United States v. At&t Inc.

Opinion

UNITED STATES DISTRICT COURT FOR THE DlSTRICT OF COLUMBIA

UNITED STATES OF AMERICA, ) Plaintiff, § v. § Civil Case No. 17-2511 (RJL) AT&T INC., et al., § Defendants. § F I L E D h/_ JUN 1 2 2018 MEMORAND_UM OPINION @9.'3§‘§ ?t?ili‘%‘l鑧l§§?'é’§ll¥l§¥a

(June LZ, 2018)

If there ever were an antitrust case where the parties had a dramatically different assessment of the current state of the relevant market and a fundamentally different vision of its future development, this is the one. Small wonder it had to go to trial l

On November 20, 2017, the U.S. Department ofJustice’s Antitrust Division brought this suit, on behalf of the United States of America (“the Government” or "the plaintiff’), to block the merger of AT&T lnc. (“AT&T”) and Time Wamer lnc. (“Time Warner”) as a violation of Section 7 of the Clayton Act, 15 U.S.C. § 18. The Government claims, in essence, that permitting AT&T to acquire Time Warner is likely to substantially lessen competition in the video programming and distribution market nationwide by enabling AT&T to use Time Warner’s “must have” television content to either raise its rivals’ video programming costs or, by way of a “blackout,” drive those same rivals’ customers to its subsidiary, DirecTV. Thus, according to the Government, consumers nationwide will be

harmed by increased prices for access to Turner networks, notwithstanding the

Government’s concession that this vertical merger would result in hundreds of millions of dollars in annual cost savings to AT&T’s customers and notwithstanding the fact that (unlike in “horizontal” mergers) no competitor will be eliminated by the merger’s proposed vertical integration

Not surprisingly, the defendants, AT&T, Time Warner, and DirecTV, strongly disagree Their vision couldn’t be more different. The video programming and distribution market, they point out, has been, and is, in the middle of a revolution where high-speed internet access has facilitated a "veritable explosion” of new, innovative video content and advertising offerings over the past five years. Trial Tr. (“Tr.”) 139711-4 (l\/lontemagno (Charter)). Vertically integrated entities like Netflix, Hulu, and Amazon have achieved remarkable success in creating and providing affordable, on-demand video content directly to viewers over the internet. l\/Ieanwhile, web giants Facebook and Google have developed new ways to use data to create effective - and lucrative - digital advertisements tailored to the individual consumer.

As a result of these “tectonic changes” brought on by the proliferation of high-speed internet access, video programmers such as Time Warner and video distributors such as AT&T find themselves facing two stark realities: declining video subscriptions and flatlining television advertising revenues. Ia’. at 3079: 18 (Bewkes (Time Warner)). lndeed, cost-conscious consumers increasingly choose to “cut” or “shave” the cord, abandoning their traditional cable- or satellite- TV packages for cheaper content alternatives available over the internet. At the same time, Facebook’s and Google’s dominant digital advertising

platforms have surpassed television advertising in revenue. Watching vertically integrated,

data-informed entities thrive as television subscriptions and advertising revenues declined, AT&T and Time Warner concluded that each had a problem that the other could solve: Time Warner could provide AT&T with the ability to experiment with and develop innovative video content and advertising offerings for AT&T’s many video and wireless customers, and AT&T could afford Time Warner access to customer relationships and valuable data about its programming Together, AT&T and Time Warner concluded that both companies could stop “chasing taillights” and catch up with the competition. 2/16/18 Hr’ g Tr. 34116 [Dkt # 67]. Those were the circumstances that drove AT&T, a distributor of content, and Time Warner, a content creator and programmer, to announce their historic $108 billion merger in October 2016 (the “proposed merger” or "challenged merger”). Those are the circumstances that cause them to claim today that their merger will increase not only innovation, but competition in this marketplace for years to come.

Section 7 ofthe Clayton Act assigns this Court the “uncertain task” ofweighing the parties’ competing visions of the future of the relevant market and the challenged merger’s place within it. United Slates v. Baker Hughes lnc., 908 F.2d 981, 991 (D.C. Cir. 1990). Nothing less than a comprehensive inquiry into future competitive conditions in that market is expected. And the Government has the burden of proof to demonstrate that the merger is likely to lessen competition substantially in that uncertain future.

Since announcing the transaction in late October 2016, defendants have delayed closing on the merger agreement for about 18 months as a result of the Government’s investigation and suit. The deal is now set to expire if not consummated on or before June

2 l, 2018 - a turn of events that would require AT&T to pay Time Warner a “break-up fee”

of $500 million. The parties have engaged in a highly accelerated discovery schedule to prepare themselves to try this case in March and April of this year. The trial itself lasted nearly six weeks. Both sides put on a case-in-chief and the Government put on a rebuttal case as well. At the conclusion ofthe trial, l advised the parties I would issue a ruling, if not an opinion, no later than June 12, 2018 so that the losing side would have the agreed- upon time remaining to pursue its appellate rights before the merger or the $500 million break-up fee went into effect.

The following islthe Court’s Opinion. Initially, l provide context for this suit by reviewing the background of the video programming and distribution industry, the proposed merger, and the procedural history of this case. Thereafter, l discuss the legal standards governing a suit under Section 7 of the Clayton Act, emphasizing in particular the considerations at play in evaluating vertical mergers. With that in place, l next analyze each of the Government’s three theories of harm to competition, balancing, as appropriate, the conceded proconsumer benefits ofthe merger with the consumer harms alleged and the evidence offered to support them. Ultimately, I conclude that the Government has failed to meet its burden to establish that the proposed “transaction is likely to lessen competition substantially.” Baker Hughes, 908 F.2d at 985.

As such, based on that conclusion, and for all the reasons set forth in greater detail in this Opinion, the Court DENIES the Government’s request to enjoin the proposed

merger.

TABLE OF CONTENTS

BACKGROUND ............................................................................................................................ 7 l. The Video Programming and Distribution lndustry .................................................................... 7 A. Video Programing and Distribution ..................................................................................... 8 1. Programmers ....................................................................................................................... 8 2. Distributors ....................................................................................................................... ll 3. Affiliate Negotiations and “Blackouts” ............................................................................ 14 B. lndustry Trends ................................................................................................................... 18 1. Rise and Innovation of Over-the-Top, Vertically lntegrated Video Content Services ..... 18

2. Declining l\/IVPD Subscriptions Resulting from an Increasingly Competitive lndustry Landscape ......................................................................................................................... 21 3. Shift Toward Targeted, Digital Advertising ..................................................................... 25 Il. The Parties and Proposed l\/Ierger ............................................................................................ 28 A. AT&T ................................................................................................................................. 28 B. Time Warner ....................................................................................................................... 30 1. Turner Networks ............................................................................................................... 31 2. HBO .................................................................................................................................. 34 C. The Proposed l\/lerger ......................................................................................................... 36 111. Procedural History .................................................................................................................. 40 A. The Investigation ................................................................................................................ 40 B. Pretrial Proceedings ............................................................................................................ 40 1. The Complaint .................................................................................................................. 40 2. Turner’s Arbitration Commitment .................................................................................... 41 3. Pre-Discovery Timeline .................................................................................................... 42 4. Discovery .......................................................................................................................... 42 5. Discovery Disputes ........................................................................................................... 44 6. Evidentiary Disputes ......................................................................................................... 46 C. The Trial ............................................................................................................................. 47 IV. Legal Standard ........................................................................................................................ 50 A. The Clayton Act ................................................................................................................. 50 B. Baker Hughes Burden Shifting Framework ....................................................................... 53 C. Antitrust Analysis of Vertical l\/lergers .............................................................................. 55

ANALYSIS ................................................................................................................................... 59 l. Market Definition ....................................................................................................................... 61 ll. Conceded Consumer Benefits of Proposed l\/Ierger ................................................................. 66

lll. The Government Has Failed to l\/leet lts Burden to Show That the Proposed l\/lerger ls Likely to Substantially Lessen Competition by lncreasing Turner’s Bargaining Leverage in Affiliate Negotiations ............................................................................................................ 68

A. Background of lncreased-Leverage Theory of Harm ......................................................... 71

`1. B. The Government’s So-Called “Real-World Objective Evidence” ls lnsufficient to Support lts lncreased-Leverage Theory of Harm ............................................................................ 74

1. Evidence Regai'ding the Popularity of Turner Content ls of Limited Probative Value in Evaluating the Contention That Turner Will Gain lncreased Leverage Due to the Proposed Merger ............................................................................................................... 75

2. Defendants’ Own Statements and Documents Provide Little Support for the Contention That Turner Will Gain lncreased Leverage Due to the Proposed l\/lerger ........................ 79

3. Third-Party Competitor Witness Testimony Provides Little Support for the Contention That Turner Will Gain lncreased Leverage Due to the Proposed l\/lerger ........................ 91

4. Real-World Evidence lndicating That Prior Veitical lntegration of Programmers and Distributors Has Not Affected Affiliate Fee Negotiations Undermines the Government’s lncreased-Leverage Theory of Harm ................................................................................ 99

C. The Government’s Expert Testimony ls Also lnsufficient to Support lts Increased- Leverage Theory of Harm ................................................................................................ 109

1. The Evidence ls insufficient to Support Professor Shapiro’s Conclusion That the l\/lerger Will Increase Turner’s Bargaining Leverage and, in Turn, Affiliate Fees ..................... 1 11

2. The Evidence ls lnsufficient to Support the lnputs and Assumptions lncorporated into Professor Shapiro"s Bargaining l\/lodel ........................................................................... 1 18

lV. The Government Has Failed to l\/leet lts Burden to Show That the Proposed l\/lerger ls Likely to Substantially Lessen Competition on the Theory That AT&T Will Act to Harm Virtual MVPDS Through lts Ownership of Time Warner Content .................................... 150

V. The Government Has Failed to l\/leet lts Burden to Show That the Proposed l\/lerger ls Likely to Substantially Lessen Competition on the Theory That AT&T Will Restrict Distributors’ Use of HBO as a Promotional Tool ............................................................... 165

CONCLUSION ........................................................................................................................... 170

BACKGROUND I. The Video Programming and Distribution lndustryl The structure of the video programming and distribution industry generally resembles the “three-stage chain of production comprised of manufacturers, wholesalers, and retailers that typifies the distribution of many, if not most, physical goods in the U.S.

economy.’ Christopher S. Yoo, Vertz`cal lntegratz`on and Media Regu[ation in the New Economy, 19 Yale J. Reg. 171, 220 (2002). Here, that three-stage chain of production and distribution involves "content creation, content aggregation, and content distribution.” Proposed P`indings of l~`act ofthe United States (“Gov’t PF()F”) H 8 [Dkt. # 128].2 Television content begins at the manufacturing level. Although video programming is often created by studios (such as Time Warner’s Wamer Bros.), some networks or distributors “produce content for themselves” or, in the case of live sporting events, license the rights to broadcast the events from the various sports leagues. See Tr. 80112-16

(Fenwick (Cox)). At the second level, programmers (such as Time Warner’s Turner or

Home Box Office ("HBO”)) aggregate content into a network or network group and then

l For consistency throughout this Opinion, l will use the phrase “video programming and distribution industry” to include creation, packaging, and distribution of professionally produced video content Of particular relevance here, the Court’s definition of "video programming and distribution industry” encompasses programmers such as Turner; traditional multichannel video programming distributors (“l\/lVPDs") such as cable and satellite companies; virtual multichannel video programming distributors (“virtual MVPDS” or “vl\/lVPDs”) such as DirecTV Now and DISH’s Sling; and subscription video on demand services (“SVODS”) such as Netflix, Hulu, and Amazon Prime. By contrast, l use the phrase “pay-TV” to refer only to the packaging and delivery of linear _ or “live” - television content That phrase encompasses only l\/iVPDs and virtual MVPDS.

2 Many materials before the Court contain confidential business information or other proprietary data; such submissions were typically filed under seal with an accompanying redacted version accessible to the public. The Court has made great effort to refrain from quoting or otherwise including confidential business information in this Opinion, opting instead to refer generally to the exhibits or information filed under seal.

license those networks to video distributors, like AT&T’s DirecTV. See, e.g., id. at 80:4- 9; Plaintiff`s Exhibit (“PX”) 456-4 to 10. At the third level, distributors bundle and distribute networks to their subscribers Tr. 80:4-9 (Fenwick (Cox)).

Some subscription-based video programming services are “vertically integrated," meaning, in this context, that those services create or aggregate their content offerings and then distribute those offerings directly to consumers [d. at 3081:18-25 (Bewkes (Time Warner)); see Defs.’ Proposed Findings of Fact (“Defs.’ PFOF”) 11 12 [Dkt. # 120]. Examples of those services include Netflix, Hulu, and Amazon Prime. Tr. 3155:22-23 (Bewkes (Time Warner)). Traditional video programmers, such as Turner, generally lack such “soup to nuts” integration of content creation and distribution; they are instead reliant upon video distributors to deliver their content offerings to consumers Id. at 3388:6-7 (Stephenson (AT&T)); See id. at 485:1-486:6, 612:17-20 (l\/lartin (Turner)). Because the Government’s claims center on the proposed combination of Time Warner’s video programming with AT&T’s video distribution, my background review focuses on those facets of the video programming and distribution industry.

A. Video Programing and Distribution

1. Programmers

Traditional programmers, such as Turner, acquire and aggregate video content. Ia’. at 80:4-16 (Fenwick (Cox)). Generally, programmers do not offer their content directly to consumers See, e.g., id. at 485:1-486:6, 612:1-20 (l\/lartin (Turner)). lnstead, they package video content into networks - in Turner’s case, networks such as TNT, TBS, and CNN - and then license the rights to display those networks to video distributors PX459-18; Tr.

80:6-9 (Fenwick (Cox)). As such, Turner and its programming competitors may be thought of as content “wholesaler[s]” in that they are typically reliant upon third-party video distributors to get their offerings to consumers in the downstream market. Tr. 612:3-4, 17- 20 (l\/lartin (Turner)).

Most programmers make money in two primary ways, and Turner is no exception First, programmers receive payments from distributors, known as “affiliate fees,” in exchange for granting distributors the rights to display the programmers’ content. See, e.g., id. at 604:21-23, 610:20-23. Affiliate fees are memorialized in affiliate agreements, which specify the “net effective rate" a programmer charges for a network on a per-subscriber, per-month basis. Id. at 987:5-17 (Breland (Turner)). Rates typically increase year-over- year, pursuant to what are called “escalator” clauses Ia’. at 91 :6-10 (Fenwick (Cox)); ial at 2728:19-23 (Katz). Affiliate fees have been “going up” over the past decade industrywide, due at least in part to rising costs of making “higher quality” content. Tr. 2562:9-2563:25 (Carlton); cf. id. at 1495:12-16 (Sutton (HBO)) (“So the cost it takes`to make shows, shows like the shows we make, has escalated significantly.”). Affiliate fees vary, however, based on the size ofthe distributor; specifically, in order to incentivize and reward wide distribution, programmers typically provide distributors with “volume discounts” on affiliate fees, meaning that the more subscribers a distributor has, the more a programmer’s net effective rates will decline. See, e.g., id. at 987:25-988:13 (Breland (Turner)) (explaining variance in rates between “small,” “medium,” and “large” MVPDs and virtual MVPDS); id. at 291 1121-23 (Holanda (RCN)) (describing “‘volume discounts”

of larger distributors); PX127-2 (showing rate differentials). 9

Secona’, and as any television viewer can attest, programmers sell advertising slots on their networks to advertisers See Tr. 3179:23-3181:6 (Bewkes (Time Warner)). For decades, television advertising has followed the same playbook. See id. at 3086:9-10. During each hour of television, there are roughly eighteen minutes of advertisements See id. at 609:23-610:4 (l\/lartin (Turner)). Distributors sell advertisements for only two of those minutes; the programmer sells ads for the remaining sixteen minutes See id. Advertising fees vary by the channel and the time of day an ad airs Id. at 625:4-1 l. As with affiliate fees, the broader a program’s audience, the more advertising revenue for Turner: as Chairman and CEO John Martin explained with regard to Turner’s advertising strategy, “our goal is to have our networks in front of as many eyeballs as possible." Id. at 605:7-8.

The classic model oftelevision advertising is limited in two ways First, in deciding the placement of commercials to be seen by a wide audience, programmers generally must rely on general demographic data, such as age range, about the typical audience for a given program. See id. at 625:4-6. Second, and as a result, programmers have no choice but to saturate all viewers of a program with the same, undifferentiated ads - despite knowing that the selected ad will be of little interest to some number of those viewers See id. at 3087:1-8 (Bewkes (Time Warner)).

ln the past, Turner’s total revenues have been split roughly equally between affiliate fee revenues and advertising revenues See id. at 3088:10-12; PX456-8. For present purposes, however, the key point is this: both the affiliate fee and advertising revenue

streams depend upon broad distribution of programmers’ networks to consumers See, e.g.,

Tr. 604:17-18 (l\/lartin (Turner)) (“l believe that distribution is the most important variable for success for any programmer."); id. at 3078:17-20 (Bewkes (Time Warner)) ("Q: What are the key drivers ofthe Turner business? A: Well, the Turner business, first, we need to get it on every distribution platform so that we can have subscriber fees and advertising revenues”). For that reason, Turner executives aim to "achieve wide distribution” of their networks Post-Trial Brief ofthe United States (“Gov’t Post-Tr. Br.”) 6 [Dkt. # 126]; see also, e.g., Tr. 3120:3-7 (Bewkes (Time Warner)); (“So we try everything to stay on all of our channels, Turner, HBO, everything, to keep them on there. And that`s very important to us lfthey’re not on there, we’re not only losing the subscriber fees; we’re losing the advertising revenues.”); cf. id. at 90:1-2 (Fenwick (Cox)) (“We are dealing with network

groups where their goal [is] a hundred percent distribution.").

2. Distributors Today, there are three categories of key players in the distribution of professionally produced video content: (1) “traditional” multichannel video programming distributors (“l\/lVPDs”); (2)“virtual” MVPDs; and (3) subscription video on demand services (“SVODs”). See Tr. 485:1-487:13 (l\/lartin (Turner)); Gov’t PFOF 1111 9, 14, 19. Fz`rst, there are traditional MVPDs. Those distributors include direct broadcast satellite providers, such as DlSH or AT&T’s DirecTV; cable television providers, such as

Comcast,3 Charter Communications (“Charter”), or Cox Communications (“Cox”);

3 ln 2009, Comcast announced its intent to acquire ownership ofNBCUniversal (“NBCU”), a media and entertainment company that owns the NBC and Telemundo networks as well as Universal Pictures and Universal Studios. See Compl. 1111 16-17, United States v. Comcast Corp., 808 F. Supp. 2d 145 (D.D.C. 201 1) (No. 1 1-cv-106) [Dkt. # 1]. Although the Government, through the Antitrust Division, filed an action

“overbuilders,” such as RCN; or “telcos,” such as AT&T’s U-verse or Verizon Fios. See Gov’t PFOF 1111 9, 43-45; Defs’ PFOF 1 34. All ofthose services offer live - or "linear” - television content as well as libraries of licensed content available for viewing on demand, typically in exchange for a monthly subscription fee. See Tr. 81:1-82:8 (Fenwick (Cox)); id. at 471:12-16, 638:16-22 (Martin (Turner)); id. at 1185:22-1186:1 (Warren (Turner)). Satellite distributors such as DirecTV and DlSH operate nationally, whereas cable companies telcos, and overbuilders distribute video content regionally; in any given local area, however, the incumbent cable operator is typically the dominant MVPD. See id. at 408;1-3 (Schlichting (DISH)). Consumers’ choices of traditional MVPDs are therefore dictated by geography. See id. at 2187:3-23 (Shapiro); Gov’t PFOF iii 43-46. Consumers often subscribe to traditional MVPDS as part of a “bundle” of various services, which may include, for example, a single price offering for cable, wireless internet, and home or

mobile phone services See Tr. 2784:21-25 (Rossi). Of the approximately 90 million

claiming that the transaction would violate Section 7 ofthe Clayton Act - an action, as fortune would have it, also assigned to this Court - the Government also urged me to approve the transaction pursuant to a final judgment containing various “remedies” that it represented would “diminish[] Comcast’s ability to use [NBCU’s] programming to harm competition.” Competitive lmpact Statement 3, 7, Comcast Corp., 808 F. Supp. 2d 145 [Dkt. # 4]. Those remedies related to procedures set forth in a related FCC order governing the transaction, including, as especially relevant here, requirements that Comcast-NBCU: 1) submit to “baseball style arbitration,” at the distributor’s option, in the event the parties were unable to reach a carriage agreement, 7/27/11 Hr’g Tr. 7:4-7 [Dkt. # 38], and 2) “continue to provide” video programming to the distributor “pursuant to the terms of any existing agreement until the arbitration is completed,” Competitive lmpact Statement Ex. A, at 24, Comcast Corp., 808 F. Supp. 2d 145 [Dkt. # 4-1]. At a hearing to discuss the proposed final judgment, counsel for the Government asserted that, “especially in cases of vertical mergers, conduct remedies” such as the ones proposed “can be a very useful tool to address the competitive problems while preserving competition and allowing efficiencies” that “may result from the transaction.” 7/27/1 1 Hr’g Tr. 15:16-21. Ultimately, l approved the Government’s proposed finaljudgment with a few modifications to allow me to better monitor the implementation ofthe remedies imposed as part ofthejudgment. See generally Comcast Corp., 808 F. Supp. 2d 145. The transaction proceeded and today Comcast-NBCU operates as a “vertically integrated” programmer and distributor. See Tr. 882:14-16 (Rigdon (Comcast)).

American households that still receive television content from providers in the pay-TV industry, a substantial majority do so through traditional MVPDs. See Gov’t PFOF 1111 9, 13. That number is steadily declining however, as consumers shift towards lower-cost virtual MVPDS or SVODs. See Tr. 3450:7-14 (Stephenson (AT&T)); id. at 3157:5-13 (Bewkes (Time Warner)).

Secona’, there are virtual MVPDS, which began to arrive in the marketplace in early 2015. See id. at 235:18-22 (Schlichting (DlSH)). Like traditional MVPDS, virtual MVPDS distribute linear channels and on-demand content to subscribers for a subscription fee; unlike traditional MVPDs, virtual MVPDs offer their services over the internet. rather than through proprietary infrastructure such as satellite networks or cable lines Gov’t PFOF 1111 14, 15. Because they offer their services over the internet, virtual MVPDs offer service nationwide, either via the web or mobile apps. See PX8-18. Examples of virtual l\/IVPDs include DirecTV Now, DlSH’s Sling, Sony’s Playstation Vue, Hulu Live, Google’s YouTube TV, FuboTV, and Philo. See id. at 18-19; Defs.’ PFOF 11 8. As their names suggest, some virtual MVPDS are associated with companies that operate traditional MVPDS. Each virtual MVPD competes with traditional MVPDs for subscribers and, increasingly, virtual MVPDs are gaining market share on traditional MVPDS due in part to their ease of use and lower-cost offerings See, e.g., Tr. 448:24-449:2 (Schlichting (DISH)); id. at 607:17-20 (Martin (Tumer)); id. at 1829:3-12 (Merrill (AT&T)). Therefore, despite their relatively recent vintage, virtual MVPDs already have millions of subscribers

See id. at 2019:20-2020:18 (Bond (NBCU)).

Thz`ra’, there are SVODs, a category that includes Netflix, Hulu, and Amazon Prime. SVODs generally do not offer live, linear programming such as live sporting events or live news See id. at 487:1-16 (Martin (Turner)). lnstead, they have large libraries of original and acquired content, accessible by a viewer on demand at any time. See ial. at 486:12-17. The leading SVODS are vertically integrated and invest billions of dollars in creating original programming See ia’. at 3081:13-25 (Bewkes (Time Warner)); ia’. at 3388:8-9 (Stephenson (AT&T)). By way of example, Netflix alone spends more on content than all of Time Warner. See i`a’. at 2456:13-14 (Carlton); see also ia’. at 1053:2-7 (Breland (Turner)) (Netflix will spend "almost 558 billion” on content “[t]his year”). As with virtual MVPDs, SVODS offer low-cost subscription plans as compared to traditional MVPDs and continue to gain market share in the video programming and distribution industry. lndeed, while traditional l\/IVPDs are losing subscribers at a steady clip, Netflix added 2 million

subscribers in the last quarter alone. See ia’. at 3450:1 1-12 (Stephenson (AT&T)).

3. Affiliate Negotiations and “Blackouts”

As previously discussed, the schemes under which programmers extend licensing rights to MVPDs and virtual MVPDs are governed by detailed contracts known as affiliate agreements See PX456-8; Tr. 80:4-9 (Fenwick (Cox)); ial. at 485:1-486:6 (Martin (Turner)). Those agreements describe the precise rights granted by the programmer, and contain numerous terms and conditions See, e.g., PX409. Although the “rate"` or payment amount is an important feature of any affiliate agreement, Tr. 90:5-10 (Fenwick (Cox)), "these deals are complicated” and “start with a hundred plus open issues,” ia'. at 459:24-

25. (Schlichting (DISH)); see also id. at 1690:23-25 (York (AT&T)) (“There’s literally 14

hundreds ofitems that go on kind ofa priority list on what’s the right deal."). Those issues can include digital rights, "windows” (z'.e., limitations on when certain content can be aired), “TV Everywhere” rights (i.e., the rights for subscribers to access content away from home on an authorized device), volume discounts, and penetration rate requirements among others See, e.g., ia’. at 90:5-14, 101:19-23 (Fenwick (Cox)); Gov’t PFOF 1111 11, 105; PX409-14. At least in the case of Turner, affiliate agreements also include inost- favored-nation (“l\/IFN”) clauses, which generally require the programmer to extend to the distributor certain types of terms given to another distributori See Tr. 1024:6-14 (Breland (Turner)) (describing l\/lFNs). Affiliate agreements run "between five and eight years on average.” Icl. at 87:9-11 (Fenwick (Cox)).

Because wide distribution maximizes programmers’ two income streams ~ affiliate fees and advertising revenue » programmers like Turner bargain for terms aimed at promoting that distribution To start, Turner seeks to license “every network” it owns Ia'. at 606:6-8 (Martin (Turner)).4 ln addition, Turner negotiates for guarantees of particular “penetration rates” - the percentage of a given distributor’s subscribers who receive a given channel. Ia’. at 1023:10-16 (Breland (Turner)).

Given the duration of the contract and the rights at issue, a single affiliate agreement

can dictate the transfer of upwards ofa billion dollars between programmer and distributor.

4 That said, in the case of DlSH’s virtual MVPD, Sling, Turner did license only its "core"’ networks - CNN, TBS, TNT, and Cartoon Network. See Tr. 236:23-24 (Schlichting (DISH)). As Time Warner CEO Jeff Bewkes testified, because 85 to 90% of Turner’s revenue comes from four networks, Turner is well situated to offer skinnier bundles See id. at 3126:22-3127:3 (Bewkes (Time Warner)); See also ia’. at 584:18-24 (Martin (Turner))(same). By contrast, NBCU’s revenues are spread more evenly across its more than one-dozen networks See id. at 3127:6 (Bewkes (Time Warner)).

See, e.g., PX144-21, 48. lt is thus no surprise that witnesses described affiliate agreement negotiations as “very tough" and “intense and aggressive.” Tr. 1022:25-1023:2 (Breland (Turner)); id. at 3251:24-25 (Stankey (AT&T)); see also Gov’t PFOF 11 104. Although the negotiations themselves typically last several months, closing a deal often “come[s] down to the last day and sometimes the last handful of minutes” Tr. 1093:14-16 (Breland (Turner)); see also z`a’. at 87:14-19 (Fenwick (Cox)). Negotiations involving programmers with multiple networks, such as Turner, are particularly “time consuming.” Id. at 87:17 (Fenwick (Cox)).

Affiliate negotiations are also idiosyncratic, varying from programmer to programmer and distributor to distributor. The Government’s chief economic expert, Professor Carl Shapiro, recognized as much at trial. Noting that “bargaining is a dark art in many ways,” Professor Shapiro acknowledged that negotiations may turn on myriad “unpredictable factors,” including the “personalities” at the table and other "hairy stuff`." [cl. at 2213:12, 2294:18-2295:6 (Shapiro). This dynamic flows from the “multitude" of considerations that inform each negotiation [d. at 1690:15 (York (AT&T)). With so many factors and priorities, and with such high stakes, it should be no surprise that terms and conditions vary across affiliate agreements See, e.g, z`a’. at 1681 : 16-17 (“[W]e do hundreds ofdeals, and we have hundreds of flavors ofmost favored nations.”). ln short, as Professor Shapiro explained, “the real world is messy and it’s imperfect.” Ial. at 2210:22-23 (Shapiro). 4

Sometimes, negotiations between programmers and distributors reach an impasse.

lfa negotiation is ultimately unsuccessful, the distributor will lose the rights to display the

programmer’s content to its customers - a situation known in the industry as a programming “blackout,” or “going dark.” See la’. at 129:4-9 (Fenwick (Cox)). Blackouts have negative consequences for programmers and distributors alike. On the programming side, a blackout causes a programmer to suffer immediate (and unrecoverable) losses of both advertising and affiliate fee revenue. See, e.g., l`a’. at 1094:21-1096:18 (Breland (Turner)). On the distributor side, a blackout may lead a distributor to lose subscribers or may prevent the distributor from attracting new subscribers See Gov’t PFOF 11 1 19; see also, e.g., id. at 864:12-23 (Rigdon (Comcast)); ial. at 1348:3-7 (Montemagno (Charter)) (discussing PX373). Because blackouts are almost always negative events for both programmers and distributors, “at the end of the day . . . [t]here’s no benefit for anyone to walk away” without an affiliate agreement Tr. 89:23-90:4 (Fenwick (Cox)). Therefore. bargains between programmers and distributors are almost always struck in order to avoid long-term blackouts See id. at 138:13-15; ia’. at 1027:4-7 (Breland (Turner)); z`al. at 1359:14-15 (l\/lontemagno (Charter)); icl. at 3124:4-7 (Bewkes (Time Warner)).

That is not to say, however, that blackouts are irrelevant to the negotiating dynamic. Rather, in what can best be thought of as an elaborate and stylized Kabuki dance, the evidence shows that “almost every negotiation” involves both programmers and distributors threatening blackouts, especially when one side is seen as demanding terms that are out ofline with the market. Ial. at 1026:17-20 (Breland (Turner)); cf ia’. at 376:22- 377:11 (Schlichting (DISH)). To better understand how to assign the “right value” to a particular deal, programmers and distributors might perform “drop” or “go dark” analyses

to estimate the potential impact of a blackout on the programmer’s advertising or affiliate

fee revenues or on the distributor’s customer base. Ia’. at 1343:11-16 (l\/lontemagno (Charter)); see also z'a’. at 1348:3-10 (discussing PX373); id. at 862:19-863:3 (Rigdon (Comcast)); ia’. at 1029:10-1030:11 (Breland (Turner)) (discussing PX144).

Nevertheless, given the negative consequences for both sides from a blackout, “the reality" is that “virtually every"l bargaining impasse between a programmer and distributor “is resolved after requiring either no blackout or a short-term blackout."" Ia'. at 2396:1-5 (Shapiro). Indeed, in recent memory, Turner networks have been blacked out only twice, both for roughly one-month periods See ia’. at 2357:15-23; Defs.’ PFOF 1111 139-143. Permanent blackouts, the evidence shows, are a vanishingly rare occurrence; the record indicates that Turner has never engaged in a long-term blackout with a distributor. See Tr. 2394:8-11 (Shapiro) (acknowledging that “in the real world there has never been a permanent blackout of the Turner networks”).

B. lndustry Trends

ln recent years, traditional programmers, including Turner, and MVPDsr including DirecTV, have been faced with a number of interrelated industry trends that are particularly

relevant to the challenged merger. l will review three of those trends in turn.

1. Rise and lnnovation of Over-the-Top, Vertically lntegrated Video Content Services

Traditional programmers and distributors are experiencing increased competition from innovative, over-the-top content services, including virtual MVPDs and SVODs See infra p. 24 n.5. Those web-based companies are harnessing the power ofthe internet and

data to provide lower-cost, better-tailored programming content directly to consumers The

dramatic growth ofthe leading SVODS in particular, including Netflix, Hulu, and Amazon Prime, can be traced in part to the value conferred by vertical integration ~ that is, to having content creation and aggregation as well as content distribution under the same roof. See, e.g., Tr. 3080:8-3085:21 (Bewkes (Time Warner)).

As relevant to the video programming and distribution market, vertical integration provides two notable advantages to content services First, vertical integration reduces the "bargaining friction” inherent in the arm’s-length affiliate negotiations that govern the exchange of rights between traditional programmers and distributors See, e.g., z`cl. at 3104:18-3107:13; ia’. at 1684:25-1685:13 (York (AT&T)). As numerous witnesses discussed, bargaining friction refers to the difficulty inherent in assigning value to and negotiating over new, innovative content rights, like “TV Everywhere,” download rights, and “4K” high resolution See ia’. at 1685:22-1686:7, 1688:6-13 (York (AT&T)); 1`a’. at 3104:18-25 (Bewkes (Time Warner)); z'a’. at 3222:4-3223:2 (Stankey (AT&T)). AT&T executive Daniel York testified, for example, that DirecTV has attempted, with limited success (and considerable delay), to obtain such rights from programmers through arm’s length-negotiations See id. at 1685:24-1686:22 (York (AT&T)). RCN CEO Jim Holanda joined York in discussing the way in which bargaining friction hindered RCN`s negotiations over TV Everywhere rights See ia’. at 2968:25-2971:14 (Holanda (RCN)). Further, DirecTV Now’s affiliate agreements require it to restrict the number of viewers who can stream or access programs simultaneously on its platform. See lcl. at 1687:10-14 (York (AT&T)). And when DirecTV floated the concept of "DirecTV mobile" - a pay-

TV subscription exclusively for mobile devices - that was “dead on arrival."’ Icl. at

1687:15-25. By contrast, with control over the creation and use of large amounts of original content, SVODS have driven much of the recent innovation in the video programming and distribution industry. See ia’. at 1685:7-13; ia’. at 639:1-8 (Martin (Turner)). Thesecompanies have, for example, developed download rights, allowing users to view their content anywhere without wireless access See z`a’. at 1688116-18 (York (AT&T)).

Second, and relatedly, SVODS’ ability to distribute their content directly to consumers over the internet gives them superior access to customer data. SVODs are able to use that customer data to inform their strategy and improve the customer’s experience in a number of ways See l`a’. at 3081:21-25 (Bewkes (Time Warner)); see also z`a’. at 3388:6-3389:8 (Stephenson (AT&T)). SVODS can use data about viewing habits to determine what programs are popular, and create more of that type of content. See la’. at 2452:21-2453:3 (Carlton); ia’. at 3245:16-20 (Stankey (AT&T)). ln addition, data informs marketing decisions, and allows SVODs to recommend content to users based on their revealed preferences z'.e,, the shows they have watched in the past. See ia’. at 3080:19- 3081:12 (Bewkes (Time Warner)). Even more, data can inform scheduling choices, and enhance efforts at recapturing consumers who disconnect. See z`a’. at 3245:16-20 (Stankey (AT&T)); ia’. at 3081:4-12 (Bewkes (Time Warner)). Finally, and as discussed in more detail below, to the extent SVODs incorporate advertising into their platforms, data allows

those ads to be more targeted and thus more lucrative.

2. Declining MVPD Subscriptions Resulting from an lncreasingly Competitive lndustry Landscape

At trial, witness after witness acknowledged that MVPD subscriptions are on the decline. See, e.g., id. at 633:5-15 (Martin (Turner)); ia’. at 891:18-22 (Rigdon (Comcast)); 2229:21-22 (Shapiro); 3369:13-16 (Stankey (AT&T)); id. at 3450:15-3451:1 (Stephenson (AT&T)); see also PX63-36. Those declines "started faster” than many in the industry anticipated Tr. 3369:13-16 (Stankey (AT&T)) (discussing the “inflection change"` where the “decline of the traditional pay-TV bundle started faster than [AT&T] assumed”). To illustrate, in 2016 AT&T’s traditional MVPDS lost 133.000 customers', last year, DirecTV alone lost 1.2 million subscribers See ia’. at 3004:6-8 (Christopher (AT&T)); z`a’. at 3450:7- 9 (Stephenson (AT&T)).

The decline in traditional MVPD subscriptions is just one symptom of the increasingly competitive nature of the video programming and distribution industry. lndeed, several witnesses testified that competition in the industry is more intense today than ever before. See, e.g., ia’. at 1398:24-25 (l\/lontemagno (Charter)) (video distribution business is “more competitive now than l’ve ever experienced in my career"’); z`a’. at 2134:1- 3 (Sejen (Cable ONE)) (“Q: In your 31 years in the industry, have you ever seen it more competitive at the distribution level‘? A: No.”); [a’. at 2950:2-6 (Holanda (RCN)) (“Q: And so in the course ofthis 30 years that you have been in the business, the video distribution market today is more competitive than at any point that you can recall, true? A: True.”);

id. at 3213:9 (Stankey (AT&T)) (competition in industry is “at an all-time high”); ia’. at

2476:1-9 (Carlton) (“new entrants” in market such as “Netflix” are “making the market more competitive”).

l\/lore specifically, the decline of traditional MVPD subscriptions reflects the growing popularity of virtual MVPDs and SVODs See, e.g., PX153-3. On that score, two rising trends are worth noting: cord-cutting and cord-shaving A household “cuts the cord" when it discontinues MVPD services altogether. whether traditional or virtual MVPDs See id. at 605:23-606:4 (Martin (Turner)); id. at 2505:10-20 (Carlton). As Professor Carlton relayed, SNL Kagan estimates that roughly twenty percent of American households have cut the cord, discontinuing traditional MVPD services Id. at 2505:12- 20. This number, high as it is, continues to grow, See ia’. at 2466:4-10; see also z`a’. at 891:18-22 (Rigdon (Comcast)); cf. ia’. at 2948:20-2949:3 (Holanda (RCN)). That said, those households have not exited the entertainment field altogether. See id. at 3450:2-6, 12-14 (Stephenson (AT&T)). lnstead, many have gravitated to vertically integrated SVODs. See PX153-3; see also Tr. 3449:12-24, 3450:7-12 (Stephenson (AT&T)). Consumers, particularly young people, find SVODs attractive, with their improved user

interfaces, premium content, and lower price points See, e.g., Tr. 639:1-8 (Martin

(Turner)); id. at 3449:12-18 (Stankey (AT&T)). On a similar note, a household “shaves the cord” when it departs a traditional MVPD for one of the many virtual MVPDs, which, again, typically carry smaller bundles of networks at lower price points Gov’t PFOF 11 16; Defs’ PFOF 11 21. l\/lany other consumers have shaved the cord, reducing, but not eliminating, their consumption of MVPD services See, e.g., Tr. 606:2-4 (Martin (Turner)).

