis of PIIE 2, Dr. Dwyer compares prices taken from a “pre” period of August 2005 and September 2005 to prices taken from a nine-month “post” period of May 2006 to January 2007. That is, both the “pre” and “post” periods for PIIE 2 are completely within the alleged conspiracy period.113 106. For this reason, all price changes between the “pre” and “post” periods for PIIE 2 likely reflect changes in economic conditions, not the effect of the alleged conspiracy. For example, in the “pre” period before PIIE 2, the pulp price was either $600 or $605 per ton; in the “post” period after PIIE 2, the pulp price was in the range of $665 per ton to $750 per ton. Thus, the “impact” that Dr. Dwyer measures from PIIE 2 reflects the effect of changes in economic conditions, but cannot measure the effect of the alleged conspiracy because both the “pre” and “post” prices reflect the alleged conspiracy overcharge.   3. Dr. Dwyer’s Methodology Measures “Impact” Even in Periods During Which the PPW Index is Unchanged or Falling. 107. In his impact analysis, Dr. Dwyer claims that when a price increase is “successful” – as measured by the PPW index – the prices paid by all or nearly all putative class members increase. Dr. Dwyer’s analysis can be summarized as follows: (1) the alleged conspiracy resulted in an industry-wide price increase, as measured by the PPW index; (2) the increase in the PPW index is associated with an increase in the price paid by all or nearly all putative class members as measured by Dr. Dwyer (i.e., the customer paid more for at least one product in at least one location);114 and (3) therefore, the alleged conspiracy impacted all or nearly all putative class members. 113. Dr. Dwyer combines the first two “successful” price increases into PIIE 1. For PIIE 1, the “pre” period consists of the period December 2003 and January 2004, and the “post” period is July 2004 to March 2005. Thus, the “pre” period is before the start of the alleged conspiracy, and the “post” period is during the allege conspiracy. For all of the other PIIEs, both the “pre” and “post” periods are during the alleged conspiracy period. 114. Because Dr. Dwyer’s methodology defines “impact” based on a price increase for any single product, he would treat customers 1 and 2 as “impacted” under the following FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 49 of 94 PageID #:21864 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 48 - 108. As I have discussed, customer-level transaction prices often move independently of the PPW index and thus transaction prices can increase (or fall) even if the PPW index does not change. In this section of my report, I show that Dr. Dwyer’s methodology is flawed because it finds “impact” for a large number of customers even when the PPW index is unchanged or falling. Thus, Dr. Dwyer cannot conclude that an increase in the PPW index caused all or nearly all customers to pay more because Dr. Dwyer’s methodology fails to account for the fact that his approach would find that a large number of customers would have been “impacted” even if the PPW index had not increased. 109. To illustrate, suppose that there are 100 customers and a price increase is announced. Shortly after the price increase announcement, the PPW index increases from $500 to $550 per ton. Suppose that Dr. Dwyer’s methodology finds that 90 percent of those customers are “impacted” after the price increase announcement. Suppose also that had there been no price increase announcement, and the PPW index had remained constant at $500 per ton during this period, Dr. Dwyer’s methodology would nonetheless find that 36 customers were “impacted.” In that case, Dr. Dwyer could not conclude that the increase in the PPW index is what “impacted” all or nearly all customers (i.e., 90 customers) because Dr. Dwyer’s methodology implies that 40 percent of those customers (i.e., 36 customers) would have been “impacted” whether or not there had been a price increase announcement followed by an increase in the PPW index. Hence, Dr. Dwyer’s methodology would only be capable of linking the increase in the PPW index to increased prices for 54 customers (i.e., 90 minus 36 (...continued) conditions. Suppose customers 1 and 2 each buy one unit of product A and one unit of product B before and after a “successful” price increase. Suppose the price customer 1 pays for product A increases by $1 but the price it pays for product B falls by $10; similarly, the price customer 2 pays for product B increases by $1 but the price it pays for product A falls by $10. In this example, the average price paid by customers 1 and 2 for both products A and B falls. Similarly, the average price paid for products A and B for both customers 1 and 2 falls. Nevertheless, using Dr. Dwyer’s methodology, both customers are “impacted” because each customer pays more for one product (A in the case of customer 1; B in the case of customer 2). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 50 of 94 PageID #:21865 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 49 - customers).115 Thus, Dr. Dwyer’s methodology would not be isolating and measuring the “impact” of the increase in the PPW index because that methodology would find that many customers would be “impacted” even absent the PPW index increase. As I show in this section of my report, Dr. Dwyer’s methodology measures precisely this kind of effect of nonconspiratorial variations in individual product prices for many customers. 110. To show that Dr. Dwyer’s methodology finds “impact” that cannot be caused by an increase in the PPW index, I apply his methodology to data from three periods during which the PPW index was unchanged or fell: (1) the year 2003 (a year when no conspiracy is alleged); (2) the year 2011 (another year in which no conspiracy is alleged); and (3) the period February 2005 – June 2005, a period during the alleged conspiracy before and after an “unsuccessful” price increase. For each time period, Dr. Dwyer’s methodology implies that about 36 to 39 percent of customers are “impacted” even though there are no price increases as measured by the PPW index during those periods. That is, Dr. Dwyer’s methodology finds “impact” even when the PPW index does not change (or, in some months, falls). As a result, the “impact” that Dr. Dwyer’s methodology finds during the periods of “successful” allegedly conspiratorial price increases cannot be attributed solely to the alleged conspiracy because his methodology finds “impact” even when the PPW index is unchanged or falling. Findings of “impact” when the PPW index is not increasing must reflect non-conspiratorial variations in individual product prices. Thus, for any particular customer, Dr. Dwyer’s methodology cannot distinguish the “impact” of a “successful” allegedly conspiratorial price increase from the “impact” of non-conspiratorial variations in individual product prices. I now illustrate this point with data from the three periods I previously discussed. 115. Furthermore, because Dr. Dwyer’s methodology does not estimate individual customer price changes but for an increase in the PPW index, it cannot distinguish which 36 customers would be “impacted” in the absence of an increase in the PPW index from the 54 customers that are “impacted” because of the PPW index increase. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 51 of 94 PageID #:21866 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 50 - 111. In 2003, before the start of the alleged conspiracy, there was no increase in containerboard prices as measured by the PPW index. In January 2003, the PPW index was $385 per ton. The PPW index fell to $375 per ton in February 2003 and remained at that level through July 2003. The PPW index fell to $365 per ton in August 2003 and remained at that level through November 2003. In December 2003, the PPW index fell to $355 per ton. To test Dr. Dwyer’s methodology (i.e., to determine whether it would measure any “impact”), I compare customer prices in a “pre” period of January 2003 to a nine-month “post” period of March 2003 to November 2003 (i.e., to check whether one or more prices in the “post” period are higher than the corresponding “pre” period price despite the decline in the PPW index during 2003). 112. Dr. Dwyer’s methodology applied to the January 2003 – November 2003 time period implies that 37 percent of all customers were “impacted” during this period, despite the fact that the PPW index was lower in every month of the “post” period than in the “pre” period (and the fact that there is no price conspiracy alleged during the period). The “impact” on these customers, because it occurred during a period with no increase in the PPW index prior to the alleged conspiracy, must reflect non-conspiratorial variations in individual product prices. The percentage of “impacted” customers is substantially higher for “corrugated” (e.g., sheets and boxes) customers – 38 percent – than for “rollstock” (i.e., containerboard) customers – 22 percent, which indicates that there is more non-conspiratorial variation in prices for corrugated products than for containerboard. See Table 2.116 116. I present Table 2 and subsequent tables in the same format as Dr. Dwyer’s Exhibit 6. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 52 of 94 PageID #:21867 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 51 - Table 2 Dr. Dwyer's Exhibit 6 Impact Analysis Applied to the Period January 2003 through November 2003 Number of  Assessed Customers Defendant Type All Defendants All Defendants All Defendants   GP IP/WY NP PCA SSCC TI All Corrugated Roll Stock   All All All All All All 23,469 23,014 455   2,145 9,083 243 4,167 6,101 1,730 Number of Increase  Total Post‐Event Customers  Customers Total Post‐Event  Dollars Paid by  with  as Percentage  Dollars Assessed  Increase  Post‐Event  of Assessed  (In Millions) Customers Price Increases Customers (In Millions) 8,729 8,631 98   808 3,147 92 1,592 2,456 634 37% 38% 22%   38% 35% 38% 38% 40% 37% $1,970.3 $1,753.5 $216.9 $1,031.2 $972.3 $58.9   $275.2 $749.2 $5.6 $209.2 $480.8 $250.4   $122.0 $341.8 $1.5 $110.6 $301.7 $153.6 Percentage of  Assessed Post‐ Event  Dollars Paid by  Increase  Customers 52% 55% 27%   44% 46% 26% 53% 63% 61% Percentage of  Total Post‐Event  Post‐Event Dollars  Showing  Dollars Showing  Price Increases  Higher Prices for  Increase  (In Millions) Customers $172.4 $165.5 $6.9   $21.3 $68.5 $0.3 $18.0 $45.2 $19.1 17% 17% 12%   17% 20% 22% 16% 15% 12% Source: Dr. Dwyer's backup materials. 113. I also test Dr. Dwyer’s methodology in a post-conspiracy period to determine whether it would measure any “impact.” The PPW index was unchanged at $640 per ton during all of 2011 (indeed, the PPW index was at $640 per ton for 29 consecutive months – from April 2010 to August 2012). To test Dr. Dwyer’s methodology for this period, I compare customer prices in a “pre” period of January 2011 to a nine-month “post” period of March 2011 to November 2011. Dr. Dwyer’s methodology applied to this time period implies that 39 percent of customers were “impacted” during this period, despite the fact that the PPW index was unchanged throughout the year (and that there is no price conspiracy alleged during the period). Again, the percentage of “impacted” customers is substantially higher for corrugated products (40 percent) than for containerboard customers (12 percent), which indicates that there is more non-conspiratorial variation in prices for corrugated products than for containerboard. The “impact” on these customers, because it occurred during a period with no increase in the PPW index after the alleged conspiracy, must reflect non-conspiratorial variations in individual product prices. See Table 3. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 53 of 94 PageID #:21868 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 52 - Table 3 Dr. Dwyer's Exhibit 6 Impact Analysis Applied to the Period January 2011 through November 2011 Number of  Assessed Customers Defendant Type All Defendants All Defendants All Defendants   GP IP/WY NP PCA SSCC TI All Corrugated Roll Stock   All All All All All All 21,638 21,224 414   2,030 7,734 1,120 5,088 2,974 2,692 Number of Increase  Customers  Customers with  as Percentage  Post‐Event  of Assessed  Price Increases Customers 8,479 8,430 49   688 2,734 529 2,026 1,266 1,236 39% 40% 12%   34% 35% 47% 40% 43% 46% Total Post‐ Event  Dollars  Assessed  (In Millions) $3,570.3 $3,218.4 $351.9 Total Post‐Event Dollars Paid by  Increase  Customers (In Millions) $2,052.6 $1,985.9 $66.7   $1,145.3 $1,068.8 $50.3 $313.7 $487.8 $504.3   $634.7 $531.2 $39.4 $179.1 $340.2 $328.1 Percentage of  Assessed Post‐ Event  Dollars Paid by  Increase  Customers 57% 62% 19%   55% 50% 78% 57% 70% 65% Percentage of  Total Post‐Event  Post‐Event Dollars  Showing  Dollars Showing  Price Increases  Higher Prices for  Increase  (In Millions) Customers $424.4 $410.0 $14.4   $71.1 $132.1 $14.2 $36.5 $60.6 $109.9 21% 21% 22%   11% 25% 36% 20% 18% 34% Source: Dr. Dwyer's backup materials. 114. Finally, I test Dr. Dwyer’s methodology after an “unsuccessful” price increase (as determined by Dr. Dwyer) during the alleged conspiracy period. In particular, I use as a “pre” period February 2005 (i.e., the month before price announcement 3 in Table 1) and the threemonth period May through July 2005 as the “post” period (i.e., the three-month period beginning with the “following” month).117 I limit the “post” period to three months because a “successful” price increase was implemented in October 2005.118 In the “pre” period (February 2005), the PPW index was $450 per ton; the PPW index was $435 per ton in May and June 2005 and $405 in July 2005. Dr. Dwyer’s methodology applied to this time period implies that 36 percent of customers were “impacted” during this period, despite the fact that the PPW index was lower in every month of the “post” period than in the “pre” period. Again, the percentage of “impacted” customers is substantially higher for corrugated products (37 percent) than for containerboard 117. I do not analyze the other five “unsuccessful” price increases because the “post” period of each overlaps with the implementation of a “successful” price increase. 