Consumers intent on shaving the cord have an increasing array of virtual MVPD services

from which to choose - services that operate nationwide over the internet. See z`al. at 2949:15-18 (Holanda (RCN)). Consumers may choose to subscribe to a less expensive, “skinny bundle,” i.e., one with fewer networks, and then supplement that bundle with subscriptions to SVODS like Netflix and Hulu. Cf id. at 2984:13-20 (SEALED); ial. at 3506:24-3507:2 (Stephenson (AT&T)).

Ofcourse, when a household departs a traditional MVPD, whether for an SVOD or a virtual MVPD, that subscriber loss affects the traditional MVPD in the form of lost margins on subscription fees See, e.g., PX456-56; Tr. 3450:7-14 (Stephenson (AT&T)); ia'. at 2219:13-21 (Shapiro). Such losses may also affect programmers in the form of declining affiliate fee revenues as well as stagnating or declining viewership. See, e.g., 1`a’. at 3088:22-3089:1 (Bewkes (Time Warner)) (SVODs and other new competitors are "bleeding away our viewers”); PX153-3. Turner, for example, projects that its domestic subscription revenue growth will decrease to low single digits in each year from 2018 to 2022. See Tr. 647:3-11 (Martin (Turner)) (discussing Defendants’ Exhibit (“DX") 781- 21). lncreased competition from SVODs also means that more original, high-quality programming is being produced - a trend that increases the costs of securing the talent and rights necessary to make such programming See ia’. at 1494:15-21, 1495:12-16 (Sutton (HBO)) (“There was a time when very few people were making the kind of shows we make. Now, it seems that almost every week, there’s an announcement of somebody else making it. . .. [A]s l’ve mentioned, Netflix; llqu makes shows and so does Priine Video. . . . So the cost it takes to make shows, shows like the shows we make, has escalated

significantly” because “more people are bidding for the talent involved.”); PX153-6; cf.

Tr. 633:16-18 (Martin (Turner)) (“[T]he number of professionally produced television shows in the United States has doubled in the last five years alone.”).5

lt is therefore no surprise that programmers and distributors alike have noted the competitive threat posed by SVODs. After all, as Nobel laureate Bob Dylan correctly observed: “You don’t need a weatherinan to know which way the wind blows." Subterranean Homesick Blues. At trial, numerous witnesses from defendants testified that SVODs present a broad-range of competitive challenges See, e.g., Tr. 3088:22-3089:25 (Bewkes (Time Warner)) (over-the-top companies are “‘bleeding away our viewers because they’re offering competitive video that has these advantages because they know what to put in front of you individually, and we don’t”); ia’. at 3213:3-9, 3214:8-10 (Stankey (AT&T)) (“The time-and-attention competition now from the likes of Facebook, from the likes ofGoogle, from the likes ofNetflix . . . . l started asking myself, what should the business do to respond to the changing environment that we’ve heard about in this courtroom, the dawn of these new services coming from the likes [of] Netflix and Google?”). Third-party witnesses from AT&T’s competitor distributors also testified to the role of SVODs in the increasingly competitive industry landscape. See id. at 860:24- 861 :9 (Rigdon (Comcast)) (“[A]n SVOD service like in Netflix provides a wide array of

entertainment choices So people have limited time in the day. So where they"re going to

5 Although the Government asserts that “consumers ofMultichannel Video Distribution are largely insensitive to price changes"’ as reflected by their continued payment of increased subscription costs, Gov`t PFOF 1135, at trial there was near-uniform testimony that “consumers are up to here with subscription prices” and that “it’s getting harder and harder” for distributors to pass their increased costs along, Tr. 3089:6-1 l (Bewkes (Time Warner)); id. at 3446:1-4 (Stephenson (AT&T)). That consumers are at a "gag point” when it comes to traditional MVPD subscription costs is further illustrated by the continued decline in subscriptions nationally. ld. at 140:13-15 (Fenwick (Cox)); id. at 3450:7-9 (Stephenson (AT&T)).

spend their time for entertainment in that respect Netflix competes with traditional TV providers.”); id. at 1395:12-21 (l\/lontemagno (Charter)) (Charter’s competitors include "the Googles and the Amazons and the Netflix”); see also DX921-35 (DlSH "face[s] significant competition” from other companies including, among others “Netflix, Hulu,

Apple, Amazon, Alphabet . . . ."').6

3. Shift Toward Targeted, Digital Advertising Finally, and again as a result of the rising influence of innovative, web-based competitors the advertising landscape has shifted away from reliance on television advertising offered by programmers to highly-targeted digital advertising See Tr. 3088:3-

6 (Bewkes (Time Warner)) (noting that advertisers are shifting their “ad budgets which

6 ln the face ofall that, the Government continues to insist that SVODs are merely “complement[s]" or “adjunct[s]” to traditional MVPDS, rather than competitors of traditional MVPDS; a few of the Government’s third-party competitor witnesses testified to the same. Gov’t PFOF 11 36. l agree with defendants that the Government’s arguments (and the corresponding witness testimony) on that score defy reality, as demonstrated by the evidence adduced at trial. The evidence clearly showed that the leading SVODS - as vertically-integrated entities that produce and distribute their own award-winning content ~ fiercely compete both with programmers such as Turner and HBO and with traditional MVPDs and virtual MVPDs. lndeed, industry data refiects that large percentages of MVPD customers have chosen to “cut the cord” and receive content exclusively from SVODs. See supra pp. 22-24.

To be sure, the Government contends that, notwithstanding the increasing prevalence of SVODs, “[e]ven programmers believe MVPDs are likely to remain highly profitable in the future.” Gov’t PFOF11 13. That proposition rests on a document from l\/Iay 2016. Id. (citing PX78). As the Court learned at trial, however, the industry has undergone significant changes since mid-2016, diminishing the persuasiveness ofthat statement and others like it. To takejust one example7 video programming margins are declining, a fact that presents an obvious threat to future MVPD profitability Tr. 3853:18-19 (Shapiro) (“l think it is not disputed that the video margins are going down.”). And while the Court accepts that traditional MVPDs continue to have a substantial subscriber base, and indeed may currently constitute a distinct submarket, see infra pp. 61-66, it is inescapable that SVODS have played a large role in causing the demand for and continued purchase of traditional MVPD subscriptions to "declin[e] at a rapid pace.” Tr. 3450:7-3451:1 (Stephenson (AT&T)). To ignore those industry trends - trends that are transforming how consumers view video content and blurring the lines between programming distribution, and web-based competitors _ would be to ignore the Supreme Court’s direction to examine this case with an eye toward the “structure, history, and probable future” of this fast-changing industry. United States v. Gen. Dyna)nics Corp., 415 U.S. 486, 498 (1974) (internal quotation marks omitted). l, of course, cannot do that 1

are finite, to the digital platforms at Google and Facebook” and “away from television advertising in general"); PX456-56 (“The advantages of digital advertising... have resulted in advertisers shifting more of their advertising budgets from traditional television advertising to digital advertising.”). The share of U.S. spending on digital advertising exceeded spending on television advertising in 2016. See DX746A-2; Tr. 3092:15-19 (Bewkes (Time Warner)). Digital advertising revenue is expected to further eclipse television advertising revenue in the coming years See Tr. 3092:22-3093:1 (Bewkes (Time Warner)).

Why the rush from television ads to digital ones? Simply put, digital ads are more efficient. Through their access to and use of consumer data, Google and Facebook are better able to discern the purchasing preferences and interests of individuals viewing particular online content. See id. at 623:2-13 (Martin (Turner)); id. at 3087:16-3088:2 (Bewkes (Time Warner)); id. at 3243:5-10 (Stankey (AT&T)). They can use that information t`o infer what types of ads would most interest those users See id. at 3087:16- 3088:2 (Bewkes (Time Warner)). And they can tailor digital advertisements to those users based on those preferences See id. at 623:8-13 (Martin (Turner)). Best of all from an advertiser’s perspective, Google, Facebook, and other entities engaged in digital advertising have confirmatory data that demonstrates whether particular ads were effective. See id. at 623:14-22. l

Although traditional programmers like Turner maintain “massive inventories of advertising,” they lack the type of fine-grained data necessary to generate targeted ads Id.

at 3392:10-13 (Stephenson (AT&T)). Under the “spray and pray” approach, programmers 26

instead sell ads based on “broad demographic data” about the viewers of a particular program. Id. at 3760:20-24 (Athey). As a result, consumers regularly see ads for things that do not interest them, and advertisers pay to show ads that they know will be ineffective in motivating many in the audience See id. at 3087:1-8 (Bewkes (Time Warner)). As Turner CEO John Martin put it, “there’s been a long saying in the advertising industry where the advertiser would always say, l know l’m wasting half of my money, ljust don’t know which half.” ]d. at 685:20-23 (Martin (Turner)).

The shift toward digital advertising has been extremely profitable for the tech giants ~ Google and Facebook, in particular. lndeed, those two entities account for roughly 60% ofU.S. digital advertising See id. at 3746:16-22 (Athey). And they are growing at a rapid pace: Google’s advertising revenue has “almost tripl[ed]” between 2012 and 2017. while Facebook’s advertising revenue went from $4 to 340 billion in the same period. [d. at 3097:2-1 1 (Bewkes (Time Warner)) (discussing DXD122). l

By contrast, the rise of digital advertising has been costly to Turner and other programmers that rely on television advertising as a major source of revenue. See id. at 3088:3-21; cf PX456-25. ln 2017, for example, Turner’s advertising revenue declined by 2% relative to the previous year. See Defs.’ PF()F 11 31 (citing PX456-65); Tr. 3097:14- 20 (Bewkes (Time Warner)). ln light of the dual-revenue-stream business model of programmers witnesses testified that declines in television advertising revenue will produce a predictable result: it will place more pressure on affiliate fees meaning that programmers will increase the fees charged for their content. See, e.g., Tr. 3088:16-21

(Bewkes (Time Warner)). For that reason, leff Bewkes CEO of Time Warner. explained

that the explosion of digital advertising is “actually bad for” video distribution consumers “because it means that the financial support for all this programming on all these different channels gets pushed over toward subscription prices And that’s a problem, because we

think consumers are up to here with subscription prices.” Id. at 3089:6-11.

ll. The Parties and Proposed Merger A. AT&T

AT&T is a “leading provider of communications and digital entertainment services in the United States and the world.” PX455-7. As a distribution company, AT&T is in what its Chairman and CEO Randall Stephenson calls “the connectivity business.” Tr. 3378:23-24 (Stephenson (AT&T)). Although originally known for its "voice telephone" service, AT&T also provides wireless service, broadband service, and pay-TV service to consumers See id. at 3377:23-25, 3379:12-15. AT&T, however, does not create any significant television or movie content. See id. at 3245:24-25 (Stankey (AT&T)).

AT&T has two traditional l\/IVPD products: DirecTV and U-verse. Defs.’ PFOF 1134. DirecTV, acquired by AT&T in 2015, is a “satellite-based MVPD service that operates by transmitting programming from satellites to rooftop dishes installed at the customers’ homes.” Id.; see Tr. 3206:21-22, 3207:21-23 (Stankey (AT&T)); PX455-ll to 12. U-verse, by contrast, is a “telco” MVPD service that operates “[o]ver the same line that [] deliver[s] your telephone service."` ]d. at 3384;1-2 (Stephenson (AT&T)); Defs’ PFOF 11 34. Between DirecTV and U-verse, AT&T has approximately 25 million video distribution subscribers today, making it the largest provider of traditional MVPD services

See PX455-11; Tr. 3384:13-14 (Stephenson (AT&T)). 28

Despite that substantial traditional MVPD subscriber base, AT&T witnesses testified that they believe the company’s future lies in the use ofonline and mobile wireless connections to access premium video. As John Stankey, the AT&T executive who will be tasked with running lime Warner should the merger proceed, explained, AT&T acquired DirecTV in 2015 not in an effort to double down on the satellite business - a concededly mature and indeed declining asset - but to “pick up a lot of new customers that we could work on migrating” to new, innovative products necessary to compete in the future. Tr. 3207:18-3208:2, 3209:4-7 (Stankey (AT&T)). ln late 2016, AT&T launched one such product, DirecTV Now. See, eig., id. at 1824:23-24 (Merrill). DirecTV Now is a virtual MVPD and, as such, carries fewer channels than DirecTV or other traditional MVPDs; is offered at a lower price-point; and is delivered over the internet. See id. at 1825:1-3; id. at 3385:5-3386:10 (Stephenson (AT&T)). Today, and in large part due to significant promotional efforts and high-level support for the product’s launch, DirecTV Now has grown to more than one million subscribers See id. at 3386:2-3 (Stephenson (AT&T)); id. at 1825:12-1826:8, 1827:18-1828:2(Merri11(AT&T)).

AT&T Chairman and CEO Randall Stephenson testified that DirecTV Now plays to AT&T’s strong suit, namely its 100-million plus wireless subscriber base. See id. at 3379:19-20, 3385:9- 14 (Stephenson (AT&T)). With customers increasingly turning to cell phone and mobile devices to access video content, fully “[h]alfofthe volume on [AT&T`s] network is video.” Id. at 3382:5-6. Stankey noted that AT&T welcomes this trend, as it results in users purchasing larger data plans and acquiring more devices See id. at

3254:15-22 (Stankey (AT&T)). AT&T’s next major initiative, fifth generation or "5G" 29

wireless is calculated to increase video consumption even more. See id. at 3383:3-14 (Stephenson (AT&T)). As Stephenson explained to the Court, “[w]hat we’re all working towards is creating [$]35 and $15 bundles And that’s where the world is moving . . . .” ld. at 3506:23-25. To that end, Stephenson continued, AT&T has plans to launch a new product called AT&T Watch, through which customers will be able to receive “real skinny bundle[s]” of programming for 815 per month or, in the case of “AT&T wireless unlimited customer[s] . . . for free.” Id. at 3434:12-3435:4.

B. Time Warner

Time Warner, by contrast, is in the entertainment business lt has three distinct units: Warner Bros., Turner, and HBO. See PX459-l8 (Turner), -22 (HBO), -24 (Warner Bros.). Turner operates among other things ten linear cable networks that televise scheduled video programming around the clock. See id. at 18; Defs’ PFOF 11 7.7 HBO is a premium, subscription-based video service that offers movie and television shows including a significant amount of original content. See PX459-22. Unlike Turner, which collects both programming fees and advertising revenue, HBO relies solely on subscription payments to operate. See id. at 23; PX456-67; compare Tr. 604:21-23 (Martin (Turner)), with id. at 1450:12-17, 1493:15-17 (Sutton (HBO)). Warner Bros. operates a studio that creates movies television programs and other kinds of video content that are licensed both

to Time Warner’s other businesses and to third parties See PX459-24.

7 Those networks are TNT, TBS, CNN, CNN Espar`iol, CNN lnternational, Cartoon Network/Adult Swim, TruTV, TCl\/l, Boomerang, and HLN. See Defs’ PFOF11 7.

The Government’s claims in this case implicate Turner and HBO. Those business

units are therefore discussed in more detail below.

1. Turner Networks

The Turner networks are central to the Government’s primary theory of harm, and thus warrant the greatest attention here. Turner’s business model is simple: distribute its content as broadly as possible in order to maximize the dual income streams of affiliate fees and advertising revenue. See Tr. 3078:17-20 (Bewkes (Time Warner)). Historically, Turner has relied on unaffiliated third parties to distribute its content to consumers See id. at 485:1-18, 612:3-4 (Martin (Turner)). Those include traditional MVPDS, such as cable companies and satellite companies See id. at 485:1-18. ln recent years Turner has distributed its content to consumers through virtual MVPDs as well. See id. at 485:19- 486:6; Gov’t Post-Tr. Br. 6.

lndustry participants view Turner content as popular and valuable, primarily for Turner’s broadcast rights to live sports and for CNN’s live news See, e.g., Tr. 2112:24- 21 13 :12 (Sejen (Cable ONE)) (agreeing that “sports programming” is “[t]he only thing that was unique” to TBS and TNT); id. at 245:7-23 (describing TBS and TNT’s “important sports” and CNN’s “news”). CNN is the second-rated news network, and a top-seven ranked network by viewership. PX8-35; Tr. 717:5-8 (Hinson (Cox)). ln the sports domain, Turner has long-term contract rights to show portions of NCAA March l\/ladness, the NBA Playoffs, and certain games ofthe Major League Baseball Playoffs. See PX8-35; Tr. 533 :3- 12 (Martin (Turner)); see generally Gov’t PFOF 1111 82-86, 88 (reviewing Turner’s sports

rights). TBS and TNT are “by far and away"vthe two most popular Turner networks due 31

to their sports content. Tr. 471:17-20 (Martin (Turner)). Not surprisingly perhaps TBS and TNT rank in the top ten most profitable cable networks Id. at 471:21-24; see also Gov’t PFOF 1111 25, 27.

Reflecting that popularity, Turner enjoyed rate increases from every major MVPD in the last five years See Tr. 998:20-22 (Breland (Turner)); see also Gov’t PFOF 1 97. Turner executives testified that those rate hikes were due in part to a multi-year plan to “catch up” to competitors’ price increases after years ofbelow-market increases Tr. 644:1- 18 (Martin (Turner)). As such, Turner projects that its_ rate increases will slow to the low single digits from 2018 to 2022. See id. at 647:3-11 (discussing DX781). `That slowing rate-increase trend is consistent with Turner’s declining viewership numbers See id. at 2458:5-8, 22-24 (Carlton); see also PX153-3 to -4; PX456-22. Turner networks account for only 8% ofpay-TV viewership, down from 10% in 2011. See Tr. 2458:22-24 (Carlton) (discussing DXD109). When internet-based distribution is added to the mix, Turner’s share shrinks to 6% of viewership for 2017. See id. at 2458:13-15.

The growth in digital advertising has also posed a particular challenge for Turner. Today, “advanced advertising” makes up less than 5% of Turner’s ad revenue - and it shows ld. at 680:4-7 (Martin (Turner)). Turner’s ad revenues have fiatlined. See PX456- 65. This is because, as a “stuck in the middle wholesaler,” Turner for the most part lacks customer relationships which supply critical data concerning consumer preferences - data that can be used to tailor advertisements to the end user. See Tr. 641:13-25 (Martin (Turner)); id. at 3087:16-3088:2 (Bewkes (Time Warner)). Without such data, Turner

cannot tailor ads to particular consumers making its ads less valuable than those carried

on Google or Facebook. See id. at 623:5-16 (l\/Iartin (Tumer)); cf id. at 3771:12-23 (Athey).

At trial, the Court learned that Turner has attempted workarounds to improve its data and sharpen its advertisements Turner has tried, for example, to purchase data from third parties but that data was not sufficiently granular. See, e.g., id. at 3100:2-4 (Bewkes (Time Warner)). Time Warner also considered buying technology companies but concluded that the companies’ data was insufficient, and came without any guarantee of long-term access See id. 3102:9-3103:6. Finally, Turner has attempted to obtain rights to customer information through affiliate negotiations See id. at 3100:16-22; cf. id. at 92:19- 24 (Fenwick (Cox)). The record reflects however, that such efforts generally have been unsuccessful due to the bargaining friction of hotly contested affiliate negotiations and the fact that distributors consider their customer data proprietary. Id. at 955:10-18 (SEALED); cf id. at 1022:2-20 (Breland (Turner)); Defs.’ PFOF 11 16.8

ln an effort to break out of its “trapped wholesaler” role, Turner has made recent efforts to launch its own direct-to-consumer content offerings The most notable ofthose offerings are Film Struck, Boomerang, and Bleacher Report Live. See Tr. 588:8-16, 666:10-12 (Martin (Turner)). FilmStruck, which allows viewers to access classic movies as well as independent films has approximately 100,000 subscribers Boomerang, which

offers a library of children’s content and cartoons has around 150,000 subscribers Defs."

8 Turner has been able to negotiate for the rights to limited data from Hulu’s and YouTube’s virtual MVPDs. See Gov’t PFOF 11 336. As relevant here, however, that data relates only to the viewing patterns of those who view Time Warner content. That is a limited picture, as such data does not allow Turner to discern what its viewers are watching on competing channels which could help develop a fuller picture of viewer preferences Tr. 3101:13-22 (Bewkes (Time Warner)).

PFOF 11 15. Those figures are ofcourse microscopic in comparison to Netflix"s 125 million subscribers and Amazon’s 100 million Prime subscribers with access to video content. See

Tr. at 3099:6-12 (Bewkes (Time Warner)); id. at 3389:22-25 (Stephenson (AT&T)).9

2. HBO

HBO has a different business model than Turner. As a premium network, HBO offers high-quality programming that is supported by subscriber fees rather than advertising Tr. 1450:12-17 (Sutton (HBO)); see also PX456-67. lndeed, HBO has no advertising inventory at all. Id. ln addition, and unlike the Turner networks which appear in base cable or satellite packages HBO is typically an “add-on.” Id. at 3073:14-15 (Bewkes (Time Warner)); see id. at 1451:16-18 (Sutton (HBO)).10 HBO offers popular movies and television shows including a significant amount of original content. See PX459-22.

Without advertising HBO’s business model is even more reliant on broad distribution: “the more, the better,” according to Time Warner CEO leff Bewkes Tr. 3070:3-8 (Bewkes (Time Warner)). HBO content reaches consumers in four ways: (i) through MVPDs; (ii) through virtual MVPDS; (iii) through SVODs; and (iv) through

HBO’s proprietary over-the-top product, HBO Now. Id. at 1494:1-8, 1451:13-23 (Sutton

9 Tumer’s l,illiputian direct-to-consumer subscriber numbers on their face, discredit the Government’s assertion that “Turner is also not the `trapped wholesaler’ it claims to be."` Gov’t PFOF 11 30.

10 The Government states that “pay-TV packages include linear TV programming on-demand content, and typically premium channels like HBO.” Gov’t PFOF 11 12 (emphasis added). However, no matter how many premium channels “like HBO” may be available on such packages HBO itself has historically had only a 30% national penetration rate. See Tr. 1529:16-17 (Patel (AT&T)); id. at 3073:22- 23 (Bewkes (Time Warner)).

(HBO)). ln each case, the end-customer accesses HBO by way of a distributor - even for HBO Now, which is sold by digital distributors like Apple and Amazon. See id. at 1491 :6- 1 1. As with Turner, the fact that HBO relies on third parties to distribute its programming means that Time Warner lacks critical data about the preferences and viewing habits of HBO’s subscribers See id. at 3084:14-24, 3098:13-16 (Bewkes (Time Warner)).

HBO faces an array of competitors in the field of premium content creation and programming There are premium television networks like Showtime, Starz, and Epix. and online offerings such as Netflix, Amazon Prime, and Hulu. See id. at 1492:20-23 (Sutton (HBO)). What’s more, Disney has launched, and Apple appears poised to launch, a premium, direct-to-consumer service. See id. at 1492:22-24; id. at 1396;21-25 (Montemagno (Charter)). All ofthose rivals feature high-quality, premium content, and thus compete directly with HBO. See, e.g., id. at 1494216-23 (Sutton (HBO)). lndeed, Netflix’s programming budget alone is more than twice the size ofHBO’s Id. at 3099:13- 15 (Bewkes (Time Warner)).

ln this highly competitive environment, and lacking direct relationships with its viewers HBO “[a]bsolutely” depends on MVPD promotions to maximize its distribution Id. at 1496:16-17 (Sutton (HBO)); cf id. at 1528:25-1529:4 (Patel (AT&T)). As HBO President Simon Sutton explained, “our whole business is relying on our affiliates to promote us lf we can’t do that, then our entire business model is destroyed.” Id. at 1508:14-16 (Sutton (HBO)). For that reason, HBO seeks to structure its affiliate

agreements so as to "incent" distributors to maximize HBO’s distribution Id. at 1456:8-

10. Specifically, as distributors add HBO subscribers "they generally pay less on the increment.” Id. at 1455:18-19.

C. The Proposed Merger

On October 22, 2016, AT&T announced its plan to acquire Time Warner. Answer 18 [Dkt. # 20]. lnclusive of debt. the transaction is valued at approximately $108 billion Id.

At trial, the evidence showed that defendants view the proposed merger as an essential response to the industry dynamics described above - that is the increasing importance of web- and mobile-based content offerings; the explosion in targeted, digital advertising; and the limitations attendant with AT&T’s and lime Warner’s respective business models See generally Defs’ PFOF 111149-62 (discussing various proconsumer rationales for the proposed merger). The proposed merger would do so, defendants’ executives asserted, through vertical integration of the companies’ complementary assets: Time Warner’s popular content and significant advertising inventory, and AT&T’s consumer relationships customer data, and large wireless business

As a traditional programmer, Time Warner generally lacks access to valuable information about its viewers ~ it is as mentioned, akin to a “stuck in the middle wholesaler.” Tr. 641 :13-25 (Martin (Turner)). That is because it is the video distributors ~ not Turner ~ that own the customer relationships and, therefore, the customer data. See supra pp. 20, 25-28. Although Time Warner has “massive inventories of advertising,” it does not “know who the customer is . . . They don’t know who they are, they don’t know

what they’re watching” Id. at 3392:10-13 (Stephenson (AT&T)). Without information

about who its customers are and what their content preferences may be, Time Wamer is disadvantaged vis-a-vis SVODs, such as Netflix, Hulu, and Amazon Prime, and web companies such as Facebook and Google, when it comes to its ability to cater programming or advertisements to viewers See supra pp. 20, 25-28. As AT&T CEO Randall Stephenson explained, without consumer relationships and access to data, Time Warner’s “large load of advertising inventory [is] being under utilized.” Tr. at 3394:1-2 (Stephenson (AT&T)); see also id. at 3771 :12-23 (Athey) (confirming that AT&T’s digital, data-driven advertising prices are 60% higher than Nielsen-based ads because the former have “finer demographics that are offered for targeting”).

As a video distributor, AT&T generally lacks control over the video content it offers See id. at 3219:1-3 (Stankey (AT&T)) (“What we don’t have is we didn’t have programming We didn’t have the flexibility to change the product, and that’s what the guys on the other side had.”). AT&T also has access to only limited advertising inventory. Cf. id. at 3393:1-11 (Stephenson (AT&T)); id. at 609:23-610:4 (Martin (Turner)). When AT&T seeks to negotiate with programmers for rights to provide or experiment with innovative content offerings it typically encounters significant bargaining friction that renders those efforts unsuccessful See supra pp. 19-20.

By acquiring Time Warner, AT&T executives testified, the company will immediately gain access to high-quality content and an extensive advertising inventory, See Tr. 3408:3-10 (Stephenson (AT&T)). Using its wireless network, AT&T intends to distribute Time Warner content through mobile devices With such strong industry

tailwinds in favor of mobile video consumption, this strategy will increase viewership,

making Time Warner content “worth far more.” Id. at 3393:24-25; cf. 891 :23-25 (Rigdon (Comcast)) (confirming “increasing trend in the consumption of video over mobile devices”). At the same time, AT&T will bring to bear its consumer relationships and data to begin to tailor Time Warner’s advertising and increase its value. See id. at 3394:3-18 (Stephenson (AT&T)).

As the Government concedes that access will inure right away to the benefit of AT&T’s current video distribution subscribers ln particular, the Government’s own expert predicts that, due to a standard benefit of vertical integration, AT&T’s DirecTV and U- verse customers will pay a total of about $350 million less per year for their video distribution services See infra pp. 66-68. AT&T executives testified about the other efficiencies that would redound to the benefit of AT&T subscribers should the merger be approved. Of most relevance here, with the Time Warner assets and without the interference of bargaining friction, AT&T will be able to deliver content to its customers in more innovative ways The merged entity could, for instance, gather and edit individual news clips from CNN throughout the day - all tailored to a given user’s interests - and deliver that news to the wireless customer for viewing on his or her fifteen-minute break. See Tr. 3220:21-3221:9 (Stankey (AT&T)). According to AT&T executive John Stankey, that opportunity represents “a new customer at a new moment doing something that wasn’t being done otherwise.” Id. at 3221 :13-14. Stankey testified that the absence of bargaining friction will also enable AT&T and Time Warner to pursue broader introduction of new

technologies such as “4K” high-resolution programming See id. at 3222:4-22.

AT&T will also, with their customers’ permission, use consumer data to develop targeted ads thereby increasing the value of Time Warner’s ad inventory. See id. at 3391:12-22, 3393 :4-9 (Stephenson (AT&T)). AT&T witnesses testified that, in their view, the Time Warner ad inventory is of sufficient scale to warrant the development of a “programmatic advertising platform” through which AT&T can deploy its data to create a marketplace of data-informed advertising inventory for use by Time Warner and third- party programmers alike. Id. at 3243:14-3244:8 (Stankey (AT&T)). At the same time, new, tailored forms of mobile content delivery ~ like the CNN clips teased above - will create additional advertising opportunities See id. at 3221:10-11. Those opportunities Time Warner and AT&T witnesses testified, will lead to higher ad revenues that will alleviate pressure on the programming side and lower the price of video distribution to consumers All ofthose steps defendants asserted, will allow AT&T to imitate the highly successful, data-driven entities in the video programming and distribution and advertising markets

ln addition, ownership of Time Warner content will allow AT&T to more efficiently pursue what it sees as the future of the video programming and distribution industry: increased delivery of content via mobile devices such as cell phones See id. at 3381:24- 3382:2, 3393:13-25 (Stephenson (AT&T)). AT&T’s vast wireless business - a business that, iftaken separately, “would be number 37 in the Fortune 500” - has over 100 million subscribers Id. at3379:20-24;seeia’. at3208:21-23 (Stankey(AT&T)). AT&T executives testified about their vision for using those wireless connections to “transform the way we

deliver video to customers [to] make the video far more portable."` Id. at 3208:20-22

(Stankey (AT&T)); see id. at 3393:13-25 (Stephenson (AT&T)). To sum it up, in the words of AT&T Chairman and CEO Randall Stephenson, defendants view the proposed merger as a “vision deal” reflecting a belief "that distribution of [Time Warner’s] content to wireless will drive the value ofthe content up,” and that “the ability to pair our data with

[Time Warner’s] advertising inventory will drive value.” Id. at 3402:24-3403:6.

III. Procedural History A. The Investigation

Following the announcement of the deal in October 2016, the Department of Justice’s Antitrust Division conducted an investigation of the proposed merger’s competitive effects Defs’ PFOF 11 2. The investigation lasted more than one year. ]d. During that investigatory phase, the Government took approximately 20 depositions and received roughly 25 million pages ofdocuments. Despite the investigation’s vast scale and obvious importance, defendants had scarce visibility into the process They could not access the Government’s materials during the course of the investigation See 12/21/ 18 Hr’g Tr. 12:1-12 [Dkt. # 56]. Nor could they attend, let alone ask questions during the depositions that took place during the investigation See id.

B. Pretrial Proceedings

1. The Complaint

On November 20, 2017, the Govemment, acting through the Department of Justice, filed this lawsuit against AT&T, DirecTV, and Time Warner to enjoin the proposed merger under Section 7 of the Clayton Act, 15 U.S.C. § 18. See Compl. 11 48. Thirty-seven

members of the Department of Justice, including Assistant Attorney General for Antitrust

l\/lakan Delrahim, signed the Complaint. [d. at 23. ln its prayer for reliefi the Government asked that defendants AT&T and Time Warner “be permanently enjoined from carrying out the proposed merger and related transactions” or “carrying out any other agreement, understanding or plan by which AT&T would acquire control over Time Warner or any ofits assets; or merging.” Id. 1148.

2. Turner’s Arbitration Commitment

About one week after the Government filed its Complaint, Turner sent a letter and an accompanying list of terms and conditions to approximately 1,000 video distributors See, e.g., PX490; PX491; Tr. 1181:11-16 (Warren (Turner)). ln the letter, Turner represented that it was “irrevocably offering to you this agreement to engage in AAA arbitration, subject to the conditions below.” PX490. "`This agreement,” the letter continued, “also provides you with the right to continued carriage of the Turner Networks . . . pending the arbitration in the event ofa failure to agree upon renewal terms” Id. The agreement specifies that once arbitration is invoked by a distributor, Turner must continue to provide carriage on the same terms and conditions in effect at the expiration of its existing contract with the distributor, subject to the right to receive a “true-up” - make-up payments in essence - based on the arbitrator’s award. PX491-3 to -4, §§ B.l-.3. ln other words the commitment guarantees that no blackout of Turner content can occur once arbitration is invoked. See, e.g, Tr. 2653:21-23 (Katz). The proposed arbitration agreement incorporates by reference the choice-of-law provisions in the underlying

affiliate agreements PX491-2, 11 7.

3. Pre-Discovery Timeline

Defendants filed their answer on November 28. 2017. See generally Answer. AT&T and Time Warner also announced that they had agreed to extend the merger agreement through April 22, 2018. See PX456-2. Defendants swiftly moved for a trial date and, along with the Government, for a protective order. See Defs.’ Mot. to Set Trial Date [Dkt. # 221; Defs’ l\/lot. to Enter Protective Order [Dkt. # 23]; Pl.’s Mot. to Enter Protective Order [Dkt. # 24]. On December 8, 2017, l issued a protective order governing the designation and use of confidential information See Protective Order [Dkt. # 371. On December 21, 2017, l issued a Case l\/lanagement Order ("Cl\/lO") [Dkt. # 541 and Scheduling Order [Dkt. # 55], which, among other things set the trial for March 19, 2018 and stated that there would be no dispositive motions That same day, to allow for the possibility of the March 19, 2018 trial and the ruling to follow, AT&T and Time Wamer extended yet again the drop-dead date ofthe merger from April 22, 2018 to June 21, 2018. See PX456-2. lf the deal is not consummated by then, the merger agreement specifies that AT&T will be required to pay Time Wamer a break-up fee of$500 million See PX451- 87. ln the event of a favorable judgment, defendants agreed “not to consummate or otherwise complete the challenged acquisition until 12:01a.m. on the sixth calendar day

following entry of such judgment.” CMO 11 3.

4. Discovery

Given the stakes and the June 21, 2018 drop-dead merger deadline, the parties proceeded through discovery on an expedited basis Fact discovery began in late

December. and concluded in mid-February. The Government began producing third-party

documents collected during the investigation to defendants before the New Year. The parties exchanged preliminary fact witness lists in early January, and final fact witness lists one month later. They spent the intervening time on a forced march of depositions The exchange of initial expert reports took place in early February, with rebuttal reports due at the end of that month. Supplemental discovery closed on February 28, 2018` and expert discovery did so on March 9, 2018. The Scheduling Order set additional deadlines for pre- trial motions Daul)ert motions and pre-trial submission of final exhibit lists just before the March 19 start date for trial.

l provided detailed prescriptions concerning discovery in this compressed time period. The Cl\/lO limited each side’s final trial witness list to 30 fact witnesses Cl\/lO 1 12. The Government and defendants each had a maximum of 15 interrogatories and seven requests for admission Id. 11 14(d), (e). The CMO restricted each side to 150 hours of party-depositions plus 100 hours of non-party depositions Id. 11 16. The Cl\/lO did not preclude the taking of a deposition of someone already deposed during the investigation phase. Id. There were no limits on the number of requests for production Id. 11 14(a).

The parties achieved herculean feats during that time. Beyond the 25 million pages of documents produced during the Government’s investigation, an additional 7.5 million pages of documents were produced during discovery. 2/2/18 Hr’g Tr. 13:10-13 [Dkt. # 66]. Dozens ofthird parties received Rule 45 subpoenas See 1/5/18 Hr’g Tr. 7:18-21 [Dkt. # 611. The Government noticed more than 40 depositions of defendants’ witnesses Id. at

9:13-15.

5. Discovery Disputes

Rather than appointing a special master to handle discovery related issues l relied upon the seasoned counsel on both sides ofthis case to work together to resolve discovery disputes as they arose. Although counsel generally were successful in doing sol two notable pre-trial issues were brought to this Court for resolution The first, which arose in mid-January, concerned the disclosure of third-party data collected in prior Government investigations and still in the Government’s possession The second flash point, which took place closer to trial, involved discovery requests in support of defendants’ selective prosecution claim.

ln a January 18, 2018 letter and during a status hearing held the next day, defendants raised an issue related to the production of historical video programming pricing data in Government files - data that the Government had apparently obtained via prior merger investigations See 1/19/18 Hr’g Tr. 6:14-9:23 [Dkt. # 63]. To that point, the Government had resisted defendants’ production requests arguing that the Antitrust Civil Process Act, 15 U.S.C. § 1313, required it to obtain consent from each ofthe third parties that originally had produced the information in question See id. at 13:14~15. No third party had given consent, the Government continued; nor did those parties continue to possess some or all of the requested information due to the passage of time since those earlier investigations See id. at 7:12-16, 8:17-20, 15:19-25.

Stuck in a seeming game of document "hot potato,” defendants asked this Court to direct the Government to provide copies of the pricing data to the third parties that

originally produced it. Id. at 16:11, 18:23-25. Such an order would in turn enable

defendants to subpoena the information directly from the third parties Following oral argument on the issue, l ordered the Government to seek consent from the relevant third parties and to produce the requested information to those third parties by a date certain 1/22/18 Order [Dkt. # 621. The Government complied with this Order and defendants apparently were able to obtain the pricing data at issue. 2/2/18 Hr’g Tr. 6:2-5.

The case sailed along until mid-February, when the parties raised an issue related to defendants’ contemplated motion for discovery on their “selective enforcement” claim and their attendant inclusion of Assistant Attorney General l\/lakan Delrahim on their trial witness list. The Court held a hearing and heard oral argument on that dispute. See generally 2/16/18 Hr’g Tr. [Dkt. # 67]. ln that hearing defendants made an oral motion to compel production of privilege logs relating to their selective enforcement defense See id. at 22:17-23. The Governmentq for its part, made an oral motion to strike defendants` outstanding discovery and interrogatory requests for logs listing (i) all written communications about the proposed merger between the White House and the Attorney General’s Office, (ii) all written communications about the White House’s views of the proposed merger between the Attorney General’s Office and the Antitrust Division, and (iii) all oral communications about the proposed merger between the White House and the Antitrust Division. See id. at 46:8-20, 54:13-55:14. During the hearing defendants agreed to strike l\/lr. Delrahim from their witness list subject to the right to call him at trial for good cause. Id. at 36:17-37:4. A few days later, after considering the parties’ arguments at the hearing l issued a l\/lemorandum Gpinion denying defendants’ oral motion to compel and

granting the Government’s oral motion to strike. 2/20/18 l\/lem. Op. & Order 6 [Dkt. # 681. 45

As set out more thoroughly in that opinion, l concluded that defendants had failed to meet the rigorous standard for obtaining discovery on their selective enforcement defense. See

id. at 4.

6. Evidentiary Disputes

As with most trials featuring large volumes of documentary evidence, evidentiary issues were heavily litigated in this case. lndeed. l set aside the first two days ofthe trial to address evidentiary issues Not surprisingly, each side vacillated between arguing for exclusion of documents as prejudicial or irrelevant, on the one hand, or for admission of documents because such concerns are inapplicable in bench trials on the other. While keenly aware ofthe principles governing evidentiary rulings in bench trials in this case, l did not have the luxury ofblanketly admitting a mass of documentary evidence and sorting through it after trial.ll The compressed timeline and novel, complicated nature of the case instead necessitated that l make individualized rulings on relevance and admissibility Cf. l\/lanual for Complex Litigation § 12.5.

For this reason, l generally instructed the parties to seek admission of documents through sponsoring witnesses in order to facilitate determinations of relevancy or to

establish the foundation necessary for nonhearsay or hearsay exceptionsl Witnesses

ll Nor did defendants broadly stipulate to the admission of the Government’s proffered documentary evidence, as defendants seem to have done in recent antitrust cases in our Circuit. The parties also did not introduce their experts’ reports into evidence; instead, they rested on the experts’ trial testimony.

'2 There was not a uniform rule mandating sponsorship of documents by witnesses I tookjudicial notice, for example, ofcertain statements made by DirecTV and AT&T before the FCC without sponsoring witnesses See Tr. 3966:5-3967:22. in the same way, l was mindful that some documents such as a slide presentation known at trial as “version 41,” would not constitute hearsay, as they were introduced to establish the intent ofthe parties rather than for the truth of the matter asserted

would be able to contextualize and explain the technical and lengthy documents at issue, which might otherwise be misunderstood or selectively cited in post-trial briefs As such, l instructed the parties to introduce documents through sponsoring witnesses recognizing that doing so would extend, somewhat, the length of the trial. ln the end, the parties agreed to abide by that approach. See, e.g., 3/19/18 Hr’g Tr. 6:17-22 (afternoon session) (Government agrees to “add[] some additional witness and [to] talk[] with the defendants about that with regard to sponsorship issues”). '3

C. The Trial

The trial began on l\/larch 19, 2018 and ended with closing arguments on April 30, 2018.14 Over that period, there were 23 days of proceedings

The Government called 20 fact witnesses and two expert witnesses in its case-in- chief. Of the fact witnesses 11 were employees of defendants and 9 were employees of third parties The Government’s chief economic expert was Professor Carl Shapiro. Professor Shapiro is a Ph.D. industrial economist who currently holds a professorship at the University of California, Berkeley. Professor Shapiro has served in various positions

in the federal government, including most recently as Deputy Assistant Attorney General

13 Negotiations between the parties further winnowed the evidentiary disputes See, e.g., 3/19/18 Hr’g Tr., PDF at p. 7 (morning session). The parties also heeded warnings from the Court during initial evidentiary hearings as to the likely inadmissibility of certain documents For instance, after a warning as to the likely admissibility of newspaper clippings defendants did not seek admission of those documents at trial. See 3/20/18 Hr’g Tr. 51 16-20 (afternoon session) (advising defendants that the Court “usually [does not] allow news articles [to be] introduced into evidence I’ll wait to see what you’ve got . . . but l’m giving you fair notice here").

14 On March 9, 2018, the parties each filed a brief, laying out their theories of the case [Dkt. ## 75, 76, 771. On l\/larch 13, 2018, the parties filed a Statement of Evidentiary ijections under seal. [Dkt. # 861. The same day, the parties filed a loint Statement on the Burden of Pi'oof at Trial, which set forth each side’s views ofthe legal standards and burden of proofapplicable to this case [Dkt. # 871.

for Economics at the Antitrust Division in 2009 through 2011 and as a member of the President’s Council of Economic Advisers in 201 1 and 2012. He has testified in a number of antitrust matters including several antitrust trials in our Circuit. The Government also called Professor John Hauser from the Massachusetts lnstitute of Technology to testify about a survey he designed and performed and on which the Government relies

For their part, defendants called three expert and three fact witnesses Chief among their experts was University of Chicago Professor Dennis Carlton, who provided rebuttal testimony to Professor Shapiro. Professor Carlton has served as an economics professor within the University of Chicago since 1976, teaching in the economics department, business school, and the law school. Like Professor Shapiro, Professor Carlton is a seasoned expert witness who himself has served as Deputy Assistant Attorney General for Economics at the Antitrust Division from 2006 to 2008. Defendants also called Professor l\/lichael Katz from the Haas School of Business at the University of California, Berkeley, and Professor Peter Rossi from the UCLA’s Anderson School of l\/lanagement. Defendants called Professor Katz to testify about the effect ofarbitration and the FCC’s program access rules and called Professor Rossi to testify about survey methods and to rebut testimony concerning surveys and studies on which the Government relied. As their fact witnesses defendants called J eff Bewkes Chairman and CEO of Time Warner, and Randall Stephenson, Chairman and CEO of AT&T, to testify regarding their decision to merge Defendants also called John Stankey, a senior executive at AT&T responsible for planning

and integration of the proposed merger. Stankey, who will be running Time Warner should

the merger be allowed to occur, testified about the rationale for the merger as well as the synergies and efficiencies that would result from the merger.

The Government’s rebuttal case consisted of testimony from three experts First, the Government called Ronald Quintero, an accounting and financial consultant, to testify as an expert witness on defendants’ claims that the challenged merger will result in a number of procompetitive synergies Next, the Government called Professor Susan Athey, an economics of technology professor at the Stanford Graduate School of Business to testify regarding defendants’ proffered “content intelligence” synergies Finally, the Government closed out its rebuttal presentation by recalling Professor Shapiro to defend and further explain his case-in-chieftestimony in the face ofdefendants’ various criticisms

To say the trial was well staffed would be an understatement Thirty-two lawyers entered appearances for the Government, and 14 did so for defendants Evidentiary disputes were handled on a case-by-case basis as issues arose ln order to accommodate the confidentiality interests ofthird parties counsel agreed to craft their questions so as not to elicit sensitive business information, and, on three occasions l had to close the courtroom to the public following factual proffers by the Government as to the need for doing so.15 ln total, l admitted into evidence over 3,000 pages of documents broken up

into over 120 exhibits The trial transcript itself exceeds 4,300 pages in length.

15 The Court explained to the parties that it appreciated both the public’s interest in open judicial proceedings and the importance to the Government’s case ofthird-party testimony and the need to maintain confidentiality Consistent with these competing interests and applicable case law, the Court advised the parties that, when seeking to close the courtroom, they would first need to make a proffer explaining the necessity of doing so. Cf. 28 C.F.R. § 50.9 (2017) (reciting “‘the vital public interest iri open judicial proceedings” and stating the policy that DOJ counsel “shall not move for or consent to closure of a

On May 3, 2018, a mere one week after the close of evidence, the parties filed their proposed Findings of Fact and Conclusions of Law, totaling nearly\400 pages in length, as well as briefs that synthesized their arguments On the last day oftrial, l advised the parties that it would issue a ruling by lurie 12, 2018 in order to avoid running afoul of the defendants’ merger deadline of June 21, 2018 and to provide the losing party sufficient

time to preserve its appellate rights

IV. Legal Standard A. The Clayton Act

The Government seeks to enjoin the proposed merger on the basis that it violates

Section 7 of the Clayton Act, 15 U.S.C. § 18. See id. § 25 (authorizing United States to

proceeding” unless “[n]o reasonable alternative exists for protecting the interests at stake" and “[fjailure to close the proceedings will produce . . . [a1 substantial likelihood of denial of the right . . . to a fair trial").

ln order to accommodate those confidentiality interests counsel agreed to craft their questions so as not to elicit sensitive business information See Tr. 692:14-16 (“[G]overnment’s counsel has got this choreographed approach here to get this information from you under oath without revealing it to the public.”); see also, e.g., id. at 99:12-14 (SEALED). Counsel routinely asked witnesses to point to or confirm for the Court the contents of documents under seal. See, e.g., id. at 119:1-21, 124:18-125213 (Fenwick (Cox)); 535:11-22, 662:7-20 (Martin (Turner)); id. at 1095:19-1096:7 (Breland (Turner)); id. at 3011:9-21 (Christopher (AT&T)); id. at 3529:18-3530:10 (Quintero). lndeed, the Government succeeded in eliciting considerable testimony from a third-party witness - this time from AT&T’s competitor, Cox - by way of a single exhibit. See, e.g., id. at 689:18-20 (Hinson (Cox)) (“Your Honor, l’d like to mark Plaintiffs Exhibit, it’s got some confidential information that Mr. Hinson can point to.”); see generally id. at 692:25-708:14; see also PX523. ln those instances the Court, but not the public, had access to the referenced documents ln the same way, counsel asked witnesses to describe the contents at an appropriate level of generality. See id. at 259111-13 (Schlichting (DISH)); id. at 1278:13-1279:21 (Bewley (Altman Vilandrie)).

Through skillful lines ofinquiry and the use ofexhibits and demonstratives, this approach resolved most confidentiality-based concerns For several witnesses the Government initially raised the possibility of going into closed session, before later declining to seek to do so. See, e.g., Tr. 439:14-16 (SEALED). Other times the Government elected to establish the factual proffer necessary to close the courtroom. To take one example ofthe way in which - when it chose to do so _ the Government developed the need for closing the courtroom, Government counsel confirmed with NBCU’s l\/ladison Bond in open court that he felt constrained by confidentiality obligations with respect to at least six different items See, e.g., id. at 1992:2-1992:8; id. at 1993:24-1994:6 (Bond (NBCU)).

seek equitable relief to restrain a pending acquisition that violates Clayton Act). As relevant here, Section 7 “prohibits acquisitions including mergers ‘where in any line of ` commerce or in any activity affecting commerce in any section of the country, the effect of such acquisition may be substantially to lessen competition.”’ FTC v. H.J. Heinz Co., 246 F.3d 708, 713 (D.C. Cir. 2001) (quoting 15 U.S.C. § 18). The Government "has the ultimate burden of proving a Section 7 violation by a preponderance of the evidence." UnitedStates v, H & R Block, Inc., 833 F. Supp. 2d 36, 49 (D.D.C. 201 1) (internal quotation marks omitted); see also Proposed Conclusions of Law of the United States (“Gov"t PCOL”) 11 24 [Dkt. # 127]. Accordingly, the Government’s “failure of proofin any respect will mean the transaction should not be enjoined."` FTC v. Arch Coal, Inc., 329 F. Supp. 2d 109, 116 (D.D.C. 2004).

By using “the words ‘may be substantially to lessen competition”’ in Section 7, Congress indicated “that its concern was with probabilities not certainties.” FTC v. Whole Foods Mki., Inc., 548 F.3d 1028, 1042 (D.C. Cir. 2008) (emphasis omitted) (quoting Brown Shoe Co. v. United Siaies, 370 U.S. 294, 323 (1962)). Although certainty of harm is not necessary to prove a Section 7 violation, neither is the "mere possibility” ofharm sufficient Heinz, 246 F.3d at 713 (quoting S. Rep. No. 1775, at 6 (1950)); see also Baker Hughes. 908 F.2d at 984 (“Section 7 involves probabilities not certainties or possibilities"`). Rather, to grant injunctive relief under the Clayton Act, the Court rnust conclude that the

Government has introduced evidence sufficient to show that the challenged “transaction is

likely to lessen competition substantially.” Baker Hughes 908 F.2d at 985.16 As part of satisfying that burden, Section 7 “demand[s] that a plaintiff demonstrate that the substantial lessening of competition will be ‘sufficiently probable and imminent’ to warrant relief.” Arch Coal, 329 F. Supp. 2d at 115 (quoting United States v. Marine Bancorporation, 418 U.S. 602, 623 n22 (1974)).

ln assessing the Government’s Section 7 case, the court must engage in a

"‘comprehensive inquiry’ into the ‘future competitive conditions in a given market,

keeping in mind that “the Clayton Act protects ‘competition,’ rather than any particular

16 lt is undisputed that the Government has the burden of proving a Section 7 violation The Government’s view on what measure of proof that burden requires however, has been somewhat of a moving target. ln some instances the Government mirrors defendants’ position that Section 7 requires a showing that the challenged transaction is “likely” to harm competition; in others the Government states that it must show a “reasonable probability” or “appreciable danger"’ of harm to prevail Compare Compl. 11 44 (“The effect ofthe proposed merger would be likely to lessen competition substantially” in the relevant markets.), and Gov’t PFOF 20 (“The proposed merger would likely substantially lessen competition” in the relevant markets.) (capitalization altered), with Gov’t Post-Tr. Br. 13 (disputing that the “llnited States must show that harm is ‘likely”’), and Gov’t PCOL 115 & n.l (reciting a purportedly more lenient "reasonable probability"' standard). ln the final analysis each alternative formulation appears aimed at clarifying the central point that Section 7 does not require “certain" harm, but instead permits courts to use predictivejudgment to “arrest anticompetitive tendencies in their `incipiency"’ United Slates v. Penn-()lin Cheni. Co., 378 U.S. 158, 171 (1964) (quoting UnitedSIates v. Philadelphia Nal’l Bank, 374 U.S. 321, 362 (1963) (internal quotation marks omitted)). Thus, it is not surprising that courts have used these terms interchangeably See, e.g., Hosp. Corp. ofAin v. FTC, 807 F.2d 1381, 1389 (7th Cir. 1986) (noting that Section 7 requires “an_appreciable danger” of anticompetitive consequences and concluding in same paragraph that Commission had adequately demonstrated that the “challenged acquisitions are likely to foster collusive practices harmful to consumers”); Anthem, 236 F. Supp. 3d at 215 (citing with approval other court’s use of “reasonably likely” formulation later concluding that “[p]laintiffs have carried their burden to establish that the merger is likely to harm competition”).

For present purposes l need not further toil over discerning or articulating the daylight, if any, between “appreciable danger, probable,” “reasonably probable," and “likely" as used in the Section 7 context. That is because even assuming that the “reasonable probability” or "appreciable danger" formulations govern here and require more than a “mere possibility,” but less than a “more likely than not" showing of harm, but see Baker Huglies, 908 F.2d at 991 (describing “the ultimate issue” in a Section 7 case as “whether [the proposed] transaction is likely to lessen competition substantially" (emphasis added)); Anthein, 236 F. Supp. 3d at 215 (“merger is likely to harm competition”); United States v. Aetna, Inc., 240 F. Supp. 3d 1, 9 (D.D.C. 2017) (“the proposed merger is likely to substantially lessen competition”); FTC v. Siaples, Ine., 190 F. Supp. 3d 100, 110 (D.D.C. 2016) (“proposed merger is likely to reduce competition”), my conclusions regarding the Government’s failure of proof would remain unchanged for all ofthe reasons discussed below.

55 34

competitor.” United States v. Aeina, 240 F. Supp. 3d 1, 18 (D.D.C. 2017) (quoting Baker Hughesi 908 F.2d at 988, 991 n.12). “[O]nly . . . examination ofthe particular market - its structure, history and probable future - can provide the appropriate setting for judging the probable anticompetitive effect of the merger.” United States v. General Dynamics Corp., 415 U.S. 486, 498 (1974) (quoting Brown Shoe, 370 U.S. at 322 n.38). “Hence, antitrust theory and speculation cannot trump facts”; the Government must make its case “on the basis ofthe record evidence relating to the market and its probable future.” Arcli Coal, 329 F. Supp. 2d at 116-117.

B. Baker Hughes Burden Shifting Framework

As the above discussion displays Section 7 vests courts with the “uncertain task" of“making a prediction about the future.” Baker Hughes, 908 F.2d at 991; United States v. Anthe)n, Inc., 236 F. Supp. 3d 171, 191 (D.D.C. 2017). To say the least: that is no easy assignment! ln such a setting and in the absence of a crystal ball, “allocation of the burdens of proof assumes particular importance.” Baker Hughes, 908 F.2d at 991. To further assist courts in this prospective inquiry, our Circuit has set forth a burden shifting framework for use in determining whether a proposed transaction violates the Clayton Act. See, e.g., id. at 982-83.

Under that framework, the Government must first establish its prima facie case by 1) identifying the relevant product and geographic market and 2) showing that the proposed merger is likely to “substantially lessen competition” in that market. Id. at 982i 991; see also Arch Coal, 329 F. Supp. 2d at 117; Gov’t PCOL 11 24. lfthe Government satisfies its

prima facie burden, the burden then shifts to defendants to “provide sufficient evidence

that the prima facie case ‘inaccurately predicts the relevant transaction’s probable effect on future competition”’ United States v. Anilzem, Inc., 855 F.3d 345, 349 (D.C. Cir. 2017) (quoting Balcer Hughes, 908 F.2d at 991). One way defendants may do so is to offer evidence that “post-merger efficiencies will outweigh the merger’s anticompetitive effects.” Heinz, 246 F.3d at 721. lfthe defendants put forward sufficient evidence to rebut plaintiffs prima facie case, "the burden of producing additional evidence of anticompetitive effect shifts to the [government], and merges with the ultimate burden of persuasion which remains with the [government] at all times.” Anihem, 855 F.3d at 350

(quoting Baker Hughes, 908 F.2d at 983).17

17 Defendants assert that the burden-shifting framework is inapplicable to vertical merger cases where no market-concentration-based presumption of harm attaches As such, defendants argue that the Government has the burden to account for all of defendants’ proffered efficiencies as part of making its prima facie case l am skeptical of this position, both as a matter of law and logic. Cf Heinz, 246 F.3d at 720 (discussing “efficiencies defense” as a component ofthe defendants’ case); 4A Areeda & Hovenkamp, Antitrust Law 11 970c. But given that the “ultimate burden” of proving a Section 7 violation rests with the plaintiff, H & R Block, Inc., 833 F. Supp. 2d at 49, any debate over burden shifting “may be somewhat academic,” as defense counsel conceded, 3/20/18 Hr"g Tr. 67:6-7 (morning session); cf Baker Hughes, 908 F.2d at 991 (deeming “the distinction between” the “burden of production” and “the ultimate burden of persuasion” as “always an elusive distinction in practice”). That is especially so here, where, as will become evident, the Court’s ruling does not turn on the efficiencies offered by defendants in their affirmative case, but rather on its conclusion that the Governmenl’s evidence, as “undermined and "discredit[ed]" by defendants’ attacks is insufficient to “show[] a probability of substantially lessened competition," and thus that the Government has “failed to carry its ultimate burden of persuasion.” Baker Hughes, 908 F.2d at 983, 990-91.

l will nevertheless pause to mention briefly why l am confident that defendants will achieve considerable efficiencies beyond those conceded by the Government. At trial, defendants presented the Court with documentary and testimonial evidence concerning efficiencies likely to fiow from the proposed merger. The efficiencies defendants explain, come both on the “cost” side, and on the “revenue” side By defendants’ calculations cost synergies will total $1.5 billion and revenue synergies $l billion on an annual basis See Tr. 3234:17-3235:14 (Stankey (AT&T)). On the cost side, AT&T’s John Stankey testified that the marriage of AT&T and Time Warner will lead to the elimination of redundant positions in each company, achievement of certain economies of scale, and insourcing of services that the acquired entity currently acquires from vendors See id. at 3235:22-3240:1. And on the revenue side, AT&T and Time Warner expect to see the gains in innovation - particularly by way of a new programmatic advertising platform - that motivated the merger in the first place See id. at 3229:20-25, 3240:2-3246:9.

Putting aside the revenue synergies which, by their nature, are more uncertain, l have a high degree of confidence that defendants will generate most, if not all, of the predicted $1.5 billion in annual cost

C. Antitrust Analysis of Vertical Mergers

ln the typical horizontal merger case under Section 7, the Government’s path to carrying its prima facie burden is clear: by putting forward statistics to show that the proposed "merger would produce a firm controlling an undue percentage share of the relevant market, and would result in a significant increase in the concentration of firms in that market,” the Government triggers a “‘presumption’ that the merger will substantially lessen competition.” Heinz, 246 F.3d at 715 (internal quotation marks and alterations omitted) (quoting United States v. Philadelphia Nai ’l Bank, 374 U.S. 321, 363 (1963)); see also, e.g., Anthein, 236 F. Supp. 3d at 209; Aelna, 240 F. Supp. 3d at 43; H & R Block, ]nc., 833 F. Supp. 2d at 72.

ln this case, however, the “familiar” horizontal merger playbook is of little use. Baker Hughes, 908 F.2d at 982. That is of course, because the proposed transaction between AT&T and Time Warner is a vertical merger - i.e., one that involves “firms that do not operate in the same market” and thus “produce[s] no immediate change in the level of concentration in any relevant market.” Dept. of Justice & Fed. Trade Comm’n, Non-

Horizontal Merger Guidelines §4.0 (June 14, 1984) (“Non-Horizontal l\/lerger

savings by 2021. See id. at 3234:13-20. AT&T derives its prediction through the same rigorous analytical process applied in each of its mergers See id. at 3226:1-3229:3; see also DX658. l\/lost recently, in the acquisition of DirecTV, AT&T exceeded cost synergy predictions which now total $2 billion annually Tr. 3229:4-8, 3369:21-3370:4 (Stankey (AT&T)). lndeed, it is uncontested that AT&T has a strong record of meeting similar cost synergy estimates in past mergers See id. at 3229:2-3, 3229:9; see also id. at 3226:3- 5. That "analogous past experience" serves to “substantiat[e]” defendants’ “efficiency claims" leaving this Court with little doubt that AT&T will stay on its projected track. Dep’t ofJustice & Fed. Trade Comm’n, Horizontal l\/lerger Guidelines § 10 (Aug. 19, 2010). Thus while not necessary to my final judgment in this case, defendants have presented persuasive probative evidence that the merger will produce even more efficiencies than those accounted for in this Opinion. As such, no further “troll[ing] the lnternet” by l\/lr. Quintero would likely convince the Court otherwise ! Tr. 3605:25 (Quintero).

Guidelines”).18 The parties therefore agree that in this case “there is-no short-cut way to establish anticompetitive effects as there is with horizontal mergers."' Joint Statement on the Burden of Proof at Trial (“Joint Statement”) 3 [Dkt. # 87]; see 4A Areeda & Hovenkamp, Antiirust Law 11 1000a (“[T]he basic economic reason for limiting horizontal mergers is well-founded and rather generally accepted: horizontal mergers increase market concentration and high market concentration can substantially lessen competition among rivals particularly with respect to price Unfortunately, there is no comparable theoretical basis for dealing with vertical mergers.”).

With no presumption of harm in play, the Government concedes that, to satisfy its burden here, it must make a “fact-specific” showing that the effect of the proposed merger `“is likely to be anticompetitive"' Joint Statement 3-4. Such a showing is “necessarily both highly complex” and “institution specific.” David T. Scheffman & Richard S. Higgins Vertical Mergers.' Theory and Policy, 12 Geo. Mason L. Rev. 967, 967 (2004); see also Gov’t PCOL 1125 (collecting sources for proposition that “vertical mergers are judged on a case-by-case basis” based on consideration of “case-specific evidence of a danger of future competitive harm”). Of particular relevance here. the Government states that a vertical merger may “act as a clog on competition” by giving the merged firm “control of a competitively significant supplier.” Gov’t PCOL 11 46 (quoting Brown Shoe, 370 U.S. at

324). Such a situation would occur, the Government continues if the merged firm were to

18 Although the Guidelines are not binding on this Couit, our Circuit has noted that they are "a helpful tool, in view of the many years of thoughtful analysis they represent, for analyzing proposed mergers.” Anthern, 855 F. 3d at 349 (citing Baker Hughes, 908 F.2d at 98/5-86). As the Non-Horizontal l\/lerger Guidelines make reference to concepts contained within the Horizontal l\/lerger Guidelines l will cite to both as appropriate

withhold a source of supply from its rivals or otherwise foreclose access to the source "on competitive terms,” such as by causing its rivals to “pay[] more to procure necessary inputs” which in turn could “harm[] competition and consumers.” Id. 111146, 57-58 (emphasis omitted) (quoting Yankee Entrn’t & Sports Network, LLC v. Cablevision Sys. Corp., 224 F. Supp. 2d 657. 673 (SiD.N.Y. 2002); Sprint Nextel Corp. v. AT&T, [ne., 821 F. Supp. 2d 308, 330 (D.D.C. 2011)).

Further complicating the Government’s challenge is the recognition among academics courts and antitrust enforcement authorities alike that “many vertical mergers create vertical integration efficiencies between purchasers and sellers.” l\/lichael H. Riordan & Steven C. Salop, Evaluaiing Vertieal Mergers.' A Posi-Cliicago Approach, 63 Antitrust L.J. 513, 519 (1995).19 The proposed merger reflects that principle: the Government’s chief economic expert, Professor Shapiro, predicts that the merger, if consummated, would lead to $352 million in annual cost savings on the part of AT&T’s customers See Tr. 2252:19-21 (Shapiro); infra pp. 66-68; see also Gov’t PF()F 111 222- 223 (EDl\/l effect is ‘“generally accepted as a potential procompetitive benefit resulting from vertical mergers”).

As the Government also notes the “principal objective of antitrust policy is to

maximize consumer welfare by encouraging firms to behave competitively” Gov`t PCOL

19 See also Robert H. Borl<, The Aniitrust Paradox 227 (2d ed. 1993) (“Vertical mergers may cut sales and distribution costs facilitate the flow of information between levels of the industry . . .[,] create economies of scale in management, and so on.”); Ernest Gellhorn et al., Antitrusr Law and Economics 41 l (5th ed. 2004) (discussing the “[v]arious efficiency rationales” that “can motivate vertical mergers”); cf Naiianal Fuel Gas Supply Corp. v. FERC, 468 F.3d 831, 840 (D.C. Cir. 2006) (“[V]ertical integration creates efficiencies for consumers"’).

11 4 (quoting Antlzem, 855 F.3d at 366 (emphasis and internal quotation marks omitted)); see id. (“Section 7 proscribes mergers with the potential to harm the competitive process and thereby result in harm to consumers including higher prices . . . .”). As such, any proper assessment of a proposed merger, Professor Shapiro testified, must consider both the positive and negative “impact[s] on consumers” by “balancing” the proconsumer, “positive elements” of the merger against the asserted anticompetitive harms See Tr. 2182:12-20, 2253:4-5 (Shapiro); see also id. at 2461:22-2462:5 (Carlton) (“Well, Professor Shapiro is looking at the [e]ffects on consumer prices That seems the right thing to do. . . .[W]e want to see what’s going to be the result on the end price that consumers pay”); cf. Gov’t PFOF 11 223 (discussing fact that Professor Shapiro accounted for EDl\/l effects). ln view of that “somewhat different” analysis applicable to vertical mergers Tr. 2182:16-18 (Shapiro), it is perhaps little surprise that the Department of Justice’s Non- Horizontal l\/lerger Guidelines recognize that vertical mergers “are less likely than

horizontal mergers to create competitive problems Non-Horizontal l\/lerger Guidelines § 4.

Given all ofthe competing considerations at play, “the analysis ofvertical mergers” has been described as “much more complex than the analysis of horizontal mergers" Scheffman & Higgins, Veriical Mergers, 12 Geo. l\/lason L. Rev. at 967. Things are made more difficult still by the lack of modern judicial precedent involving vertical merger

challenges - a dearth of authority that is unsurprising considering that the Antitrust

Division apparently has not tried a vertical merger case to decision in four decades l See

Defs’ Proposed Conclusions of Law (“Defs.’ PCOL”) 1 32 [Dkt. # 1201; 2/16/18 Hr’g Tr. 13:24-14:1.

To sum up, the Court accepts that vertical mergers “are not invariably innocuous" but instead can generate competitive harm “[i]n certain circumstances.” Non-Horizontal l\/lerger Guidelines §§ 4, 4.2; Gov’t PCOL 122.20 The case at hand therefore turns on whether, notwithstanding the proposed merger’s conceded procompetitive effects the Government has met its burden of proof of establishing through "case-specific evidence." that the merger of AT&T and Time Warner, at this time and in this remarkably dynamic industry, is likely to substantially lessen competition in the manner it predicts Gov’t PCOL 1 25. Unfortunately for the Government, for the following reasons it did not meet

its burden

ANALYSIS

The challenged vertical merger here would unite Time Warner, a creator and supplier of popular video content, with AT&T, a large downstream purchaser and distributor of video content. The Government concedes that the challenged merger, like

most vertical mergers will result in significant benefits to customers of the merged

20 The Court therefore declines defendants’ invitation to adopt either a per se rule or a presumption that would apply to most vertical mergers See Pre-Tr. Br. of Defs. 29 [Dkt. # 771. To be sure, the standard for which defendants advocate aligns with the views ofa number of authorities includingjudges from this Circuit. See, e.g., Robert Bork, The Antitrust Paradox 245 (“[l]n the absence of a most unlikely proved predatory power and purpose, antitrust should never object to the verticality of any merger."); Comcast Cable Cornms., LLC v. FCC, 717 F,3d 982, 990 (D.C. Cir. 2013) (Kavanaugh, J., concurring) (“[A]bsent market power, vertical integration and vertical contracts areprocompetilive.”) (citing Douglas H. Ginsburg Vertical Restrainls.' De Facto Legality Under the Rule of Reason, 60 Antitrust L..l. 67, 76 (1991)). Tempting though it may be to agree with my appellate brethren l need not, and will not, go that far to resolve this case.

company Specifically, the Government’s lead expert, Professor Carl Shapiro, estimates that the merger will cause AT&T to lower the price of DirecTV, resulting in $352 million in annual savings for DirecTV’s customers See Tr. 2252:19-20 (Shapiro).

Notwithstanding those conceded consumer benefits the Government contends that the challenged merger is “likely to lessen competition substantially,” Baker Hughes, 908 F.2d at 985, and thus should be enjoined under Section 7, see Compl. 1 10. The challenged merger would likely result in a substantial lessening of competition according to the Government, in three “mutually reinforcing” ways Gov’t Post-Tr. Br. 7.

Firsi and foremost, the Government argues that the challenged merger would enable Turner to charge AT&T’s rival distributors - and ultimately consumers ~ higher prices for its content on account ofits post-merger relationship with AT&T. See, e.g. Compl. 11 36- 38; Gov’t PFOF 11 226, 231-32; Gov’t Post-Tr. Br. 1-2. Second, the Government contends that the challenged merger will substantially lessen competition by creating an increased risk that the merged firm will act, either unilaterally or in coordination with Comcast- NBCU, to thwart the rise of the lower-cost, consumer-friendly virtual MVPDs that are threatening the traditional pay-TV model. See Compl. 1140-41; Gov’t PFOF 1278. Finally, the Government alleges that the merged entity could harm competition by preventing AT&T’s rival distributors from using HBO as a promotional tool to attract and retain customers See Compl. 1 39; Gov’t PFOF 1 234.

ln the remainder of this section l will analyze each of those theories of harm to competition lnitially, l will set forth the relevant market definition which incorporates

the Government’s proposed product and geographic markets Next. l will discuss the

conceded consumer benefits associated with the proposed merger. l\/lindful of those conceded benefits and the need to balance them against the Government’s allegations of consumer harm, l will then evaluate whether the Government has carried its burden to show a likelihood that the challenged merger will result in a substantial lessening of competition For the reasons discussed in detail below, l have concluded that the answer to that question

isno!

I. Market Definition

Typically, “[m]erger analysis starts with defining the relevant market"' in which to assess the alleged anticompetitive harms FTC v. Sysco Corp., 113 F. Supp. 3d 1, 24 (D.D.C. 2015) (citing United States v. Marine Bancorporaiion, 418 U.S. 602, 618 (1974)). The relevant market comprises two parts: a product market and a geographic market. Anthein, 236 F. Supp. 3d at 193. Here, the Government defines the primary relevant product market as the “l\/lultichannel Video Distribution” market, and the relevant geographic markets as the approximately 1,200 local markets in which residents have access to video offerings from the same set of multichannel video programming distributors Gov’t PFOF 11 31, 38-41. Both ofthose proposed markets find support, the Government contends in Professor Shapiro’s expert analysis see Tr. 2184:22-2188:4 (discussing hypothetical monopolist test, among other things), as well as the Brown Shoe “practical indicia,” see 370 U.S. at 325 (listing “industry or public recognition of the submarket,” “the product’s peculiar characteristics and uses” and "distinct customers” and "distinct prices” ofthe product as relevant to product market determination); Gov"t PFOF

1111 32-36. 61

Horizontal merger cases often “to a great extent . . . hinge[] on" market definition because such definition affects the ultimate market concentration statistics associated with a proposed transaction FTC v. Siaples, Inc., 970 F. Supp. 1066, 1073 (D.D.C. 1997). For that reason market definition is often heavily contested in horizontal merger cases turning on fine-grained economic analyses of"SSNlPs” and cross-elasticity of demand. See, e.g., Anthe)n, 236 F. Supp. 3d at 193-198; FTC v. Staples, Inc., 190 F. Supp. 3d 100, 116-127 (D.D.C. 2016); Syseo Corp., 113 F. Supp. 3d at 24-48. Happily, l need not delve deeply into those concepts here The proposed vertical merger, as discussed, does not “involve an increase in market concentration," and defendants for all of their objections to the Government’s case, have not meaningfully challenged the Government’s proposed product or geographic markets Joint Statement 3; see Tr. 2186:25-2187:2, 2188:2-4 (Shapiro). l will thus accept the Government’s proposed product and geographic markets for purposes of this case, and briefly discuss the basics of those markets - as well as the role of the product market as it relates to my analysis ofthe Government’s claims of harm - below.

Product Markel. The Government’s prirnary product market is the market for multichannel video distribution Multichannel video distribution as defined by the Government, involves the distribution of live, or “linear,” video programming networks as well as on-demand content, to subscribing consumers Gov’t PFOF 1 31; Trial Br. of the United States (“Gov’t Pre-Tr. Br.”) 22 [Dkt. # 76]. As relevant here, the sellers in that product market are: 1) MVPDs including cable television providers such as Comcast, Cox, and Charter; direct broadcast satellite providers such as DirecTV and DlSH, which

operate nationally; telecommunications providers or “telcos” such as Verizon F ios and

AT&T’s U-verse; and overbuilders such as RCN; and 2) virtual MVPDs, including Sony’s Playstation Vue, Hulu Live, Google’s YouTube TV, DirecTV Now, and DlSH`s Sling. As discussed, virtual MVPDs provide the same live-TV services as do traditional MVPDs but do so over the internet rather than by way of a dedicated transmission path that they control. See Gov’t PFOF 1 15. Although the majority of U.S. households (approximately 90 million) currently receive linear video programming through traditional MVPDs, id., and a majority are likely to continue to do so, there is no debating that the number of MVPD subscribers is “declining unequivocally” as consumers increasingly turn to virtual MVPDs and SVODs for their video content needs Tr. 3451:22-23 (Stephenson (AT&T)); see id. at 3449:12~3451:1 (“DirecTV lost 1.2 million subscribers in 2017. The whole system, pay TV, cable, satellite, lost 3 million.”); see also id. at 2948:1 1-24 (Holanda (RCN)); PX455- 136 to -137.