118. Following Dr. Dwyer’s methodology, if a price for a particular customer/product/location is not available for February 2005, I use prices as far back as September 2004 as a “pre” price. The PPW index was unchanged from September 2004 through February 2005. Because the “post” period in this analysis is only three months long, my results likely are conservative (i.e., it is less likely that Dr. Dwyer’s methodology will measure “impact” in a three-month “post” period than in a nine-month “post” period). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 54 of 94 PageID #:21869 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 53 - customers (16 percent), which indicates that there is more non-conspiratorial variation in prices for corrugated products than for containerboard. See Table 4. Table 4 Dr. Dwyer's Exhibit 6 Impact Analysis Applied to a Period after an "Unsuccessful" Price Increase in March/April 2005 Number of  Assessed Customers Defendant Type All Defendants All Defendants All Defendants   GP IP/WY NP PCA SSCC TI All Corrugated Roll Stock   All All All All All All 26,817 26,326 491   244 11,557 288 5,653 6,044 3,031 Increase  Total Post‐Event Number of Customers Total Post‐Event  Dollars Paid by  Customers with as Percentage of  Dollars Assessed  Increase  Post‐Event  Assessed  (In Millions) Customers Price Increases Customers (In Millions) 9,749 9,672 77   62 3,902 90 2,123 2,466 1,106 36% 37% 16%   25% 34% 31% 38% 41% 36% $1,037.2 $876.4 $160.7 $542.7 $498.2 $44.5   $50.3 $396.3 $4.6 $126.1 $260.4 $199.4   $9.2 $194.8 $1.1 $70.5 $161.3 $105.8 Percentage of  Percentage of  Total Post‐ Post‐Event Dollars  Assessed Post‐ Event  Showing  Event  Dollars Showing  Higher Prices for  Dollars Paid by  Price Increases  Increase  Increase  (In Millions) Customers Customers 52% 57% 28%   18% 49% 24% 56% 62% 53% $118.9 $112.4 $6.5   $2.8 $46.5 $0.3 $19.9 $29.7 $19.7 22% 23% 15%   31% 24% 28% 28% 18% 19% Source: Dr. Dwyer's backup materials. 115. Thus, applying Dr. Dwyer’s methodology to three different periods – one prior to the beginning of the alleged conspiracy; one after the alleged conspiracy; and one at the time of an “unsuccessful” price increase – during which industry prices as measured by the PPW index were unchanged or falling produces “impact” for 36 to 39 percent of putative class members. Dr. Dwyer claims that 92 percent of putative class members “were impacted, that is, showed elevated prices following a PIIE. . . . Overall, this analysis shows that these Defendants' price increase implementations systematically pervade class member purchases, supporting the conclusion that all or nearly all class members were impacted.”119 But using the results of my analyses as a guide, Dr. Dwyer’s methodology would find “impact” for about 40 percent (e.g., 36 to 39 divided by 92 percent) of these putative class members even in the absence of a “successful” price increase announcement. Thus, Dr. Dwyer’s analysis is not capable of distinguishing “impact” caused by an allegedly conspiratorial price increase from “impact” caused by non-conspiratorial variations in individual product prices. 119. Dwyer Report, ¶ 38. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 55 of 94 PageID #:21870 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 54 - 4. Dr. Dwyer’s Methodology Implies that Many Customers are “Negatively Impacted” Even in Periods During Which the PPW Index is Increasing. 116. As a further illustration that Dr. Dwyer’ methodology is flawed and that non- conspiratorial variations in individual product prices have a substantial effect on Dr. Dwyer’s “impact” methodology, I use Dr. Dwyer’s methodology to measure “negative impact.” In particular, I define a customer as receiving a “negative impact” if it pays less for one or more products in one or more locations before and after the PIIEs Dr. Dwyer defines (i.e., I repeat Dr. Dwyer’s analysis but instead identify customers as “negatively impacted” if they paid less for one or more products in one or more locations). Such negative impacts cannot be the result of a “successful” conspiratorial price increase, and so instead must reflect non-conspiratorial variations in individual product prices. 117. Applying Dr. Dwyer’s methodology to the “successful” price increases, I find that 35 percent of customers were “negatively impacted” by defendants’ “successful” price increases. That is, because of non-conspiratorial variations in individual product prices, 35 percent of customers paid less for one or more products in one or more locations despite a “successful” price increase. As with my analyses of “impact” during periods in which the PPW index is unchanged or falling, the percentage of “negatively impacted” customers is higher for corrugated products (35 percent) than for containerboard customers (27 percent), which further indicates that there is more non-conspiratorial variation in prices for corrugated products than for containerboard. See Table 5. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 56 of 94 PageID #:21871 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 55 - Table 5 Dr. Dwyer's Exhibit 6 Impact Analysis Used to Identify "Negative Impact" Number of  Assessed Customers Defendant Type All Defendants All Defendants All Defendants   GP IP/WY NP PCA SSCC TI All Corrugated Roll Stock   All All All All All All 78,224 77,134 1,090   3,455 38,029 1,930 16,748 9,467 8,595 Number of Customers with Post‐Event  Price Decreases 27,331 27,039 292   1,221 11,609 879 5,846 3,979 3,797 Decrease  Total Post‐Event Customers Total Post‐Event  Dollars Paid by   as Percentage of  Dollars Assessed  Decrease  Assessed  (In Millions) Customers Customers (In Millions) 35% 35% 27%   35% 31% 46% 35% 42% 44% $14,619.5 $12,548.9 $2,070.6 $10,728.2 $9,479.3 $1,248.9   $3,021.9 $5,216.7 $146.3 $1,569.6 $2,197.9 $2,467.1   $2,329.8 $3,306.3 $106.0 $1,144.3 $1,726.3 $2,115.7 Percentage of  Assessed Post‐ Event  Dollars Paid by  Decrease  Customers 73% 76% 60%   77% 63% 72% 73% 79% 86% Total Post‐Event  Dollars Showing  Price Decreases  (In Millions) Percentage of  Post‐Event Dollars  Showing  Lower Prices for  Decrease  Customers $1,407.1 $1,299.6 $107.5   $220.1 $474.0 $28.9 $159.0 $238.6 $286.6 13% 14% 9%   9% 14% 27% 14% 14% 14% Source: Dr. Dwyer's backup materials. 5. Dr. Dwyer’s Impact Analysis Fails to Adjust for Non-Conspiratorial Changes in Economic Conditions and Non-Conspiratorial Variations in Individual Product Prices. 118. As I have discussed, Dr. Dwyer’s impact analysis fails to distinguish changes in prices caused by non-conspiratorial changes in economic conditions from changes in price (if any) caused by the alleged conspiracy. Also as I have discussed, Dr. Dwyer’s methodology implies that a substantial number of customers are “impacted” even in periods during which the PPW index is unchanged or falling, and many customers are “negatively impacted” in periods during which the PPW index is increasing. Those findings indicate that the prices of individual containerboard and corrugated products vary over time relative to the PPW index for a variety of non-conspiratorial reasons. In this section of my report, I show that the potential magnitude of ignoring these two factors – the effect of non-conspiratorial changes in economic conditions; and the effect of non-conspiratorial variations in individual product prices – is substantial. 119. I control for changes in economic conditions and the effect of non-conspiratorial variations in individual product prices as follows:  First, to control (to some extent) for cost changes, I estimate containerboard and corrugated product but-for prices based on changes in one important cost factor – the price of pulp. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 57 of 94 PageID #:21872 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 56 -  Second, to control (to some extent) for non-conspiratorial variations in individual product prices, I calculate the weighted average change in price (across all of the customer’s purchases, locations and PIIEs) and compare that weighted average price to the equivalent weighted average but-for price. 120. I adjust for the effect of changes in the price of pulp by estimating a simple model in which the “dependent variable” is the logarithm of the PPW index, and the only “explanatory variable” is the logarithm of the pulp price.120 I base my model of the PPW index on the price of pulp because the PPW index tends to move together with the price of pulp, a key input into the production of containerboard; the correlation coefficient between the two series is 0.89. See Figure 3. Figure 3 I find that there is a statistically significant relationship between the PPW index and the pulp price, and the regression implies that a 10 percent change in the pulp price is associated with 120. I use the results of this regression for illustrative purposes only. I do not intend this model to be a complete description of the determinants of the PPW index; in general, the PPW index will depend on a variety of factors other than the pulp price. I note, however, that the pulp price explains about 77 percent of the variation in the PPW 42 lb. linerboard index during the period I analyze. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 58 of 94 PageID #:21873 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 57 - an increase in the PPW index of about 7.2 percent.121 For the purpose of my analysis, I assume that the change in the but-for price of containerboard and corrugated products equals the change in the PPW index implied by the PPW/pulp regression. 121. For example, in the case of PIIE 1, Dr. Dwyer compares prices in January 2004 to prices in September 2004, December 2004 and March 2005.122 In January 2004, the U.S. pulp price was $560 per ton; the pulp prices in September 2004, December 2004 and March 2005 were $625 per ton; $620 per ton; and $650 per ton. Because pulp prices in the three “post” months were 11.6 percent; 10.7 percent; and 16.1 percent higher than in the “pre” month, I estimate that the but-for price would have increased by 8.3 percent; 7.7 percent; and 11.5 percent from the January 2004 level.123 122. To illustrate how I control for non-conspiratorial variations in individual product prices, suppose that a putative class member purchases one unit each of two corrugated products, A and B, both before and after a “successful” price increase. Suppose that, after the “successful” price increase, the price of product A is $10 and the price of product B is $5. Suppose further that, adjusting for changes in pulp prices, the but-for price (i.e., the but-for price derived from my PPW/pulp price model) of product A would have been $10.50 (i.e., $0.50 above 121. The estimated coefficient on ln(pulp price), which implies that a 10 percent increase in the pulp price is associated with an increase of the PPW index of about 7.2 percent (i.e., 1.100.73 = 1.072). At his deposition, Dr. Harris testified: “Q. Is it your opinion that costs play any role in the price of containerboard between 2004 and 2010? A. Yeah, absolutely. I mean that’s exactly why Dr. Dwyer put it in his model. Q. Other than Dr. Dwyer’s analysis, do you have any independent opinion about the magnitude of that role? A. No. In order to have an opinion about the magnitude [of the effect of cost on price], you’d have to have an empirical model, and I don’t have that.” (Harris dep., at 385: 13 – 23). 122. If prices for those months are not available, Dr. Dwyer searches for prices in earlier months. See Dwyer Report, ¶ 37 (“The pre-event price is the monthly net price in the closest month up to (and including) the month prior to the PIIE as listed in Exhibit 5. The comparisons then are to the monthly net prices for that same product within 3 months, 6 months, and 9 months after the PIIE.”). 123. In comparison, the PPW index increased by 26.8 percent between the “pre” and “post” period (i.e., the PPW index was $355 per ton in January 2004 and $450 per ton in each of the “post” period months). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 59 of 94 PageID #:21874 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 58 - the actual price) and the but-for price of product B would be $4.90 (i.e., $0.10 below the actual price). 123. If “impact” is defined as “actual price is above but-for price for one or more products,” this customer would be “impacted” because the actual price of product B is higher than its but-for price, even though the actual price of A is less than its but-for price. As I have discussed, the evidence shows that non-conspiratorial changes in price are common, so that both differences between actual and but-for prices (i.e., -$0.50 for product A and $0.10 for product B) could reflect such variations. Because of the possibility of such non-conspiratorial price variations, it is reasonable to define a customer as “impacted” only if the average price it paid for all of its purchase is more than the average but-for price of those purchases. Thus, to “average out” the effect of non-conspiratorial variations in price, I calculate the sales-weighted average price across all of a customer’s purchases. For example, if the customer purchased equal quantities of products A and B, the actual average price it paid is $7.50, and the average but-for price is $7.70. In this example, the customer’s average price for both products is less than its average but-for price, which indicates that it was not “impacted” by the price increase. 124. Making these two changes to Dr. Dwyer’s analysis – that is, controlling (to some extent) for non-conspiratorial changes in a single cost (pulp), and for non-conspiratorial variations in individual product prices -- reduces the share of “impacted” customers from 92 percent to 66 percent. See Table 6. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 60 of 94 PageID #:21875 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 59 - Table 6 Dr. Dwyer's Exhibit 6 Impact Analysis Aggregated by Data Source and Customer ID Number of  Assessed Customers Defendant Type All Defendants All Defendants All Defendants   GP IP/WY NP PCA SSCC TI All Corrugated Roll Stock   All All All All All All 78,224 77,134 1,090   3,455 38,029 1,930 16,748 9,467 8,595 Number of Customers with Post‐Event  Price Increases 51,387 50,409 978   2,226 24,276 861 11,813 6,072 6,139 Increase Customers  as Percentage of  Assessed Customers 66% 65% 90%   64% 64% 45% 71% 64% 71% Total Post‐Event  Dollars Assessed  (In Millions) $14,617.6 $12,547.0 $2,070.6 Total Post‐Event Percentage of  Dollars Paid by  Assessed Post‐Event  Increase Customers Dollars Paid by  (In Millions) Increase Customers $11,140.1 $9,152.6 $1,987.5   $3,020.0 $5,216.7 $146.3 $1,569.6 $2,197.9 $2,467.1   $2,428.7 $3,954.9 $95.9 $1,202.1 $1,677.8 $1,780.6 76% 73% 96%   80% 76% 66% 77% 76% 72% Source: Dr. Dwyer's backup materials. 6. Dr. Dwyer’s Methodology Does Not “Assess” Impact for About Two-Thirds of Putative Class Members. 125. In addition to the flaws in Dr. Dwyer’s impact analysis that I have discussed, Dr. Dwyer’s analysis fails to “assess” impact for about 150,000 of the 230,884 putative class members.124 At his deposition, Dr. Dwyer stated that he had not analyzed these approximately 150,000 putative class members: Q. A. Q. A. With respect to the customers that you weren't able to measure impact, have you done any examination of those customers? No, I have not. And do you know if the customers for whom you did measure impact are representative of the customers for which you didn't assess impact? [Objection omitted] I have no reason to believe that they're not representative.125 124. Dr. Dwyer conducts his analysis on what he calls “class member identifications.” A single customer may be associated with more than one “class member identification. See Dwyer Report, fn 20 (“Across defendant datasets, the total number of class member identifications is 230,884. For almost all defendants, no information was provided to unify separate customer identifiers (e.g., customer numbers) when those customers were part of the same company. This is but one reason why class member counts are likely overstated here, and why the share of class members showing evidence of impact is likely understated.”). For ease of reference, I use the term “customer” or “putative class member” to refer to Dr. Dwyer’s “class member identifications.” 125. Dwyer dep., at 276: 13 – 24. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 61 of 94 PageID #:21876 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 60 - 126. Dr. Dwyer has no basis for his claim that the “assessed” putative class members are representative of the “unassessed” putative class members. Dr. Harris conducted no analysis to determine whether the putative class members he “assessed” are representative of the those he did not “assess.” In fact, the approximately 150,000 “unassessed” customers are systematically different from the approximately 78,000 assessed customers. Because assessed customers necessarily purchase both before and after a “successful” price increase, Dr. Dwyer has no basis to claim that those customers are representative of “unassessed” customers, whose purchase patterns over time are different. In particular, many of the unassessed putative class members made purchases only during periods after “unsuccessful” price increases (i.e., periods of “cartel instability”) when the PPW index was falling or unchanged. Dr. Dwyer has no basis to claim that the assessed customers are representative of unassessed putative class members that made purchases only during these periods. 127. For example, the six-month period April 2005 to September 2005 was during and after an “unsuccessful” price increase, and before the next “successful” price increase (in October 2005). During this six-month period, the PPW index fell from $450 per ton to $395 per ton. Similarly, the eight-month period May 2009 to December 2009 was after an “unsuccessful” price increase, and before the next “successful” price increase (in January 2010). During this eight-month period, the PPW index fell from $540 per ton to $530 per ton. In total, 10,495 putative class members only made purchases during one of these 14 months.126 In addition, 2,906 putative class members made purchases in two or more of these 14 months, but not in any other month.127 Dr. Dwyer provides no basis to assume that these customers were impacted by the alleged conspiracy because the timing of “assessed” customers’ purchases – before and after at least one “successful” price increase is not representative of their purchase 126. That is, these customers made purchases in only one month of the period December 2003 – October 2010. 127. As I have discussed, RISI excludes certain “spot” purchases from its PPW indices. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 62 of 94 PageID #:21877 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 61 - patterns – only during and after “unsuccessful” price increases while the PPW index was constant or falling. 128. In sum, Dr. Dwyer’s impact analysis is unreliable for a number of reasons: (1) the PPW index does not provide evidence of class-wide impact because transaction prices do not move in lock-step with that index, as Dr. Dwyer implicitly assumes; (2) Dr. Dwyer measures only the change in prices over time, but fails to compare actual prices to but-for prices; (3) Dr. Dwyer’s methodology implies that a substantial number of putative class members would be “impacted” even during periods when the alleged conspiracy does not increase the PPW index; (4) Dr. Dwyer’s methodology also implies that a substantial number of putative class members are “negatively impacted” by “successful” price increases; (5) Dr. Dwyer’s methodology fails to find “impact” for about a third of “assessed” putative class members when it is modified to control for non-conspiratorial changes in a key cost and non-conspiratorial variations in individual product prices; and (6) Dr. Dwyer’s methodology does not “assess” impact for about two-thirds of putative class members. VI. PLAINTIFFS PROVIDE NO ECONOMIC ANALYSIS THAT RELIABLY MEASURES AGGREGATE DAMAGES. 129. Plaintiffs rely on an estimate of aggregate damages produced by Dr. Dwyer. Dr. Dwyer claims that he has “employed standard econometric methods, multiple regression analysis, on a class-wide basis to demonstrate, preliminarily, that such methods may be used to quantify aggregate overcharge damages to the class.”128 Dr. Dwyer’s analysis of aggregate damages is unreliable for a number of reasons: (1) Dr. Dwyer uses statistical approaches that lack an economic foundation; (2) Dr. Dwyer inconsistently applies his statistical methodology; moreover, if he had applied his methodology consistently, it would have produced materially 128. Dwyer Report, ¶ 12. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 63 of 94 PageID #:21878 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 62 - different results; (3) Dr. Dwyer’s model is not “stable” and the results of his models are sensitive to reasonable changes. A. Dr. Dwyer Uses Statistical Approaches to Estimating Damages that Lack an Economic Foundation. 130. Dr. Dwyer’s damages analysis is based on statistical techniques that lack an economic foundation. Dr. Dwyer’s regression analysis includes two separate steps – a “principal components” analysis for creating potential explanatory variables and a “stepwise” procedure to choose which variables to include in his damages models. 131. Dr. Dwyer uses a statistical technique known as “principal components” to create potential explanatory variables from a variety of economic variables (e.g., measures of cost and demand).129 Principal components analysis is a statistical technique that creates weighted averages of a set of variables in such a way that the resulting weighted averages (i.e., the “principal components”) are not correlated with each other. Each principal component is a weighted average of economic variables (such as measures of supply and demand). Dr. Dwyer includes 150 variables in his analysis so each principal component is a weighted average of all 150 variables (the principal components differ from each other because they apply different weights to each of the 150 original variables). Thus, an increase in the value of any particular principal component reflects changes (either increases or decreases) in up to 150 variables. Because each principal component is the weighted average of 150 variables, it will not generally be possible to give an economic interpretation to any of the principal components in Dr. Dwyer’s analysis.130 129. See Dwyer Report, ¶ 65. Dr. Dwyer testified that he used principal components to address collinearity issues. See Dwyer dep., at 122: 9 – 13 (“Q. Is it correct that the reason why you used the principal components analysis was because you were concerned about collinearity issues across your independent variables? A. That’s correct.”). 130. See, for example, Peter Kennedy, A Guide to Econometrics, 3rd ed. (“Kennedy”), at 182 (“The variables that are collinear could be grouped together to form a composite index FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 64 of 94 PageID #:21879 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 63 - 132. To illustrate with a simplified example, suppose that the value of a principal component is the weighted average of one demand variable and one supply variable, and can be written as: Principal Component = 5 x D – 2.5 x S, where D measures a demand variable and S measures a supply variable. If D increases by 2, the value of the principal component increases by 10 (i.e., 5 x 2 = 10). Similarly, if the value of S falls by four, the value of the principal component increases by 10 (i.e., -2.5 x -4 = 10). Finally, if the value of D increases by three and the value of S increases by 2, the value of the principal component again increases by 10 (i.e., 5 x 3 – 2.5 x 2 = 15 – 5 = 10). Suppose that a model of the type used by Dr. Dwyer implies that a 10-point increase in the value of the principal component is associated with a five percent increase in the average FCP price. In that case, it is not possible to interpret the five percent increase in average FCP price that results from the 10-point increase in the value of the principal component as either a demand-induced or supplyinduced increase because that increase could be the result of an increase in demand with no change in supply; a decrease in supply with no change in demand; or an increase in both demand and supply (or other combinations of changes in supply and demand). 133. The second step in Dr. Dwyer’s analysis is the use of a “stepwise” selection procedure to choose the variables that are included in his damages model.131 Stepwise procedures are computer algorithms that, in effect, allow the computer to choose which (...continued) capable of representing this group of variables by itself. Such a composite variable should be created only if the variables included in the composite have some useful combined economic interpretation; otherwise the empirical results will have little meaning. For example, in undertaking a study of the effect of marketing activity on consumer demand, a researcher might find that variables representing different dimensions of marketing activity are highly collinear; some combination of these variables could be interpreted as a ‘marketing variable’ and its use in the model would not confuse the meaning of the empirical results. The most popular way of constructing such a composite index is to use the first principal component of the variables in question.”). Dr. Dwyer did not, however, use principal components analysis in this way because he did not limit the variables in his analysis to those that “have some useful combined economic interpretation.” 131. See Kennedy, at 52 (“A variant of OLS called stepwise regression is to be avoided.”). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 65 of 94 PageID #:21880 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 64 - variables to include in the model based on their statistical properties independent of their economic rationale. 134. Neither of these approaches – the use of principal components as implemented by Dr. Dwyer, nor the use of a stepwise selection procedure – is highly regarded in the economics profession. For example, not one article in the major economics journal I co-edit has published an article that uses Dr. Dwyer’s approach during the last 10 years. B. Dr. Dwyer Applies his Statistical Methodology Inconsistently; A Consistent Application of his Methodology Produces Materially Different Results. 135. Even if Dr. Dwyer’s approach were valid (and I explain later in this section why it is not), Dr. Dwyer does not consistently apply either portion of his methodology. If he had done so, his results would have been materially different. A consistent application of Dr. Dwyer’s approach finds no damages for corrugated products and reduces his estimate of containerboard damages by up to 75 percent. 136. As I have discussed, Dr. Dwyer applies his principal components analysis to a list of 150 explanatory variables.132 In general, one principal component can be created for every variable – thus, 150 variables can be transformed into 150 principal components.133 Dr. Dwyer could have included all 150 principal components into his damages model, but instead only includes 133 of the 150.134 That is, 17 principal components are not included as potential 132. Dr. Dwyer begins with 24 potential explanatory variables, and then adds squared versions for six of these variables. For each of these 30 variables, Dr. Dwyer includes in his principal components analysis the contemporaneous value of the variable and four lagged values, resulting in 150 total explanatory variables. 133. One principal component can be created for each variable included in the analysis unless one or more of the variables included in the analysis is a perfect linear combination of two or more other variables in the analysis. 