As the above discussion indicates the Government’s proposed product market focuses on the downstream distribution of live-TV content to consumers_a focus that excludes both the upstream programming market and the market for SVODs such as Netflix, Hulu, and Amazon Prime. See, e.g., id. at 2184:22-2185:5 (Shapiro); cf. Gov’t PCOL 138 (disputing need to “define an ‘upstream’ programming market”).21 That product market definition appears to reflect the Government’s (and Professor Shapiro’s)

projections regarding where the challenged merger’s ultimate “net harm"’ to consumers ~

21 The Government also asserts_that a broader market of“All Video Distribution” - which includes SVODS in addition to MVPDs and virtual MVPDs a constitutes a relevant product market. See Gov’t PFOF 137 (citing Tr. 2184:18-2185:17) (Shapiro). For simplicity’s sake, this discussion mirrors the Government’s focus on the multicliannel video distribution market. Cf. Gov’t Pre-Trial Br. 22.

i.e., the predicted increased costs to “multichannel video subscribers” ~ will result. Cf. Gov’t PFOF 1 231. lmportantly, however, accepting the Government’s proposed product market does not mean that Turner’s position in the upstream programming market is irrelevant to evaluating the Government’s theories ofharm in this case Nor does it require this Court to ignore the rising role of SVODs in the broader multichannel video programming and distribution market. That is because the Government’s proffered increased-leverage theory, not to mention its other theories of harm, incorporates those factors in at least three different ways

Firsl, as will become clear in the ensuing discussion examining the importance of lurner’s content to distributors in the upstream programming market is a necessary (but not sufficient) step in evaluating the Government’s increased-leverage theory Cf. Gov’t PFOF 11 69-102 (proposing findings of fact to support assertion that the “merger would enable AT&T to harm competition because MVPDS and virtual l\/lPVDs need Turner content to compete effectively"’). Second, the bargaining model from which the Government’s measures of consumer harm are derived itself accounts for the increasing role of SVODs and “cord cutting” in the market, as those trends affect the amount of benefits that AT&T could expect to receive under the Government’s increased-leverage theory See, e.g., Tr. 2242:2-18 (Shapiro) (discussing role of“cord cutting” in calculating the bargaining model’s “diversion rate" input); id. at 2504:11-2506:24 (Carlton) (explaining why cord cutting “matters a lot” to bargaining model). Third and finally, the Government has argued that certain documents reflect an intent on the part of defendants

to use the proposed merger to act consistently with the Government’s increased-leverage

theory of harm, among other theories See Gov’t PCOL 1 51 (stating in relation to "[d]efendants’ internal documents,” that “[e]vidence of anticompetitive intent can also form the basis of a court’s prediction ofharm”). To appropriately evaluate the strength of such evidence, however, l must be able to put it in the context of other documents and statements related to the various rationales for the proposed merger including of most relevance here, defendants’ asserted desire to compete with SVODS and other technology companies amid “the ongoing revolution in video programming and distribution” Defs.’ PFOF16; see also Tr. 3079:18-3080:2 (Bewkes (Time Warner)). Therefore, although the Government is of course correct that the refrain “‘we are getting killed by new competition

in different markets is no “defense to an illegal merger,” Gov’t Post-Tr. Br. 21, l simply cannot evaluate the Government’s theories and predictions of harm, as presented by the Government at trial, without factoring in the dramatic changes that are transforming how consumers view video content.

Geographic Markets. The Government has identified over 1.100 local multichannel video distribution markets as the relevant geographic markets See Gov’t PFOF 1 41. These local markets which the Government calls “Local Footprint 0verlap Zones,” represent each local geographic area in which “residents have access to video offerings from the same set ofMVPD competitors” Id.; see Tr. 2187:3-25 (Shapiro). The localized geographic markets reflect the reality that, due to limitations of the physical transmission paths maintained by many of the providers in the multichannel video

distribution market, the mix of MVPDs and virtual MVPDs available to a consumer varies

based upon where that consumer lives See Gov’t PFOF 1 40. As such, the Government

contends that the asserted “effects of the proposed merger” will vary depending “on the market shares of the various MVPDs and virtual MVPDs in [a] region,” and that analyzing the local markets is therefore appropriate Id. The Government has not relied upon harm in any particular local market as the basis for enjoining the merger, however. lnstead, the Government’s expert “aggregated” all of the alleged harms in the local markets in order to derive a total measure of nationwide economic harm. Gov’t PFOF 13 (“Relevant downstream geographic markets are local, but they can be aggregated for analytical convenience."); see Tr. 2255:1-2256:15 (Shapiro) (providing aggregate estimates of

consumer harm nationwide).

II. Conceded Consumer Benefits of Proposed Merger

Vertical mergers often generate efficiencies and other procompetitive effects See supra pp. 53-57 & nn. 17, 19. The proposed merger is no exception lndeed, the Government concedes that this case implicates one “standard benefit” associated with vertical mergers the elimination of double marginalization (“EDM”). Tr. 2438:6 (Carlton); Gov’t PFOF 1 222.

As relevant here (and at the risk of oversimplifying things), double marginalization refers to the situation in which two different firms in the same industry, but at different levels in the supply chain each apply their own markups (reflecting their own margins) in pricing their products See Tr. 2251:15-25 (Shapiro). Those “stacked” margins are both incorporated into the final price that consumers have to pay for the end product. Id. at

2251 :24. By vertically integrating two such firms into one, the merged company is able to

“shrink that total margin so there’s one instead of two," leading to lower prices for consumers Id. at 2252:1-3. EDM is therefore, procompetitive

ln the context of a Time Warner and AT&T combination EDM will play out as follows Prior to the merger, AT&T must pay lime Warner a certain price to display Turner content to its DirecTV customers Id. at 2251:19-25. The price that AT&T pays includes Time Warner’s profit margin that is an amount over and above the marginal cost of the programming Id. After the vertical integration of AT&T and lime Warner, however, AT&T will no longer need to pay Turner’s profit margin to display Turner content. See id. at 2252:1-3; id. at 2438:9-15 (Carlton). ln effect, that means that AT&T’s marginal cost oflicensing Turner content will be lower, which in turn renders distribution of Turner to its DirecTV customers more profitable Id. at 2438:13- 15 (Carlton). With its profits increased, AT&T would have the “incentive to get more customers and in particular AT&T’s price, the DirecTV price will go down to consumers” Id. at 2438:16-18.

According to the Government’s expert, Professor Shapiro, EDM would result in AT&T lowering the price for DirecTV by a “significant” amount: $1.20 per-subscriber, per-month. Id. at 2252:6-7 (Shapiro). All told, those savings to AT&T’s customers add up to $3 52 million annually See id. at 2252:19-21. Those savings moreover, would begin flowing to AT&T’s customers “pretty quickly” after consummation of the merger. Id. at 2446:4-5 (Carlton).

All sides agree that any proper antitrust analysis of the proposed merger must account for those “positive elements of the merger in terms of DirecTV, having lower

costs.” Id. at 2182:12-13 (Shapiro); cf. Gov’t PFOF 11222-23. ln other words to 67

understand whether the proposed merger will harm consumers Professor Shapiro explained, it is necessary to “balance” whether the Government’s asserted harms outweigh the merger’s conceded consumer benefits Tr. 2180:24, 2181:1-6 (Shapiro); see id. at 2182:11-21 (“So l’m going to need to trade those off. This is somewhat different than horizontal merger analysis We’re talking about vertical merger analysis here.”). With that important principle in mind, l will now examine whether the Government has met its

burden under Section 7.

III. The Government Has Failed to Meet Its Burden to Show That the Proposed Merger Is Likely to Substantially Lessen Competition by Increasing Turner’s Bargaining Leverage in Af`f`iliate Negotiations

The Government’s primary theory of harm to competition focuses on the challenged merger’s integration of Turner’s important video content ~ content that includes among other things the networks CNN, TNT, and TBS - with AT&T’s video distributors U-verse and DirecTV.22 Specifically, the Government contends that, should the challenged merger proceed, Turner’s relationship with AT&T will enable Turner to extract greater prices from AT&T’s rival distributors for its “must-have” content than it could without the merger. See, e.g., Compl. 11 31-38. The Government argues that distributors would then pass on those price increases to their subscribers resulting in an increase of hundreds ofmillions of dollars in annual consumer payments ld. 1 39; Gov’t PFOF 11 231-232.

According to the Government, it carried its burden to support its increased-leverage

theory of harm to competition by offering what it refers to as “real-world objective

22 For purposes of this section the Court at times refers to AT&T’s collective distribution offerings as “DirecTV.”

evidence” - namely, statements contained within defendants’ prior regulatory filings and internal business documents as well as testimony from third-party competitor witnesses Gov’t PC()L 1 21. To further corroborate its increased-leverage theory and predict the consumer harm that would be generated, the Government also relied on testimony and economic modeling proffered by Professor Carl Shapiro. Professor Shapiro opined that a post-merger Turner would be able to extract greater affiliate fees from distributors due to increased bargaining leverage lurner would gain on account of its relationship with AT&T. Citing the results of his economic models Professor Shapiro predicts that such increased leverage would lead to total, annual consumer harms that outweigh the conceded 8352 million in annual cost savings that the proposed merger would generate for AT&T’s customers See, e.g., Tr. 2253:4-15 (Shapiro).

Not surprisingly, the defendants vigorously disagree with the Governmentis increased-leverage theory of harm. To start, defendants argue that the Government has failed to put forward any “meaningful real-world evidence"’ to support the premise that a post-merger Turner would benefit from increased bargaining leverage with distributors on account ofits relationship with AT&T. Defs"PFOF 1 81. lf anything defendants argue, analysis of real-world pricing data demonstrates that prior instances of vertical integration in this industry have not produced the increased-leverage effects that the Government predicts Id. 11 95-102. Defendants also challenge Professor Shapiro’s testimony, arguing that it lacks sufficient basis in the facts ofthis industry and reflects results based on a model

riddled with improper inputs and faulty assumptions Id. 11 86-94, 105, l 1 1-13, 188, 204.

ln evaluating these competing contentions the Court unfortunately does not have the luxury of looking to judicial precedents applying the increased-leverage theory in the context of a Section 7 challenge to a vertical merger. lndeed, the Government has not pointed to any prior trials in federal district court in which the Antitrust Division has successfully used this increased-leverage theory to block a proposed vertical merger as violative of Section 7. Cf. Tr. 2390:2-4 (Shapiro) (noting with respect to proffered economic bargaining model, that "[w]hat’s less common is to use it to evaluate a merger or a vertical merger especially”); Defs.’ PCOL132. Thus in this matter of first impression l must determine whether the evidence adduced at trial is sufficient to support the Government’s assertion that Tumer will likely gain increased bargaining leverage in affiliate negotiations on account ofthe proposed merger and, if so, whether any increased distributor or consumer costs stemming from the increased bargaining leverage will result in a substantial lessening of competition under Section 7.

Having heard and considered the evidence adduced at trial, 1 conclude that the Government has failed to clear the first hurdle of showing that the proposed merger is likely to increase Turner’s bargaining leverage in affiliate negotiations l thus need not consider the separate legal question of whether any effects associated with the Government’s increased-leverage theory would result in a substantial lessening of competition for

purposes ofthe Clayton Act’s prohibitions23 Before explaining that conclusion l need to

23 On that score, defendants argue that “even taken at face value, the Government’s projected price effects do not state a claim under the Clayton Act."' Defs.’ PCOL 159 (capitalization altered); see also id.1131-33. ln particular, defendants point out that the miniscule per-consumer price increases of approximately 27-cents per month relied on by the Government would not prevent AT&T’s rival distributors from competing in the marketplace or otherwise “impair[] their ability to discipline” AT&T’s

briefly review the basics of affiliate negotiations and the Government’s increased-leverage theory of harm. With that background established, l will examine the evidence put forward by the Government to support its argument that the challenged merger would likely increase Turner’s bargaining leverage with distributors and thereby enable it to secure greater affiliate fees than it could without the merger. Ultimately, as l will explain the Government’s proof at trial falls far short of establishing the validity of its increased- leverage theory

A. Background of Increased-Leverage Theory of Harm

As previously discussed, the terms under which distributors may license and display programmers’ content are set through a “very tough” series of affiliate negotiations Tr. 1023:2 (Breland (Turner)); see supra pp. 14-18. As with any type of bargaining each party to an affiliate negotiation attempts to take advantage ofits points ofleverage, and "reaching a deal in the end can come down to a battle ofthe competing bargaining leverages" Tr. 1025:20-22 (Breland (Turner)); Gov’t PFOF 1 154. ln the event an affiliate negotiation is

unsuccessful, the distributor will lose the rights to display the programmer’s content to its

prices indeed, they claim that competition would be promoted by the challenged merger’s conceded vertical integration effect oflowering AT&T’s prices to its projected consumers Id. 11 31-32; cf Conicasi Cable Cornrns., LLC v. FCC, 717 F.3d 982, 990 (D.C. Cir. 2013) (Kavanaugh, J., concurring) ("Vertical integration and vertical contracts become potentially problematic only when a firm has market power in the relevant market.”).

For the reasons given by defendants the Court harbors serious doubts that the Government’s proffered affiliate fee increases to AT&T’s rivals or the resulting 27-cent per-month subscriber cost increases would, if proven constitute a "substantial lessening of competition” for purposes of Section 7. 15 U.S.C. § 18. As just noted, however, l need not rest this opinion on that legal conclusion That is because, for all ofthe reasons provided in the section that follows the Government has failed to carry its burden to put forward adequate evidence to show that there are likely to be any price increases (much less price increases that outweigh the conceded EDM benefits to consumers) either to AT&T’s rival distributors or their subscribers under its increased-leverage theory

customers Such a situation is known in the industry as a programming “blackout," or "going dark.” Tr. 129:4-9 (Fenwick (Cox)).

-Blackouts have significant, if not “catastrophic,” negative consequences for programmers - in the form of lost advertising and affiliate fee revenues Id. at 1128:7-12 (Breland (Turner)); Defs.’ PFOF 11 76-77. Distributors for their part, may lose subscribers See generally, e.g., Tr. 2197:4-2198:2 (Shapiro). ln “almost every negotiation” therefore, programmers and distributors threaten blackouts in an attempt to gain concessions Id. at 1026:17-1027:3 (Breland (Turner)); cf. id. at 367:1-22, 376:22- 377:11 (Schlichting (DISH)). Given that blackouts are negative events for both programmers and distributors however, deals between programmers and distributors are invariably struck in order to avoid long-term blackouts See id. at 138:13-15 (Fenwick (Cox)); id. at 1027:4-7 (Breland (Turner)); id. at 1359:14-15 (Montemagno (Charter)); id. at 3124:4-7 (Bewkes (Time Warner)). lndeed, when it comes to Turner, the record shows that there has never been a long-term blackout of the Turner networks See id. at 2357:12- 14 (Shapiro) (“Q: But to be sure there`s never been a long-term blackout of Turner, right‘? A: No . . . .”); Defs.’ PFOF 1 94. That fact is by no means lost on either side

That background brings us to the Government’s increased-leverage theory Notably, under that theory, the Government does not allege that a post-merger Turner would be incentivized to start actually engaging in long-term blackouts with distributors That is so, as Professor Shapiro concedes because withholding Turner content would not be “profitable” to the merged entity given the attendant losses in significant advertising

and affiliate fee revenues See Tr. 2293:9-17 (Shapiro). ln other words and in contrast to

a prevalent theory of vertical merger antitrust harm, Turner will not “‘foreclose" downstream distributors from accessing Turner content. See id. at 2218:15-16 (“This is not a foreclosure-withholding story”); cf Brown Shoe, 370 U.S. at 323-24 (stating that “[t]he primary vice ofa vertical merger or other arrangement tying a customer to a supplier is that, by foreclosing the competitors of either party from a segment of the market otherwise open to them, the arrangement may act as a clog on competition” (internal quotation marks omitted)).

lnstead, the Government’s increased-leverage theory of harm posits that Turner’s bargaining position in affiliate negotiations would improve after the merger due to its relationship with AT&T. That is so, the Government argues because lurner and its distributor counterparties would recognize that, should Turner fail to strike a deal and engage in a long-term blackout with a distributor, Turner would no longer face the mere downside oflosing affiliate fees and advertising revenues See, e.g., Gov’t Post-Tr. Br. 1- 2. Rather, some of those losses would be offset, according to the Government. by new benefits to AT&T’s video distribution companies via the following chain of events: 1) some of the rival distributor’s customers would depart or fail to join the distributor due to the missing Turner content; 2) some portion ofthose lost customers would choose to sign up with AT&T’s video distributors (which would have Turner); and 3) AT&T would profit from those gained subscribers See generally Tr. 2197:15-2198: 12 (Shapiro). As a result, the Government predicts that Turner’s downside position in the event of a blackout would improve as a result of the proposed merger. That improved downside position according

to the Government, would in turn enable Turner to demand higher prices for its content in

post-merger affiliate fee negotiations with distributors - price increases that would ultimately be passed on to consumers See Compl. 1 38.

At trial, the Government relied on two primary categories of evidence to support its increased-leverage theory of harm, First, it offered so-called “real-world objective evidence” - namely, statements contained within defendants’ prior regulatory filings and internal business documents as well as testimony from third-party competitor witnesses Gov’t PCOL 21. Second, the Government called an expert, Professor Carl Shapiro, to testify about its increased-leverage theory, which is based on an economic theory of bargaining known as the Nash bargaining theory, and to estimate the consumer harm associated with the increased-leverage theory Gov’t PFOF 1 201, For the following reasons neither category of evidence was effective in proving the Government’s increased- leverage theory Accordingly, as to this theory, the Government has failed to meet its burden of proof to show that the merger is likely to result in a substantial lessening of competition

B. The Government’s So-Called “Real-World Objective Evidence” Is Insufficient to Support Its Increased-Leverage Theory of Harm

To support its increased-leverage theory of harm, the Government first points to various pieces of the so-called “real-world objective evidence” it offered at trial. Gov’t PCOL 21. That evidence primarily consisted of defendants’ ordinary course-of-business documents and excerpts of regulatory filings submitted by defendants in prior administrative proceedings as well as the testimony ofthird-party witnesses from Al&l’s

rival distribution companies Of particular importance here, the Government’s so-called

real-world evidence was directed at explaining and establishing two main concepts First, the Government sought to establish the importance of Turner content to distributors and the resulting leverage Turner enjoys in affiliate fee negotiations Second, the Government relied on this so-called “real-world objective evidence" to substantiate its prediction that Turner’s leverage with distributors would increase as a result of Turner’s post-merger relationship with AT&T. Neither, however` provided persuasive support for the

Government’s increased-leverage theory of harm. How so?

1. Evidence Regarding the Popularity of Turner Content ls of Limited Probative Value in Evaluating the Contention That Turner Will Gain lncreased Leverage Due to the Proposed l\/Ierger

At trial, much time was spent debating the “must-have” status of Turner’s programming content. According to the Government, distributors literally “‘must have”’ Turner’s content in order “to compete effectively” in the video distribution industry Gov’t Post-Tr. Br. 4; see also id. at 6 (“Distributors don’tjust want this specific input to compete effectively, they truly need it."); Gov’t PFOF 23 (similar). Defendants countered that the term “must have” is simply a marketing phrase used to mean "popular” and, similarly, that Turner content is not actually necessary to allow distributors to operate their businesses successfully See Defs.’ PFOF 1 179.

Based on the evidence, l agree with defendants that Turner’s content is not literally “must have” in the sense that distributors cannot effectively compete without it. The evidence showed that distributors have successfully operated, and continue to operate,

without the Turner networks or similar programming Cf Tr. 351:5-25 (Schlichting

(DISH)) (discussing fact that DlSH’s virtual MVPD, Sling offers packages without broadcast stations and CBS); PX144-121 (listing “[p]ast [n]etwork [d]rops” by distributors). lndeed, Stefan Bewley. a consultant who generated a slide deck with recommendations for Charter’s use in evaluating its relationships with programmers indicated that “Charter would be better off and would save a lot of money [by] canceling Turner.” Tr. 1336:10-12 (Bewley (Altman Vilandrie)). Sling President Warren Schlichting acknowledged DISH founder and chairman Charlie Ergen made similar statements to the investment community See, e.g., id. at 365:17-366:1 (Schlichting (DISH)) (conceding that Ergen stated in investor call that a Turner blackout would be “slightly cash positive for us from a cash-flow perspective"’).

l therefore give little credit to blanket statements by third-party competitor witnesses indicating that the entire “viability of [their] video model” could depend on whether they offer Turner programming Id. at 128:21 (Fenwick (Cox)); see also id. at 697:2-19 (Hinson (Cox)) (claiming that, without Turner, Cox would lack “the ability to compete” and that their customers would “go somewhere else”). Such statements were largely unaccompanied by any sort of factual analyses or, worse, contradicted by real-world examples from the witnesses themselves See, e.g., id. at 128:22-129:20 (Fenwick (Cox)) (neither she nor others at Cox had done analysis of potential subscriber losses in Turner blackout); id. at 2947:1-13 (Holanda (RCN)) (“Q: And so today, you’re not offering this Court any empirical data or any real-world evidence of subscriber losses if RCN didn’t have Turner, right? A: No, not our company”). Compare id. at 242:14-15, 352:5-7

(Schlichting (DISH)) (“[l]f you don’t have March l\/Iadness" games half of which are 76

carried by lurner, "you’re not in the pay-TV business”), and id. at 245:14-15 (“Q: How about CNN, why is CNN must have? A: Well, imagine coming around to midterm elections without CNN, right.”), and id. at 242:16-243:1 (“ABC, NBC, CBS, Fox and Time Warner are the five groups that you, you just, it’s very hard to have a pay-TV service without them.”), with id. at 352:1-19 (conceding that DlSH’s Sling does not carry CBS, which offers the other half of the l\/larch l\/ladness games), and id. at 360:18-24, 388:10-389:5 (acknowledging that DlSH went dark with CNN at time of 2014 midterm' elections and suffered only negligible subscriber loss), and id. at 351:1 1-21 (admitting that Sling Orange package lacks all of the “broadcast stations [and] CBS”).2"1

Nor does those witnesses’ (or, for that matter, defendants’) use of the term “must have” to describe Turner content change things lndeed, the evidence indicated that the term “must have” is a marketing phrase used by virtually every programmer to suggest that its content is popular with viewers See, e.g., id. at 549:19-20 (Martin (Turner)) ("‘l\/lust have" is another way of saying we have popular programming.”); id. at 899:13-16 (Rigdon (Comcast)) (agreeing that “must have is just a term of art that means something is

popular”); id. at 1092:18-24 (Breland (Turner)) (“[M]ust have means it’s popular . . . . l

2‘1 The “must have” status of Turner content also varies based on whether the content is available for viewing through other means such as over the internet. Former Cable ONE negotiator Randy Sejen testified, for example, that subscriber losses from a blackout of Turner’s live baseball content were mitigated by the fact that “consumers were able to wire around” the blackout by “accessing mlb.com ifthey needed to see a paiticular playoff game." Tr.2117:21-2l18:20(Sejen(CABLE ONE)). Along those same lines Sejen testified that the online availability of March l\/ladness basketball games could potentially “address the sort of must-have nature"` ofthat content. See id. at 2121 :l 1-16, 2123:1-5. l received similar evidence indicating that the availability of HBO’s content through online, direct-to-consumer platforms has lowered the value of HBO programming ~ and thus its leverage - in the eyes of distributors See, eig.i DX709.

don’t in a literal sense mean that 1 must have this content or l can’t be successful.”); id. at 2130:23-2131:6 (Sejen (Cable ONE)) (agreeing that he would “expect to hear" all programmers pitch their content as “must-have” and that he would “kind of take that with a grain of salt”).

That said, l do nonetheless accept the Government’s contention that Turner has popular content ~ especially live sporting events and live news - and, as a result, enjoys bargaining leverage with distributors See Gov’t PFOF 11 70-102 (summarizing evidence regarding Turner’s importance to distributors); id. 11103-177 (summarizing evidence supporting proposition that “Turner’s valuable content gives it leverage in negotiations" with distributors). lmportantly, however, accepting that straightforward proposition - i.e., that popular programmers such as Turner are able to demand more for their content than less popular programmers - does not prove that the challenged merger would harm competition pursuant to the Government’s increased-leverage theory of harm. To prove its increased-leverage theory, in other words it is not sufficient for the Government to put forward evidence that Turner has important content and thus bargaining leverage - that fact is true today, pre-merger. Rather, the Government’s increased-leverage theory posits that Turner’s pre-merger bargaining leverage would materially increase as a result of its post-merger relationship with AT&T and that, as a result, distributors would cede greater affiliate fees than they would absent the merger.

To support that_contention at trial, the Government primarily relied on defendants’ own statements and documents as well as testimony of third-party competitor witnesses

most (but not all) of whom expressed concern regarding the challenged merger’s potential

effects on their businesses Neither category of evidence, however, is persuasive in proving that Turner’s post-merger negotiating position would materially increase based on its

ownership by AT&T.

2. Defendants’ Own Statements and Documents Provide Little Support for the Contention That Turner Will Gain lncreased Leverage Due to the

Proposed l\/lerger

According to the Government, defendants’ own prior statements and ordinary course business documents “recognize that vertical integration poses a threat to competition” and, thus provide convincing support for the Government’s bargaining leverage claim. See Gov’t PFOF 11 47-58. The Government points to statements made by defendants in the context of prior regulatory proceedings and statements contained in internal documents such as slide decks and emails created by various individuals within the defendant companies Neither category, however, was of any particular probative value How so?

As a general matter, the Government is undoubtedly correct that "ordinary course- of-business documents including those generated by the defendants” can be probative of whether a proposed merger is likely to result in competitive harm, Gov’t PCOL 1 49. But as with any other piece of documentary evidence assessing the probative value of defendants’ own documents and statements requires an examination of the context, circumstances and foundation of the proffered evidence As such, with few exceptions the Court denied the Government’s requests to admit into evidence and cite in post-trial

briefing a number of company documents for which there was no accompanying

background or foundation testimony See supra pp. 46-47 & nn 11-13. With the benefit of foundational testimony l have considered all of the documentary and testimonial evidence from defendants’ files and witnesses upon which the Government relied at trial. Having done so, l nonetheless conclude that the proffered statements and documents admitted are of such marginal probative value that they cannot bear the weight the Government seeks to place on them.25

First, the Government argues that defendants’ statements “made in external filings with governmental authorities” are evidence of defendants’ “understanding of the anticompetitive effects that result from this transaction” Gov’t PCOL 152. The statements in particular upon which the Government relies were made either in comments or supporting expert reports filed by AT&T or DirecTV, in the course of the following FCC proceedings l) the 2010 review ofthe Comcast-NBCU merger, see PXl (DirecT\/);

PX441 (DirecTV); 2) the 2012 proceeding to determine, inter alia, whether to allow one

25 Before proceeding further, the Court notes a bit ofconfusion in the Government’s position about the role of defendants’ alleged “anticompetitive intent” in assessing the likely harms associated with the challenged merger. Gov’t PCOL 1 51. ln opening arguments counsel for the Government stated, in reference to the predictive exercised called for by Section 7, that “courts don’t focus on intent. What they focus on is effects effects in the market.” Tr. 10:15-16. But the Government’s post-trial brief cites cases for the proposition that “[e]vidence of anticompetitive intent can also form the basis ofa court’s prediction of harm,” while at the same time noting that “absence of evidence demonstrating anticompetitive intent . . . suggests nothing.” Gov’t PCOL 1 51 & n.12.

The Court need not toil to reconcile those positions or parse the state of our Circuit’s current case law on the issue Compare Whole Foods Mkt., 548 F.3d at 1047 (Tatel, J., concurring in the judgment) (“[T]he Supreme Court has clearly said that ‘evidence indicating the purpose ofthe merging parties where available, is an aid in predicting the probable future conduct ofthe paities and thus the probable effects of the merger.” (emphasis and internal quotation marks omitted) (quoting Brown Shoe, 370 U.S. at 329 n.48)), with id. at 1057 (Kavanaugh, l., dissenting) (“[I]ntent is not an element ofa § 7 claim. . . ." (citing A.A. Poultry Farms, Inc. v. Rose Acre Farms, [nc., 881 F.2d 1396, 1402 (7th Cir. 1989)) (‘“Firms need not like their competitors they need not cheer them on to success a desire to extinguish one’s rivals is entirely consistent with, often is the motive behind, competition.”)). That is because, as discussed below, here there is nothing akin to the direct, anticompetitive intent evidence ofthe other cases cited by the Government in its post-trial brief.

of the FCC’s program access rules to sunset, see PX2 (AT&T); PX442 (AT&T); PX443 (DirecTV); 3) the 2014 annual video competition proceeding see PX444 (AT&T); and 4) the 2014 review of the AT&T-DirecTV merger, see PX467 (AT&T and DirecT\/).26 Not surprisingly, the Government contends that these prior statements show that defendants have previously recognized the validity of applying its increased-leverage theory to affiliate fee negotiations See, e.g., Gov’t Post-Tr. Br. 2. But with that said: so what? Although l agree that a few of the proffered statements might be somewhat probative of the Government’s increased-leverage theory, that limited probative value cannot, and does not, overcome the numerous insufficiencies with the Government’s case discussed below.

ln particular, in examining defendants’ prior regulatory filing statements 1 am mindful of the considerations discussed in the context of the third-party competitor testimony See infra pp. 91-99. When AT&T and DirecTV made many ofthe proffered

regulatory filings they acted as competitors to (or customers of) distributors whose

26 Just prior to the close of evidence, when the Government moved the Court to takejudicial notice of certain enumerated regulatory filings l noted that the materials filled a notebook that is about “4 inches thick of paper.” Tr. 3942:4-5. Given the complex analyses and arguments contained within the voluminous filings l noted that the Government was “at an absolute minimum . . . going to have to isolate and identify as to each document which statement or statements” it thought were relevant to the case for purposes of clearing Federal Rule of Evidence (“FRE”) 403. Id. at 3943:23-3944:3. ln response, counsel for the Government stated that the “memorandum that l handed up isolates and lists the specific statements and l’m happy to limit to those that are identified on page 3 and 4." Id. at 3945:1 1-13. ln its post-trial papers the Government nonetheless appears to argue that the entire expert reports appended to the prior regulatory filings are admissible under FRE 801 (d)(2) as adoptions ofdefendants. See Gov’t PCOL 1 54 & n. l 3. That is largely beside the point, however. That is because the Court declines to admit those portions of the proffered expert reports and filings not “identified on page 3 and 4” of the Government’s motion under FRE 403. Id. at 3945:11-13. ln my judgment, evaluating the coinplicated, fact-specific arguments and analyses contained with those filings and reports would essentially require a trial within a trial (recall that not even the expert reports in this case were offered into evidence by the parties), the result of which would produce evidence that is only marginally probative for all ofthe reasons discussed below.

competitive positions would be affected by FCC review. For that reason alone, l am hesitant to assign any significant evidentiary value to those prior regulatory filings

Finally, with respect to this particular categories of statements 1 particularly decline to place much stock in the statements related to the sunsetting of the FCC’s ban on exclusive contracting between certain programmers and distributors See, e.g., PX2. PX442. Many of those statements relate to the issue of withholding content - something the Government’s own expert concedes would not occur as a result of the proposed merger. Compare PX2-4 (“[V]ertically integrated programmers continue to have the incentive and ability to use (and indeed have used whenever and wherever they can) that control as a weapon to hinder competition to their down-stream cable affiliates by withholding popular programmingfrom competing MVPDs.”) (emphasis added), with Tr. 2218:13-17 (Shapiro) (“l’m not saying that after the merger, Turner will deny its content to the other distributors T his is not a foreclosure-withholding story.”) (emphasis added). Generic statements about “mushroom[ing]” bargaining power of all programmers are similarly unhelpful to evaluating the Government’s particular claims in this case PX444-3 to -4.

That brings us to select statements made by DirecTV or AT&T that relate to vertically integrated programmers’ ability to raise content prices and the use of the Nash bargaining model to estimate increased affiliate fees See, e.g., PX1-17, -83 ("[V]ertical integration of programming and distribution can if left unchecked, give the integrated entity the incentive and ability to gain an unfair advantage over its rivals.”)_; PX441-5 (noting “voluminous economic and other evidence that the proposed transaction would

enable Comcast to raise the prices paid by its MVPD rivals for NBCU programming”); 82

PX443-79 (“[V]ertically integrated MVPDs have an incentive to charge higher license fees for programming that is particularly effective in gaining MVPD subscribers than do non- vertically integrated MVPDS.”). According to the Govemment, those statements show that defendants recognize the validity of applying this bargaining model to estimate the impact ofAT&T and lime Warner’s vertical integration on affiliate fee negotiations Please ! Generic statements that vertical integration “can” allow the integrated entity to gain an “unfair advantage over its rivals” PX1-17 (emphasis added), do not come close to answering the question before the Court in relation to the Government’s increased-leverage theory: whether the Government has carried its Section 7 burden to show, through proof at trial, that Time Warner will gain increased bargaining leverage in affiliate negotiations on account of the proposed merger and, if so, whether that increased bargaining leverage would result in increased distributor or consumer costs that would constitute a substantial lessening of competition under Section 7. Cf. In re Applications of Comcast Corp., 26 FCC Rcd. 4238 124 (2011) (noting differences in FCC’s “public interest~” review and D()J’s burden for “block[ing] a transaction” under Section 7). Similarly, the arguments that the Comcast-NBCU merger would harm distributors or consumers (as well the projections of harm) were, of course, informed by the state of the market at the time of the proceeding and the particular inputs to the models presented to the FCC. See, e.g., id. app. B (Technical Appendix) (setting out various formulae and inputs used to model potential economic harm). Given all that, defendants’ specific predictions regarding the ability of a merged Comcast-NBCU to leverage price increases by threatening to withhold the

particular programming at issue is not particularly probative of whether a merged AT&T-

Time Wamer could do the same with its programming in today’s more competitive marketplace Compare id. 1 41 (“We do not determine at this time whether online video competes with MVPD services”), with Gov’t PFOF 11 14-18 (detailing role of virtual MVPDs in “distribut[ing] linear channels and on demand content to subscribers over the internet”). l\/loreover, as discussed in more detail below, defendants’ expert Professor Carlton concluded in an econometric analysis of content pricing following the Comcast- NBCU merger that, contrary to the predictions offered by competitors in the regulatory filings the merger did not cause content prices to increase See infra pp. 100-105.

That said, the Court agrees with the Government that the fact that defendants previously submitted expert reports or commentary sponsoring the use of the Nash bargaining model in the context of affiliate fee negotiations counts as a mark (albeit a faint one) against defendants’ attempts to disavow the applicability of the Nash bargaining theory in this case. Unfortunately for the Government, however, my conclusion that the Government has failed to provide sufficient evidentiary support to show the Nash bargaining theory accurately reflects post-merger affiliate negotiations or the proffered bargaining model in this case does not turn on defendants’ protestations that the theory is “preposterous,” "ridiculous” or “absurd.” Gov’t PFOF 1 47 (quoting Tr. 50:18 (Defs.’ Opening); id. at 3119:19-24 (Bewkes (Time Warner)); id. at 3430:1-11 (Stephenson (AT&T)). lt rests instead on my evaluation of the shortcomings in the proffered third-party competitor testimony, see infra pp. 91-99; the testimony about the complex nature of these negotiations and the low likelihood of a long-term Turner blackout, see infra pp. 14-18,

115-117 & nn34-36; and the fact that real-world pricing data and the experiences of 84

individuals who have negotiated on behalf of vertically integrated entities all fail to support the Government’s increased-leverage theory, see infra pp. 99-108. Therefore, even assigning some probative weight to the statements made by defendants in prior regulatory proceedings those statements do not come close to providing a sufficient evidentiary basis to prove the viability of the Government’s increased-leverage theory in this case.27 Second, to prove its increased-leverage theory, the Government relies upon random

statements from defendants’ “‘ordinary course” business documents including employees’

27 The Government takes its regulatory filings argument one step further in its post-trial briefing asserting for the first time, that defendants’ prior regulatory statements should result in them being judicially estopped from denying basic predicates ofthe increased-leverage theory of harm, Gov’t PCOL 11 74-75. To say the least, that argument is a stretch. As the Supreme Court has explained, the "equitable doctrine” ofjudicial estoppel may be “invokedl by a court at its discretion” to guard against a party’s "improper use ofjudicial machinery” to gain an “unfair advantage.” New Harnpshire v. Maine, 532 U.S. 742, 750-51 (2001) (internal quotation marks omitted). To appropriately applyjudicial estoppel against a party, the “party’s later position must be ‘clearly inconsistent’ with its earlier position”; courts also consider whether the party has “succeeded in persuading a court to accept that party’s earlier position” or would “derive an unfair advantage or impose an unfair detriment on the opposing party if not estopped." Id. (internal quotation marks omitted).

Applying those factors l easily conclude that estoppel is not appropriate here To start, the cited prior regulatory comments are not “clearly inconsistent" with defendants’ current positions predicting that a different vertical transaction made at an earlier time period and in a less-competitive market, will shift bargaining outcomes is not inconsistent with arguing that the Government has failed to carry its burden of proof to show at trial that a different transaction proposed in the context of an even more competitive market, is likely to similarly shift outcomes (much less substantially lessen competition). Maine, 532 U.S. at 750; Jankovic v. Int ’l Crisis Grp., 822 F.3d 576, 586 (D.C. Cir. 2016) (declining to apply estoppel when party’s position was not inconsistent). Although that consideration alone is fatal to the Government`s estoppel argument, the Court further notes that the equities also weigh against applying estoppel here The Government investigated the proposed merger for approximately one year before filing its suit. Disputes regarding the applicability ofan increased-leverage theory as applied to the transaction have been front and center in the litigation and were fully aired at trial. Given all that, l am hard pressed to understand how the Government would suffer an “unfair detriment” if defendants are not estopped; if anything it would seem manifestly unfair to defendants to accept the Government’s post-trial estoppel argument that much ofthe trial evidence can be ignored and indeed substituted with decades-old regulatory filings Thus, even assuming that estoppel can be applied based on statements contained within third-party regulatory comments to prior administrative proceedings but see Abtew v. U.S. Dep’t ofHome/and Sec., 808 F.3d 895, 899-900 (D.C. Cir. 2015) (“[T]he rule ofjudicial estoppel ‘generally prevents aparty from prevailing in one phase ofa case on an argument and then relying on a contradictory argument to prevail in another phase.”’ (emphasis added) (quoting Maine, 532 U.S. at 749)), the Court declines the Government’s last- minute invitation to estop defendants here,

emails and internal slide decks lndeed, the Government even featured many such statements (or, more accurately, snippets of such statements) in its Complaint and pre-trial filings However, as became clear at trial, when live witnesses take the stand a trial by slide deck leaves much to be desired l

Exemplary of this problem is a series of Government exhibits containing emails and drafts of slide decks generated prior to a merger integration meeting in 2017. See PX31; PX184; PX189; PX363. The Government has emphasized statements excerpted from those slide decks contending before, during and after trial that they highlight AT&T’s “core belief’ that the merger would help it preserve the role of “[t]raditional Pay-TV” as a "cash cow business to AT&T for many years to come” by ensuring "stability through the slow, structural decline of the industry” PX363-12 to -13; see, e.g,, Compl. 1 3 (“As AT&T/DirecT\/’s strategic merger documents state, after the merger, disruption need not occur immediately -the merged firm ‘can operate [its] pay-TV business as a ‘cash cow’ while slowly pivoting to new models”); Gov’t Pre-Tr. Br. 2-3 (same).