134. At his deposition, Dr. Dwyer testified that he had “not looked into whether that selection [i.e., of only 133 principal components] is something that I have control over. I'm using a default program to compute principal components. There's a – there's a number that summarizes the – and allows you to rank by this number the principal components in terms of the variance. Did they explain that? It's called an eigenvalue. As that declines, there's some point at which the program determines that it's sufficiently close to zero that it FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 66 of 94 PageID #:21881 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 65 - explanatory variables in Dr. Dwyer’s analysis. Because of the way principal components are estimated, the last 17 explain relatively little of the variation in the original set of 150 explanatory variables, but any one of those 17 excluded principal components could help explain the average transaction price. Put differently, the last 17 principal components could have more explanatory power in the regression model than at least some of the 133 principal components that Dr. Dwyer includes in his analysis. For this reason, in my analysis, I include all 150 principal components in the list of potential explanatory variables.135 137. The stepwise procedure that Dr. Dwyer uses (known as “forward” selection) begins by identifying the potential variable with the most “explanatory power” for the dependent variable in the regression. The model then identifies the variable that adds the second most explanatory power (given the variables that already are included in the model), and continues adding variables so long as the explanatory power added by the last variables exceeds a particular threshold.136 Dr. Dwyer explains: (...continued) cannot – it would not estimate another principal component based on that” (Dwyer dep., at 128: 17 – 129: 4). The section of the manual on “principal components analysis” for the STATA software package used by Dr. Dwyer states: “components(#) and mineigen(#) specify the maximum number of components (eigenvectors or factors) to be retained. components() specifies the number directly, and mineigen() specifies it indirectly, keeping all components with eigenvalues greater than the indicated value. . . . mineigen(#) sets the minimum value of eigenvalues to be retained. The default is 1e-5 or the value of tol() if specified.”). The value used to determine whether to stop estimating principal components (i.e., the “eigenvalue”) is not related to the explanatory power of a principal component in a regression on a different variable (in this case, the FCP or ICP price). Thus, if certain principal components are excluded from the analysis, as in Dr. Dwyer’s analysis, then the analysis will pay no attention to the explanatory power of those excluded principal components. 135. When I run Dr. Dwyer’s model with all 150 principal components as potential explanatory variables, the BIC selection criterion selects principal components 134, 136, 137, 138, 139, 140, 142 and 143 for the FCP model, and principal components 134, 135, 136, 137, 138, 139, 140, 141, 142 and 143 for the ICP model. That is, Dr. Dwyer’s own methodology (with the BIC selection criterion) includes several of the principal components that he did not include in his analysis. 136. In general, the incremental explanatory power of a variable depends on what variables are already included in the model because of correlation between variables. A criticism of stepwise procedures is that the order in which variables are included in the analysis affects the results of the model. The principal components in Dr. Dwyer’s model are not FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 67 of 94 PageID #:21882 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 66 - In order to determine which of the economic factors identified above best explain prices, I use a stepwise model selection procedure to identify principal components with the most explanatory power. In this way, the model selection process is fully reproducible and is not influenced by subjective determinations, nor by implications with respect to the estimated conspiracy effect of the selection of some economic factors versus others.137 Dr. Dwyer uses two “model selection criteria” (known as “AICC” and “BIC”) to select variables to include in his damages model. The selection criteria differ by how they determine when to stop adding variables, and so, in general, the two selection criteria will not choose the same number of explanatory variables to include.138 At his deposition, Dr. Dwyer testified that the use of these model selection criteria means that his model selection process is “not influenced by subjective determinations.”139 138. Despite his claim, however, Dr. Dwyer does not rely on either model selection criteria to determine all of the variables in his analysis. Dr. Dwyer applies the model selection criteria to the 133 principal components he estimates and 11 additional “monthly indicator variables.”140 Dr. Dwyer does not, however, apply the same model selection criteria to the “conspiracy indicator” he uses to measure damages, nor does he apply the model selection criteria to five variables he uses as an “overall measure of inflation.”141 Instead, Dr. Dwyer (...continued) correlated with each other (i.e., the variables are “orthogonal” to each other), but the month indicators, the inflation variables and the conspiracy indicator (which Dr. Dwyer’s approach cannot include in his principal components analysis) are not orthogonal to each other or to the principal components (i.e., those additional variables may be correlated with each other and the principal components) and therefore Dr. Dwyer’s regression is subject to the criticism that the order in which variables are added to the model affects the results of the model. 137. Dwyer Report, ¶ 66 (emphases added; footnote omitted). 138. The two selection criteria add variables to the model in the same order, but the two criteria generally do not add the same number of variables. 139. See Dwyer dep., at 136: 15 – 19 (“Q. And is it the fact that the computer makes these selections why you view this approach as not influenced by subjective determinations? A. Yes, it is.”). 140. See Dwyer Report, ¶ 64. Dr. Dwyer does not include the monthly indicator variables in his principal components analysis, although he could have. Because the monthly indicator variables are not included in the principal components analysis, those variables generally will be correlated to at least some degree with each of the principal components. 141. See Dwyer Report, ¶ 63. Dr. Dwyer includes in his model the monthly “Producer Price FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 68 of 94 PageID #:21883 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 67 - forces each model to include the conspiracy indicator and all five inflation variables, even if his model selection criterion would not include them; that is, the selection criteria are applied only to the 133 principal components and the monthly indicator variables, but not to the conspiracy indicator or inflation variables. 139. Dr. Dwyer’s implementation of his damages model means that every model he estimates must include the conspiracy indicator and all five inflation variables, even if, according to the model selection criteria he uses, those variables would be excluded from the model because of their low explanatory power. That is, Dr. Dwyer requires his model to include the conspiracy indicator even if that variable would not otherwise be selected by his model selection criteria. In other words, despite choosing to use the stepwise selection methodology because it is “not influenced by subjective determinations,” Dr. Dwyer nonetheless made the subjective determination that every model must include the conspiracy indicator variable (and the five inflation variables) even if the objective model selection criterion would have excluded them. 140. If Dr. Dwyer had implemented his methodology consistently, he would have applied his model selection criteria to all the potential explanatory variables in his analysis – the 150 principal components; the monthly indicator variables; the inflation variables; and the conspiracy indicator. I have repeated Dr. Dwyer’s analysis with the following two changes: (a) I include all 150 principal components as potential explanatory variables; and (b) I include all potential explanatory variables in the stepwise procedure. That is, I do not force the model to include the conspiracy indicator and the five inflation variables, allowing the computer algorithm to operate free from any subjective determinations as to which variables to include in the model. 141. Dr. Dwyer estimates separate models for what he calls “Final Containerboard Products” or “FCP” (i.e., all corrugated products except sheets) and what he calls “Intermediate Containerboard Products” or “ICP” (i.e., containerboard and sheets). For both FCP and ICP, he (...continued) Index” and four lags of that index; thus, Dr. Dwyer uses five variables to control for overall inflation. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 69 of 94 PageID #:21884 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 68 - estimates a model using both the AICC and BIC model selection criterion. I find that the conspiracy indicator is not chosen for inclusion in either of Dr. Dwyer’s FCP models (i.e., using either AICC or BIC as a model selection criterion). That is, when I allow the computer algorithm to choose which variables to include in the model based only on their explanatory power (and not Dr. Dwyer’s subjective decision to force the conspiracy indicator into the model), the conspiracy indicator variable is not chosen. The AICC criterion chooses 68 variables, but not the conspiracy indicator. Similarly, the BIC criterion chooses 142 variables, but not the conspiracy indicator.142 In Dr. Dwyer’s analysis, damages associated with putative class members’ purchases of FCP are about 80 percent of the total.143 142. In his ICP analysis, Dr. Dwyer estimates a conspiracy effect of 4.05 percent using the AICC criterion and 4.04 percent using the BIC criterion. When I allow the computer algorithm to choose which variables to include in the model based only on their explanatory power, the conspiracy indicator is chosen using both model selection criteria. The results using the AICC criterion are similar (an estimated conspiracy effect of 3.82 percent), but the estimated conspiracy effect using the BIC criterion falls by over 75 percent, to 0.92 percent. See Table 7.  142. Dr. Dwyer also presents what he refers to as “10 PC” models (see Dwyer Report, Exhibit 11). Dr. Dwyer’s “10 PC” models are not based on a statistical selection criterion. Instead, in these models, Dr. Dwyer includes the first 10 principal components chosen by the AICC specifications (see Dwyer Report, ¶ 68). That is, the “10 PC” models each include 10 principal components, as well as all five inflation variables and the conspiracy indicator variable (plus the month indicator variables selected by the AICC criterion in the AICC model). That is, as with the AICC and BIC models, Dr. Dwyer “forces” the conspiracy indicator to be included in each of his “10 PC” models. 143. See Dwyer Report, Exhibit 12. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 70 of 94 PageID #:21885 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 69 - Table 7 Dr. Dwyer's Damage Regression Model Results 150 Principal Components with All Variables Included in Stepwise Procedure Product AICC BIC Final Containerboard Products Overcharge Regressors Observations Adjusted R‐squared None Selected 68 144 99.6% None Selected 142 144 100.0% Intermediate Containerboard Products Overcharge P‐value Regressors Observations Adjusted R‐squared 3.82% 0.000 55 144 99.1% 0.92% 0.000 142 144 100.0% Source: Dr. Dwyer's backup materials.      Note: P‐values are based on HAC standard errors. C. Even if Dr. Dwyer’s Approach were Reasonable – Which it is Not – the Models he Estimates are Not Stable and Sensitive to Reasonable Changes. 143. The purpose of an econometric model in a price-fixing case is to determine the price increase caused by the cartel, not simply a price change that coincides with the period of the alleged conspiracy but is caused by other factors such as an increase in costs. This means that any econometric model must be carefully constructed to be sure that it actually measures price changes that are caused by an alleged conspiracy and is not misattributing price changes arising from other changes in the economic environment to the alleged conspiracy. A reliable economic model should have the property that: (1) it is “stable” over time; and (2) its results are not sensitive to reasonable changes.144 As I explain in this section of my report, Dr. Dwyer’s models are not stable, and his results are sensitive to reasonable changes. 144. If the results of a model are sensitive to reasonable changes, economists often refer to such models as not being “robust.” FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 71 of 94 PageID #:21886 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 70 - 1. Dr. Dwyer’s Models are Not Stable. 144. To model transaction prices reliably with the type of model used by Dr. Dwyer, an economist needs to know that the underlying economic pricing relations, which the model uncovers, are “stable” between the alleged conspiracy and non-conspiracy periods, in the sense that the same economic factors matter, and that the effect of those factors on price are the same throughout the time period over which the model is estimated, so that any differences in prices between these two periods can reasonably be attributed to the effects of the cartel. This means that any model based on a benchmark period or periods that is used to assess causal impacts of an alleged conspiracy is based on the assumption that there is structural stability of the model across the non-conspiracy periods and the alleged conspiracy period. If a model is not structurally stable, there is no reason to believe that it can measure the effects of the cartel with any confidence. Each of the models used by Dr. Dwyer is based on the assumption of stability – in particular, that the relationship between transaction price and explanatory variables is the same in the alleged conspiracy and benchmark periods (except for the conspiracy indicator).145 145. To illustrate the importance of the stability assumption with a simple example, suppose that a model of the relationship between the weight and height of athletes is first estimated using a group of sumo wrestlers, and then estimated using a group of NBA players. The relationship between weight and height will differ substantially depending on whether it is based on sumo wrestlers or NBA players. That is, the relationship between weight and height will not be stable across the two groups of athletes. If the sumo wrestler model is used to predict weights of NBA players, it will give unreliable results. Similarly, if the NBA player model is used to predict weights of sumo wrestlers, it will give unreliable results. If a model of the 145. See Dwyer dep., at 90: 16 – 20 (“Q. When you run a regression, are you making the assumption that the causal factors have the same impact over the period of time you're studying? A. On average, yes.”). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 72 of 94 PageID #:21887 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 71 - relationship between weight and height is estimated using both sumo wrestlers and NBA players, the results of that model will generally produce unreliable predictions for both groups of athletes. 146. The stability assumption that Dr. Dwyer makes can be tested empirically using a standard statistical technique known as the “Chow test.”146 For the purpose of conducting stability tests, I adopt Dr. Dwyer’s models as estimated.147 A Chow test cannot be reliably conducted on Dr. Dwyer’s AICC and BIC models (because they contain so many explanatory variables), so I only test the stability of Dr. Dwyer’s “10 PC” models.148 For both Dr. Dwyer’s FCP and ICP models, I find that the Chow test rejects the stability assumption. Because Dr. Dwyer’s models are not stable, the overcharge estimates they produce are not reliable. 2. The Results of Dr. Dwyer’s Models are Sensitive to Reasonable Changes. 147. Using econometric models to estimate overcharges requires economists to make a series of decisions, many of which have no single correct answer. Hence, all such models should be subject to sensitivity testing, in which one evaluates the effect of making reasonable changes to the model. Sensitivity tests should always be performed to make sure that reasonable changes in the model do not have substantial effects on the results. If such changes do have substantial effects, then the model’s results are unreliable. As I have 146. See Dwyer dep., at 91: 25 – 92: 4 (“Q. Are you aware of any tests that are used to test the stability of a model over time? A. Sure. Q. What are those? A. One is the Chow test.”). 147. That is, I adopt Dr. Dwyer’s choice: of how to process the raw data; which explanatory variables to include; how to construct those variables; of assumptions about the error term. I do not, however, necessarily endorse Dr. Dwyer’s models or assumptions for the reasons I discuss in this report. 148. Dr. Dwyer’s AICC and BIC models contain 62 to 108 explanatory variables. A Chow test involves estimating models in which each explanatory variable is allowed to have a different coefficient in each relevant time period. In this case, that would involve estimating three coefficients (pre-alleged conspiracy, during alleged conspiracy, and postalleged conspiracy) for each variable, or 186 to 324 variables. Because each model is estimated based on only 144 variables, a model with more than 144 variables will, by necessity, fit the data perfectly. For this reason, I only estimate Chow tests based on Dr. Dwyer’s “10 PC” models (which contain only 18 or 20 explanatory variables). FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 73 of 94 PageID #:21888 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 72 - discussed, Dr. Dwyer’s model is unreliable because it is not stable. Even if I ignore this fundamental flaw, I find that his results also are unreliable because his (flawed) model is sensitive to reasonable changes. a. Dr. Dwyer’s Results are Sensitive to the Level of Aggregation He Uses to Measure Transaction Prices. 148. Dr. Dwyer assumes that the overcharge percentage is the same across all defendants. In particular, Dr. Dwyer estimates a single FCP model and a single ICP model that applies to all defendants, and uses those models to estimate damages for each defendant separately. For example, Dr. Dwyer claims that damages associated with GP’s sales are $671.32 million.149 But Dr. Dwyer estimates defendant-specific damages by assuming that the overcharge percentage he estimates for FCP sales and ICP sales are the same across all defendants. Dr. Dwyer could have evaluated this assumption, but did not do so.150 149. To test Dr. Dwyer’s assumption that the overcharge percentage for FCP and ICP sales is the same across defendants, I use Dr. Dwyer’s methodology (including his subjective decision to force the conspiracy indicator variable into his models) and data to estimate separate FCP and ICP models for each defendant (for the purpose of this analysis, I ignore the instability of Dr. Dwyer’s models).151 I find that Dr. Dwyer’s methodology, applied to GP’s data, produces a negative estimated overcharge for FCP using the AICC criterion. Using the BIC criterion, the stepwise selection procedure selects as many variables as there are monthly observations (and thus perfectly fits the data), further demonstrating the unreliability of Dr. 149. See Dwyer Report, Exhibit 12. 150. See Dwyer dep., at 114: 19 – 21 (“Q. You did not run a separate regression across the data for each defendant? A. That’s correct.”). 151. All of the FCP GP transactions in Dr. Dwyer’s analysis are from June 2004 and later. Thus, a GP-only FCP regression based on Dr. Dwyer’s data does not include a “preconspiracy” benchmark. Thus, in his aggregate FCP model, Dr. Dwyer is, in effect, estimating GP’s damages, in part, by comparing GP’s prices during the alleged conspiracy period to a benchmark that consists of GP’s (and other defendants’) prices in the alleged post-conspiracy and the prices of other defendants only in the alleged pre-conspiracy period. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 74 of 94 PageID #:21889 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 73 - Dwyer’s approach.152 For ICP, Dr. Dwyer’s methodology produces an estimated overcharge of 2.19 percent using the AICC criterion. Using the BIC criterion, the stepwise selection procedure again selects as many variables as there are monthly observations (and thus perfectly fits the data).153 See Table 8.   Table 8 Dr. Dwyer's Damage Regression Model Results GP Only Product AICC Final Containerboard Products Overcharge P‐value Regressors Observations Adjusted R‐squared ‐1.41% 0.000 53 103 99.8% Intermediate Containerboard Products Overcharge P‐value Regressors Observations Adjusted R‐squared 2.19% 0.000 65 131 99.9% Source: Dr. Dwyer's backup materials. Note: P‐values are based on HAC standard errors. 150. Similarly, I find that Dr. Dwyer’s methodology produces negative overcharges for SSCC sales of FCP (based on the BIC criterion).154 I also find that Dr. Dwyer’s methodology produces negative overcharges for Norampac sales of ICP (based on the BIC criterion). 151. I conclude that Dr. Dwyer’s models are sensitive to reasonable changes in the level of disaggregation and thus are not reliable. 152. I also estimate the GP FCP model in a different standard statistical software package – SAS instead of STATA – which uses a different version of the BIC criterion. The SAS version of the BIC criterion selects fewer variables, and produces a negative estimated overcharge (-1.32 percent). 153. The SAS version of the BIC criterion selects fewer variables for the GP ICP model, and produces an estimated overcharge of 0.27 percent. 154. Using the BIC criterion for the SSCC ICP model, the stepwise selection procedure selects as many variables as there are monthly observations (and thus perfectly fits the data). The SAS version of the BIC criterion selects fewer variables for the SSCC ICP model and produces a negative estimated overcharge. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 75 of 94 PageID #:21890 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 74 - b. Dr. Dwyer’s Results are Sensitive to the Definition of the Alleged Conspiracy Period. 152. Dr. Dwyer assumes that the relevant conspiracy period is February 15, 2004 through November 8, 2010.155 The original complaint in this case claimed that the conspiracy began, instead, in August 2005.156 To test the sensitivity of Dr. Dwyer’s results to the start date of the alleged conspiracy, I repeated his analysis with a conspiracy indicator that is “turned on” in August 2005 (again, for the purpose of my analysis, I ignore the instability of Dr. Dwyer’s models). Using this alternate definition of the alleged class period, I find that Dr. Dwyer’s methodology produces a negative estimated overcharge for both ICP models (AICC and BIC), and a negative estimated overcharge for the FCP model based on the AICC criterion. For the FCP model based on the BIC criterion, the estimated overcharge is only 0.24 percent (and not statistically significantly different from zero). See Table 9.  155. See Dwyer dep., at 89: 3 – 16 (“Q. For your class period, I just want to make sure I understand whether you have studied or considered any other period or just took it as an assumed period from February 15, 2004 through November 8, 2010. Can you tell me whether you did any analysis or any study of a different period or just took this as an assumption? A. I took that as an assumption. Q. If you are asked to re-run your models on a different proposed class period, you could do it; correct? [Objection omitted] A. Yes, I could.”). 156. See Kleen Products LLC v. Packaging Corporation of America, et al., Consolidated and Amended Complaint For Violation of the Sherman Act, ¶ 1. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 76 of 94 PageID #:21891 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 75 - Table 9 Dr. Dwyer's Damage Regression Model Results Original Alleged Conspiracy Period Product AICC Final Containerboard Products Intermediate Containerboard Products Overcharge BIC ‐0.86% 0.24% P‐value Regressors Observations Adjusted R‐squared 0.000 61 144 99.6% 0.3063 1 128 144 100.0% Overcharge P‐value Regressors Observations Adjusted R‐squared ‐4.22% 0.000 58 144 99.0% ‐5.26% 0.000 47 144 98.6% Source: Dr. Dwyer's backup materials. Note: P‐values are based on HAC standard errors. 1  P‐value based on Newey‐West standard errors. 153. Thus, Dr. Dwyer’s results are sensitive to whether the period February 15, 2004 to July 2005 is considered to be part of the alleged conspiracy. Thus, had plaintiffs not changed their alleged conspiracy period, Dr. Dwyer’s model would have found no damages for containerboard and sheets, and inconsistent results for corrugated products other than sheets (e.g., boxes), with a maximum (and not statistically significantly different from zero) overcharge of 0.24 percent. 154. In sum, Dr. Dwyer’s damages methodology is unreliable for numerous reasons. In particular, Dr. Dwyer’s damages methodology: (1) is based on statistical techniques not typically used in modern economic analysis; (2) does not apply those techniques consistently; if Dr. Dwyer had applied those techniques consistently, his results would have been materially different; (3) the results of Dr. Dwyer’s models are not stable and so his damage estimates are not reliable; (4) even if the instability of Dr. Dwyer’s models is ignored, Dr. Dwyer’s results are sensitive to reasonable changes and so are not reliable. FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 77 of 94 PageID #:21892 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 - 76 - September 18, 2014 __________________________ Dennis W. Carlton FILED UNDER SEAL -- CONTAINS INFORMATION SUBJECT TO PROTECTIVE ORDER Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 78 of 94 PageID #:21893 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 Exhibit 1 DENNIS WILLIAM CARLTON Senior Managing Director September 2014 Business Address: Compass Lexecon 332 South Michigan Avenue Chicago, Illinois 60604 Email Address: dcarlton@compasslexecon.com (312) 322-0215 EDUCATION Ph.D., MASSACHUSETTS INSTITUTE OF TECHNOLOGY, Cambridge, Massachusetts: Economics, 1975. M.S., MASSACHUSETTS INSTITUTE OF TECHNOLOGY, Cambridge, Massachusetts: Operations Research, 1974. A.B., HARVARD UNIVERSITY (Summa cum laude): Applied Math and Economics, 1972. EMPLOYMENT COMPASS LEXECON (formerly Lexecon), Chicago, Illinois (2008 – present) Senior Managing Director; LEXECON INC., (1977 – 2006), President 1997 – 2001, Senior Managing Director 2003 - 2006 UNIVERSITY OF CHICAGO, Booth School of Business, David McDaniel Keller Professor of Economics (2011 – present); Katherine Dusak Miller Professor of Economics (2008 – 2011); Professor of Economics (1984 – 2008); Law School, Professor of Economics (1980 – 1984); Department of Economics, Assistant Professor (1976 – 1979); Associate Professor (1979). U.S. DEPARTMENT OF JUSTICE, Washington, District of Columbia (2006 – 2008) Deputy Assistant Attorney General for Economic Analysis, Antitrust Division MASSACHUSETTS INSTITUTE OF TECHNOLOGY, Cambridge, Massachusetts, Department of Economics (1975 – 1976) Instructor in Economics OTHER PROFESSIONAL EXPERIENCE HARVARD UNIVERSITY, Public Policy Summer Course in Economics (1977), Professor BELL TELEPHONE LABORATORIES (Summers 1976, 1977) JOINT CENTER FOR URBAN STUDIES OF M.I.T. AND HARVARD UNIVERSITY, Cambridge, Massachusetts (1974 - 1975) CHARLES RIVER ASSOCIATES, Cambridge, Massachusetts (Summers 1971, 1972) Research Assistant Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 79 of 94 PageID #:21894 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 FIELDS OF SPECIALIZATION Theoretical and Applied Microeconomics