At trial, however, we learned that those statements were drafted by a lower-level AT&T employee who had nothing to do with the substance ofthe decision to acquire Time Warner, see Tr. 1777:16-1778:3 (Manty (AT&T)), and in any event, were contained in a preliminary draft and were subsequently removed or changed, see id. at 1732:25-1733:25. T`o be sure, Government counsel endeavored to characterize that subsequent change as a nefarious “sanitization” by lawyers but testimony indicated that the "whole deck changed” as a result ofthe parlor room process and its attendant legal review. See id. at 1738:7-13,

1744:8-13. Compare PX363 (Apr. 8, 2017), and PX31 (Apr. 9, 2017), with PX189 (Apr. 86

18, 2017). ln the final analysis no upper-level AT&T witness testified to ever having viewed or otherwise relied on the draft statements To say the least, their probative value was minimal.

As it turned out, much of the Government’s proffered “ordinary course" evidence went the way of those draft slide deck statements Compare Tr. 1713:20-23, 1714:3-6 (Gibson (AT&T)) (confirming that internal AT&T documents stated that “NBCU could become a more formidable negotiating power” and that “[c]ontent costs could increase” as a result of the expiration of the Comcast-NBCU consent decree) (internal quotation marks omitted), with id. at 1712:14, 25, 1714:1-2, 9-10 (testifying that the document in question represents a “draft understanding of some pretty complicated merger conditions” designed to “brainstorm the what-ifs” of what Comcast-NBCU “may be able to do” that the team “‘hadn’t finished”), and id. at 1715:20-21. 1717:17-18 (email chain PXl 1, contains "first very rough understanding of” Comcast-NBCU merger conditions by “individual who reported to me regarding merger conditions for the first time”). See also id. at 1770:25- 1771:12, 1772:16-25 (Manty (AT&T)) (showing that PX184, although sent to two AT&T senior vice presidents in July 2016, was generated in 2014 by team of lower-level AT&T employees and consulting firm members). l need not recount all of the examples here Suffice it to say that l find that the Government frequently “overemphasized the importance and relevance” of the excerpts from defendants’ documents given that many of them, the testimony revealed, contained "informal speculation” about “rationales for the merger" or were generated by individuals "who had no decision-making role or authority in relation

to the merger.” H & R Block, 833 F. Supp. 2d at 77 n.30; cf Dep’t ofJustice & Fed. Trade 87

Comm’n, Horizontal l\/lerger Guidelines § 2.2.1 (Aug. 19, 2010) (“Horizontal l\/lerger Guidelines”) (“The Agencies give careful consideration to the views of individuals whose responsibilities expertise, and experience relating to the issues in question provide particular indicia of reliability.”).

ln a few instances however, the Government sought to draw evidentiary support from some of AT&T CEO Randall Stephenson’s own statements and notes The Government pointed, for example, to an email that Stephenson sent upon being informed by lime Warner CEO leff Bewkes that “Time Warner had ‘taken a 10% stake in Hulu’ and that Hulu was going to launch a virtual MVPD.” Gov’t PFOF 1 51 (alteration omitted) (quoting PX47). ln response to Bewkes’ statement that he did not think the announcement would impact AT&T’s relationship with Time Warner, Stephenson stated that it was “hard to imagine how it won’t impact all of our relationships,” continuing that AT&T is “trying to figure out how we navigate a very new world where you folks are going around us while trying to preserve the old revenue streams and business models from us” PX47. At trial` Stephenson testified that his email indicated his concern that DirecTV Now. the new virtual MVPD AT&T was “standing . .. up” at around that same time, would get the “same access” as one ofits virtual competitors Hulu. Tr. 3475:21-22, 3477:6-7. ln this Court’s view, expressing concern about how a rival virtual MVPD’s relationship with lime Warner could affect AT&T’s nascent DirecTV Now platform does little to prove how AT&T would likely behave in the event of a vertical integration

The Government also relies on notes that Stephenson drafted to himself in

preparation for an AT&T Board of Directors Meeting to discuss the merger. See Gov’t

PF()F 1 52. ln those notes Stephenson listed the following as a discussion point: "How can you advantage your own distribution (TV, BB, Wireless) without harming TW position as a wide distributor of content to other SVOD, cable networks and broadcast networks” DX609-8. The Government argues that this bullet point reflects “exactly the theory of the government’s case: use content to advantage distribution” Tr. 3980:4-5 (Gov’t Closing); see also Gov’t PFOF 11 52-53. Not so. At trial, Stephenson testified credibly that the point of that note was to frame a discussion with his Board “that if there is a thought process that says we’re going to use this content to enhance the distribution business that means you’re going to have to limit the distribution” and that “is counter is how you create value in one ofthese businesses.” Tr. 3407:16-21. That testimony mirrors the contents ofa letter sent by Stephenson to all AT&T officers shortly after the announcement of the proposed merger. ln that letter, known among those in defendant companies as the “‘Magna Carta”’ of the merger, Stephenson writes “[t]o lime Wamer employees We will continue to distribute lime Warner content broadly across the industry ln fact, we want to extend its distribution deeper into mobile so all wireless companies become distribution points for Time Warner content.” DX625-1; see also Tr. 3408:16-22 (Stephenson (AT&T)).

To be sure, the Government impugns Stephenson’s explanation calling it “curious” and credulity “strain[ing]” in light ofthe testimony given about the other notes on the same page See Gov’t PFOF 1 53; Tr. 3980:21, 3981:9 (Gov’t Closing). But even should l fail to credit Stephenson’s explanation about that particular pre-Board-meeting bullet point, the contents of that bullet point fail to meaningfully advance the Government’s case To start,

as we learned at trial, there are a number of ways in which AT&T could "advantage [its]

own distribution” through use of Time Warner content without acting in accordance with the Government’s increased-leverage theory of anticompetitive harm. See, e.g., Tr. 3220:21-3221:20 (re-stacking and re-editing personalized sets of CNN news clips for access on mobile devices); id. at 3222:4-22 (shooting producing and broadcasting live sporting events in 4K resolution); id. at 3223:13-3224:4 (integration of social media and multi-screen functionality with content).

ln short, despite the Government’s efforts to paint a contrary picture, this is not a case containing direct, probative evidence of anticompetitive intent on the part of high- level executives within the merging company Cf., e.g., Whole Foods Mkt., 548 F.3d at 1044-45 (Tatel, J., concurring in thejudgment) (discussing “Project Goldmine,” as well as other merger-related documents in which Whole Foods CEO stated, among other things that company to be acquired is the “only existing . . . springboard for another player to get into this space” and that “[e]liminating” the company “means eliminating this threat forever, or almost forever”). Stephenson’s statements and the Government’s other proffered documentary evidence instead suggest, at the very most, that AT&T (or its third- party consultants) recognized that one possibility of uniting content and distribution would be to withhold or otherwise limit content from other distributors in an attempt to benefit AT&T’s distribution platforms But evidence indicating defendants’ recognition that it could be possible to act in accordance with the Government’s theories of harm is a far cry from evidence that the merged company is likely to do so (much less succeed in generating anticompetitive harms as a result). Cf. Baker Hughes 908 F.2d at 984 (“Section 7 involves

probabilities not certainties or possibilities.”). That is especially true when the

Government’s documentary evidence is weighed against the considerable contrary evidence - including other evidence related to the motivation for the challenged merger - that came out at trial. See, e.g., Defs.’ PFOF 11 49-62 (collecting evidence regarding the proposed merger’s ability to “‘enable the combined company to respond to the challenges posed by the current transformation ofthe video marketplace and, in so doing bring better products and better value to consumers”); see also supra pp. 36-40. Thus taking such documentary evidence for all it’s worth, that evidence is only marginally probative of the

viability of the Government’s increased-leverage theory of harm.

3. Third-Partv Competitor Witness Testimonv Provides Little Support for the Contention That Turner Will Gain lncreased Leverage Due to the

Proposed Merger

ln further support of its bargaining leverage claim, the Government called a number of third-party witnesses from AT&T’s competitor video distribution companies to the stand. Although such companies are “customers” that purchase Turner content, Tr. 18:15 (Gov’t Opening), all of them are also competitors of AT&T’s video distribution services See, e.g., Tr. 82:7-8 (Fenwick (Cox)); id. at 263:19-24 (Schlichting (DISH)). Not surprisingly, most of the third-party competitor witnesses testified that they oppose the challenged merger for a number of reasons According to the Government, that “direct industry evidence” supports its bargaining claim by describing “how the merger would increase Time Warner’s leverage over distributors.” Gov’t Post-Tr. Br. 8. l disagree For the reasons discussed below, the third-party competitor witness testimony fails to provide

meaningful, reliable support for the Government’s increased-leverage theory

As has been observed in the context of other merger cases l start by noting the difficulty of determining just how much weight to give the proffered third-party competitors’ concerns about the challenged merger. On the one hand, such testimony can provide the Court with insight into the nature of the industry and a proposed transaction’s potential effects in the market. See Gov’t PCOL 11 48-49. On the other hand _ and particularly in the context of a vertical merger case where, as here, upstream customers are downstream competitors - there is a threat that such testimony reflects self-interest rather than genuine concerns about harm to competition Cf Arch Coal, 329 F. Supp. 2d at 145 (citing 2A Areeda & Hovenkamp, Antitrust Law 1 538b, at 239 (“‘subjective’ testimony by customers” is “often unreliable”)); Horizontal Merger Guidelines §2.2.2 (noting possibility that customers may voice opposition to merger “for reasons unrelated to the antitrust issues raised by that merger”); Tr. 2462:14-23 (Carlton) (noting that a “rival doesn’t want to see a transaction that makes it[s] competitor more efficient,” even though such a result may be “good for consumer[s]”). As in any Section 7 case, however, the central issue here is whether the Government has proffered sufficient support for the anticompetitive effects it asserts it is not about protecting AT&T’s rivals from any and all competitive pressures they would experience should the merger go through. Cf. Aetna, 240 F. Supp. 3d at 18 (“[T]he Clayton Act protects ‘competition,’ rather than any particular competitor.”) (citing Baker Hughes, 908 F.2d at 988, 991 n12). Caution is therefore necessary in evaluating the probative value of the proffered third-party competitor testimony Cf. Ken Heyer, Predicting the Competitive Effects of Mergers by Listening to

Customers, 74 Antitrust L.J. 87, 127 (2007) (“ln evaluating the likely competitive 92

consequences of proposed mergers competition authorities and courts properly weigh the totality of the evidence, refusing to take the views expressed by customers at face value and insisting that customer testimony be combined with economic evidence providing objective support for those views . . . .”).

For starters l would note that not all third-party witnesses provided testimony supportive of the Government’s predictions that Turner’s post-merger bargaining leverage would increase as a result of its relationship with AT&T. F or example, when Comcast lead negotiator Gregory Rigdon was asked whether he believed the merger would increase Turner’s bargaining leverage, he answered in the negative noting that he didn’t “have any reason to believe that it will impact my negotiations with lurner or HBO.” Tr. 884:5-6 (Rigdon (Comcast)). Thus the evidence indicates that AT&T’s largest video distribution competitor - and thus a significant source of harm in Professor Shapiro’s model, see, e.g., id. at 2665:3-7 (Katz) - does not anticipate changing its negotiating strategy with respect to a post-merger Turner. Along those same lines Randy Sejen a recently-retired negotiator from Cable ONE, testified that when negotiating with a programmer, “it doesn’t matter to us who owns the network.” Id. at 2102:6-7 (Sejen (Cable ONE)). ln short, the Government’s third-party competitor witnesses were not consistently concerned regarding Turner’s ability to demand increased affiliate fees post-merger.

lt is the case, however, that other third-party competitor witnesses expressed “concern about the increased” bargaining leverage or other competitive gains on the part of Turner “that will result from the proposed transaction" Arch Coal, 329 F. Supp. 2d at

145. Their testimony, however, suffered from shortcomings that, when viewed in light of

my fundamental concerns with crediting the “subj ective views of customers in the market,” id., undermine the probative value of their evidence in supporting the Government’s predictions of Turner’s increased-bargaining leverage

l\/luch of the third-party competitor testimony l heard consisted of speculative concerns regarding how the witnesses thought Turner might act in negotiations after the merger. Some witnesses simply accepted key assumptions ofthe Government’s increased- leverage theory without any supporting analysis or data. For example, testimony from the Government’s lead-off witness Cox negotiator Suzanne Fenwick, helps to illustrate both of those problems When asked on direct examination about her views of the proposed merger, Fenwick stated that she is “very concerned” that, post-merger, Cox would be presented by Turner with “a horribly ugly deal and that when faced with that deal, we have to think about that if we do go dark, they have a benefit in picking up Cox customers” via DirecTV. Tr. 107:18-21 (Fenwick (Cox)). Fenwick continued that, as a result of that "benefit that is created in this merger that isn’t there today,” the negotiating "leverage changes” and that AT&T “has a different incentive now than they had before" ~ namely, the incentive to “pick up customers” lost by Cox in a Turner blackout. Id. at 107:12-14, 108:7-9,148:1-2.

Fenwick’s speculation about how Turner might act relies on certain key assumptions for which she had no factual basis lndeed, the amount of customers that distributors would lose as a result of a Tumer blackout (not to mention the resulting “benefit" to AT&T), is

one of the central disputes in this case. Without offering any supporting analysis Fenwick

simply assumes those figures to be in line with the Government’s predictions a point highlighted by the following exchange during cross-examination:

Q: So let’s talk about that. How many customers are going to leave [Cox] even with the reduction in your price to your cable subscribers how many?

A: We don’t know.

Q: Have you tried to compute it?

A: l have not.

Q: You have no idea?

A: We believe that it’s a large number.

Q: l know you believe that, but do you have any evidence any information any hard facts?

A: l don’t have a churn analysis for you, no.

Q: Do you think you had an obligation in giving testimony to oppose a merger of this importance that you would do some homework and run some numbers?

A: No, we felt like ourjob was to point out how the leverage changes Q: So you think you could just come in here and give your opinion that the leverage is going to change and you’re going to lose all ofthese customers even though you have no idea how many customers you’re going to lose and you’ve never done a single bit of quantitative analysis is that true? A: Sure.

Id. at 141:1-142:5; see also id. at 147;22-148:10. Testimony from other third-party witnesses suffered from similar problems DlSH

Sling president Warren Schlichting testified that the merger would “kind of throw[] the

card table up in the air” by placing Turner in a “‘win win” situation where they “can raise

prices and make more money and make us less competitive, or they can raise, they can present onerous terms that we can’t accept.” Id. at 261:24-25, 262:8-22 (Schlichting (DISH)). That was so, according to Schlichting because DlSH would lose “a lot of subs" in the event of a Turner blackout and most of those lost subscribers “‘would accrue to [DirecTV’s] benefit.” Id. at 262:19-21. RCN CEO Jim Holanda testified that he feared his company would lose access to certain lime Warner programming rights even though he had no ‘,‘empirical data or any real-world evidence of subscriber losses if RCN didn’t have Turner.” Id. at 2947:10-13 (Holanda (RCN)). Just as with Fenwick’s testimony Schlichting’s and Holanda’s contentions about Turner’s post-merger position - including the amount of subscribers they would lose and AT&T would gain - assume away many of the disputed issues in this case Cf. id. at 404:22-405:3 (Schlichting (DlSH)) ("Q: You don’t have any calculations about how many subs DlSH would lose or Sling would lose if there were a blackout let’s say today . . . A: No.”).

Some third-party competitor testimony even contradicted the testimony of the Government’s lead expert, Professor Carl Shapiro. Cf Staples, 970 F. Supp. at 1085 (declining to “give . . . much weight” to party’s testimony that was “contradicted by other evidence” submitted by the party). For example, Schlichting’s testimony regarding Turner’s increased post-merger leverage assumes that Turner would profit from, or at the very least would be willing to accept, a long-term blackout of DISH. See, e.g., Tr. 263:10- 12 (Schlichting (DlSH)) (stating that Turner may be incentivized to blackout DlSH because “it’s always it’s more lucrative to take subs than it is to, you know, collect programming

programming fees”); id. at 264:6-8 (“Q: So you would expect to be more likely to go dark

[with Turner] if the merger goes through‘? A: l would.”). Tom l\/lontemagno, a lead negotiator for Charter, testified similarly He noted that his concern with the challenged merger is “mainly around what’s going to happen with excessive price, pricing increases” and specifically whether Charter will “lose access to critically important content that AT&T make take exclusive away from our customers and make it harder for [Charter] to compete.” Id. at 1350:12-15, 1352:1-3 (Montemagno (Charter)). The assumptions reflected by that testimony - namely, that a post-merger Turner could and would go dark with DlSH or Charter - run directly contrary to Professor Shapiro’s testimony that a post- merger Turner would not be incentivized to blackout or otherwise withhold its content from distributors See id. at 2293:3-4, 14-15 (Shapiro) (Turner will “continue to license Turner content” to distributors after the merger); id. at 2218:13-21 (“l’m not saying that after the merger, Turner will deny its content to the other distributors”). lndeed, when asked whether he was “aware” of Professor Shapiro’s opinion that “it would not be profitable for the merged company to withhold the Turner Networks from DlSH and other distributors" Schlichting admitted that he was not, Id. at 417:13-17, 418:15-16 (Schlichting (DlSH)). Other concerns raised by the third-party competitors were not particularly germane to the Government’s Section 7 allegations in this case. Charter’s l\/lontemagno, for example, noted his concerns that the merger would harm Charter’s competitive position due to the bundling of the Turner networks and the ability of DirecTV to use advertising to appeal to Charter’s customers See id. at 1405:13-18 (Montemagno (Charter)). On cross-examination however, l\/lontemagno conceded that “none of those issues are a result

of this merger,” but instead “all exist in the marketplace today.” Id. at 1407:12-18; see

also id. at 1407:19-23 ("Q: And AT&T, DirecT\/, if it wanted to buy ads on Turner or anybody else in order to try to lure your customers away, they could do that today they could do that yesterday couldn’t they? A: They can buy them yes.”). Holanda grounded RCN’s concerns about the challenged merger in a prior experience with Comcast-NBCU and negotiations over RCN’s "broadcast basic” package Id. at 2920:6-23, 2921;2-6 (Holanda (RCN)). But that experience is not especially probative of the Government’s increased-leverage theory, given that the Turner networks do not include major broadcast programming and, in any event, that penetration rates exist in the pre-merger market. See id. at 2955:9-12.

Finally - and perhaps unsurprisineg given that a post-merger Turner, like a pre- merger Turner, would stand to suffer large losses in affiliate fee and advertising revenues in the event of a blackout ~ the record is barren of any contentions by the third-party competitors that they would actually give in to any price increases demanded by Turner as a result of its purported increase in post-merger leverage Schlichting never testified, for instance, that DlSH would in fact pay more to Turner for its content as a result of the merger, noting instead that “l don’t think we’ve quite figured out what we would do" during post-merger negotiations with Turner. Id. at 264:11-12 (Schlichting (DISH)). The lack of real-world evidence that Turner would likely be successful in obtaining increased fees from virtually every distributor (as Professor Shapiro’s model projects) due to its relationship with AT&T is yet another strike against the Government’s increased-leverage theory of competitive harm, Cf. Anthem, 855 F.3d at 360 (describing as “farfetched” the

assumption that contractual negotiations will lead to the same outcome “in every instance,"

especially in light of the fact that contracts at issue were “customized relationship-driven contracts” (internal quotation marks and alteration omitted)).

ln the final analysis the bulk of the third-party competitor testimony proffered by the Government was speculative, based on unproven assumptions or unsupported - or even contradicted - by the Government’s own evidence Especially in view ofthe fact that the third-party competitor witnesses have an incentive to oppose a merger that would allow AT&T to increase innovation while lowering costs such testimony falls far short of persuasively “show[ing] that this merger threatens” to harm competition by allowing

Turner to wield increased bargaining leverage Gov’t Post-Tr. Br. 8.

4. Real-World Evidence lndicating That Prior Vertical lntegration of Programmers and Distributors Has Not Affected Affiliate Fee Negotiations Undermines the Government’s lncreased-Leverage Theorv of Harm

For the reasons discussed above, the Court is not convinced that the “real-world objective evidence” offered by the Government provides sufficient support for its increased-bargaining leverage claim. That conclusion is further bolstered by evidence relating to three prior instances of vertical integration in the video programming and distribution industry: 1) News Corp., a programmer, acquiring part of DirecTV in 2003 and then spinning it off in 2008; 2) the 2009 split of Time Warner, a programmer, from Time Warner Cable, a MVPD; and 3) the 201 1 combination ofComcast, a distributor, and NBCU, a programmer. See Defs.’ PFOF 1 96; Tr. 2440:4-8 (Carlton). According to defendants the econometric analysis of their chief economic expert, Professor Dennis

Carlton and witness testimony both provide significant, real-world evidence indicating

that, contrary to the Government’s increased-leverage theory those prior instances of vertical integration did not affect affiliate fee negotiations or content prices For the

following reasons the Court agrees with defendants

a. Professor Carlton ’s Econometric Analyses ofPrior Vertical Transactions Found No Statistically Significant E/§fects on Content Pricing

When it comes to evaluating the antitrust implications of proposed mergers both Professor Shapiro and Professor Carlton recognize that empirical analysis of prior, similar transactions can be “convincing evidence” Tr. 2526:13 (Carlton); see id. at 3885:25- 3886:20 (Shapiro) (agreeing with the “general thrust” of statement that “compar[ing] the observed changes from completed mergers against premerger predictions” is the “most direct way” to gauge the “reliability ofdifferent methods ofevaluating proposed mergers"); cf. Horizontal l\/lerger Guidelines § 2.1.2 (“The Agencies look for historical events or ‘natural experiments,’ that are informative regarding the competitive effects of the merger. For example, the Agencies may examine the impact of recent mergers entry, expansion or exit in the relevant market.”). ln this case, however, neither the Government nor Professor Shapiro presented original analysis of any prior vertical transactions in this industry See Tr. 2337:11-13 (Shapiro) (“l did not end up doing my own separate analysis" of transactions analyzed by Professor Carlton); id. at 2473:22-25 (Carlton) (“Professor Shapiro did no econometric analysis of any ofthe data as far as l can tell.”); see also Defs.’

PFoF 1111 96, 99.28

28 lndeed, when asked in discovery whether it had a position on whether these transactions affected content prices the Government cited to one FCC study related to the News Corp.-DirecTV transaction and

Defendants by contrast, did seek to analyze the available pricing data resulting from prior instances of vertical integration Although they initially had trouble obtaining some of the relevant pricing data from the Government or third-parties see supra pp. 44-45, they were eventually able to obtain the data after seeking relief from this Court, see id.; 1/22/18 Order. Defendants’ lead economic expert, Professor Dennis Carlton then analyzed that third-party pricing data, among other proprietary and public-source data in his possession to test whether it is “true that content prices are higher on a network when it’s sold by someone who’s vertically integrated.” Tr. 2470:10-12 (Carlton). Specifically, Professor Carlton performed a “regression analysis or an econometric analysis which is a statistical attempt to answer the question precisely.” Id. at 2473:1-2. ln running his regressions Professor Carlton used different “statistical techniques to analyze the problem in a variety ofways" Id. at 2473:7-8.

All of that analysis Professor Carlton testified, generated “completely consistent" results across all three examples he considered: “There’s absolutely no statistical basis to support the government’s claim that vertical integration in this industry leads to higher content prices.” Id. at 2473:13, 2440:13-15; see id. at 2470:13-17, 2476:22-24. The “bulk of the results” Professor Carlton explained, “show no statistically significant result at all,” although “many do show a decrease” in content prices Id. at 2477:7-12 (emphasis added).

l\/loreover, Professor Carlton noted that his results are particularly “compelling” in light of

stated that, beyond that study, “the United States does not, at this time, have a position as to whether any prior vertical integration between a programmer and a distributor resulted in higher video programming fees” or “higher prices for consumers” than “would have prevailed absent the integration.” DX893-28 to - 29.

the fact that the industry, as reaffirmed by numerous witnesses at trial, is “more competitive” today than at the time of the prior transactions he analyzed Id. at 2476:6-9; see also id. at 1398:24-25 (l\/lontemagno (Charter)) (video distribution business is “more competitive now than l’ve ever experienced in my career"); id. at 2134:1-3 (Sejen (Cable ONE)) (“Q: ln your 31 years in the industry, have you ever seen it more competitive at the distribution level‘? A: No.”); id. at 2950:2-6 (Holanda (RCN)) (“Q: And so in the course of this 30 years that you have been in the business the video distribution market today is more competitive than at any point that you can recall, true? A: True.”); id. at 3213:9 (Stankey (AT&T)) (competition in industry is “at an all-time high”). ln short, based on his analysis Professor Carlton stated that there has been “nothing like” the price increases predicted by Professor Shapiro following prior instances of vertical integration of programmers and distributors Id. at 2470:19-20 (Carlton).

Although the Government and Professor Shapiro sought to undermine the basis for Professor Carlton’s conclusions at trial, those efforts were unavailing Professor Shapiro, for his part, critiqued Professor Carlton for relying on faulty data and attempting to draw conclusions from prior transactions that are not comparable to the challenged merger. Focusing on Professor Carlton’s reliance on SNL Kagan data, Professor Shapiro stated that such data is “pretty poor” because it relies on “public sources” and reports content costs “to all of the distributors on average” Id. at 3831:11-18 (Shapiro). Of course, Professor Carlton testified that he relied not only on SNL Kagan data, but also on data from third parties such as DirecTV, DlSH, and Charter - all of which, when analyzed, showed no

statistical pricing effects associated with the relevant prior instances of vertical integration

Id. at 2470:4-12 (Carlton). Taking Professor Shapiro’s critiques of the SNL Kagan data on their own terms however, those critiques miss the mark. For one thing even Professor Shapiro acknowledged that SNL Kagan data is “commonly used" by individuals in the industry Id. at 3889:3 (Shapiro); see also, e.g., id. at 1073:20-1074:4 (Breland (Turner)).29 l\/loreover, it was SNL Kagan data that formed the basis of the only study of prior harm cited by the Government and Professor Shapiro. Id. at 3889:4-9 (Shapiro) (agreeing that FCC study that he “relied on” in his expert report was “based on Kagan data”); Gov’t Post- Tr. Br. 16 (citing same FCC Study); DX893-28 (Gov’t answer to interrogatory citing same FCC study); see also Tr. 2467:21-2468:9 (Carlton). For those reasons Professor Shapiro’s criticisms of defendants’ prior transaction data does not, in this Court’s view, detract from Professor Carlton’s expert opinion that defendants’ evidence related to the prior transactions is “especially probative” when considering the Government’s claims of harm. Id. at 2475:21-22 (Carlton); see id. at 2441:13-20 (“lgnoring that evidence is a big mistake”).

Professor Shapiro and the Government also denounced Professor Carlton’s analysis on the basis that the prior vertical transactions are not sufficiently similar to the challenged merger. They pointed out, for example, that two of the prior transactions involved regional cable distributors (Comcast and Time Warner Cable), whereas the challenged merger involves DirecTV, which operates nationally Regional operation means Professor

Shapiro testified, that one would “not expect[] to see evidence of post-merger price

29 One more witness testified to this fact in sealed testimony Tr. 930:17-18 (SEALED). 103

increases beyond the overall industry increases" because “most of the MVPDs . . . don`t compete with Comcast,” for example Id. at 2338:8-13 (Shapiro); cf. id. at 2558:18- 2559:15 (Carlton). Professor Carlton explained, however, that the regional versus national distinction is “irrelevant” when it comes to his analysis of DirecTV and DlSH prices that is so, Professor Carlton stated, because those two satellite companies compete "everywhere” the regional cable companies operate and it is the “national share” that matters to Professor Shapiro’s bargaining model. Id. at 2474:11-17, 2560:5-1 1 (Carlton). To the extent the Government is now arguing that one would not expect to see any increased-leverage harm due to Comcast’s status as a regional distributor, l simply note that the Government argued to the contrary prior to this case See, generally, e.g., Compl., Comcast Corp., 808 F. Supp. 2d 145 (No. 11-cv-106).

Finally, the Government and Professor Shapiro note that the prior vertical transactions all were “remediated” by regulatory or court-ordered conditions - conditions that will not apply to the challenged merger. Tr. 3830:20 (Shapiro). Professor Carlton agrees that, in theory his study’s conclusions would be affected if the conditions associated with the prior transactions were not “sufficiently similar” to those at issue here, Id. at 2558:12-15 (Carlton). l will thus briefly address Turner’s 2017 arbitration offer and its relation to the conditions on the Comcast-NBCU transaction

The arbitration proceedings envisioned by Turner’s offer are similar in many of “the fundamental ways” to those blessed by the FCC, DOJ, and this Court in the Comcast- NBCU merger. Defs.’ PFOF 1 214 (citing Tr. 2680:1-9 (Katz)); see also id. 1 225. l\/lost

notably both arbitration arrangements are “baseball-style”: each party puts forward a final

offer before knowing about its counterparty’s offer, and the arbitrator chooses between those two. Tr. 2680:1-9 (Katz). ln addition both sets of arbitration arrangements contain “standstill provisions” which prevent the blackout of content while the arbitration is pending Id. They also both set out “fair market value” as the standard, and have similar discovery procedures Id. at 2680:1-13. As Professor Katz testified, “the objective is the

same The overall structure the same. So they are similar overall. Id.; see also id. at 2958:12-16 (Holanda (RCN)). Given all of these similarities l conclude that Professor Carlton’s econometric analysis of the pricing effects of the Comcast-NBCU combination can be afforded probative weight in predicting the potential pricing effects of the challenged merger.30

To sum it up, neither the Government nor Professor Shapiro has given this Court an adequate basis to decline to credit Professor Carlton’s econometric analysis And that analysis according to Professor Carlton definitively shows that prior instances of vertical

integration in the video programming and distribution industry have had no statistically

significant effect on content prices

30 The parties spent a good deal of the trial debating the finer points of Turner’s November 2017 arbitration offer, made shortly after the filing of the Complaint in this case The Government asserts that the arbitration commitment must be ignored or, at the very least, must be proven binding and effective by defendants while defendants describe its absence from Professor Shapiro’s model as a critical weakness in the model’s design and the Government’s prima facie case Compare Gov’t Post-Tr. Br. 21-22, with Post- Trial Br. of Defs. (“Defs.’ Post-Tr. Br.”) 14. For purposes of this discussion as explained below, 1 have confidence that Turner’s arbitration offer will have real-world effect and, thus that it is appropriate to consider Professor Carlton’s econometric analysis of the Comcast-NBCU transaction See infra n.51.

b. Executives from Vertically lntegrated Programmers and Distributors Testifi`ed T hat Vertical Integration Does Not Affect Ajj"i`liate Fee Negotiations

Professor Carlton’s analysis of prior vertical integration is further reinforced, defendants contend, by the consistent testimony of Comcast-NBCU and Time Warner executives that the integration of programming and distribution does not affect affiliate negotiations lagree.

Defendants first point to the testimony from Madison Bond, who has served as a lead negotiator for NBCU during the past seven years when the company has been vertically integrated with Comcast. When questioned by defense counsel about his prior negotiations on behalf of NBCU, Bond testified that he “never once took into account the interest of Comcast cable in trying to negotiate a carriage agreement.” Tr. 2014:22-24 (Bond (NBCU)). Consideration of potential Comcast gains during an NBCU blackout "doesn’t factor at all” into his negotiations Bond continued, nor has anyone from Comcast "ever asked” him “to think about that.” Id. at 2015:1, 2015:10-12. Bond’s statements were similar to testimony given by Comcast’s chief negotiator, Greg Rigdon who testified that he has never suggested, or seen a Comcast document suggesting that NBC “should go dark on one of [Comcast’s] competitors because then [Comcast] might pick up some subscribers” or that NBCU should “hold out for a little bit more in affiliate fees because

that will harm” Comcast’s competitors Id. at 882:22-24, 883:1-1 1 (Rigdon (Comcast)).31

31 ln response the Government asks this Court to ignore the import of that testimony from the Comcast and NBCU witnesses on the basis that the conditions governing the Comcast-NBCU transaction would have prevented any coordination between the programming and distribution components and thus rendered such conversations between the two pointless See Gov’t Post-Ti'. Br. 19 n14. Pleasel The Comcast and NBCU witnesses’ testimony aligns with testimony from witnesses not subject to the FCC

lime Warner executives testified similarly about their time at the company when it was vertically integrated with lime Warner Cable. Recalling that period, lime Warner CEO leff Bewkes testified that he was not aware of any lime Warner negotiator “articulating this theory of added incentive or added ability to leverage a price increase” because lime Warner was “vertically integrated with lime Warner Cable.” Id. at 3121 :22- 3122:8 (Bewkes (Time Warner)). Turner CEO John Martin, who served as CFO of lime Warner Cable at the time it was vertically integrated with lime Warner, testified along the same lines as did lurner lead negotiators Coleman Breland and Richard Warren See id. at 601:10-602:15 (Martin (Turner)) (“Q: Did you ever hear anyone say that Turner would have more leverage because lime Warner Cable and Turner were in the same family? A: No, l did not.”); id. at 1129:6-12 (Breland (Turner)) (“l’ve been in Turner when we were a vertically integrated company and had a sister company called Time Warner Cable And l can tell you at no time during my tenure there did anyone ask me to consider in my negotiations and how l dealt with other distributors the outcome and impact at lime Warner Cable . . . .”); id. at 1190:14-15 (Warren (Turner)) (noting when asked about Government’s increased-leverage theory that “[w]e didn’t do that when we were part of

Time Wamer Cable”). l\/lartin also testified that Time Warner’s content prices did not

order’s conditions and is also entirely consistent, as subsequently discussed, see infra pp. 1 14-1 17, with the goal of companywide profit maximization. See Tr. 601:10-602:15 (Martin (Turner)); id. at 1129:6-12 (Breland (Turner)); id. at 1190:14-15 (Warren (Turner)); cf id. at 2102:6-11 (Sejen (Cable ONE)) (“l mean it doesn’t matter to us who owns the network . . . . lt really doesn’t matter.”). For that reason among others see infra nn 34, 36, l decline the Government’s invitation to disregard the Comcast and NBCU witnesses’ testimony referenced in this section

decrease following the spin-off of lime Warner Cable. See id. at 603:24-604:1 (l\/lartin (Turner)).

The Government seems to believe that any “post-merger” testimony given by lime Warner executives should be "discount[ed]” as potentially biased because it was given by interested employees of a defendant company Gov’t PCOL 1 56. Poppycock l The testimony at issue does not involve promises or speculations about the employees’ future post-merger behavior. Rather, it is testimony about what these executives previously experienced when working within a vertically integrated company That testimony regarding executives’ prior experiences in the industry is uniform among all testifying witnesses and unrebutted by the Government; moreover, it finds independent support in the analysis performed by Professor Carlton For those reasons l decline the Government’s request to discount it.

To be sure neither Professor Carlton’s econometric analysis nor the testimony discussed above provides “perfect evidence” of what will happen as a result of the challenged merger. Tr. 2475:15-17 (Carlton). But when weighed against the relatively weak documentary and third-party testimonial evidence proffered by the Government in support of its increased-leverage theory the real-world evidence indicating that vertical integration has not affected content prices or affiliate negotiations further undermines the

persuasiveness of the Government’s proof.

C. The Government’s Expert Testimony ls Also Insufficient to Support Its Increased-Leverage Theory of Harm

ln addition to offering the so-called “real-world objective evidence” set out above the Government called noted antitrust economist, Professor Carl Shapiro, to testify in support of its increased-leverage theory Professor Shapiro first discussed the academic underpinnings of the theory explaining that it was grounded in an economic concept known as the Nash bargaining theory Thereafter, Professor Shapiro opined that Turner’s post-merger leverage would increase pursuant to those economic principles ln order to predict the increased distributor costs and consumer harms that would result from Turner’s increased post-merger leverage Professor Shapiro constructed economic models Acknowledging that proper antitrust analysis of a proposed vertical merger requires balancing the merger’s proconsumer benefits with its harms see supra pp. 52-54, Professor Shapiro testified that the challenged merger would result in annual consumer cost increases that would far outweigh the $350 million in annual EDl\/l savings he conceded the merger would generate He thus concluded, based on his economic modeling that the merger was likely to cause a substantial lessening of competition by increasing consumer costs as a result of Turner’s increased bargaining leverage

At trial, defendants mounted a series of attacks on Professor Shapiro’s analysis They challenged Professor Shapiro’s threshold contention that the economic theory of Nash bargaining can accurately predict the dynamics and final fee structure of complex affiliate fee negotiations They also asserted that the theory as applied here, rests on

improper assumptions - including the notion that Turner could gain increased leverage

from threatening a long-term blackout ~ that negate its usefulness in evaluating the real- world effects of the proposed merger. Finally defendants both through their own experts and their examinations of industry witnesses argue that Professor Shapiro’s inputs are faulty and note further that use of the proper inputs would cause the model to predict that the merger will have a net benefit to consumers rather than a net harm. As will become clear in the section that follows l largely agree with defendants’ various critiques of Professor Shapiro’s testimony

For starters l couldn’t help but notice that the more and more questions were raised during the trial about the reliability of Professor Shapiro’s theory and model, the more the Government appeared to be minimizing the importance of his analysis Cf. Defs.’ Post-Tr. Br. 10 (noting Government’s attempt to “retreat from the model” in its closing argument). lndeed, during its closing argument, the Government touched on Professor Shapiro’s model relatively briefly arguing that it simply “confirmed what the industry witnesses had already explained.” Tr. 4000:5-6 (Gov’t Closing). And the Government’s post-trial filings for their part, all but ask the Court to overlook any failings of the model, arguing that “Section 7 does not require any quantification of harm from a price increase” and that "it would be perverse to penalize a plaintiff that does provide a quantification of the potential price increase." Gov’t PCOL 1 20; see also Gov’t Post-Tr. Br. 15 (“[D]efendants’ critique of

Professor Shapiro’s model misses the bigger picture: the model is but one part of Professor

Shapiro’s opinion and his opinion is one part of the United States’ evidence.”).32 Go figure !