Industrial Organization ACADEMIC HONORS AND FELLOWSHIPS Keynote Address, International Industrial Organization Conference, 2014 The 2014 Distinguished Fellow, Industrial Organization Society Award for Best Antitrust Economist, Global Competition Review, 2014 Keynote Address, Sixth Annual Federal Trade Commission Microeconomics Conference, 2013 Heath Memorial Lecture, University of Florida, 2013 Award (w. Mark Israel) for Best Antitrust Analysis in Litigated Cases, Global Competition Review, 2013 Keynote Address, 21st Annual Workshop of the Competition Law & Policy Institute of New Zealand, 2010 Keynote Address, Japanese Symposium on Competition, sponsored by Japan Fair Trade Commission, 2009 Recipient of Inaugural Robert F. Lanzilotti Prize, awarded by the Industrial Organization Society for Best Paper in Antitrust Economics, 2008 Keynote Address to Israel Antitrust Conference, 2008 Lewis Bernstein Memorial Antitrust Lecture, Washington, D.C., 2006 Distinguished Visitor, University of Melbourne, April 2005 Milton Handler Lecture, New York, 2004 Keynote Address to the International Competition Network, Mexico, 2004 Alexander Brody Distinguished Lecture, Yeshiva University, 2000 Ph.D. Thesis chosen to appear in the Garland Series of Outstanding Dissertations in Economics Recipient of the 1977 P.W.S. Andrews Memorial Prize Essay, best essay in the field of Industrial Organization by a scholar under the age of thirty National Science Foundation Grant, 1977 - 1985 Recipient of Post-doctoral Grant from the Lincoln Foundation, 1975 National Science Foundation Fellowship, 1972 - 1975 Phi Beta Kappa, 1971 John Harvard Award, 1970 Detur Book Prize, 1969 Edwards Whitacker Award, 1969 M.I.T., National Scholar Award, 1968 PROFESSIONAL AFFILIATIONS AND ACTIVITIES Appointed Member of the ABA Transition Task Force, Antitrust and Consumer Protection, 2012 Advisory panel to the Department of Justice and the FTC on the merger guidelines, 2010 Co-editor, Journal of Law and Economics, 1980 - present Visiting Committee, MIT, Department of Economics, 1995 - 2011 Member, Advisory Board, Economics Research Network, 1996 - present Member, Advisory Board of Antitrust and Regulation Abstracts, Social Science Research Network, 1998 - present Advisory Board, Massachusetts Institute of Technology, Department of Economics, 1999 - present Editorial Board, Competition Policy International (CPI), 2010 – present, Co-Editor, Competition Policy International (CPI), 2004 – 2009 Member, Economic Task Force – Antitrust Division, American Bar Association, 2010 – present Advisory Board, Journal of Competition Law and Economics, 2004- present Adjunct Scholar, American Enterprise Institute for Public Policy Research, 2007 – present -2- Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 80 of 94 PageID #:21895 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 Deputy Assistant Attorney General for Economic Analysis, Antitrust Division, U.S. Department of Justice, 2006 - 2008 Presidential Appointment to the Antitrust Modernization Commission, 2004 - 2007 Invited Panelist at Public Hearing on the Retail Banking Sector Inquiry: Payment Cards, before the European Commission in Brussels, Belgium, July 17, 2006. Consultant on Merger Guidelines to the FTC, 2003 Professor, George Mason Institute for Judges, October 2001 Chairman, FTC Round Table on Empirical Industrial Organization (September 11, 2001) Participant in the Round Table on the Economics of Mergers Between Large ILECS before the Federal Communications Commission, February 5, 1999 Member, Steering Committee, Social Science Research Council, Program in Applied Economics, 1997 - 1999 Participant in roundtable discussions on "The Role of Classical Market Power in Joint Venture Analysis," before the Federal Trade Commission, November 19, 1997 and March 17, 1998. Participant in meetings with Committee of the Federal Reserve on Payment Systems, June 5, 1997 Associate Editor, Regional Science and Urban Economics, 1987 - 1997 Resident Scholar, Board of Governors of the Federal Reserve System, Summer, 1995 Accreditation Committee, Graduate School of Business, Stanford University, 1995 Associate Editor, The International Journal of Industrial Organization, 1991 - 1995 Editorial Board, Intellectual Property Fraud Reporter, 1990 - 1995 Consultant on Merger Guidelines to the U.S. Department of Justice, 1991 - 1992 Member, Advisory Committee to the Bureau of the Census, 1987 - 1990 National Bureau of Economic Research, Research Associate Member, American Economics Association, Econometrics Society BOOKS Market Behavior Under Uncertainty, Ph.D. Thesis, Massachusetts Institute of Technology (September 1975); Garland Publishing (1984). Modern Industrial Organization, Scott, Foresman & Co., co-authored with Jeffrey Perloff, first edition (1990), (Chapter 17 of first edition reprinted as “The Economics of Information” for the University of Connecticut, Food Marketing Policy Center (1989)), second edition (1994), translated into Chinese, French, Hungarian and Italian; Addison Wesley Longman, third edition (2000), fourth edition (2005), translated into Chinese (2009). RESEARCH PAPERS "The Equilibrium Analysis of Alternative Housing Allowance Payments," (with Joseph Ferreira) Chapter 6 of Analysis of a Direct Housing Allowance Program, The Joint Center for Urban Studies of M.I.T. and Harvard University, (July 1975). "Theories of Vertical Integration," presented at Fourth Annual Telecommunications Conference. Appears in a volume of Proceedings of the Fourth Annual Telecommunications Conference, Office of Telecommunications Policy, (April 1976). "Uncertainty, Production Lags, and Pricing," American Economic Review, (February 1977). "Selecting Subsidy Strategies for Housing Allowance Programs," (with Joseph Ferreira) Journal of Urban Economics, (July 1977). -3- Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 81 of 94 PageID #:21896 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 "Peak Load Pricing With Stochastic Demand," American Economic Review, (December 1977). (Reprinted in Economic Regulation edited by P.L. Joskow, Edward Elgar Publishing Limited, 1998 and Reprinted in The Economics of Public Utilities edited by Ray Rees, Professor of Economics at the University of Munich, Germany, 2005.) "The Distribution of Permanent Income," Income Distribution and Economic Inequality, edited by Zvi Griliches, et al. (Halsted Press, 1978). "Vertical Integration--An Overview," in Congressional Record Hearings on the Communications Act of 1978. Bill H.R. 13105, (August 3, 1978). "Market Behavior with Demand Uncertainty and Price Inflexibility," American Economic Review, (September 1978). "Vertical Integration in Competitive Markets Under Uncertainty," Journal of Industrial Economics, (March 1979). Awarded the P.W.S. Memorial Prize for the best essay in the field of Industrial Organization by a scholar under the age of thirty. "Valuing Market Benefits and Costs in Related Output and Input Markets," American Economic Review, (September 1979). "Contracts, Price Rigidity and Market Equilibrium," Journal of Political Economy, (October 1979). "Why New Firms Locate Where They Do: An Econometric Model," in Studies in Regional Economics, edited by W. Wheaton, (Urban Institute, 1980). "Benefits and Costs of Airline Mergers: A Case Study," (with W. Landes and R. Posner) Bell Journal of Economics, (Spring 1980). (Reprinted in "Air Transport" in Classics In Transport Analysis series, edited by Kenneth Button and Peter Nijkamp, 2001.) "The Limitations of Pigouvian Taxes as a Long Run Remedy for Externalities," (with G. Loury) Quarterly Journal of Economics, (November 1980). "The Law and Economics of Rights in Valuable Information: A Comment," Journal of Legal Studies, (December 1980). "Price Discrimination: Vertical Integration and Divestiture in Natural Resources Markets," (with J. Perloff) Resources and Energy, (March 1981). "The Spatial Effects of a Tax on Housing and Land," Regional Science and Urban Economics, (November 1981). "Comments on Weicher," Journal of Law and Economics, (December 1981). Comment, in Sherwin Rosen ed. Studies in Labor Markets, University of Chicago Press, (1981). "Planning and Market Structure," in The Economics of Information and Uncertainty, edited by J.J. McCall, University of Chicago Press, (1982). "The Disruptive Effect of Inflation on the Organization of Markets," in Robert Hall, ed. The Economics of Inflation, University of Chicago Press, (1982). "The Need for Coordination Among Firms With Special Reference to Network Industries," (with J. M. Klamer) University of Chicago Law Review, (Spring 1983). -4- Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 82 of 94 PageID #:21897 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 "A Reexamination of Delivered Pricing," Journal of Law and Economics, (April 1983). "Futures Trading, Market Interrelationships, and Industry Structure," American Journal of Agricultural Economics, (May 1983). "The Regulation of Insider Trading," (with D. Fischel), Stanford Law Review, (May 1983), reprinted in J. Macey ed., Classics in Corporate Law and Economics, Edward Elgar Publishing (2008), reprinted in part in Roberto Romano, Foundations of Corporate Law, Oxford University Press (1993), Foundation Press (2010 forthcoming), and Corporate Law Series – Insider Trading, Edward Elgar Publishing (2011 forthcoming). "The Location and Employment Choices of New Firms: An Econometric Model with Discrete and Continuous Endogenous Variables," The Review of Economics and Statistics, (August 1983). "Economic Goals and Remedies of the AT&T Modified Final Judgment," (with W. Lavey), Georgetown Law Review, (August 1983). "Equilibrium Fluctuations When Price and Delivery Lags Clear the Market," Bell Journal of Economics, (Autumn 1983). "Energy and Location," Energy Costs, Urban Development, and Housing, Brookings Institution, (1984). "Futures Markets: Their Purpose, Their History, Their Growth, Their Successes and Failures," Journal of Futures Markets, (September 1984). (Reprinted in Futures Markets edited by A.G. Malliaris and W.F. Mullady, Edward Elgar Publishing Limited, 1995; and in Classic Futures: Lessons from the Past for the Electronics Age, edited by Lester Telser, Risk Books, 2000.) “The Economics of Gray-Market Imports,” (with C. DeMuth), written for the Coalition to Preserve the Integrity of American Trademarks (COPIAT), (May 1985). "The Limitation of Pigouvian Taxes As A Long Run Remedy for Externalities: Extension of Results," (with G. Loury) Quarterly Journal of Economics, (August 1986). "The Rigidity of Prices," American Economic Review, (September 1986). "The Theory and The Facts of How Markets Clear: Is Industrial Organization Valuable for Understanding Macroeconomics?" in Handbook of Industrial Organization, eds. Schmalensee and Willig, (1989). "Market Power and Mergers in Durable-Good Industries," (with R. Gertner), Journal of Law and Economics, (October 1989). “Comments on Vertical Integration and Market Foreclosure,” Brookings Papers on Economic Activity: Microeconomics, (1990). Book Review of Tirole's “The Theory of Industrial Organization”, Journal of Political Economy, (June 1990). "The Genesis of Inflation and the Costs of Disinflation: Comment," Journal of Money, Credit & Banking, (August 1991, Part 2). "The Theory of Allocation and its Implications for Marketing and Industrial Structure: Why Rationing is Efficient," Journal of Law and Economics, (October 1991). -5- Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 83 of 94 PageID #:21898 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 "The Economics of Cooperation and Competition in Electronic Services Network Industries," in Economics of Electronic Service Networks, Wildman Steven ed., Praeger Press, (1992). "Merger Policy and Market Definition Under the EC Merger Regulation," (with W. D. Bishop). Conference on Antitrust in a Global Economy, Fordham Corporate Law Institute, (1994). "The Antitrust Economics of Credit Card Networks," (with A. Frankel) Antitrust Law Journal, (Winter 1995). "Economic Organization and Conflict," Journal of Institutional and Theoretical Economics, (March 1995). "Antitrust and Higher Education: Was There a Conspiracy to Restrict Financial Aid?" (with G. Bamberger and R. Epstein) The Rand Journal of Economics, (Vol. 26, No. 1, Spring 1995, pp. 131-147). "The Competitive Effects of Line-of-business Restrictions in Telecommunications," (with K. Arrow and H. Sider), Managerial and Decision Economics, (Vol. 16, pp. 301-321, 1995). (Reprinted in Deregulating Telecommunications - The Baby Bells Case for Competition, edited by Richard S. Higgins and Paul H. Rubin, John Wiley & Sons Ltd., 1995.) "The Antitrust Economics of Credit Card Networks: Reply to Evans and Schmalensee," (with A. Frankel), Antitrust Law Journal, (Spring 1995). "Antitrust and Payment Technologies," (with A. Frankel), Review, Federal Reserve Bank of St. Louis (November/December 1995). "Antitrust Policy Toward Mergers When Firms Innovate: Should Antitrust Recognize the Doctrine of Innovation Markets?" Testimony before the Federal Trade Commission Hearings on Global and Innovation-based Competition (October, 1995). "You Keep on Knocking But You Can't Come In: Evaluating Restrictions on Access to Input Joint Ventures," (with S. Salop), Harvard Journal of Law & Technology, (Volume 9, Summer, 1996). (Reprinted in e-Commerce Antitrust & Trade Practices, Practicing Law Institute, 2001.) "Comments on Causes and Consequences of Airline Fare Wars," Brookings Papers on Economic Activity: Microeconomics, (1996). "A Critical Assessment of the Role of Imperfect Competition in Macroeconomics," in Market Behavior and Macro Economic Modeling, Brakman, Van Ees, & Kuipers (eds.), MacMillan Press (1997). "Price Rigidity," Business Cycles and Depressions, David Glasner ed., Garland Publishing, Inc., (1997). "Communication Among Competitors: Game Theory and Antitrust," (with R. Gertner and A. Rosenfield), George Mason Law Review, (1997). (Reprinted in e-Commerce Antitrust & Trade Practices, Practicing Law Institute, 2001.) "Comments on Born and Viscusi," Brookings Papers on Economic Activity: Microeconomics, (1998). "Antitrust and Higher Education: MIT Financial Aid (1993)," (with G. Bamberger), The Antitrust Revolution, in eds. J. Kwoka and L. White, (Oxford University Press, 3rd edition 1999). -6- Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 84 of 94 PageID #:21899 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 "Market Power and Vertical Restraints in Retailing: An Analysis of FTC v. Toys 'R' Us," (with H. Sider), The Role of the Academic Economist in Litigation Support, edited by Daniel Slottje, North Holland, (1999). “The Economics of Religion, Jewish Survival and Jewish Attitudes Toward Competition on Torah Education,” (with A. Weiss), Journal of Legal Studies, (2001). (Reprinted in Essential Readings on Jewish Identities, Lifestyles and Beliefs, edited by Stanford M. Lyman, Gordian Knot Books, 2003). “A General Analysis of Exclusionary Conduct and Refusal to Deal -- Why Aspen and Kodak are Misguided,” Antitrust Law Journal, (2001). (Reprinted in e-Commerce Antitrust & Trade Practices, Practicing Law Institute, 2001.) “The Lessons from Microsoft,” Business Economics, (January 2001). "Lessons from Halacha About Competition and Teaching," (with A. Weiss), Center for Business Ethics Social Responsibility, http://besr.org/library/competition.html, (March 2001). "The Choice of Organizational Form in Gasoline Retailing and The Costs of Laws Limiting that Choice," (with A. Blass), Journal of Law and Economics, (October 2001). Reprinted in Franchise Contracting and Organization, edited by Francine Lafontaine, Elgar Publishing, (2005). "Should The Merger Guidelines Be Scrapped? Introduction to a Debate," in Symposium On The Antitrust Analysis Of Mergers: Merger Guidelines vs. Five Forces, 33 U. WEST L.A. L. REV. (2001). "Free Riding and Sales Strategies for the Internet," (with J. Chevalier), The Journal of Industrial Economics, (December 2001). “The Strategic Use of Tying to Preserve and Create Market Power in Evolving Industries,” (with M. Waldman), The Rand Journal (Vol. 33, No. 2, Summer 2002). (Reprinted in B. Klein and A. Lerner eds. Economics of Antitrust Law, Edward Elgar Publishing Ltd, 2008, and Recent Developments in Monopoly and Competition Policy, The International Library of Critical Writings in Economics, edited by George Norman, Edward Elgar Publishing Ltd, 2008.) "The Competitive Effects of Fannie Mae," (with D. Gross and R. Stillman) in Housing Matters: Issues in American Housing Policy, Fannie Mae (January 2002, reprinted 2004). "Intellectual Property, Antitrust and Strategic Behavior," (with R. Gertner), in eds. Adam Jaffee and Joshua Lerner, Innovation Policy and the Economy, Volume 3, MIT Press (2003). "Airline Networks and Fares," (with G. Bamberger), Handbook of Airline Economics, 2nd ed., Darryl Jenkins, ed., McGraw Hill (2003). "Contracts that Lessen Competition -- What is Section 27 for, and How Has it Been Used?" (with David Goddard), in Mark N. Berry and Lewis T. Evans eds., Competition Law at the Turn of the Century: A New Zealand Perspective, Victoria University Press (2003). Interview, Economists’ Roundtable, Antitrust Magazine, (Spring 2003). “The Relevance for Antitrust Policy of Theoretical and Empirical Advances in Industrial Organization,” (Fall 2003), George Mason Law Review. -7- Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 85 of 94 PageID #:21900 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 "The Control of Externalities in Sports Leagues: An Analysis of Restrictions in the National Hockey League," (with A. Frankel and E. Landes), Journal of Political Economy, (February 2004), reprinted in Recent Developments in the Economics of Sport, edited by W. Andreff (forthcoming). "An Empirical Investigation of the Competitive Effects of Domestic Airline Alliances," (with G. Bamberger and L. Neumann), Journal of Law and Economics, Vol. 47, No. 1, (April 2004, pp. 195-222). “Why Barriers to Entry are Barriers to Understanding,” American Economic Review, (May 2004). “Using Economics to Improve Antitrust Policy,” Milton Handler Lecture, Columbia Business Law Review, (June 2004). “The Proper Role for Antitrust in an International Setting,” (Keynote address: Second Annual Conference of the International Competition Network (ICN), Merida City, Mexico (June 25, 2003), appears as Appendix to “Using Economics to Improve Antitrust Policy”, Columbia Business Law Review (June 2004). “Econometric Analysis of Telephone Mergers,” (with H. Sider) pp. 373-395 in American Bar Association, Econometrics: Legal, Practical, and Technical Issues, (2005). “How Economics Can Improve Antitrust Doctrine Towards Tie-in Sales,” (with M. Waldman), Competition Policy International, (Spring 2005). Preface to: “Law and Economics of the Mexican Competition Laws,” by Francisco Gonzalez de Cossio (2005). “Transaction Costs, Externalities and “Two-Sided” Payment Markets,” (with A. Frankel), Columbia Business Law Review, No. 3, Vol. (2005). “Predation and the Entry and Exit of Low-Fare Carriers,” (with G. Bamberger), in Advances in Airline Economics: Competition Policy and Antitrust, Darin Lee, ed., (2006). “Why Tie An Essential Good,” (with Michael Waldman), in Hahn R. ed., Antitrust Policy and Vertical Restraints, AEI-Brookings, (July 2006). “Market Definition: Use and Abuse,” Competition Policy International (Spring 2007) Interview with Deputy Assistant Attorney General, The Antitrust Source (February 2007) Separate Statement of Dennis W. Carlton, in The Report of the Antitrust Modernization Commission, (April 2007) “Does Antitrust Need to be Modernized?,” Journal of Economic Perspectives (Summer 2007) “The Year in Review: Economics at the Antitrust Division 2006-2007” (with K. Heyer), Review of Industrial Organization, (2007). “Economic Analysis of Competition Practices in the EU and the U.S.: A View from Chief Economists,” (with M. Salinger), Competition Policy International (Autumn 2007). “Merger Analysis,” Palgrave Dictionary, (with J. M. Perloff), (2008). -8- Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 86 of 94 PageID #:21901 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 “Tying,” (with M. Waldman), in W. Collins ed. Issues in Competition Law and Policy, American Bar Association, (2008). “Barriers to Entry,” in W. Collins ed. Issues in Competition Law and Policy, American Bar Association, (2008). “Product Variety and Demand Uncertainty: Why Mark-ups Vary with Quality,” (with James D. Dana Jr.), Journal of Industrial Economics (2008) “Regulation, Antitrust, and Trinko,” (With H. Sider), in eds. J. Kwoka and L. White, The Antitrust Revolution, (2008). “A Solution to Airport Delays,” (with W. Whalen, K. Heyer and O. Richard), Regulation (2008). “Should ‘Price Squeeze’ Be A Recognized Form of Anticompetitive Conduct?,” Journal of Competition Law & Economics (2008). “Safe Harbors for Quantity Discounts and Bundling,” (with M. Waldman), George Mason Law Review (2008). “Appropriate Antitrust Policy Towards Single-Firm Conduct: Extraction vs. Extension” (with K. Heyer),” Antitrust, (condensed version of subsequent paper), (Summer 2008). “Extraction vs. Extension: the Basis for Formulating Antitrust Policy Towards Single-Firm Conduct” (with K. Heyer), Competition Policy International, (Autumn, 2008). “Assessing the Anticompetitive Effects of Multiproduct Pricing,” (with P. Greenlee and M. Waldman), Antitrust Bulletin, (Fall, 2008). “The Need to Measure the Effect of Merger Policy and How to Do It,” Antitrust, (condensed version of subsequent paper), (Summer 2008). “How to Measure The Effectiveness of US Merger Policy,” http://voxeu.org/index.php?q=node/3344, (2009), and a slightly revised version appears as “Measuring the Effectiveness of US Merger Policy” in The Economists' Voice: Vol. 6: Iss. 7, Article 2, (2009). These are condensed versions of the subsequent paper. “Why We Need to Measure the Effect of Merger Policy and How to Do It,” Competition Policy International (Spring 2009). "Competition, Monopoly, and Aftermarkets," (with M. Waldman), Journal of Law, Economics and Organization, (April 2009). “Competition Policy: Beware of Using It to Harm Competition,” Fair Trade, Japan, (Spring, 2009). “Should Competition Policy Prohibit Price Discrimination?” (with M. Israel), Global Competition Review, (2009). “Merger Guidelines Revisited?” an interview, Antitrust, American Bar Association, (Fall 2009). “How Should Economic Evidence be Presented and Evaluated,” proceedings of the EU Competition Workshop, Florence, Italy, (June 2009). -9- Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 87 of 94 PageID #:21902 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 “Externalities in Payment Card Networks: Theory and Evidence: A Commentary,” The Changing Retail Payments Landscape: What Role for Central Banks?, Federal Reserve Bank of Kansas City (2010). “Why Tie a Product Consumers Do Not Use?,” (with J. Gans and M. Waldman), Recipient of Inaugural Robert F. Lanzilotti Prize, awarded by the International Industrial Organization Society for Best Paper in Antitrust Economics, (2008), American Economic Journal: Microeconomics (2010). “Mergers in Regulated Industries: Electricity,” in Competition Law and Economics: Advances in Competition Policy Enforcement in the EU and North America, A. Mateus and T. Moreira editors, (2010). “Financial Issues (Comments on Bankruptcy and Clearing Houses),” Chapter X in Competition as Public Policy, American Bar Association, (2010). “Revising the Horizontal Merger Guidelines,” Journal of Competition Law & Economics (2010), also appears in Journal of Competition Law – CADE (Brazil) Vol. 1, No. 23 (2011), also appears in Revista de Direito da Concorrencia, Conselho Administrativo de Defesa Economica, (Brazilian Government Publication), translated into Portuguese by T. Aranovich, p. 83-114 (2011). “Net Neutrality and Consumer Welfare,” (with G. Becker and H. Sider), Journal of Competition Law & Economics, (2010). “Will The New Guidelines Clarify or Obscure Antitrust Policy?,” (with M. Israel), The Antitrust Source, (2010). “Introduction to Stigler’s Theory of Oligopoly,” (with S. Peltzman), Competition Policy International, (2010). “Response to Gopal Das Varma’s Market Definition, Upward Pricing Pressure, and the Role of Courts: A Response to Carlton and Israel” (with M. Israel), The Antitrust Source, (2010). “Use and Misuse of Empirical Methods in the Economics of Antitrust,” 3(1) Competition Policy International Antitrust Chronicle, (2011). “The Economics of Patent Ambush,” Concurrences, New Frontiers of Antitrust, (2011). “Antitrust and Regulation,” (with R. Picker) in N. Rose ed., Economics of Deregulation, NBER, (forthcoming). “Proper Treatment of Buyer Power in Merger Review,” (with M. Israel), Review of Industrial Organization, (2011). “Upgrades, Switching Costs, and the Leverage Theory of Tying,” (with M. Waldman), Economic Journal, (2012). “Brantley Versus NBC Universal: Where’s the Beef?,” (with M. Waldman), Competition Policy International, (2012). “Buyer Power in Merger Review,” (with M. Coleman and M. Israel), Oxford Handbook of International Antitrust Economics, eds. R. Blair and D. Sokol, (forthcoming). “The Economics of Cartel Cases and the Use of Experts,” (with G. Bamberger and M. Israel), ABA Handbook of Cartel Enforcement, ed. S. Cherry, (forthcoming). - 10 - Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 88 of 94 PageID #:21903 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 “An Economic Interpretation of FRAND,” (with A. Shampine), Journal of Competition Law & Economics, (2013). “Economists’ Roundtable on Hot Patent-Related Antitrust Issues,” Antitrust Magazine, Vol. 27, No. 3, (Summer 2013). “Robert Bork's Contributions to Antitrust Perspectives on Tying Behavior," (with M. Waldman), Journal of Law & Economics, (forthcoming). “Patent Litigation, Standard Setting Organizations, Antitrust and FRAND,” (with A. Shampine), University of Texas Intellectual Property Law Journal, (forthcoming). “Identifying Benchmarks for Applying Non-Discrimination in FRAND,” (with A. Shampine), Competition Policy International, CPI Antitrust Chronicle, (August 2014). UNPUBLISHED PAPERS "Modeling the Housing Allowance Program," M.A. Thesis, Massachusetts Institute of Technology (September 1974). "The Cost of Eliminating a Futures Market and The Effect of Inflation on Market Interrelationships," (1984). "The Empirical Importance of Delivery Lags as an Explanation of Demand," (1984). "Statistical Supplement to The Antitrust Economics of Credit Card Networks: Reply to Evans and Schmalensee Comment, 63 Antitrust Law Journal 903 (1995)," (with Alan Frankel), (May 1997). “Antitrust and Not for Profits; The Case of Hospitals,” (with C. Capps, and G. David), mimeo (2010). EXPERT TESTIMONIAL EXPERIENCE (LAST FOUR YEARS) Declaration, Deposition, and Affidavit of Dennis W. Carlton in Re: Chicago Board Options Exchange, Incorporated, Dow Jones & Company, Inc., and the McGraw- Hill Companies, Inc., v. International Securities Exchange, LLC and the Options Clearing Corporation in the Circuit Court of Cook County, Illinois County Department, Chancery Division, No. 06 CH 24798, October 30, 2009 (Declaration), January 29, 2010 (Deposition), March 26, 2010 (Affidavit). Responses of Dennis Carlton in Re: Verizon Horizontal Merger Guidelines Review Project to the U.S. Department of Justice and Federal Trade Commission, November 9, 2009. Declaration and Reply Declaration of Dennis W. Carlton and Gary S. Becker in the matter of Preserving the Open Internet Broadband Industry Practices, before the Federal Communications Commission in Washington D.C., GN Docket No. 09-191, January 14, 2010 (Declaration), April 7, 2010 (Reply Declaration). Declaration and Reply Declaration of Dennis W. Carlton and Hal Sider in Re: Special Access Rates for Price Cap Local Exchange Carriers, AT&T Corp. Petition for Rulemaking to Reform Regulation of Incumbent Local Exchange Carrier Rates for Interstate Special Access Services, before the Federal Communications Commission in Washington D.C., WC Docket No. 05-25, January 19, 2010 (Declaration), February 24, 2010 (Reply Declaration). - 11 - Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 