With that, l will now turn to my own evaluation of Professor Shapiro’s expert testimony First, l will explain why the evidence is insufficient to support Professor Shapiro’s conclusion that this Nash bargaining theory will accurately predict an increase of Turner’s post-merger bargaining leverage in affiliate fee negotiations with distributors Second, 1 will examine Professor Shapiro’s economic bargaining model, concluding that the evidence is also insufficient to support the input values upon which he relied to generate

his predictions of harm.

1. The Evidence ls lnsufficient to Support Professor Shapiro’s Conclusion That the Merger Will Increase Turner’s Bargaining Leverage ann in Turn Affiliate Fees

Relying on a particular economic bargaining theory Professor Shapiro opines that,

due to its post-merger relationship with AT&T, Turner’s leverage in affiliate negotiations

32 To the extent the Government’s increased-leverage theory now leans more heavily for support on the industry Witness testimony and defendants documents as "framed” by Professor Shapiro’s analysis more generally Gov’t Post-Tr. Br. 8, that shift in emphasis fails to salvage its claim given the independent problems with that so-called “real-world objective evidence” set out in the section above See supra pp. 75-109; cf Defs.’ Post-Tr. Br. 10 (“But adding zero to zero is hardly a sound way to prove a price increase”). ln the Court’s view, however, it is worth noting that the Government’s retreat from Professor Shapiro’s model cannot be squared with Professor Shapiro’s testimony (seemingly approved by the Government) that to perform a valid “veitical merger analysis” under the applicable "consumer welfare" standard, it is necessary to “balance" or “tradeoff’ the merger’s proconsumer benefits with any predicted consumer harms See Tr. 2180:8-2181:8, 2182:7-21, 2253:4-5 (Shapiro). At trial, that “somewhat different"` “balancing" analysis ofthe challenged vertical merger was enabled not by the testimony of the third-party competitors or defendants’ documents and statements but by the cost-benefit predictions Professor Shapiro generated through use of his models See id. at 2182:17-18, 2252:19-2253:15. For that reason asking the Government to provide sufficient support for the proffered bargaining model is not, as the Government seems to argue penalizing them for failure to quantify the “specific magnitude of the potential harm,” Gov’t PCOL1 16, but instead is simply part and parcel ofwhat Professor Shapiro testified is necessary to determining whether the proposed vertical merger will harm consumers overall.

will increase due to a reduction in financial exposure in the event of a long-term blackout. Professor Shapiro in turn opines that, as such, a post-merger Turner would be able to secure greater affiliate fees from distributors

lt is beyond dispute that, to be probative in a particular case expert testimony must incorporate assumptions that are "reasonable” in light of the record evidence Joint Statement 8; cf. Brooke Grp. Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 242 (1993) (“When an expert opinion is not supported by sufficient facts to validate it in the eyes of the law, or when indisputable record facts contradict or otherwise render the opinion unreasonable it cannot support ajury’s verdict.”). Hewing to that rule is especially important in Section 7 cases where the Supreme Court’s observation that “only" an "examination of the particular market - its structure history and probable future - can provide the appropriate setting for judging the probable anticompetitive effect of the merger” dictates that the disputes “must be resolved on the basis of record evidence relating to the market and its probable future" General Dynamics, 415 U.S. at 498 (internal quotation marks omitted); Arch Coal, 329 F. Supp. 2d at 116-117. “Hence” to borrow a

line from one of my able colleagues “antitrust theory and speculation cannot trump facts Arch Coal, 329 F Supp. 2d at 116; accord Gov’t PCOL 1 22 n6 (quoting Steven C. Salop, Invigorating Vertical Merger Enforcement, 127 Yale L.J. 1962, 2018 (2018) (“[T]he direction ofthe net competitive effect is a question of fact, not theory . . .”). That is true no matter whether the testimony relates to a theory that is considered “inainstream” or has

been deemed applicable to different factual or economic scenarios in other proceedings

Gov’t PFOF 1202; cf Gov’t Post-Tr. Br. 9.33 Unfortunately for Professor Shapiro, the facts adduced at trial regarding the real-world operation of affiliate negotiations demonstrated that his testimony “rests on assumptions” that are "implausible and inconsistent with record evidence.” Matsushita Elec. lndus. Co. v. Zenith Radio Corp., 475 U.S. 574, 594 n.19 (1986).

lo start, various industry witnesses testified that the identity of a programmer’s owner does not affect the negotiating dynamic. lndeed, this opinion by Professor Shapiro runs contrary to all of the real-world testimony during the trial from those who have actually negotiated on behalf of vertically integrated companies While l need not repeat their testimony here l would simply note that the witnesses consistently testified that they had never considered the identity ofthe programmer’s owner in the course of affiliate fee negotiations See, e.g., Tr. 2014:22-2015:14 (Bond (NBCU)); id. at 882:22-24. 883:1-11 (Rigdon (Comcast)); id. at 3121 :22-3122:8 (Bewkes (Time Warner)); id. at 601:10-602:15

(Martin (Turner)); id. at 1129:6-12 (Breland (Turner)); id. at 1190:14-15 (Warren

11 On that score it is notable that, although the Government states that its proffered bargaining model is “a standard model that is in economics textbooks and widely used by economists" Gov’t PFOF 1 202, Professor Shapiro acknowledged that, with respect to the model, “[w]hat’s less common is to use it to evaluate a merger or a vertical merger especially,” Tr. 2390:2-3 (Shapiro).

To support Professor Shapiro’s testimony regarding economic bargaining theory and his model, the Government contends that defendants’ experts “endorsed” application of the model generally but quibbled with the model’s inputs Gov’t Post-Tr. Br. 2. That characterization is questionable especially given Professor Carlton’s extensive testimony about his conclusion that "the evidence provides no statistical support for the government’s claim that prices will rise in this transaction” ~ statistical evidence that he considers more probative in analyzing the Government’s increased-leverage theory than Professor’s Shapiro’s "quite . . .complicated economic model." Tr. 2439:19-25, 2441 :25 (Carlton); see id. at 2439:22- 2441:25. Nonetheless, it is of course the Government’s burden ~ not defendants’ ~ to sufficiently link its proffered expert testimony to the underlying facts in the industry lt is therefore no surprise that Professor Carlton spent most of his limited time on the stand discussing the econometric studies he performed, rather than cataloguing whether the facts adduced at trial support Professor Shapiro’s testimony

(Turner)). One was left to wonder why Professor Shapiro turned a blind eye to such extensive real-world experience? When l asked Professor Shapiro about the effect of that testimony on his analysis the following exchange ensued: [A]: No, 1 am aware ofthat testimony And sol think there’s a very serious tension between that testimony and the working assumption for antitrust economists that

Professor Carlton and l share; that the company after the merger will be run to maximize their joint profits

[A]: So what l’m saying is that it will be in AT&T’s interests to play this - to use this leverage in the negotiations lt will be in their interest -

The Court: So that’s an assumption that you`re making?

[A]: Yes it is Okay

The Court: But you don’t have an independent basis of evidence for that? [A]: That is fair.

The Court: That’s an economist assumption?

[A]: That is true. That is true.

[A]: Look, l think ifyou accept that, which, from my point of view, would not be in the combined interests of the new company They would be leaving money on the table The Court: Okay. [A]: lf you accept that, then this bargaining leverage would not come into play

Id. at 2199:22-2200:2, 2200:22-2201:7, 2202:6-12 (Shapiro).

The Court accepts Professor Shapiro’s (and the Government’s) argument that,

generally “a firm with multiple divisions will act to maximize profits across them.” Gov’t

Post-l`r. Br. 19; see also Tr. 2525:22-25 (Carlton). That profit-maximization premise is not inconsistent, however, with the witness testimony that the identity of a programmer’s owner has not affected affiliate negotiations in real-world instances of vertical integration Rather, as those witnesses indicated, vertically integrated corporations have previously determined that the best way to increase company wide profits is for the programming and distribution components to separately maximize their respective revenues See, e.g.. Tr. 2015:16-19 (Bond (NBCU)) (“Q: And, in fact, what you were doing is trying to maximize the revenue ofNBC as a programmer in those negotiations correct‘? A: Yes sir."); see also id. at 1129:17-18 (Breland (Turner)). ln the case of programmers that means pursuing deals “to be on all the platforms” rather than undertaking a “series of risks” to threaten a long-term blackout. Id. at 1129:17-22 (Breland (Turner)); id. at 3120:22 (Bewkes (Time Warner)). So understood, the consistent and, in this Court’s judgment, credible trial testimony is not in fact in “serious tension” with "economic logic” - just with Professor Shapiro’s opinion that the identity of a programmer’s owner influences negotiations l Id. at 2199:22-2200:2 (Shapiro); Gov’t Post-Tr. Br. 19.

Next, Professor Shapiro’s opinion that Turner"s post-merger relationship with AT&T will enable Turner to more credibly threaten a distributor with a long-term blackout in order to extract greater affiliate fees was severely undermined by defendants’ evidence that such a blackout would be infeasible See id. at 2195:4-7 (Shapiro) (“Q: Explain to His Honor why blackouts are relevant here for this discussion today A: Well, even though they don’t happen very much, that’s the key to leverage okay?”); see also id. at 2442:13-

17 (Carlton). lndeed, the evidence showed that there has never been and is likely never

going to be an actual long-term blackout of Turner content. See id. at 2218:13-23, 2357112-14 (Shapiro). Numerous witnesses explained,34 and Professor Shapiro acknowledged, that a long-term blackout of Tumer content, even post-merger, would cause Turner to lose more in affiliate fee and advertising revenues than the merged entity would gain Cf. id. at 2293:2-17. Given that, there is an insufficient evidentiary basis to support Professor Shapiro’s contention that a post-merger Turner would, or even could, drive up

prices by threatening distributors with long-term blackouts35

34 Witness after witness confirmed that blackouts - and the attendant loss of distribution - have “massive implications” for Turner. Tr. 1189:13-16 (Warren (Turner)); see also, e.g., id. at 659:22 (l\/lartin (Turner)) (“[I]t’s very bad for business to go dark.”); id. at 1128:7-1129:4(Breland(Turner))(“l lose money the minute 1 go dark. lt can be catastrophic to my business . . . .”); id. at 31 19:22-3120:22 (Bewkes (Time Warner)) (“So if our channels any ofthem, are not in some distribution offering that’s catastrophic for us We lose a lot of money . . . Due to the size of most of our distributors hundreds of millions of dollars”). During Turner’s one-month blackout with DlSH in 2014, for example, Tumer lost “[njorth of 30 million dollars” in subscriber fees and advertising revenue Id. at l 11512 (Breland (Turner)). ln order to end the blackout, Turner agreed to a temporary affiliate agreement extension that released DlSH from any obligation to pay $120 million in audit monies that lurner believed it was owed. Id. at 1118:15-19. Turner agreed to cede those funds Turner executive Coleman Breland testified, because Turner was “bleeding" and “losing a tremendous amount of money” during the blackout. Id. at 1 1 18123-24. Given all that, it is perhaps unsurprising that, for all ofthe testimony about the “very intense and aggressive” nature of affiliate negotiations id. at 3251:24-25 (Stankey (AT&T)), Professor Shapiro testified that Turner has never experienced a long-term blackout with a distributor, see id. at 2357112-14 (Shapiro).

35 To understand why note that Professor Shapiro’s opinion incorporates the “key” recognition that each side’s bargaining leverage “is based on what would happen ifthere were no deal.” Tr. 2193116-18 (Shapiro). Simply stated, ifa party’s alternative to striking a deal improves that party is more willing and able to push harder for a better deal because it faces less downside risk ifthe deal implodes. Professor Shapiro gave an example of negotiations between a seller and buyer of a used car; he noted that if the seller’s next-best offer improves he will be able to extract a higher price from the original buyer. See id. at 2213:2-10. The bargaining concept the example demonstrates Professor Shapiro explained, is that "you have more leverage now because you have a better offer. And you will be more . . . willing to apply that leverage And some of them are willing to walk away if necessary . . . [B]etter outside offers make one party stronger in those negotiations” Id. at 2213:13-20. Unlike the car seller, who might be “willing to walk away” and accept his alternative offer to sell the car for a gain however, id. at 2213:15-16, the evidence at trial indicated that Turner would not be willing to accept the “catastrophic” affiliate fee and advertising losses associated with a long-term blackout, id. at l 128110 (Breland (Turner)); see supra pp. 14-18.

lt is worth emphasizing again that Professor Shapiro does not contend that Turner’s economics are going to somehow flip after the merger ~ he acknowledged, for example, that Turner would lose over $100 million per month during a post-merger blackout with a large distributor. Id. at 2314:4-15; see also id. at 2293:3-15 (agreeing with defense counsel that Turner will “continue to license Turner content” to distributors because it would be “profitable” to do so). As a result, Professor Shapiro testified that Turner would not be incentivized to actually engage in a long-term blackout with a distributor:

l should say - l think we skipped over it. l’m not, saying that after the merger,

Turner will deny its content to the other distributors This is not a foreclosure-

withholding story . . . l considered whether there would be withholding And that

has been a concern in some private ~ prior vertical mergers And l did not think that

would happen Id. at 2218:13-21; see id. at 2443:12-15 (Carlton).

ln view of that evidence on the prospects of a long-term blackout, the lynchpin of Professor Shapiro’s testimony (and, accordingly the Government’s increased-leverage theory) is the assumption that a post-merger Turner would gain increased leverage by wielding a blackout threat that will only be somewhat less incredible That does not make

sense as a matter of logic and, more importantly that has not been supported by sufficient

real-world evidence36

16 The Court finds lime Wamer CEO leff Bewkes’ response to a question regarding the increased- leverage theory to be particularly persuasive "And the way 1-1 think it’s best the way to understand it, is if we have a risk that a thousand-pound weight might fall on us - we hope it doesn’t, but ifthat’s always there then if you said to me, well, don’t worry; it might be a 950-pound weight instead of a thousand pounds are you going to think about it differently feel differently? Are you going to take more risk that any ofthat might happen to you? Absolutely not.” Tr. 3120:23-312117(Bewkes(Time Warner)). Although not controlling the Court notes that some of Turner’s lead negotiators credibly testified to similar effect. See, e.g., id. at 1 128:7-12 (Breland (Turner)) (“The concept that Turner would push” as though "going dark is good for us 1 believe l’ve given examples today of why it’sjust the opposite l lose money the minute l

2. The Evidence ls lnsufficient to Support the lnputs and Assumptions lncorporated into Professor Shapiro’s Bargaining Model

ln order to measure the increased distributor and consumer costs associated with his prediction that Turner’s post-merger bargaining leverage would increase Professor Shapiro constructed a rather complex economic bargaining model.37 That model seeks to quantify the benefits that AT&T would gain as a result of a long-term, post-merger blackout of the Turner content on AT&T’s rival distributors According to Professor Shapiro, those benefits correspond to the increased affiliate fees that AT&T’s rival

distributors will pay as a result of Turner’s increased post-merger bargaining leverage

go dark. . . ."); id. at l 190:14-17 (Warren (Tumer)) (answering when asked whether could gain leverage by “threatening to blackout distributors” that "l don’t think that’s a realistic perspective."`).

On the stand, Professor Shapiro attempted to support his increased-leverage proposition by noting that programmers and distributors “think about what’ll happen ifthere’s a blackout” when formulating their negotiating strategy See id. at 2193123-2194:13 (Shapiro). The Government does the same in its post-trial filings See, e.g., Gov’t PFOF 11 124-153 (collecting evidence to support proposition that “MVPDS have estimated their likely subscriber losses to inform their negotiating strategy”). The evidence showed that distributors engage in that exercise “with varying degrees of sophistication.” Id. 1 124. With respect to companies that perform “go dark” analyses of the potential consequences of a blackout, the bulk of the evidence showed that negotiators relied on those analyses to get a general sense of “the value" of a programmer’s content by measuring how many customers they would lose in the event of a blackout - customer losses that, notably are not going to change as a result ofthe merger. See Tr. 935:12-16, 936:23 (SEALED); see id at 1349:15-19 (l\/lontemagno (Charter)) (reviewed the “high points" of the Altman Vilandrie go-dark analysis "[v]ery briefly”); id. at 1094121-1095:1 (Breland (Turner)) (although "you never want to go dark ifyou are a programmer,” preparing for a go dark scenario is “just prudent math"). Contrary to Professor Shapiro and the Government’s arguments such high-level evidence does not provide support for the more specific prediction that a marginal improvement in Turner’s (still unprofitable) position in a blackout would meaningfully alter Turner’s bargaining leverage

ln a similar way the Government seeks to rely on the testimony of Turner executive Coleman Breland for the proposition that "Turner bargains over price down to hundredths ofa penny,” Gov`t PFOF 1 108, and that Turner “almost went dark with lime Wamer Cable over a single penny increase on one channel in 2012,"1'd.1 158. That account of Breland’s testimony is “misleading at best.” Defs.’ PFOF 38 n.5. For the reasons set out in Defendants’ proposed findings offact, see id., Breland’s testimony does not bolster Professor Shapiro’s inodel.

37 Technically, Professor Shapiro used two models He first used an economic bargaining model to generate predicted affiliate fee increases to distributors then he plugged those distributor cost increases into a separate merger simulation model to generate his estimates for consumer cost increases See Tr. 2314:17-25 (Shapiro). As defendants’ arguments focus on the design of Professor Shapiro’s bargaining model rather than the merger simulation model, l will refer only to the bargaining model.

As Professor Shapiro explained at trial, his model relies on three primary inputs: 1) a figure for long-term subscriber loss which is the total loss of subscribers a distributor would experience in the event of a long-term blackout of Turner content; 2) the diversion rate which estimates the percentage of a distributor’s lost subscribers that would sign up for AT&T’s distribution services and 3) AT&T margin data, from which Professor Shapiro calculates a measure of profits that AT&T would derive from subscribers it gains or maintains as a result ofthe hypothesized long-term Turner blackout. See Tr. 2217:15-24 (Shapiro). After selecting and entering values for those inputs and running his bargaining model, Professor Shapiro predicts that the challenged merger would lead to annual1 net consumer harm ranging from $286.5 million to $561 million for the year 2016, with that range increasing in subsequent years See id. at 2255:14-15, 3920:6-10; id. at 2256116-20 (predicting $436 million in net consumer harm for the year 2017 and $571 million in net consumer harm for the year 2021). The low and high end ofthe ranges result from using different values - 9% and 14%, respectively - for the subscriber loss rate See id. at 2239:3- 7.

Of course both 2016 and 2017 have passed with no merger. Thus as Professor Shapiro concedes his bargaining model does not “1iterally predict[] the price increases that will occur in negotiations in the real world.” Id. at 2294:18-229511. Rather, Professor Shapiro testified that his model is designed to “evaluate the fundamental incentives and changes in the market created by the merger.” Id. at 2209:1 1-12. For that reason he stated that his model does not account for the existence of Turner’s current affiliate agreements

with distributors which will “expire in time.” Id. at 2209113-14.

Defendants attack Professor Shapiro’s bargaining model from all directions Noting that models are “only as good as the inputs” id. at 2315:1 l, defendants argue that each of Professor Shapiro’s three “very important” inputs lacks a sufficient basis in the trial evidence id. at 2315:12. Defendants also argue that Professor Shapiro’s model improperly assumes away Turner’s current affiliate agreements - agreements that will serve to significantly constrain Turner’s post-merger bargaining leverage for years .to come38 1 agree with defendants for the most part, that the inputs and assumptions of Professor Shapiro’s model are not sufficiently grounded in the evidence - a fact that "undermine[s]"

my “confidence in the reliability and factual credibility” of his projections Anthem, 855

F.3d at 363. How so?

a. T he Evidence ls Insu]j’icient to Support Professor Shapiro ’s Long- Term Subscriber Loss Rate

ln order for AT&T to benefit from a long-term Turner blackout with a rival

distributor under the increased-leverage theory a sufficient number of customers must

38 Correcting for those faults defendants argue would cause Professor Shapiro’s model to predict a net benefit to consumers on account ofthe merger. Specifically, Professor Carlton testified that when one updates or accounts for those four factors - the long-term subscriber loss rate the diversion rate, the margin data, and the presence of contracts v Professor Shapiro’s model generates an average 52-cent per-month, per-consumer benefit rather than an average 27-cent per-month, per-consumer harm, See Tr. 2516:2-6 (Carlton); see also id. at 225519-25 (Shapiro) (testifying about the "[p]redicted Turner monthly fee increases for consumers” reflected by PDXl 1, slide 1 1).

The large effects on the predicted net harm created by minor changes to Professor Shaprio’s inputs raises a separate question regarding the model’s sensitivity As Professor Carlton noted, Professor Shapiro performed no “statistical tests” to demonstrate that the “tiny percentage” increases in harm predicted by his model are “any different from zero” statistically speaking Id. at 2450116-2451:12 (Carlton). Without such statistical testing Professor Carlton testified, the predicted harms could fall within the range of zero “just because of normal fluctuations in how we estimate models in the perimeters [sic] of the model.” Id. The fact that Professor Shapiro’s model “cannot be proven to any statistical significance” provides this Court with additional cause to reject the model’s conclusions as “‘persuasive" evidence FTC v. Swedish Match, 131 F. Supp. 2d 151, 161 (D.D.C. 2000).

actually depart or decline to join the rival distributor due to its failure to offer Turner content. Professor Shapiro refers to that measure of lost customers as the "long-term subscriber loss rate.” At trial, Professor Shapiro testified that his model incorporates a low- end long-term subscriber loss rate of 9%, a number representing the combined percentage of current and potential subscribers who would either leave or decide not to sign up with a distributor in the event of a hypothetical long-term blackout of Turner content. See Tr. 2239:3-5 (Shapiro). Whether viewed as a measure of Turner’s “market power” in the programming market or not, id. at 2239118, that measure of customer loss - deemed the “long-term subscriber loss” rate by Professor Shapiro - is critical to Professor Shapiro’s bargaining model and the predicted consumer harm it generates

Of course there has never been a long-term blackout of Tumer content; Professor Shapiro thus had no “real-world” evidence on which to base his projected subscriber loss rate Id. at 2394:8-1 1. lnstead, as a basis for his chosen 9% value Professor Shapiro relied on three principal pieces of evidence: (1) a third-party consultant slide deck commissioned by Charter in late 2016; (2) his own analyses of long-term blackouts of a different programmer, Viacom, with cable distributors Suddenlink and Cable ONE; and (3) the results of an internet survey conducted by another of the Government’s testifying experts Professor John Hauser. See id. at 2225:17-2226:7. The evidence indicates however, that each of Professor Shapiro’s sources is significantly flawed Thus even taken together, they fail to establish the reliability of Professor Shapiro’s long-term subscriber loss rate

and the conclusions generated by his model.

1) Charter’s Third-Party Consultant Slide Deck

According to Professor Shapiro, a slide deck, commissioned by AT&T’s competitor Charter in late 2016 and authored by consultants at a San Francisco-based firm called Altman, Vilandrie & Company ("Altman Vilandrie”), was the “single best document and analysis” he found in coming up with a measure for the long-term subscriber loss rate. Id. at 2235:11-14. That was so, according to Professor Shapiro, because the slide deck, in contrast to the other available pieces of evidence addressed “exactly” the question of interest to his analysis the subscriber-loss effects of a long-term Turner blackout with a major distributor, as measured in both lost current customers and lost potential customers Id. at 2235:19-20, 2236120-2237:3. But although the slide deck may have analyzed the questions in which Professor Shapiro was interested (perhaps not so fortuitously see infra pp. 127-128), the evidence shows that it did so via methodologies that were significantly flawed.

Before explaining further, it is necessary to review the basics of the slide deck’s analysis Altman Vilandrie director Stefan Bewley who was responsible for supervising the project, explained that the slide deck was designed to examine “the value of content programming.” Id. at 1271:23 (Bewley (Altman Vilandrie)). The slide deck, entitled "Content Valuation Project," contains charts predicting Charter’s subscriber losses in the event of permanent blackouts with various programming networks See id. at 1249:18-21; see also PX79. To generate those loss predictions Altman Vilandrie used three different methods (1) an internet survey (2) set top box data, and (3) the so-called “hybrid” method,

which made slight adjustments to the set top box analysis See Tr. 2792:10-11, 2801:1-5.

2808:6-20 (Rossi); id. at 1271:24-1272:6 (Bewley (Altman Vilandrie)). Although Professor Shapiro praised that analysis for its apparent rigor, see id. at 2235:18-19 (Shapiro), he later conceded, despite professing that he usually does not accept data without “look[ing] into it more and figur[ing] out how reliable it is” id. at 3848:10-13, that he did not take steps to evaluate the reliability of the Altman Vilandrie data before he relied on it, id. at 3863:21-23 ("Q. So we can establish that all you did was read the report, right? A. l relied on the report. l didn"t dig behind it.”). Rather, Professor Shapiro simply incorporated the final figure included in the slide deck’s table of results See PX79-6.

Defendants’ survey and statistics expert, Professor Peter Rossi, did however examine the methods that Altman Vilandrie used to predict the reported subscriber loss rates And in testimony that largely went unrebutted,39 Rossi explained his conclusion that “[a]ll three are invalid.” Tr. 279215 (Rossi). Those conclusions which l accept, are outlined below.

First, Altman Vilandrie relied on an internet survey That survey as Professor Rossi explained, combines three different types of internet surveys - a “conjoint,” a “channel chooser,” and a “l\/Iax Diff.” Id. at 2792:13-17. ln the conjoint survey respondents view eight to ten screens and are presented with different options and pricing for bundles of video programming broadband, and telephone services the survey seeks to infer the respondent’s willingness to trade off different service features The channel chooser

survey for its part, tries to ascertain how much priority respondents give to a particular

39 The Government did not recall any Altman Vilandrie witnesses on rebuttal to answer to Professor Rossi’s critiques

cable network, And the Max Diff survey allows the respondent to rank the different networks Based on its internet survey’s combination of those approaches Altman Vilandrie calculated one set of subscriber loss figures for current and prospective video customers See PX 79-18.

Professor Rossi testified that the internet survey method was plagued by considerable flaws both in the way the questions were designed and in the way the answers to those questions were used to project subscriber loss He noted, for example that the conjoint survey’s presentation of 12 networks included only one network - CNN » owned by Turner. Tr. 2794:7-9 (Rossi). Although Professor Rossi testified to the common-sense proposition that it is “impossible” to infer the value of all ofthe remaining Turner networks just from CNN, apparently that is what Altman Vilandrie did with the results of the conjoint survey See id. at 2793:11-14, 2800:6-11. With respect to the Max Diff survey’s process for ranking channels Professor Rossi testified that such a ranking can give a sense of relative importance but cannot measure how much more or less important one network is than another; of particular relevance here moreover, the ranking methodology does not define what “important” means to a respondent, and thus says “[a]bsolutely nothing” about “whether a subscriber to Charter would leave if there was a Turner blackout.” Id. at 2795 19- 16. Finally, although Altman Vilandrie purported to combine the conjoint and Max Diff methodologies to bolster its analysis Professor Rossi testified that such methodologies "fundamentally cannot be combined” as a matter of statistical practice See id. at 2796:18-

2797:4; see id. at 2800116-17 (“lt’s literally an impossibility and there is absolutely no

way to combine these two.”).40 For those reasons Professor Rossi’s “bottom line conclusion” about the survey methodology was that it is “completely invalid.” Id. at 2800:22-24.

Second, Altman Vilandrie utilized a set top box methodology Set top box data, as should now be familiar, shows the amount oftime a particular cable set top box is tuned to specific channels See id. at 1274:16-18 (Bewley (Altman Vilandrie)); id. at 2801:5-12 (Rossi). As Bewley acknowledged, set top box data does not necessarily reflect actual viewership or correlate to a particular network’s value See id. at 1275117-22 (Bewley (Altman Vilandrie)). Professor Rossi testified similarly noting that such data, without more “cannot possibly answer the question about the effect of removing any channel or group of channels” Id. at 2802:7-8 (Rossi). ln addition because set top box data is generated by Charter’s current customers it provides no information about “prospective customers for Charter.” Id. at 2801:19-24. Notwithstanding those limitations the Altman Vilandrie slide deck purported to derive current and prospective subscriber loss figures from the set top box data by assigning differing “churn propensity” values - that is values reflecting the likelihood that a viewer will leave a distributor ~ to different levels ofviewing concentration Id. at 2802:9-2804:14; see PX79-18, -68.

Professor Rossi testified, however, that the churn propensity values and their

correlation with set top box data, are “not based on data of any kind” and instead reflect

40 Although Professor Rossi explained that Altman Vilandrie relied in part on the Sawtooth Software which incorporates some of his own innovations in survey methodology the combination procedure took place “outside of Sawtooth Software.” Tr. 2855:20-2856:6 (Rossi).

“purely assumed numbers.” Tr. 2804:12-13 (Rossi). Although the lack of empirical support is reason enough to disregard the slide deck’s analysis of the set top box data, that flaw is compounded by the particular values assigned in the churn propensity schedule ln particular, based on the schedule Altman Vilandrie predicts that the loss of a network with a specified viewing concentration or greater will always cause a distributor’s customers to leave Id. at 2807:13-20; PX79-68; DX0681-73. l agree with Professor Rossi that the upper-threshold assumption and indeed the entire set-top box methodology lacks a sufficient basis in evidence and is unreliable See Tr. 2807:13-22.

Thira’, and most importantly for purposes of Professor Shapiro’s analysis Altman Vilandrie implemented what it refers to as a “hybrid” methodology Ultimately, the April 27, 2017 slide deck upon which Professor Shapiro relied indicates that the hybrid methodology produces a video subscriber loss rate of 9% for current customers and 10% for prospective customers Tr. 238811-11, 3868:1-20 (Shapiro). Professor Shapiro testified that the 9% long-term subscriber loss rate that he incorporated into his model “reflect[s]" those results Id. at 223714-8, 2388:1-11; see PX79-6, -18.

A key problem with the design of the hybrid methodology as Professor Rossi testified, is that it blends two methods only in the sense that it alters the set top box method’s lower churn propensity threshold “based to some extent on some of the survey data.” Tr. 2808:14-18 (Rossi). ln other words the hybrid methodology can be thought of as “just a revision or alteration minor alteration to the set-top box method.” Id. at 2808119- 20. The hybrid methodology is thus plagued by the same problems as is the set top box

methodology including the fact that it “can’t say anything about prospective customers"

\

[d. at 2808:19-22. As a result, the hybrid methodology - and its associated 9% and 10% subscriber loss predictions for current and prospective customers - falters on the same grounds as the set top box methodology

l\/loreover, evidence regarding the evolution of Altman Vilandrie’s slide deck casts further doubt on the reliably of the figures associated with the hybrid methodology Specifically, the evidence shows that a “final read out” version of the slide deck sent to Charter on April 21, 2017 reported that the hybrid method produced a 5% and 6% subscriber loss rate for current and prospective customers respectively See id. at 130214- 20 (Bewley (Altman Vilandrie)); 3068:1-12 (Shapiro). Almost immediately after a meeting with Charter representatives a few days later, however, Altman Vilandrie with the "permission” of Charter, altered the results of the hybrid methodology for “just” Turner and no other programmer. Id. at 1310:XX-XXXXXXX (Bewley (Altman Vilandrie)). Those alterations led to the 9% and 10% current and prospective subscriber loss rates upon which Professor Shapiro’s analysis relied. See id. at 38_6811-12 (Shapiro); compare DX681-23, with PX79-18.

That Turner-centric turn of events is enough alone to give me pause before accepting Professor Shapiro’s reliance on the slide deck, notwithstanding the Government’s presentation of a more benign view of the slide deck’s evolution See, eig., Tr. 1327:16- 1332:4 (Bewley (Altman Vilandrie)) (testifying among other things that it was Altman Vilandrie that “proposed making an exception for Turner” based on the results ofthe hybrid methodology as compared to the results of other methodologies). ln my view, moreover,

the most troubling aspect of the testimony regarding the contested changes to the slide deck

was that Professor Shapiro was entirely unaware ofthose changes when he “first relied on the document” to perform his analysis ld. at 236518-10 (Shapiro); see also id. at 2366:4- 7. At trial, Professor Shapiro admitted that he was not aware of the alterations made to the Altman Vilandrie slide deck until his pre-trial deposition by defendants Id. at 2365:1-3. He nonetheless defended his reliance on the slide deck for the long-term subscriber loss figures in no small part based on his insistence that although the current subscriber loss figure had been altered, the prospective subscriber loss figure “was not changed here." Id. at 238811-6; see id. at 2366:9-11 (“lfl used the five percent instead, l would get a long- terin subscriber loss rate of 8.5 percent instead of nine in my calculations.”). Given that, Professor Shapiro continued, the altered current subscriber loss figure was “a lot less significant” because “it"s just one of the two components that affects the long-term subscriber rate.” Id. at 2388:1 1-15. Based on that assumption Professor Shapiro testified that, even if one accepted the original 5% existing-customer subscriber loss figure “[i]t’s not as though my number would go from . . . nine to five percent ifyou made that change lt would go from nine to 8.5” percent. Id. at 238818-10.

But Professor Shapiro “made a mistake” in so testifying a fact he was later forced to concede on rebuttal. Id. at 3868:17-20. When confronted on rebuttal with the two versions ofthe slide deck, Professor Shapiro acknowledged that the prospective subscriber loss figure had indeed been changed from an original value of 6% to the 10% value upon which he relied. Id. at 3868:1-20. He also testified, moreover, that using the original 5%

existing customer subscriber loss figure and 6% prospective subscriber loss figure would

yield a departure rate of about 5 or 6%, which in turn would “largely eliminate[] the net MVPD cost increase” he projects Id. at 3870:22-3871:3.

For all ofthese reasons l conclude that Professor Shapiro’s reliance on the projected long-term subscriber loss rates contained in the Altman Vilandrie slide deck was misplaced Given Professor Shapiro’s testimony that the slide deck was the “single most important document"` to him in calculating the long-term subscriber loss rate incorporated into the bargaining model, id. at 2360125-2361:3, that conclusion alone is all but fatal to Professor Shapiro’s analysis To the extent, however, that Professor Shapiro relied upon two other categories of evidence such evidence also fails to support his chosen long-term

subscriber loss rate.

2) Long-Term Viacom Blackouts With Suddenlink and Cable ONE

ln generating his long-term subscriber loss rate Professor Shapiro also relied on his own analysis of the effects of long-term blackouts of Viacom programming - which includes networks such as l\/ITV and Nickelodeon ~ with two MVPDs Suddenlink and Cable ONE. ln particular, Professor Shapiro opined that the Viacom blackout caused Suddenlink to lose 9.4% of its video subscribers and Cable ONE to lose 16% of its video subscribers Gov’t PFOF 1 208. The Court need not spill much ink addressing those figures because even a cursory review ofthe evidence shows that they are unreliable

With respect to Professor Shapiro’s 9.4% figure for Suddenlink, it is notable that Suddenlink itselfrepresented to the public that it suffered only a 2 to 2.5% subscriber loss

as a result of the blackout with Viacom. See Tr. 2480121-22 (Carlton). Given the unusual

nature of a long-term blackout, Charter, Comcast, and Wall Street power Citi also studied the event, the latter two concluding that Suddenlink’s subscriber loss percentage was in the “low single digits.” Id. at 2483:1-2; Defs.’ PFOF 1 150. Altman Vilandrie’s study for Charter produced similar results See PX79-6. l heard from defendants’ expert Professor Carlton that Professor Shapiro’s estimates were inflated when compared to those other reported figures due to his failure to account for the fact that the rate of subscriber loss in the video distribution industry started to increase in 2016. See Tr. 2483:16-2484:2 (Carlton); see also id. at 2490:8-10.41 When Professor Carlton corrected Professor Shapiro’s analysis to control for that trend, he generated a 4.8% subscriber loss rate for the Suddenlink-Viacom blackout, a number much more in line with industry estimates See id. at 2484:3-8.42

Professor Shapiro’s 16% subscriber loss estimate for the Cable ONE long-term blackout of Viacom is even more unreliable On that score it is sufficient to note that Randy Sejen Cable ONE’s chief negotiator, testified that “[t]he losses attributable to

Viacom are very very small . . . and were not significant.” Id. at 2123121-2124:12 (Sejen

41 Professor Shapiro omitted from his analysis of industry trends December 2016 data that showed an even steeper decline in industry subscribership. When first questioned about the decision not to include this data in his analysis Professor Shapiro did not recall that atty data was omitted, and could not provide an explanation for that omission See Tr. 3879:1-14 (Shapiro). When called back to the stand days later, Professor Shapiro recalled that he had noticed something “peculiar” about the omitted numbers Id. at 3915:9. Professor Shapiro’s testimony concerning the 2016 data was not the only time that he demonstrated a lack of familiarity with the materials he presented to the Court. See infra pp. 127-129, 139-140. To be clear, although both call into question his analysis Professor Shapiro’s lack of familiarity with the contents of his report and with his own data analysis presents a credibility problem separate from the problems with key inputs generated by outside sources like Altman Vilandrie

42 Pursuant to the parties’ representations and agreements during an April 26, 2018 bench conference related to the Suddenlink analysis the Court will strike the following lines of trial testimony from Professor Shapiro; Tr. 3926:12-13; Tr. 3917:5-7; Tr. 3878:9-10; Tr. 3877120-21;Tr. 3806:10-12.

(Cable ONE)). Specifically, Sejen noted that the Viacom blackout was “felt and absorbed” within four to six months and caused a subscriber loss ofjust 2%. See id. at 2130:1-4, 2123:21-24. Given Sejen’s testimony that Cable ONE lost only 2% of subscribers the Court has no reliable basis to accept Professor Shapiro’s calculation of a subscriber loss figure eight times that amount - and therefore rejects it in toto.

To be sure l heard evidence that, in relative terms lurner programming is more valuable than Viacom programming But that fact alone cannot make up for Professor Shapiro’s baseline failure to establish any reliable measure of subscriber losses associated with the long-term Viacom blackouts Having concluded that Professor Shapiro’s Viacom analysis lacks an adequate basis l will now turn to the last main piece of evidence he cited in support of his long-term subscriber loss figure

3) Professor Hauser’s Internet Survey

The last piece of evidence upon which Professor Shapiro based his long-term subscriber loss rate is an internet survey The internet survey was*conducted by another of the Government’s testifying experts Professor John Hauser, who heads the marketing department at the Massachusetts lnstitute of Technology Tr. 756:9-14 (Hauser). l`he survey generated a long-term Turner blackout subscriber loss percentage of 12% and a 30- day Turner blackout subscriber loss percentage ofa whopping 8.2%. Id. at 76117, 803:24- 804:3.

Although once at the forefront of the Government’s presentation see, e.g., Gov`t Pre-Trial Br. 29, Professor Hauser’s survey now finds itselfin the background, with even

Professor Shapiro minimizing his reliance on it, see Tr. 2360:22-24 (Shapiro) (“[Hauser’s]

twelve percent is corroborative lfl didn’t rely on that, if we decided that was unreliable it wouldn’t change my opinions.”). Professor Shapiro however had good reason to unhitch his analysis from Professor Hauser’s internet survey wagon: cross-examination and real- world evidence alike revealed that the survey was inherently unreliable and produced inflated results !

Before explaining that conclusion a brief review of Professor Hauser’s survey might be helpful. The survey had roughly 1,600 participants Id. at 765:11 (Hauser). Those participants were drawn from an internet panel and then broken into four groups of approximately 400 participants each: three “test” groups and one “control” group. See id. at 775:10-14, 761:21-762:5 (Hauser). The test groups were presented with an online survey in which they were presented with questions about their potential responses to Turner blackouts of varying lengths including a permanent blackout, a one-month blackout, and a one-week blackout. See id. at 775:22-776:6. The control group was not presented with any information about a blackout. See id. at 776:14-18; id. at 276818-11 (Rossi).

Defendants’ survey expert, Professor Rossi, testified that Professor Hauser’s survey is “unreliable” for any number of reasons Id. at 2768115 (Rossi). For purposes of the analysis here l need only discuss two.43 First, Professor Rossi testified that the survey was

drawn in a biased and misleading way, with the effect of overstating the importance of

43 Professor Rossi also criticized Professor Hauser for failing: 1 ) to establish that his group of survey participants constituted a representative sample ofthe population of interest, and 2) to provide a margin of error -- that is a measure of reliability ~ for his survey’s results See Tr. 2771:22-2273:21, 2775:2-6 (Rossi). Although the Court agrees that those problems are notable it sees no need to pile on by addressing them further in light ofthe two significant design flaws discussed below.

Turner content. Second, Professor Rossi testified that the survey’s centerpiece the intent- to-switch scale was confusing and skewed. See id. at 2768:12-2769:8. After considering the expert testimony as well as other evidence calling into question the results of Professor Hauser’s survey l agree with Professor Rossi’s conclusions

First, Professor Rossi faulted Professor Hauser’s survey as building in bias at the “priming” stage [d. at 2786:17. Professor Hauser testified that many television viewers think about video programming in terms of specific shows or genres not channels See id. at 817117-818:5. Professor Hauser therefore began his survey by "priming" survey respondents to connect genres of programming to specific channels through the use of network logos. See id. at 817125-818:17; see also id. at 824:15-825:6 (sports); id. at 82114- 12 (special events). According to Professor Rossi, however, Professor Hauser’s use of logos was problematic. ln particular, Rossi noted that the internet survey “tend[ed] to 1 visually overemphasize Turner content” relative to other content by for example enumerating the Turner channels in large font or inaccurately over representing the Turner networks relative to other programming Id. at 2783:12, 15-17 (Rossi); see also id. at 2787:9-2788125 (discussing DX915B).44 At one point, the survey presented respondents in the test group with large Turner logos for six straight slides despite not showing those

slides to the control group. See id. at 838:23-839:3 (Hauser); id. at 2789:25-2790:8, 24-25

44 This is not the first time Professor Hauser’s “graphic effects and presentation methods” have been called into question on this basis See Apple, Inc. v. Samsung Elecs. Co., No. ll-CV-01846-LHK, 2014 Wh 976898, at *10-*16 (N.D. Cal. Mar. 6, 2014).

(Rossi). As Professor Rossi explained, that priining tended to bias respondents in favor of indicating an intent to switch in the event ofa Turner blackout. Id. at 2790:16-17 (Rossi).

Second, Professor Rossi testified that Professor Hauser’s survey asked respondents to report their answers using a scale that was confusing and, again likely to cause respondents to overestimate their likelihood of switching distributors in the event of a Turner blackout. Professor Hauser’s survey did not squarely ask respondents whether they would switch providers in the event of a Turner blackout. lnstead, the internet survey presented respondents with, as it is known in the industry a “Juster scale” by which they answered the question “How likely are you to switch your TV provider, on a scale from 1 to 99?” DX915-152; see Tr. 788112-18, 814:1-4 (Hauser). The scale included percentages ~ 10%, 20%, 30%, etc. ~ and accompanying descriptors such as "very slight possibility"

35 66

“slight possibility some possibility" and “fair possibility.” See Tr. 813:15-814:19 (Hauser); DX915-152. The results of the luster scale were translated directly into a subscriber loss rate Thus, if each respondent rated his or her likelihood of switching at a

“very slight possibility corresponding to 10% on the luster scale Professor Hauser’s survey would spin out a subscriber loss rate of 10%. See Tr. 815:20-816:18 (Hauser). Professor Hauser’s luster scale had two critical flaws first, its text descriptions were “out of w[h]ack with the numbers,” Tr. 2778:17-21 (Rossi), and, second, luster scales are particularly unreliable in quantifying consumer choices of this kind, see id. at 2779:1- 2782:19. Professor Rossi put it in plain terms Now if l told you that 1 thought there was a very slight possibility that l would

get into a car accident driving from Washington to Baltimore on the Baltimore Washington Parkway this evening l don’t think you would say

that was one out of every ten times l attempted that. You might say one out

of every thousand or more So the text description is out of whack with the

numbers And that’s true throughout the scale Id. at 2778:12-19. Professor Rossi also testified that the survey’s text was bound to present skewed results because it “minimiz[ed] or neglect[ed] many aspects of switching costs" - that is the various costs associated with switching distributors Id. at 2783113-14 (Rossi); see also id. at 2783:19-2786:16 (detailing different kinds of switching costs including search costs transactional costs bundle-derived costs and psychological costs and concluding that Professor Hauser’s survey failed to adequately account for those costs). That problem casts further doubt on the reliability of the survey Cf H & R Block, 833 F. Supp. 2d at 66-68 (declining to rely on "customer survey[]” results in part because survey “failed to assign” adequate “pricing” data to some of participants’ response options).

l\/lore fundamentally Professor Rossi explained, luster scales are notoriously inaccurate when used “as an exact quantification" of the likelihood that a customer will engage in some future behavior. Tr. 2779116-21, 2782:2-13 (Rossi). Academic literature cited by both Professors Rossi and Hauser establishes that the average correlation for predictions of this kind falls between .3 and .6. See id. at 2779116-278015. Professor Hauser’s scale nonetheless purports to assign a correlation value of 1.0, that is aperfect linear association where intent predicts behavior virtually every time. See id. 2872115- 278112, 2872:1-4. And even that unsupported correlation “basically disappears" when

respondents are asked to predict their behavior with respect to new products or situations

- such as a permanent Turner blackout. See id. at 2780115-24.

Given the significant questions raised about the design of Professor Hauser’s survey it should come as no surprise that the survey’s results were puzzling to expert and fact witnesses alike Gregory Rigdon Comcast’s chief negotiator, responded to Professor Hauser’s one-month blackout loss estimate of 8% by noting “[T]hat seems like a big number in one month.” Id. at 897:2-3 (Rigdon (Comcast)). He gave the same answer when asked about the survey’s long-term 12% loss estimate See id. at 898:3-5 (“Q: But in terms of nay group you’ve ever seen dropped, have you ever seen anything approaching a 12 percent ~ A. That seems like a big number.”). Turner CEO lohn l\/Iartin put things a bit more strongly calling the survey’s 8% one-month blackout subscriber loss prediction “absurd.” See id. at 66019-11 (Martin (Turner)). Defendants’ expert Professor Carlton for his part, said that the 8% departure rate for one month “strikes me as way too high" and is “nothing like” the Cable ONE estimate of 1.1% to 1.2% for the actual temporary Turner drop. Id. at 2491:4-15 (Carlton). Finally, even Professor Shapiro himself noted that Professor Hauser’s one-month subscriber loss estimate of 8% “seems high.” Id. at 2360: 1 8.

Of course, if Professor Hauser’s survey generated inflated one-month subscriber loss estimates as compared to real-world evidence that fact “cast[s] doubt on what Professor Hauser is doing” with the survey design generally Id. at 2491 :4-15 (Carlton). lt is therefore small wonder why both the Government and Professor Shapiro have deemphasized the role of the Hauser internet survey All in all, l can’t help but conclude that the internet survey’s methods are unreliable and that its results fly in the face of real-

world evidence regarding the effect of programming blackouts

For all ofthe reasons discussed above the evidence is not sufficient to support the 9% long-term subscriber loss figure that Professor Shapiro utilized in his model.45 Because the Government has the burden of proof as well as the responsibility to demonstrate that its proffered expert testimony has an adequate grounding in evidence the lack of evidentiary support for Professor Shapiro’s input is fatal to the model’s probative value in predicting the asserted harm associated with the Government’s increased-leverage

theory.46

b. T he Evidence ls lnsujj”icient to Support Professor Shapiro ’s Diversion Rate

To evaluate the number of customers that AT&T stands to gain from a long-term Turner blackout with a rival distributor, it is necessary to estimate how many ofthat rivals’ customers “will end up as Direc'l`V subscribers either by moving to DirecTV or by staying

at DirecTV and not going to” the rival. Tr. 2240:9-11 (Shapiro). ln Professor Shapiro’s

43 The miniscule nature of subscriber losses resulting from the two actual instances of Turner blackouts perhaps should have alerted Professor Shapiro that something was awry with his sources The evidence showed that there have been two short-term blackouts of Tumer content with distributors l) a thirty-day blackout with Cable ONE in October 2013, which resulted in "fairly insignificant” subscriber losses in the range ofabout.6%,l`r. 2116110-13,2127:21-2128:2(Sejen (Cable ONE)); and 2) a thirty-day blackout with DlSH in November 2014, in which some Turner networks - including CNN, but not TBS or TNT ~ were blacked out, resulting in a loss of less than 1% of DlSH subscribers see id. at 388:10-38915 (Schlichting (DlSH)). Those subscriber loss figures simply cannot be squared with some of the figures represented in the sources upon which Professor Shapiro relied

46 Because the evidence does not support use of Professor Shapiro’s 9% “low end” long-term subscriber loss rate it stands to reason that the larger 14% long-term subscriber loss rate he used to generate the high end of his predicted harm range is also unsupported Tr. 3851:21-3852:8 (Shapiro). The same goes for the higher 12% and 16% long-term subscriber loss rates he used, rather curiously and contrary to the Altman Vilandrie slide deck upon which he claimed to rely to generate the predicted harms for a 2017 and 2021 market configuration See Tr. 249319-2495:18 (Carlton), Professor Shapiro"s appeal to the fact that he predicted a range of harm is therefore unavailing He is not “suffering the consequences of being conservative” in his estimates Tr. 385211-2 (Shapiro), the consequences arise because even his conservative estimate lacks sufficient evidentiary support and reliability The same can be said for the Government’s post-trial submissions regarding the "conservative[]” nature of Professor Shapiro’s analysis Gov`t Post-Tr. Br. 14.

model, that figure is known as the “diversion rate.” Id. at 2240113. The diversion rates Professor Shapiro uses differ based on geography Specifically, Professor Shapiro calculated a diversion rate for each of the local geographic markets based on an assumption that subscribers “move to the other [distributors], in each local market, to the other distributors proportional[ly] to their marketshare” Id. at 2240:23-2241:1-3, 2241 :15-20.

The parties’ main dispute related to diversion rate pertains to “cord cutting” also referred to in this context as the “outside good.” Id. at 387118-9; see id. at 2604113-17 (Carlton). As is likely familiar by now, an individual “cuts the cord” by discontinuing his MVPD subscription and opting instead to receive television programming through an internet-based SVOD like Netflix or Hulu. See supra pp. 22-23. Professor Shapiro acknowledges that, as a result of cord cutting l"[d]iversion to AT&T will be reduced to some extent because some current subscribers of a rival MVPD that would leave that MVPD due to a loss of Turner content will cancel their pay-TV service altogether” rather than “switch to AT&T or another MVPD that carries Turner.” Gov’t PFOF 1 215; see Tr. 2241:22-2242:18 (Shapiro). To account for that effect, Professor Shapiro assigns a value to cord cutting of approximately 10%. See Tr. 3871:8-15 (Shapiro).

According to defendants Professor Shapiro’s 10% figure understates the rate of cord-cutting and, accordingly results in an inflated diversion rate See Defs.` PFOF 11 182-1871 see also Tr. 2515116-20 (Carlton). Defendants insist that the proper cord- cutting rate is closer to 20%. See Defs.’ PFOF 1 185; see also Tr. 2505110-20 (Carlton). Plugging that 20% cord-cutting rate into Professor Shapiro’s model, defendants’ lead

expert Professor Carlton testified would result in a predicted net consumer benefit See

Tr. 2515:16-20 (Carlton) (if one uses 20% cord-cutting rate in Professor Shapiro’s model, then “Professor Shapiro’s 27-cent price increase on average becomes [a] 6-cent benefiti decrease"). After evaluating the evidence and the parties’ arguments on cord cutting l conclude that there is insufficient evidence to support the 10% cord-cutting figure utilized by Professor Shapiro.

The basis for Professor Shapiro’s 10% figure was the (by now discredited) Altman Vilandrie slide deck, created for Charter. See id. at 2372:8-10 (Shapiro) (“A: Well, you relied on Altman Vilandrie for what you called the outside good, correct? A: For that part, yes that’s correct.”). What l learned about the slide deck’s cord-cutting figure however, was that it was derived from the results of Altman Vilandrie’s “conjoint survey.” Id. at 2821:7-15 (Rossi). Specifically, as explained by defendants` survey expert Peter Rossi, Altman Vilandrie first looked to the measure of people who answered that they would not “take any” MVPD service in the event ofa blackout with Charter. Id. at 2821 19- 14 (Rossi); id. at 2242111-15 (Shapiro). Altman Vilandrie then took those estimates Rossi testified and "multiplied all ofthose coefficients by .6 withoutjustification” - meaning in layman’s terms that they “cam[e] up with a figure and then reduc[ed] it by 40 percent." Id. at 2821 : 14-18 (Rossi); id. at 3871116-19 (Shapiro). That reduction in turn produced Altman Vilandrie’s cord cutting estimate of 16.8%, which Professor Shapiro used to derive his ultimate cord cutting estimate 01`10%. Id. at 2372:19-237314, 3871:11-19 (Shapiro); id. at 2821116-21 (Rossi); see PX79-38.

Although Professor Shapiro testified that he was “aware” of Altman Vilandrie’s

40% reduction methodology he could not recall whether he was aware of it at the time he

relied upon Altman Vilandrie’s cord-cutting figure orjust as a result of Professor Rossi’s trial testimony Tr. 3871116-23 (Shapiro). Moreover, Professor Shapiro was unable to explain Altman Vilandrie’s choice to reduce the cord-cutting figure stating only that his “understanding is l\/lr. Bewley explained he did that based on evidence that reflected market conditions in Altman Vilandrie as part of their analysis.” Id. at 3872:4-8. The Court, however, has been unable to locate that alleged testimony in the trial record, or in the Government’s post-trial filings for that matter. Cf. Gov’t PFOF 11214-216 (discussing Altman Vilandrie’s cord-cutting figure with no reference to Bewley testimony).

lf that were not enough alone to give pause before accepting Professor Shapiro’s 10% cord-cutting estimate defendants cast additional doubt on that figure by citing to SNL Kagan data as well as to real-world evidence regarding the prevalence of cord cutting in the industry With respect to SNL Kagan data, Professor Carlton testified that the data shows that “[a]round 20 percent” of“total TV households” are "cord cutters” Tr. 2505:12- 18 (Carlton).47 SNL Kagan’s 20% figure defendants state aligns with other industry evidence about the extent of cord cutting See Defs.’ PFOF 11 183, 185. AT&T surveys of departing customers for example indicate that "25 to 30 percent” of those customers

report that they are “going to cord cutting.” Tr. 2506119-24 (Carlton). RCN CEO lim

47 To be sure the Government, through the rebuttal testimony of Professor Shapiro, attempted to rebuff Professor Carlton’s 20% cord-cutting rate Professor Shapiro pointed out that, in the context of examining the consequences of a Turner blackout, it is “pretty likely” that a departing customer would “want to go somewhere else where you can get the Turner content," Tr. 3808:11-12 (Shapiro). Thus, Professor Shapiro continued, stating that “20 percent of American households don’t have pay-TV service” overall is “beside the point.” Id. at 380815-6, 15. Were it defendants’ obligation to provide sufficient support for the departure rate in Professor Shapiro’s model, rather than Professor Shapiro’s that rebuff would perhaps be persuasive But even accepting Professor Shapiro’s point about defendants’ proposed rate that point does not prove that the departure rate he proffered had adequate evidentiary support.

llolanda testified that similar surveys by his company report that “at least half of the customers who leave RCN’s video services are leaving for OTT providers” - a number that Holanda predicts is “likely to grow in the future as l\/lillennials become more and more prominent in the marketplace.” Id. at 2948120-294913 (Holanda (RCN)). That evidence about the increasing presence of cord cutting in the market, in the Court’s view, undercuts yet another aspect of Professor Shapiro’s measures of cord cutting - namely that they apparently “declin[e] over time” because of a particular “feature of his model." Id. at 2448:7-9 (Carlton).

ln the final analysis it is the Government’s burden to adequately support its proffered model’s harm - and, necessarily the model’s inputs - through the testimony of its expert or related evidence The utter lack of explanation regarding Altman Vilandrie’s methodology for generating the cord-cutting projection upon which Professor Shapiro relied, coupled with defendants’ real-world evidence regarding the prevalence of cord cutting in the industry leaves me with little confidence in the accuracy of Professor Shapiro’s 10% cord-cutting figure As with the long-term subscriber loss estimates l therefore conclude that the Government has also failed to provide adequate support for Professor Shapiro’s diversion rate estimate and thus the model’s predicted net consumer

harm.

c. T he Evidence Is lnsufficient to Support Professor Shapiro ’s Profi`t Margin Figure

Finally, Professor Shapiro’s last input to his model is AT&T’s monthly profit

margins for its video customers See id. at 2245:7-9, 2315:12-17 (Shapiro). To calculate

those monthly video margins Professor Shapiro relied on internal AT&T figures measuring new customers’ “lifetime value” to AT&T, or “LTVs"’ Id. at 2344:12-16; id. at 2577:13- 14 (Carlton). ln particular, Professor Shapiro averaged AT&T’s reported LTVs for a three-month period ending in lune 2016. See id. at 2344:12-20, 3843:13-18 (Shapiro). That average generated a profit margin of $1,324, which Professor Shapiro used in his model to estimate the monetary benefits that AT&T would gain in the event of`a long-term Turner blackout. Id. at 3843:21-3 84414.

Defendants argue that Professor Shapiro’s 2016 LTV data is “outdated and thus not a reliable input into Professor Shapiro’s model.” Defs.’ PFOF 1 188. Defendants assert that Professor Shapiro instead should have used the “latest” available LTV figure from lune of 2017, or $821. Tr. 2508:.3 (Carlton); id. at 3844:9 (Shapiro). That $821 figure - disclosed by an AT&T witness and Professor Carlton after Professor Shapiro’s initial expert report and the close of fact discovery but before Professor Shapiro’s rebuttal report and the start oftrial - is approximately 40% lower than the 2016 margin figure used by Professor Shapiro to generate his original estimates of net consumer harm, See id. at 2448:17-2449:1 (Carlton). Defendants argue that using the $821 figure from 2017, rather than the $1,324 figure from 2016, significantly reduces the net consumer harm predicted by Professor Shapiro’s model. See id. at 2507:20-22 (“[l]f margins go down Professor Shapiro will predict lower increases in Turner content, even in his own model.”); id. at 2508:17-21 (using “the more up-to-date” profit margin figures “eliminates a large fraction

of all [of Professor Shapiro’s predicted] harms”).

At trial, each side spent much time attempting to justify or impugn Professor Shapiro’s reliance on the 2016 versus 2017 LTV data. The Government, for its part, raised questions about the genesis and legitimacy of the late-breaking 2017 margin data; on that score it requested, and was granted, the opportunity to depose the AT&T executive responsible for compiling and producing the data. Defendants on the other hand, questioned Professor Shapiro extensively about his continued reliance on the 2016 LTV data in the face of deposition testimony48 and Professor Carlton’s report, both of which disclosed updated 2017 LTV figures

While l have no reason to doubt Professor Shapiro’s good faith in continuing to rely upon the 2016 LTV data during his direct testimony for present purposes the important point is this the trial evidence indicates that Professor Shapiro’s 2016 LTV figures and thus his measure of AT&T’s margins are outdated and too high. That is true whether they are compared against the “most current finalized” lurie 2017 LTV figure ($821) cited by Professor Carlton id. at 3844:18, 3849:14-23 (Shapiro), or instead against an average of all three of the 2017 LTVs that had been finalized at the time of trial, id. at 2585113-22 (Carlton).

At trial, AT&T witness David Christopher testified about AT&T’s method for

generating the 2017 LTV data; he also confirmed the values of the finalized LTVs for

43 Specifically, David Christopher testified to the lune 2017 LTV figure during his deposition on February 14, 2018. See Tr. 3002:16-25. Although Professor Shapiro’s report cites Christopher’s deposition on the stand Professor Shapiro admitted that he did not read that deposition transcript and did not in fact know David Christopher’s role in the case See id. at 2345:17-2346:3 (Shapiro) (“Q: lfl told you that you cited to [Christopher’s1 deposition in your repoit, does that ring a bell'? A: No. Q: Well, did you read his deposition? A: l did not.").

lanuary, April, and lurie 2017. See id. at 300119-17, 3011:11-17 (Christopher (AT&T)). Although the Government rightly points out that such LTV numbers can (and, in the case of the 2017 LTVs, do) fluctuate from month to month, see id. at 3015110-24, the overall “downward trend is the same” id. at 3016:4', see also id. at 3003115-3004:15 (discussing downward pressures on LTVs). The declining state of AT&T’s 2017 LTVs, moreover, aligns with the testimony of numerous witnesses regarding the continued decrease ofvideo margins in the distribution industry See, e.g., id. at 3852:22-25 (Shapiro) ("Q: And you are aware sir, of the testimony of pretty much every other competitor witness in this case who has testified that their video margin are going down right‘? A: Yes."’).

Given that evidence it is perhaps unsurprising that even Professor Shapiro conceded during his rebuttal testimony that he "think[s] there`s some validity to using the 2017 margin instead ofthe 2016 margins” Id. at 3810:10-11; see also id. at 3843:17-18 ("[l]t would be reasonable to use the 2017 margins if one did it in the context of the rest of my analysis”); id. at 3849:5-8 (“Then when l’m given more data later and now we’ve had the trial, l understand that more; that’s why l said this time around, l could see using the 2017 data.”). Professor Shapiro also confirmed that using an average of all finalized 2017 LTVs would generate a 2016 net increase in MVPD costs of $98 million per year - a number “significantly lower” than his original estimate of $235 million in MVPD costs See id. at 3849124-3851 :3. Those lower MVPD costs in turn would decrease the predicted harm to consumers from the $.27 per-subscriber-per-inonth figure Professor Shapiro testified about

to a figure of approximately $.13 per-subscriber-per-month. See id. at 3851:6-14.

ln view of the above evidence l agree with defendants that the 2016 margin data utilized by Professor Shapiro is outdated and inflated.49 Whether one substitutes that figure for the lune 2017 LTV data or an average ofall ofthe finalized 2017 LTV data in Professor Shapiro’s model, the result is a significant decrease in the predicted amount of net consumer harm. Although that decrease standing alone does not eliminate all ofthe harms generated by Professor Shapiro’s model (just the bulk of them), it provides yet another

reason to reject the predictions offered by Professor Shapiro at trial.

49 With his model’s original reliance on the 2016 LTVs under attack, Professor Shapiro’s rebuttal testimony doubled down on an argument relating to the value ofAT&T’s existing customers The argument proceeds as follows ln addition to calculating LTVs for newly acquired video customers AT&T assigns margin values to its existing video subscribers Those values known as active customer values ("ACVs"), are generally higher than LTVs because they do not account for “subscriber acquisition costs."' Tr. 3854122- 3855:4 (Shapiro). Professor Shapiro’s long-term subscriber loss rate includes a measure of the existing customers that AT&T will retain as a result ofa long-term Turner blackout on its distribution rivals The value ofthose maintained customers Professor Shapiro opines is likely “50 percent higher” than the margin value for new-customers Id. at 2244113-21. Professor Shapiro did not, however attempt to generate or otherwise assign a “measure of the margin on the retained subscribers.” Id. at 224419-10. lnstead, his model only incorporates the margin value associated with new subscribers Id. at 2244:22. As a result, Professor Shapiro states that his "margin figure is definitely understated and substantially understated because l don’t have the proper data on the value ofthe retained customers” Id. at 2244114-17.

Therein lies the problem. Although opining about the importance ofthe value of retained customers to AT&T, Professor Shapiro undertook no analysis to incorporate that overall effect into his model. That should come as little surprise given that this “larger point"` appeared only in footnote 414 of the ninth appendix to Professor Shapiro’s 300-page expert report; nonetheless it renders his reliance on the existing- versus-new customer distinction unconvincing Id. at 3809;18, 3855:5-385615. That footnote Professor Shapiro testified, indicates that "the value ofexisting subscribers [is] between 150 and 225 percent as large as new subscribers” Id. at 3813:13-17. Beyond footnote 414’s general observation Professor Shapiro did not attempt to quantify the total dollar value of existing customers’ margins versus new customer margins much less incorporate a figure for existing customer margins into his model. Id. at 2244:22-2245:1 (“But, again the data l have available l`m using those gross add margins.”). On rebuttal, Professor Shapiro nonetheless cited that “higher itumber” as “what gives me a higher end of my range” of projected harm. Id. at 3819:25-3820:7. That does not appear to be the case: elsewhere Professor Shapiro testified that the “higher end” of the range derives from his use of a higher long-term subscriber loss rate of 14% (as compared to the 9% rate he chose to present to the Court during his direct testimony), rather than any alterations to other inputs such as the margin data. Id. at 2259:4-8 (“l realize there are ranges here These are based on we’re starting from the low end, 9 percent subscriber loss rate and projecting that. So if we started with the 14 percent, we’d have higher numbers.”); see also id. at 2239:3-7. Professor Shapiro’s belated attempts to link his point regarding the increased margins for existing customers to the high-end projections he reported, or to present those increased margins as ifthey were quantified and incorporated into his model, are thus unavailing and further undermine the credibility of his presentation

d. The Model’s Failure to Accountfor the Real- World Etects of Turner ’s Long-Term Contracts Further Undermines Its Probative Value

Turner is currently party to long-term affiliate agreements with nearly all of its distributors See Tr. 2316:3-18 (Shapiro); id. at 2444110-23 (Carlton); see also, e.g., PX211; PX410; PX422.50 Those agreements Professor Shapiro concedes will "prevent [Turner] from raising the fees for some number of years” and thus "temporarily constrain[]” his predicted effects ofthe merger in the real- orld. Tr. 2209:8-9, 16 (Shapiro). ln running his model and rendering his predictions however, Professor Shapiro curiously chose to ignore Turner’s current affiliate agreements At trial. Professor Shapiro explained - and anticipated cross-examination on - that choice by noting that his model is designed to “evaluate the fundamental incentives and changes in the market created by the merger.” Id. at 2208121-25, 2209:4-19. ln other words Professor Shapiro’s predictive exercise requires assessing “the longer term impact ofa new market structure”; factoring in Turner’s current affiliate agreements he noted, would be counterproductive because those agreements are “temporar[y]” and will “expire in time.” Id. at 2209111-19, 2320:24- 2321110.

The evidence in this case however` shows that the real-world effect of Turner’s present affiliate agreements will be rather “significant” until at least 2021. Id. at 2316114- 18. lndeed, Professor Shapiro conceded that by simply factoring in the presence of one

such affiliate agreement with a large distributor (which the Court will not name for

30 The primary exception is Charter, which has been displaying Turner content pursuant to temporary short-term extensions of the companies` affiliate agreement, which initially expired in 2016. See Tr. 1353121-135413 (l\/lontemagno (Charter)).

confidentiality purposes), the total MVPD price increase predicted by his model decreases by “about one-third” - a decrease that “take[s] away the vast majority the net effect" on MVPD monthly costs See id. at 2317:25-231816, 2319:10-16; see also id. at 2617:12- 2618:13 (Carlton) (factoring in that “one contract” reduces MVPD harm projection to "roughly a 5-cent projected price increase instead of a 27-cent price increase"). Not surprisingly Professor Carlton testified that simply by accounting for all current affiliate agreements and making no other changes to Professor Shapiro’s model, the model would generate a predicted net benefit to consumers rather than a net harm for the years 2016 and 2017. See id. at 2513:1-9 (2017) (discussing DXD116); id. at 2515:25-2516:1 (2016) (discussing DXD116).

ln other words given Turner’s existing contracts the level of post-merger harms predicted by Professor Shapiro’s existing model would not begin to phase in until at least 2021. But even Professor Shapiro concedes that 2021 is “‘getting out there a ways" and that “it gets harder" to predict actual harm that far down the line Id. at 2258:1-2, 2316:15- 2317:4-5 (Shapiro). That recognition reflects the testimony ofindustry witnesses many of whom testified that the landscape of the video distribution industry is continually changing and will continue to change as new entrants join the market. See, e.g., id. at 2456:7-11 (Carlton) (“So we have Netflix, we have Google coming in you have Amazon Prime. These are all big firms Apple and Facebook we know are coming in. . . .”); id. at 2948120- 2949:3 (Holanda (RCN)) (agreeing that migration to “OTT providers” is “likely to grow in the future as Millennials become more and more prominent in the marketplace”); cf id. at

3853118-19 (Shapiro) (“l think it is not disputed that the video margins are going down."). 147

l am thus left with projections of harm for the years 2016, 2017, and 2021 that all concede have not and will not occur in the real world due to Turner’s actual affiliate agreements See, e.g., id. at 23 1716-15 (Shapiro) (“Q: So let’s be clear about this when . . . you said $586,000,000 of annual price increase[s] to all of the MVPDs and a couple of virtual [MVPDS] in there right? A: That’s the number there Q: So just to be clear, that isn’t going to happen This isn’t going to happen let’s say in the year after the merger, right? That can’t happen A: That is true”). As such, l have no choice but to agree with Professor Carlton that Professor Shapiro’s model is “overestimating how quickly" the predicted harms “are going to start occurring.” Id. at 2444:15-23 (Carlton). To the extent. moreover, that the model projects “actual effects [that] will only occur gradually” after the largest of those agreements expires in 2021, even Professor Shapiro admits that it “gets harder” to project what the industry - and thus actual, real-world harm - will look like that far down the road Id. at 2209:17-19, 2316119-2317:5 (Shapiro); cf. id. at 235118-19 (Schlichting (DISH)) (Sling launched as the first virtual l\/lVPD in February 2015). For those reasons even putting aside the various problems with the model previously discussed,l conclude that the model’s predictions of harm are not ""sufficiently probable

and imminent to be probative in view ofthe facts ofthis case, especially “in the context" ofthe ever-increasing competitiveness of this “particular industry” Arch Coal, 329 F.

Supp. 2d at 115 (quoting Marine Bancorporation, 418 U.S. at 623 n.22); Aetna, 240 F.

Supp. 3d at 79 (quoting Brown Shoe, 370 U.S. at 321-22).

After hearing Professor Shapiro’s bargaining model described in open Courti l wondered on the record whether its complexity made it seem like a Rube Goldberg contraption. Professor Carlton agreed at the trial that that was a fair description See Tr. 2447:2-7 (Carlton). But in fairness to l\/lr. Goldberg at least his contraptions would normally move a pea from one side ofa room to another. By contrast, the evidence at trial showed that Professor Shapiro’s model lacks both “reliability and factual credibility"` and thus fails to generate probative predictions of future harm associated with the Government’s increased-leverage theory Anthem, 855 F.3d at 363. Accordingly neither Professor Shapiro’s model, nor his testimony based on it, provides me with an adequate basis to conclude that the challenged merger will lead to any raised costs on the part of distributors or consumers ~ much less consumer harms that outweigh the conceded _$350

million in annual cost savings to AT&T’s customers31

31 Although they amount to "extra icing on a cake already frosted,” there are even more reasons to be skeptical of the Government’s increased-leverage theory of competitive harm, Yates v. United States 135 S. Ct. 1074, 1093 (2015) (Kagan, l., dissenting).

First, the Court has reason to believe that, post-merger, AT&T will honor Turner’s commitment to arbitrate counterparties will agree to the terms of that commitment, and the prospect of arbitration will influence affiliate negotiations ln short, the commitment, made by Turner shortly after the filing of this suit, will have real-world effects For starters the proposed arbitration agreement is similar "`in many of the fundamental ways” to the arrangement blessed by the DOl, FCC, and this Court in the Comcast-NBCU merger. Tr. 268011-9 (Katz); see also 7/27/2011 Hr’g Tr. 714-71 1316-10, Comcast Corp., 808 F. Supp. 2d 145. Record evidence confirmed the real-world impact of an arbitration provision of this kind, giving the Court confidence both that arbitration offer will have import to negotiations and would be accepted by Turner’s counterparties See supra pp. 100-105 (reviewing econometric analysis of affiliate-agreement prices after the Comcast-NBCU merger); see also Tr. 1388:18-22 (l\/lontemagno (Charter)) (testifying to effects of arbitration in NBCU negotiations); id. at 2017:12-15 (Bond (NBCU)) (similar); id. at 121114- 122:9 (Fenwick (Cox)) (confirming that Cox had proposed arbitration “[j]ust like in Comcast case" as condition to this merger); id. at 464:17-20 (Schlichting (DISH)) (similar). Given its trial presentation 1 am hard-pressed to conclude that AT&T would (much less could) retreat from the commitment in light ofthe apparent reputational costs of doing so - costs that would imperil future negotiations in a marketplace with repeat players See, e.g., id. at 3261:23-3262:3 (Stankey (AT&T)); cf id. at 2622:4-2624:1 (Carlton).

IV. The Government Has Failed to Meet Its Burden to Show That the Proposed Merger Is Likely to Substantially Lessen Competition on the Theory That AT&T Will Act to Harm Virtual MVPDS Through lts Ownership of Time Warner Content

The Government’s second theory of competitive harm relates to virtual MVPDs Virtual MVPDs like traditional MVPDs offer consumers linear (or “live”) television programming in exchange for a subscription fee See supra pp. 1 1-13. Unlike traditional MVPDs however, virtual MVPDS transmit their video content over the internet. Id. Compared to traditional MVPDs virtual MVPDS generally offer lower-cost programming

packages to consumers those packages known in the industry as “skinny bundles ` contain fewer networks than do the larger bundles offered by MVPDs Id. Although virtual

MVPDs are of recent vintage they are quickly gaining market share in the video

Contrary to the Government’s insinuations about the reasons for the arbitration offer, moreover, the Court does not view the offer as akin to an admission by defendants that the proposed merger would lead to the anticompetitive harms that the Government posits Cf. id. at 39:1-5 (Gov’t Opening). lnstead, the Court credits lohn Stankey’s and Randall Stephenson’s testimony that the commitment was intended to “put our money where our mouth is” in showing that the proposed merger, far from being aimed at "do[ing] any ofthe things that the government allege[s],” is instead a “vision deal” being pursued to achieve "lower prices improved quality enhanced service [and] new products" [d. at 3261 1 16-326213 (Stankey (AT&T)); id at 3402:3 (Stephenson (AT&T)); see also id. at 3467:18-3468:9 (Stankey (AT&T)); id. at 3395123-25 (Stephenson (AT&T)); supra pp. 36-40.

Second, the Court observes that the Government’s increased-leverage theory fails to account for another feature ofthe market, namely the FCC’s program access rules As defendants’ expert, Professor Katz, testified, those rules are calculated to prevent precisely the kind ofharm predicted by the Government: a vertically integrated entity discriminatorily increasing programming prices on its distributor-rivals See Tr. 26931 14-2694:5 (Katz) (“They wanted to make sure that somehow control of the programmer wasn`t used to harm competition.”); 47 U.S.C. § 548(b), Q); 47 C.F.R. § 76.1001(b)(1)(i)-(ii)', see 47 U.S.C. § 548(c)(2). Those regulations are a proper subject of antitrust analysis see Verizon Comms Inc. v Law ()ffices ofCurtis V. Trinko, LLP, 540 U.S. 398, 41 1-12 (2004), and appear to be squarely on-point, at least according to the unrebutted testimony of Professor Katz. See Tr. 2693119-2694:1 (Katz) (“[T]here are two broad categories One category . . . prohibits discrimination against different distributors And the other broad category prohibits the distributor from having undue influence on the decisions ofthe programmer. So, again the idea of you don’t want the distributor telling the programmer to go do things to harm other distributors.”). Nevertheless the Government all but wishes them away - and does so with little explanation or, more importantly record evidence

programming and distribution industry See Tr. 448:24-44912 (Schlichting (DlSH)). Examples of virtual MVPDS include AT&T’s DirecTV Now, DlSH’s Sling TV, Sony’s Playstation Vue, Hulu Live Google’s Youlube TV, FuboTV, and Philo. Gov’t PFOF 1141 Defs.’ PFOF 1 8.