89 of 94 PageID #:21904 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 Report and Reply Report of Dennis W. Carlton and Gustavo E. Bamberger in Re: Air Canada. Toronto, Canada. February 5, 2010 (Report), May 18, 2010 (Reply Report). Comments of Dennis W. Carlton on Department of Justice and Federal Trade Commission’s Proposed Horizontal Merger Guidelines, June 4, 2010. Declaration and Expert Disclosure Statement of Dennis W. Carlton in Re: Light Cigarettes Marketing and Sales Practices Litigation, in the United States District Court, District of Maine, MDL Docket No. 1-09-MD-2068, January 28, 2010. Declaration and Expert Disclosure Statement of Dennis W. Carlton in Re: Karen L. Lawrence, et. al., v. Philip Morris, Inc., before the Superior Court of the State of New Hampshire, No. 09-C-518, June 28, 2010. Report of Dennis W. Carlton and Gustavo E. Bamberger in Re: Vector Limited, New Zealand, August 20, 2010. Expert Report, Rebuttal Expert Report, and Deposition of Dennis W. Carlton in Re: Delta/Airtran Baggage Fee Antitrust Litigation in the United States District Court for the Northern District of Georgia, Atlanta Division, Civil Action File Number 1:09-Md-2089-Tcb, January 7, 2011 (Expert Report), February 4, 2011 (Rebuttal Expert Report), February 24, 2011 (Deposition). Economic Analysis and Review of Economic Literature of Dennis W. Carlton and Gustavo E. Bamberger in Re: LAN/TAM Merger South America, April 12, 2011 (Economic Analysis), May 12, 2011 (Review of Economic Literature). Statement, Supplemental Statement, Reply Statement, Reply of Dennis W. Carlton to Expert Report, Deposition, Testimony, and Declaration in Re: ActiveVideo Networks, Inc. v. Verizon Communications Inc. et al.: In the U.S. District Court for the Eastern District of Virginia, No. 2:10cv248 (RAJ/FBS), April 20, 2011 (Statement), May 6, 2011 (Supplemental Statement), May 11, 2011 (Reply Statement), May 23, 2011 (Reply to Expert Report), May 26, 2011(Deposition), July 25-26, 28, 2011 (Testimony), August 24, 2011 (Declaration), November 2, 2011 (Declaration). Declaration, Reply Declaration of Dennis W Carlton, Hal Sider and Allan Shampine, in Re: the Merger of AT&T with T-Mobile: Before the Federal Communications Commission, WT Docket No. 11-65, April 20, 2011 (Declaration), June 9, 2011 (Reply Declaration). Expert Report, Expert Response Report, Expert Reply Report, and Depositions of Dennis W. Carlton in Re: DSM Desotech INC. v. 3D Systems Corporation and 3D Systems, Inc. in the United States District Court for the Northern District of Illinois, Docket No. 08C1531, May 13, 2011 (Expert Report), June 24, 2011 (Expert Response Report), August 5, 2011 (Expert Reply Report), August 26, 2011 (Deposition), October 28, 2011 (Response), November 14, 2011 (Deposition). Expert Report of Dennis W. Carlton in Re: Israel Antitrust Authority v. Bank Hapoalim, Bank Leumi, Israel Discount Bank, et al., May 31, 2011. Reply Expert Report, and Deposition of Dennis W. Carlton in Re: HTC Corporation et al. v. IPCOM GMBH & Co., KG in the United States District Court District of Columbia, Case No. 1:08-cv01897- RMC, June 24, 2011 (Reply Expert Report), August 5, 2011 (Deposition). - 12 - Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 90 of 94 PageID #:21905 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 Expert Reports, Deposition, Sur-Reply Report, and Testimony of Dennis W. Carlton in Re: TFT-LCD (Flat Panel) Antitrust Litigation, in the United States District Court Northern California San Francisco Division, No. C07-1827 SI, MDL No. 1827, July 22, 2011 (Expert Report Direct Purchaser Class Action), July 28, 2011 (Expert Report Indirect Purchaser Class Action), August 17, 2011 (Deposition), September 6, 2011 (Sur-Reply Report Indirect Purchaser Class), September 12, 2011 (Sur-Reply Report Direct Purchaser Class), February 23, 2012 (Expert Report Concerning Motorola Mobility Inc.), February 23, 2012 (Expert Report Concerning Nokia Corporation and Nokia Inc.), February 23, 2012 (Expert Report Concerning Dell Inc. and Dell Products L.P.), February 23, 2012 (Expert Report Concerning Eastman Kodak Company), February 23, 2012 (Expert Report Concerning ATS Claim, LLC; AT&T Mobility LLC et al.; Best Buy Co., Inc. et al.; Costco Wholesale Corporation; Electrograph Systems, Inc. et al.; Target Corp. et al.), April 5, 2012 (Testimony Concerning Costco Wholesale Corp v. Samsung Electronics America, Inc., Samsung Electronics Co., Ltd., and Samsung Semiconductor, Inc.) April 10, 11, and 12, 2012 (Deposition Concerning the Direct Purchaser Class), June 25, 2012 (Testimony Concerning the Direct Purchaser Class), February 22, 2013 (Expert Supplemental Report Concerning the Best Buy Co., Inc., et al. v. Toshiba Corporation, et al, matter), April 4, 2013 (Deposition Concerning the Best Buy Co., Inc., et al. v. Toshiba Corporation, et al, matter), May 13, 2013 (Direct Testimony and Expert Sur-Reply Report Concerning the Costco Wholesale Corporation v. Toshiba Corporation, et al, matter), October 23, 2012 (Deposition) October 24, 2012 (Testimony Concerning Costco Wholesale Corporation v. Sharp Electronics Corporation, and Sharp Corporation), May 20, 2014 (Expert Report Concerning the State of Washington v. Au Optronics Corporation, et al, matter), August 20, 2014 (Expert Report Concerning the State of Oregon v. Au Optronics Corporation, et al, matter). Summary of Conclusions, Expert Report, Reply Report, Expert Rebuttal Report, and Deposition of Dennis W. Carlton in Re: American Airlines, Inc., vs. Travelport Inc., Sabre, Inc., Sabre Holdings Inc., and Sabre Travel International Ltd. in the District Court of Tarrant County, 67th District Court, Cause No. 67-249214-10, August 5, 2011 (Summary of Conclusions), April 18, 2012 (Expert Report-State), July 9, 2012 (Reply Report), July 25, 2012 (Expert Report-Federal), August 27, 2012 (Expert Rebuttal Report-State), September 12, 2012 (Deposition), October 4, 2012 (Affidavit). Expert Report of Dennis W. Carlton in Re: The 2003 NPM Adjustment Proceedings Before An Arbitration Panel Pursuant To Section XI(c) Of The Master Settlement Agreement, December 19, 2011. Report of Dennis W. Carlton and Mary T. Coleman, to the Korea Fair Trade Commission, January 10, 2012. Submission, Supplemental Submission of Dennis Carlton, Charles Augustine and Gustavo Bamberger on behalf of Meridian Energy to the New Zealand Electricity Authority, February 23, 2012 (Submission), March 8, 2012 (Supplemental Submission). Expert Reports, Reply Expert Report, and Deposition of Dennis W. Carlton in Re: Apple Inc. and NeXT Software Inc., v. Motorola Inc. and Motorola Mobility Inc., March 1, 2012 (Expert Report), March 6, 2012 (Expert Report), April 15, 2012 (Reply Expert Report), August 15, 2012 (Deposition). Expert Report of Dennis Carlton in Re: The Commissioner of Competition v. VISA Canada Corporation and MasterCard International Incorporated, the Competition Act, R.S.C. 1985, c. C34, March 14, 2012. Statement of Dennis W. Carlton in the Matter of Certain Wireless Communication Devices, Portable Music and Data Processing Data Devices, Computers and Components Thereof, United States International Trade Commission, Investigation No. 337-TA-745, July 9, 2012. - 13 - Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 91 of 94 PageID #:21906 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 Expert Report of Dennis W. Carlton to the European Commission in Re: Optical Disk Drives, Case No. COMP/39639, October 28, 2012. Expert Report and Deposition of Dennis W. Carlton in Re: Rail Freight Fuel Surcharge Antitrust Litigation, MDL No. 1869, Misc. No. 07-489 (PLF), (Expert Report) January 22, 2013, (Deposition) May 14, 2013. Reply Declaration of Dennis W. Carlton and Allan L. Shampine in Re: Special Access for Price Cap Local Exchange Carriers: Before the Federal Communications Commission, WC Docket No. 0525, March 12, 2013. Expert Report of Dennis W. Carlton in Re: Aerotec International, Inc., v. Honeywell International, Inc., In the U.S. District Court for the District of Arizona Case 2:10-CV-00433-PHX-JWS, April 5, 2013. Expert Report of Dennis W. Carlton in Re: The Commonwealth of Virginia v. McKesson Corporation, Case No. C 11-02782 SI, July 15, 2013. Expert Report and Deposition of Dennis W. Carlton in Re: Mylan Pharmaceuticals, Inc., Rochester Drug Co-Operative, Inc., Meijer, Inc., Meijer Distribution, Inc., American Sales Company, LLC, Walgreen Co., Safeway Inc., Supervalu Inc., and Heb Grocery Co. LP, et al., v. Warner Chilcott Public Limited Company, et al., In the United States District Court for the Eastern District of Pennsylvania, Civ. No. 12-3824, October 18, 2013 (Expert Report), December 20, 2013 (Deposition). Expert Report of Dennis W. Carlton in Re: United States of America, et al. v. US Airways Group Inc., et al., in the United States District Court for the District of Columbia, Civil Action No. 13-1236CKK, November 8, 2013. Expert Report and Depositions of Dennis W. Carlton in Re: Polyurethane Foam Antitrust Litigation, in the United States District Court for the Northern District of Ohio, MDL Docket No. 2196, Index No. 10-MD-2196 (JZ), May 15, 2014 (Expert Report), June 11-12, 2014 (Deposition). Whitepaper on Patent Licenses Negotiated Subject to Judicial Review, submitted to the Chinese NDRC on behalf of Qualcomm, with Allan Shampine, May 16, 2014. Report of Dennis W. Carlton and Allan L. Shampine in Re: Microsoft Corporation and Nokia Corporation’s Violation of Merger Regulations: Before the Korea Fair Trade Commission, Case No. 2014GiGuel1474, July 21, 2014. Reply Expert Reports and Depositions of Dennis W. Carlton in Re: Cathode Ray Tube (CRT) Antitrust Litigation, in the United States District Northern District of California San Francisco Division, Case No. 07-5944 SC, MDL No. 1917, August 5, 2014 (Reply Expert Report Concerning the Other Direct Action Plaintiffs), August 5, 2014 (Reply Expert Report Concerning the Sharp Electronics Corp, et al. v. Hitachi Ltd. et al. and Sharp Electronics Corp. et al v. Koninklijke Philips Elecs. N.V., et al, matter), in the Superior Court of the State of California County of San Francisco, Case No. CGC-11-515784, August 5, 2014 (Reply Expert Report Concerning the State of California, et al. v. Samsung SDI, Co., Ltd. et al, matter), August 5, 2014 (Reply Expert Report Concerning the Dell Inc., and Dell Products L.P., v. Hitachi, Ltd. et al, matter), September 5, 2014 (Reply Expert Report Concerning the Viewsonic Corporation v. Chunghwa Picture Tubes, Ltd., et al. matter), (Depositions), September 16, 17, 2014. - 14 - Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 92 of 94 PageID #:21907 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94     Exhibit 2 Materials Relied Upon Kleen Products LLC, et al. v. Packaging Corporation of America, et al., Consolidated and Amended Complaint for Violation of the Sherman Act Kleen Products LLC, et al. v. Packaging Corporation of America, et al., [Proposed] Amendments to Plaintiffs’ Consolidated and Amended Complaint Plaintiffs’ Memorandum of Law in Support of Motion For Class Certification Declaration of Michael J. Harris, Ph.D. in Support of Plaintiffs’ Motion for Class Certification and Exhibits Report of Mark Joseph Dwyer, Ph.D. in Support of Plaintiffs’ Motion for Class Certification and Exhibits Deposition of Mark J. Dwyer, Ph.D. in Kleen Products LLC, et al., v. Packaging Corporation of America, et al., August 12, 2014 Deposition of Michael J. Harris, Ph.D. in Kleen Products LLC, et al., v. Packaging Corporation of America, et al., July 22, 2014 Deposition of Mollie Hilliard and Exhibits in Kleen Products LLC, et al., v. Packaging Corporation of America, et al., December 12, 2013 Corrections to the Deposition Transcript of Mollie Hilliard in Kleen Products LLC, et al., v. Packaging Corporation of America, et al., January 16, 2014 Deposition of Travis Ballard and Exhibits in Kleen Products LLC, et al., v. Packaging Corporation of America, et al., August 14-15, 2013 Corrections to the Deposition Transcripts of Travis Ballard in Kleen Products LLC, et al., v. Packaging Corporation of America, et al., October 8, 2013 Dennis W. Carlton, Robert H. Gertner and Andrew M. Rosenfield, “Communication Among Competitors: Game Theory and Antitrust,” 5 George Mason Law Review 423 (1997) George J. Stigler (1964), “A Theory of Oligopoly,” Journal of Political Economy, 72(1) Haizheng Li and Jifeng Luo, “Industry Consolidation and Price in the US Linerboard Industry,” Journal of Forest Economics 14 (2008) 93 – 115 Peter Kennedy, A Guide to Econometrics, 3rd ed. http://www.google.com/search?q=definition+ostensibly&gws_rd=ssl http://www.risiinfo.com/Marketing/OBM-MAR/RISI_ctb_price_plan_final.pdf - 1   Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 93 of 94 PageID #:21908 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94     Spreadsheet of Containerboard Capacities, By Firm AFPA00001314 AFPA00001929 AFPA00002792 AFPA00002824 AFPA00002856 - 2   Case: 1:10-cv-05711 Document #: 759 *SEALED* Filed: 09/20/14 Page 94 of 94 PageID #:21909 Case: 15-2385 Document: 57-10 Filed: 10/27/2015 Pages: 94 UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF ILLINOIS EASTERN DIVISION KLEEN PRODUCTS LLC, et al., individually and on behalf of all others similarly situated, Plaintiffs, Civil Case No. 1:10-cv-05711 v. Hon. Harry D. Leinenweber PACKAGING CORPORATION OF AMERICA, et al., Defendants. DECLARATION OF DENNIS W. CARLTON The undersigned, Dennis W. Carlton, makes this declaration based on personal knowledge and pursuant to 28 U.S.C. § 1746. I hereby state as follows: 1. My name is Dennis W. Carlton, and I have been retained by Georgia-Pacific LLC with regard to the above-referenced action. 2. The foregoing “Expert Report of Dennis W. Carlton” (“Expert Report”) submitted in connection with the above-referenced action is incorporated as if fully set forth herein. 3. I declare, under penalty of perjury, under the laws of the United States of America, that the foregoing (including my Expert Report) is true and correct. Executed on this 19th day of September 2014. ______________________________ Dennis W. Carlton