According to the Government, the challenged merger would give AT&T the "ability to harm competition by slowing the growth of emerging innovative online distributors” - that is virtual MVPDs Gov’t PFOF 104. AT&T could do so, the Government asserts either acting on its own (under the “unilateral theory") or in coordination with Comcast- NBCU (under the “coordination theory”). See Gov’t PCOL 1 63.32 Defendants counter that the evidence does not support the Government’s virtual MVPD theories Far from showing that AT&T is trying to marginalize virtual MVPDs defendants claim that the trial demonstrated that AT&T is embracing those providers - even launching and supporting a successful virtual MVPD, DirecTV Now. With respect to the supposed incentive to coordinate with Comcast, defendants argue that the Government’s theory ignores critical differences between the positions of AT&T and those of Comcast vis-a-vis virtual MVPDs as well as key limitations on the companies’ abilities to coordinate successfully For the

following reasons l agree with the defendants that the Government has failed to show a

32 lt will come as no surprise that a basic premise of the virtual claims _ as for the Government’s increased-leverage theory - is the literal “must have” nature of Tumer programming For all the reasons stated earlier in this opinion the Court is skeptical that, in the Government’s words virtual MVPDs are “dependent on programmers” like Turner. Gov’t PFOF 1 17. For instance Sling the most successful virtual MVPD, offers a package_without broadcast stations and does not offer CBS at all. See Tr. 351:12- 25 (Schlichting (DISH)). As Sling President Warren Schlichting explained, the whole point of virtual MVPDs like Sling in fact, is to carryfewer channels See id. at 236:2-6 (“Q. Do you carry all the same channels as other pay-TV services? A. Certainly not all ofthem. One ofthe places that we tried to innovate is to carry fewer channels many fewer channels.”).

likelihood that the merger would substantially lessen competition by empowering the merged company to act, either unilaterally or in coordination with Comcast-NBCU, to harm virtual MVPDs

Unilateral Theory The Government first claims that AT&T has an incentive to harm innovative virtual MVPDs and could act unilaterally on that incentive by foreclosing or restricting virtual MVPDS’ access to “must-have” Turner content. See Gov’t Post-Tr. Br. ll`. That is a curious claim, to say the least, in light of Professor Shapiro’s testimony that, in his view, “standing alone acting unilaterally the - AT&T will still want to license the Turner content to virtual MVPDs” Tr. 2260:19-21 (Shapiro) (emphasis added); see id. at 2291:8-11 (“Q: Now with respect‘to coordination you’ve made no claim that AT&T post merger would have a unilateral incentive to withhold Turner content from virtual MVPDs correct? A: Correct.”); id. at 2293:9-13 (“Q: And you’re not contending and you’ve rendered no opinion that they will withhold Turner content from MVPD[s], correct‘? A: That’s correct. Q: Or as we said unilaterally from virtual MVPDs correct? A: Also correct.”). That is so, according to Professor Shapiro, because as with traditional MVPDs it would be “profitable” for the merged entity to continue to license lime Warner content to virtual MVPDs Id. at 2293:14-17.

lf citing Professor Shapiro’s testimony weren’t enough to dispel the Government`s unilateral virtual MVPD theory defendants put forward additional evidence that AT&T would have incentive to license Time Wamer content to virtual MVPDs after the merger. For starters given lurner’s imperative of broad distribution see supra pp. 10-11, lurner

executives testified that it is important for Turner’s content to be included on virtual

MVPDS as they continue to grow in relevance With consumers choosing to cut or shave the cord, Turner has “embrac[ed] virtual MVPDs” Turner CEO lohn l\/lartin testified “because, again we need to be distributed to as full distribution as possible.” Id. at 607:13- 16 (l\/lartin (Tumer)); see also id. at 3157:22-3158:7 (Bewkes (Time Warner)) (explaining that virtual MVPDS are a favorable trend because they are "another place where we could put our networks in front of consumers”); id. at 1064:25-106513 (Breland (lurner)) (“Q. . . . 1W]hat was your strategy with respect to negotiating with the new entrant virtual MVPDS‘? A. l want to be on every platform that comes”); cf. id. at 3126:8-16 (Bewkes (Time Warner)) (stating that the Government’s coordination theory “makes no sense” because "[w]e want to be on all the virtual MVPDs"`).

The entire premise of the proposed merger ~ allowing AT&T to go mobile with video content - provides yet another reason to reject the Government’s unilateral merger theory See id. at 3393124-25 (describing plans to deploy lime Warner video content over AT&T’s wireless network in order to make that content "worth far more"); see also PX456- 3 (discussing merger strategy and AT&T “strategy of ensuring that its content is available to consumers on a wide range of distribution platforms”). AT&T’s largest business is its wireless business where it has more than 100 million subscribers [d. at 3208:19-24 (Stankey (AT&T)); id. at 3379:19-20 (Stephenson (AT&T)). On its own ifseparated from the rest ofthe corporation AT&T`s wireless business would be l"number 37 on the Fortune 500” - approximately the size of Proctor & Gamble Id. at 3379:20-3380:1 (Stephenson

(AT&T)).

Within its wireless business AT&T Chairman and CEO Randall Stephenson explained, “getting video delivered onto the mobile device” is one of AT&T’s “big focus areas” Id. at 3381:24-25; see id. at 3208:20-22 (testifying about Al&l’s goal of “transform[ing] the way we deliver video to customers [to] make the video far more portable”). Increased video consumption is lucrative for Al&T because viewers consume more data on the wireless network. This leads AT&T customers to “buy up” on data plans get more devices or connect more devices to the network - all “good for [AT&T`s] business” Id. at 3254:19-22 (Stankey (AT&T)). lndeed, “over half of all ofthe traffic on [AT&T’s] network today is video, delivering video.” Id. at 3382:4-5 (Stephenson (AT&T)).

lndustry trend-lines point toward increased video consumption in the future - and AT&T aims to ride these tailwinds See id. at 3505:21-3507:2. Right now, AT&T is working to develop fifth-generation wireless which will drive video consumption even more Id. at 3382:7-3383:5. And AT&T views mobile consumption of video, including through virtual MVPDs as a critical part of its post-merger future See id. at 3506:23-25 (“What we’re all working towards is creating [$]35 and $15 bundles And that’s where the world is moving.”). Notably, the benefits associated with AT&T customers accessing virtual MVPD content continue to accrue even when they use DirecTV Now’s competitors like Sling and Youlube TV. See id. at 3432:16-20 (“With AT&T, we’re in a unique position We like over-the-top. Over-the-top generally means in this day and age wireless People are using their wireless devices to watch video, whether it’s our video or not, we’re

somewhat ambivalent.”). All of this gives the combined entity even more reason to

distribute Time Warner content as broadly as possible in order to encourage the proliferation of virtual MVPDs As Randall Stephenson put it, the proposed merger is a "vision deal” reflecting a belief“that distribution of [Time Warner] content to wireless will drive the value of the content up” and that “the ability to pair our data with [Time Warner’s] advertising inventory” for digital ads delivered over the internet “will drive value.” Id. at 3402124-340316.

Against that evidence the Government cites a handful of AT&T documents and statements related to virtual MVPDs ~ documents the Government says show AT&T has the incentive to slow the rise of virtual MVPDs See, e.g., PX42; PX228; PX40; PX47; PX48. For multiple reasons however, l do not consider the fact that AT&T executives may have previously expressed displeasure with Turner`s relationships with its competitor virtual MVPDs to be probative of AT&T’s post-merger economic incentive to license Turner content to virtual MVPDs First, these statements shed no light on the post-merger incentive AT&T would have to maximize distribution of Turner content. As the reader now knows wide distribution is the sine qua non of the programming industry driving both subscription and advertising revenue lndeed, because of these “[gains] from trade" associated with licensing Turner content as broadly as possible Professor Shapiro himself refused to countenance the Government’s unilateral virtual MVPD theory Tr. 2293:12- 17. Second, these statements do not explain why AT&T would discard the profits

associated with increased video consumption by its 100 million-plus wireless subscribers

accessing virtual MVPD offerings ln short, the Government’s evidence on its unilateral withholding theory is fatally anemic.33

Second, from the other direction the Government advances an alternative unilateral claim: that AT&T would have the ability to break the “skinny bundle” models of virtual MVPDS by forcing those distributors to take too many Turner networks Citing the testimony of Sling’s President, Warren Schlichting the Government argues that a post- merger requirement that Sling “take eight Tumer networks instead of four would ‘break [Sling’s] model”’ and, indeed, would have a snowball effect with other programmers Gov’t PFOF 1 288 (quoting Tr. 265:17-266:8, 268:9-23 (Schlichting (DISH)).

That argument, however, ignores that Turner has less of an imperative to risk a deal with Sling (or other virtual MVPDs) by insisting on carriage of all of its networks That is so, the evidence indicates because Turner has a highly “concentrated portfolio of

networks” Tr. 55811 (Martin (Turner)), with 85 to 90% of Turner’s revenues deriving from

only four networks see Defs.’ PCOL 1 51 n.39; accord Gov’t PFOF 1 75. That fact, as

33 To the extent the Government seeks to recycle these statements for purposes of its coordination theory this evidence is unpersuasive on that count, too. The combined entity would stand to gain much from wide distribution of lime Warner content to virtual MVPDs and stand to lose much by refusing to do so. The Government’s remaining fact evidence similarly fails to establish any incentive to act, unilaterally or coordination to stifle virtual MVPDs To the extent the Government seeks to recycle the slide deck, PX184, PX543, or Schlichting’s testimony for its virtual claims that evidence remains oflimited probative value for the reasons stated above See supra pp. 86-88 (PX184, PX543); see supra pp. 75-78 (Schlichting testimony). Nor does additional speculation ofthird parties see Gov’t PFOF 11 291-292, or testimony as to the “importan[cel” of Turner content to virtual MVPDs see id. 11 293-294 _ even if presented for the first time in this section _ move the needle Altogether, the best the Government could marshal was a statement from AT&T’s lohn Stankey that “we kind of expected [Sling1 might be concerned about” AT&T attacking their skinny bundle See Tr. 3256:3-15. Such evidence on its own or in combination simply cannot countermand the prime directive of programming _ broad distribution - not to mention AT&T’s independent incentive to grow video consumption on its wireless network, see supra pp. 153-155.

lime Wamer CEO leff Bewkes noted ~ means Turner is “better placed” to succeed in the skinny bundle model. Tr. 3126:22. The Government’s skinny bundle point also overlooks the fact that Turner ~ like other programmers - already fights tooth and nail to get all of its networks into all of the packages of every distributor. See id. at 433118-21 (Schlichting (DISH)); id. at 606:5-11 (l\/lartin (Turner)). Simply put, the Government has not produced sufficient evidence to show that the challenged merger is likely to make a meaningful difference to that dynamic.34 F or all of the above reasons l conclude that the Government has failed to meet its burden on its claims arising from AT&T’s asserted potential to unilaterally harm virtual MVPDS through its post-merger control of Turner content. Coordination Theory. The Government posits that the challenged merger would also create a likelihood that AT&T would coordinate with Comcast-NBCU to harm virtual MVPDs ln contrast to the unilateral withholding claim just discussed, the Government did at least attempt to provide some expert support for this coordination claim. See id. at 2261:14-20 (Shapiro). Unfortunately for the Government, however, neither that expert testimony nor its other evidence is even close to sufficient to support its coordination claim.

How so?

34 ln support of the notion that virtual MVPDs need Turner networks (again, in the most literal sense), the Government points to a statement by lohn l\/lartin Turner’s Chairman and CEO, that Sling would be “shit without Tumer.” Gov’t PFOF 1 156 (quoting PX4). This statement does not accomplish the work that the Government thinks it does For starters as discussed above the very “skinny bundle” concept embraces fewer networks - even fewer popular ones - with the knowledge that some consumers will welcome the trade of fewer networks for a lower subscription fee And second, it should come as no surprise that - even iri colorful language ~ executives would be avid boosters for their companies’ products ln the final analysis the Government’s repeated use of this John l\/lartin quote see Tr. 1213-7 (Gov`t Opening), 17-18 (Gov’t Closing), calls to the mind one Court’s admonition “rummageling] through business records” for “tidbits that will sound impressive (or aggressive)” undermines efforts to ensure “accuracy of decisions.” A.A. Poultry Farms, Inc., 881 F.2d at 1402.

A proposed merger may violate Section 7 by "enabling or encouraging post-merger coordinated interaction among firms in the relevant market that harms [consumers]." Gov’t PCOL 167 (quoting FTC v. OSF Healthcare Sys., 852 F. Supp. 2d 1069, 1086 (N.D. lll. 2012)). Such coordinated conduct need not constitute an illegal agreement under Section 1 ofthe Sherman Act, but instead can comprise instances oftacit coordination Cf Heinz. 246 F.3d at 715 (coordinated effects can occur “either by overt collusion or implicit understanding”). ln order to assess whether a merger will lead to an unacceptable risk of competition-stifling coordination courts evaluate various “market conditions on the whole.” H & R Block, 833 F. Supp. 2d at 77 (citation omitted). ln short, that analysis involves consideration of whether would-be coordinators could wield anticompetitive power “by recognizing their shared economic interests and their interdependence with respect to price and output decisions.” Brooke Grp., 509 U.S. at 227. Not so here l

As it does for its other claims the Government relies on a key assumption when pressing its theory of coordinated effects Here the Government assumes that, “`[a]s the only two vertically integrated traditional MVPDs Comcast and AT&T would share an incentive to slow the entry and growth of virtual MVPDS.” Gov’t PFOF 1 299. To act on

that incentive the Government further asserts the companies could mutually forbear”’

from licensing their programming content without any communication between them.”’ Id. (quoting Tr. 2265:5-2265:6 (Shapiro)). Not only is that theory overly speculative it

ignores key differences between AT&T and Comcast that undermine the Government’s

argument

First, the Government has failed to put forward sufficient evidence to show more than a theoretical “possibility” of coordination Cf. Baker Hughes, 908 F.2d at 984 (“Section 7 involvesprobabilities, not certainties or possibilities”). lndeed, the Court need look no further than the testimony of Professor Shapiro in that regard When questioned at trial about the Government’s coordinated effects theory Professor Shapiro conceded that he had no “way of accessing [sic] the probability"` of coordination and thus had not attempted to “quantifiy] any risk whatsoever” that the predicted coordination "could occur.” See Tr. 2291:XX-XXXXXXX (Shapiro).33 Accordingly, Professor Shapiro confirmed that he was “not in a position to say” that coordination is “more likely to happen than not," and indeed was not even prepared to say that there’s a “one percent chance that coordination will happen” as a result ofthe challenged merger. Id. at 2292:6-13. Given

that testimony and the lack of “a detailed theory” with respect to coordination l can

33 The Government insists that it need not introduce quantitative evidence in support of the coordinated effects theory See Gov’t PCOL171. The suggestion ofcourse is that the Court should steer clear of imposing a requirement that the Government make a numbers-based showing on coordinated effects Let me be clear. The Government here has failed to carry its burden on the coordination theory not because there is some per se requirement of quantitative analysis Rather, the Government has failed to carry its burden because it has not put forward persuasive evidence - in any form - that AT&T and Comcast have the incentive or, given market constraints the ability to coordinate in the manner predicted

There is one more point. The cases cited by the Government do involve quantitative showings ln each one the Court made or adopted a threshold quantitative assessment as to market concentration See H & R Block, 833 F. Supp. 2d at 71-72 (applying Herfindahl-Hirschmann lndex to determine market concentration); OSF Healthcare 852 F. Supp. 2d at 1078-80 (same); see also Hosp. Corp., 807 F.2d at 1384 (accepting “FTC’s figures" as to “highly concentrated market"). That determination in turn triggered the ""ordinary presumption of collusion’ that attaches to a merger in a highly concentrated market." H & R Block, 833 F. Supp. 2d at 77 (quoting Heinz, 246 F.3d at 725). And with that presumption in place the burden shifted to defendants to rebut the case by “produc[ing] evidence of ‘structural market barriers to collusion’ specific to [the relevant] industry that would defeat” the presumption Id. (quoting Heinz, 246 F.3d at 725). Thus, the Government’s insinuation that past coordinated-effects challenges were tried without resort to quantitative analysis is simply misleading ln short, the Government cannot evade its burden of proof on the “ultimate issue [ofl whether the challenged acquisition is likely to facilitate collusion” Hosp. Corp., 807 F.2d at 1384; Gov’t PCOL 1 71, by simply stating that it "does not need to quantify the potential harm,” Gov’t PCOL 1 71.

sympathize with Professor Carlton’s reaction: “l’m not quite sure what l’m supposed to rebut on [t]his.” Id. at 2454:1-10 (Carlton).

Second, the Government’s argument regarding the incentive of AT&T and Comcast to coordinate to harm virtual MVPDs ignores that both stand to lose large amounts of affiliate fee and advertising revenues by withholding their content from virtual MVPDs See supra pp. 10-11; Tr. 3126:8-16 (Bewkes (Time Warner)) (stating that the Government’s coordination theory “makes no sense"` because “[w]e want to be on all the virtual MVPDs”); id. at 2020:5-18 (Bond (NBCU)) (“Q: Why have you decided to license your networks to each of those virtual MVPDS? A: Well, simply we’re interested in getting the most amount of distribution that we can get, and they represent an important new pathway of distribution As l said. they now have well over three million subscribers in total. . . . [l]fwe were not on those platforms we would have you know, three million less subs fewer subs.”). Unsurprisingly, NBCU has licensed its content to each virtual MVPD. See id. at 2019:15-202012 (Bond (NBCU)). The Government has not explained why either company would be willing to forgo those affiliate fees and advertising revenues from virtual MVPDs Nor has the Government proffered any expert analysis for example of how those economics could, or would, change assuming a coordinated blackout of both Turner and NBCU.

Third, and critically the Government’s argument also ignores key differences between the two companies - differences that AT&T executives believe give AT&T a competitive advantage over Comcast moving forward in this new era of rising virtual

MVPD prevalence AT&T’s John Stankey who will be responsible for running lime 160

Wamer should the challenged merger proceed, emphatically (and credibly) stated at trial that he could not “even imagine” aligning with Comcast given the companies" history of dealings adding “l’m not going to cooperate with somebody l don’t like.” Id. at 3255:2- 325612 (Stankey (AT&T)). AT&T CEO Randall Stephenson testified similarly responding to a question about the Government’s coordination theory as follows "You probably have to live in this industry every day like l do to appreciate what a stretch that is We compete with Comcast in the marketplace The individual that runs communication company he wakes up every day trying to think, how do l win in the marketplace against Comcast?” Id. at 3431125-3432:5 (Stephenson (AT&T)).

The most obvious “advantage” AT&T has over Comcast when it comes to virtual MVPDs is that, unlike Comcast, and as discussed at length above AT&T has a vast wireless business with over 100 million customers Id. at 3432:2-7; id. at 3208119-24 (Stankey (AT&T)); see also id. at 3432:17-22 (Stephenson (AT&T)) (“Over-the-top generally means in this day and age wireless People are using their wireless devices to watch video, whether it’s our video or not, we’re somewhat ambivalent. We’d rather it be our video; but either way it serves our interests for people to watch video over our wireless network.”); see also supra pp. 153-155. The reasons to encourage not quash, virtual MVPDs unilaterally become even more compelling in the context ofa coordination claim with Comcast - a competitor that is much more beholden to legacy cable infrastructure and the traditional MVPD business model. See id. at 343212-12 (Stephenson (AT&T)); cf. id. at 3255118-22 (Stankey (AT&T)) (“We don’t want to cooperate with Comcast to play their

game We want to figure out how we use our mobile devices and our mobile network to

change the game . . . ."`); id. at 3208:19-24 (“[O]ne of the clear objectives [for AT&T in acquiring DirecTV] was to start to transform the way we deliver video to customers [to] make the video far more portable start to emphasize the fact that we could use our 100 million wireless subscribers to be able to do things differently which is dramatically different than Comcast.”).

The Government does not dispute that AT&T’s wireless business confers strong incentives to maximize distribution to virtual MVPDs Nor can it be questioned that AT&T’s strong positioning in the world of mobile content distribution gives it a powerful disincentive to work with Comcast to stifle those mobile providers of video. AT&T has plainly positioned itself to ride industry tailwinds in support of mobile consumption of video. As l ohn Stankey explained, AT&T acquired DirecTV in 2015 not in order to double down on the satellite business a concededly mature and declining asset, but to “pick up a lot of new customers that we could work on migrating” to new products Id. at 3207:18- 20 (Stankey (AT&T)); see also id. at 3207:21-3208:2. lndeed, as soon as the merger closed, AT&T began renegotiating DirecTV’s contracts to allow for a mobile direct-to- consumer option DirecTV Now. AT&T knew that it was “in a foot race to basically start to change the product to be able to catch the next wave whatever that next wave was going to be And we didn’t expect that we were going to continue to see traditional pay-TV subscribers” increasing Id. at 3209112-16. Nowhere does the Government explain why AT&T would deploy valuable lime Warner content to prop up a rival’s business model,

while harming its own Go figure !

This fundamental problem of incentives and profitability buries the Government’s claim. lt is beyond dispute that neither the proffered concentration in the MVPD market (which, by the way, will be the same post-merger), see Gov’t PFOF 1 306, nor the importance of Turner and NBCU content, see id. 1 307, nor some transparency in “'key information" see id. 11 308-310, nor any other of the Government’s evidence on the coordination theory (alone or in combination), can establish a "risk of coordination" unless the parties have an incentive or interest to collude in the first place

Even assuming contrary to the evidence that AT&T would want to coordinate with Comcast under the Government’s theory the staggered, lengthy industry contracts would make that coordination strategy extremely risky See id. at 643:20-64412 (Martin (Turner)) (testifying that “because ofthe length ofthese contracts because they’re typically years in length,” a strategy set “in 2013” would “begin to show up in ’ 15, ’16 and ’17”); id. at 87:9- 1 1 (Fenwick (Cox)) (testifying that affiliate agreements run "between five and eight years on average”). Under the Government’s coordination theory one party - AT&T or Comcast - would have to “jump first,” giving up valuable programming rights on the hope that the other, in some years’ time would elect to do the same lndeed, this barrier to coordination is so great as to put to rest the notion not only that AT&T and Comcast would have the incentive to coordinate but that the post-merger marketplace would afford them the ability to do so. Whether by way oftacit coordination or an illegal agreement, putting such blind faith in one’s chief competitor strikes this Court as exceedingly implausible l lndeed, the decision to “not to renew [a] license or not to license to a new virtual MVPD and wait and

see if the other did it," as Professor Shapiro proposes would enhance the other party`s

position in its next round of negotiations with the virtual MVPD at issue Tr. 2264:14- 2265:13 (Shapiro). As Charter’s Tom l\/lontemagno explained, if a distributor goes dark with one network group, that distributor is in “a vulnerable spot, and l feel like l sort of have to do the deal” when another network group threatens a blackout. Id. at 1404:13-15. The result would be forgone revenue for a period of years with AT&T’s chief competitor gaining outsized profits in the next round of negotiations The Government puts forward no persuasive reason why AT&T and Comcast would engage in such conduct.

The fundamental difference in incentives between AT&T and Comcast vis-a-vis virtual MVPDs the barrier to coordination in the form of long-term contracts coupled with the fact that the Government has provided no evidence to show how the benefits of a coordinated blackout would outweigh the companies’ resulting losses of affiliate fee and advertising revenues leave me completely unable to accept the Government’s coordinated

effects theory.36

36 ln support ofits coordination theory the Government points to past communication between Dan

York of AT&T and counterparts at other distributors in the Los Angeles market concerning the Sportsnet LA network, See Gov’t PFOF111 311-312; Tr. 2081:9-2081116 (York (AT&T)); PX462. These instances are only weakly probative of future coordination involving as they do, a different market, distinctive factual setting and different distributors ln all respects this evidence cannot overcome AT&T’s strong disincentives to coordinate with Comcast detailed in this section Cf. H & R Block, 833 F. Supp. 2d at 77- 78 (detailing “highly persuasive historical act of cooperation” between the same two parties at the center of post-merger coordination allegations). The same goes for inquiries by York concerning Verizon Fios packages or evidence regarding lohn Harran’s conversations with his counterpart and “good friend" at NBCU. See Defs.’ PFOF 1 2911 Gov’t PFOF1313.

V. The Government Has Failed to Meet lts Burden to Show That the Proposed Merger Is Likely to Substantially Lessen Competition on the Theory That AT&T Will Restrict Distributors’ Use of HBO as a Promotional Tool

The Government’s final theory centers on HBO. On this score the Government alleges that the combined entity will have the “incentive and ability” to prevent rival distributors from using HBO as a promotional tool to attract and retain customers See

Gov’t Post-Tr. Br. 9-10; Compl. 1 39.37 Under this theory the combined entity would

31 ln its proposed conclusions of law, the Government describes its theory that the merged entity might “restrict the use of HBO as a competitive tool.” Gov’t PCOL 1 61; see also Tr. 3993:7-10 (Gov’t Closing) (“lt means that ifthis merger goes forward, then the combined firm could limit the use of HBO as a competitive tool, if that competition threatens to impact AT&T.”). Under this theory HBO is a "competitive tool"’ insofar as it is used by distributors for discounts promotions marketing and ad campaigns See Gov’t PCOL161 (predicting that AT&T will have an incentive “to restrict the use of HBO as a competitive tool, and thereby impair the competitive process and deny consumers the benefits of discounted HBO and otherpromotions" (emphasis added)); see also Gov’t PFOF 1 234 ("Overall, HBO is a highly valuable brand, which currently engages in significant promotional activities with MVPDs both AT&T and its rivals”). This is consistent with the way in which Professor Shapiro viewed the theory See Tr. 2290:25-2291:3 (Shapiro) (“Q. The only theory of harm that you considered relating to HBO is this issue that perhaps some promotional, some promotion of HBO might be curtailed, right? A. That’s fair.”). lt is also consistent with the way in which the Government’s Complaint and Pre-Trial Brief characterized the theory See Compl. 1 39(“1\/1VPDs . . . today use HBO as a tool to entice new customers and to dissuade unhappy customers from leaving and switching to a rival MVPD. . . . After the merger, however, the merged firm would have the incentive and ability through contractual restrictions to impede rival MVPDs from using HBO to compete against AT&T/DirecT\/.”); Gov’t Pre-Trial Br. 39 (“HBO could limit approvals for the use of HBO in marketing and promotions by DirecTV`s rivals in a number of ways including forms of subtle or targeted obstruction.”).

The Government’s proposed findings of fact, like its closing argument, appear to advance a considerably broader theory on the ways in which HBO could limit the terms ofits distribution post-merger. Such a theory would go well beyond restricting promotion-related terms See Gov’t PFOF 1 267 (listing ways in which HBO could restrict distributors’ offerings of HBO to customers); Tr. 3975:11-19 (Gov’t Closing) (same). l\/Iost troubling is the Government’s suggestion based solely on the testimony of l\/lartin Hinson of Cox, that the combined entity could “withhold[] HBO entirely.” Gov’t PFOF 1 267 (citing Tr. 703125-704:18(Hinson(Cox)). Professor Shapiro himselfdisavowed this very theory of withholding HBO content; “Q. You don’t claim that post-merger HBO will be withheld from any MVPD, correct‘? A. Correct." Tr. 2290:15-18 (Shapiro)v Professor Shapiro similarly disavowed any claim that HBO`s price would increase on account ofthe merger. See id. at 2290:21-23.

For the reasons discussed in this Part, the Government has failed to prove that the merged entity has an incentive to restrict rival distributors’ use of HBO for promotions To the extent that the Government suggests that AT&T will withhold HBO content altogether, will delay access to HBO content, will increase penetration rate requirements or will engage in any other potentially anticompetitive conduct that falls outside the proffered promotion-withholding scheme the Court holds that, in light of the sparse supporting

“foreclos[e] competitors of the purchasing firm in the merger from access to a potential source of supply or from access on competitive terms.” Gov’t PCOL 1 61 (quoting Yankees Entm’t & Sports Network, 224 F. Supp. 2d at 673). The basic idea, the Government tells us, is that rival distributors’ use of HBO in promotions will tend to draw potential customers to those MVPDs and away from AT&T, thereby giving AT&T reason to withhold or restrict its consent to use HBO in marketing discounts and bundles See Gov’t PFOF 1 234. At the risk of stating the obvious this is a gossamer thin claim.

The Government has failed to meet its burden of proof on this theory for two independent reasons First, the Government has failed to show that the merged entity would have any incentive to foreclose rivals’ access to HBO-based promotions This is because the Government’s promotion-withholding theory conflicts with HBO’s business model, which remains “heavily dependent” on promotion by distributors Tr. 3074:5-6 (Bewkes (Time Warner)). HBO does not run ads leaving subscription fees as its overwhelming source of revenue See id. at 3070:3-5; PX456-67. This makes HBO a volume-based business in which more subscribers means more revenue See Tr. 3070:3- 8, 3072:7-9 (Bewkes (Time Warner)). And because HBO continues to rely on distributors to reach the end-user, witnesses testified that HBO needs MVPD promotions in order to achieve this volume See, e.g., id. at 3128:16-3129:8; id. at 1496110-17 (Sutton (HBO)); see also id. at 1508:14-16 (“[O]ur whole business is relying on our affiliates to promote us

lfwe can’t do that, then our entire business model is destroyed.”); cf. id. at 1528:25-1529:4

evidence and Professor Shapiro’s disavowal ofthose theories the Government has failed to meet its burden of proofthat such conduct would likely result from the proposed merger.

(Patel (AT&T)). The Government simply fails to explain why AT&T wouldjeopardize - much lessjettison -the promotional model on which HBO “absolutely” depends.38 Id. at 1496:16-17 (Sutton (HBO)).

Second, the Government fails to establish that HBO promotions are so valuable that withholding or restricting them will drive customers to AT&T.39 Put differently the Government has failed to show that the marketplace substitutes for HBO are “inferior, inadequate or more costly.” Gov’t PCOL 1 62 (internal quotation marks omitted). Third- party distributor witnesses testified that, for example their companies had reduced the use of HBO in promotions see Tr. 950:22-95117 (SEALED); id. at 2135:17-22, 2135:24- 2136:1 (Sejen (Cable ONE)). An executive from RCN said that his employer used HBO for promotions only because of the “economic incentives” offered by HBO to do so. See id. at 2971:16-23 (Holanda (RCN)); cf id. at 2136:15-19 (Sejen (Cable ONE)). A Comcast executive confirmed that Netflix is a “substitute" for HBO that Comcast has incorporated into its set top box and includes in marketing See id. 886:8-22 (Rigdon (Comcast)). This is all consistent with other evidence adduced at trial, which showed that distributors’ choice

of which premium content provider to use for promotions may vary based on a number of

38 As an add-on HBO is low-hanging fruit for customers looking to shave monthly cable bills Cf Tr. 2137:3-6 (Sejen (Cable ONE)). This results in high “churn” making HBO that much more reliant on promotions to maintain subscriptions See id. at 2316:10-12; id. at 2972:20-24 (Holanda (RCN)). ln these promotions HBO depends on distributors because “the distributor . . . owns the relationship with the customer.” Id. at 1528:22-1529:4(Pate1 (AT&T)).

39 The Court is aware that, in the most technical sense HBO has the “ability” to withhold certain promotions by way of its contract-based approval process under which HBO must bless distributors’ use of HBO trademarks and talent for us in promotions This fact alone however, does not establish that AT&T would be able to "impair the competitive process” Gov’t PCOL161. For its theory the Government must also show that HBO has an incentive to act anticompetitiver and that only "inferior, inadequate or more costly” substitutes for HBO promotions exist in the marketplace id. 1 62 (citation omitted). The Government has failed to make these showings

factors See id. at 1526:17-25 (Patel (AT&T)).60 lndeed, the evidence at trial further showed that MVPDS are hardly limited to premium content providers like HBO, Showtime, and Netflix in their choice of promotional tools to the contrary distributors have been known to bundle services with gift cards price discounts higher broadband speeds additional telephone lines video on demand films devices such as iPads, and free installations or equipment [d. at 717:15-25 (Hinson (Cox)); id. at 2972:1-6 (Holanda (RCN)); id. at 1497:5-10 (Sutton (HBO)).

Although this promotion-withholding theory made only a very brief appearance at trial, the Government asserts that this theory of harm constitutes an independent basis for blocking the merger. Gov’t PCOL 11 61-62; Gov’t Pre-Trial Br. 40.61 But in support of

this theory the Government has brought to bear little evidence indeed As with its primary

66 After a trial replete with evidence on evolving hyper-competitive marketplace conditions the notion that Netflix is an adequate substitute for HBO should come as no surprise “There was a time." HBO President Simon Sutton explained, "when very few people were making the kinds of shows we make [at HBO]. Now, it seems like almost every week, there’s an announcement of somebody else making it." Tr. 1494:13-21. Netflix now has a programming budget that more than doubles HBO’s id. at 3099:13-15 (Bewkes (Time Warner)), and Netflix and HBO openly compete "in litany different ways,” including for “the talent to make the same shows,” id. at 1493118-149413 (Sutton (HBO)). And when measured by number of subscribers both Netflix and Amazon are “eclipsing HBO.” DX709-3. lndeed, one of the Government’s experts in an improper communication sent to Government attorneys during the course of his testimony in violation of the Court’s witnesses rule, forwarded a YouTube video describing Netflix as one ofthe “top-ten . . . monopolists you’ve never heard of.” See Tr. 3602:17-360317, 360417-25 (Quintero). Put simply HBO is in the fight ofits life!

61 The Government appears to suggest that incentive to engage in anticompetitive conduct - without any demonstration as to the probability of acting on that incentive - is sufficient reason to block a proposed merger. See Gov’t PCOL 1 61 (“ln this action the effect of the merger may be to lessen competition substantially by incentivizing the merged firm to restrict the use of HBO as a competitive tool, and thereby impair the competitive process and deny consumers the benefits of discounted HBO and other promotions”). This proposition seems impossible to square with the legal standards governing Section 7 actions which require a probability of anticompetitive effects See supra pp. 50-52 & n.16. Because the Government has failed to establish that the merged entity will have any incentive to withhold HBO promotional rights the Court need not answer the question whether the existence of such an incentive without more would be sufficient to show that the proposed merger would substantially lessen competition for purposes of Section 7.

increased-leverage claim of harm, the Antitrust Division decided to spill most of its ink developing undisputed facts - HBO is popular, see Gov’t PFOF 11 28, 235-242, valuable see id. 11 28, 235, 243-252, and an effective promotional tool for MVPDs see id. 11 253- 258. The Government also relays the undisputed fact that HBO, as a matter of contract, retains significant control over the way in which its “trademarks or . . . talent” are used in those promotions Tr. 1458:10-13 (Sutton (HBO)); see Gov’t PFOF 11 269-270 (discussing approval process for use of HBO in promotions). lt did not, however, come to Court with economic evidence of any kind, see Tr. 229114-7 (Shapiro), and proffered only bare conjecture about how there may be “like a thumb on the scale” in favor of the Government’s promotion-withholding stratagein id. at 2267:8-21; see also id. at 2267:3- 7. As such, the Government’s evidence is too thin a reed for this Court to find that AT&T has in that well-worn turn-of-phrase either the “incentive” or the “ability” to withhold HBO promotional rights in order to “lessen competition substantially.” Gov’t PCOL161. For these reasons it is small wonder that Professor Shapiro himself refused to endorse the theory testifying that, in his view as an economist, such a ploy “[o]n its own . . . would not have such a big impact, that it would substantially lessen competition” Tr. 2275:24- 2276:13 (Shapiro).

For these two, independent reasons the Government has failed to provide sufficient evidence to support its final theory in this case Accordingly, l reject outright the assertion that the combined entity would likely restrict HBO as a promotional tool in order to harm

AT&T’s distribution rivals and thereby lessen competition in the marketplace

CONCLUSION

The parties have waged an epic battle under extremely restricted deadlines to litigate and try this historic vertical merger case Each side’s evidence and theories have been subjected to cross-examination and the rigors of the Rules of Evidence and Civil Procedure. lt has been a herculean task for all the parties and the Court.62 Each side has had its proverbial day in Court The Court has now spoken and the defendants have won But, the process is not quite over yet 1

There is a grave and understandable fear on the part of the defendants that the Government will now seek to do indirectly what it couldn’t accomplish directly by seeking a stay ofthis Court’s order pending an appeal to our Circuit Court.

The consequences of receiving such a stay would cause irreparable harm to the defendants in general and AT&T in specific. First, it would effectively prevent the consummation ofthe merger by the lune 21, 2018 break-up date for the deal. Second, it would cause AT&T to have to pay the $500 million break-up fee it will owe to lime Warner ifthe deal is not consummated by that date Those two consequences of course would occur regardless of whether this Court’s decision were later upheld following appellate review. ln this Court’s judgment, a stay pending appeal would be a manifestly unjust outcome in this case

The Government has had this merger on hold now since October of 2016 when it

launched its investigation ln that 18-plus month period, the companies have twice

62 See, e.g., WDH & RSC at W.R. 6326. 170

extended the break-up date to accommodate the Government’s litigation of this case During that same period, the video programming and distribution industry has continued to evolve at a breakneck pace The cost to the defendants and the Government to investigate litigate and try this case has undoubtedly been staggering - easily in the tens of millions of dollars

lf the Government were to ask me to stay this Court’s ruling l would, under the law, have to weigh whether the Government has a strong likelihood of success on the merits and would suffer irreparable harm should the stay be denied among other things Well, suffice it to say - as my 170-plus page opinion makes clear - l do not believe that the Government has a likelihood of success on the merits of an appeal. And in my judgment given that our Circuit Court has never hesitated to unwind an unblocked merger if the law and facts warrant doing so, there would be no irreparable harm to the Government ~ only to the defendants ~ if my ruling were stayed As such, l could not, and would not, grant such a stay in the first instance

That of course is not to suggest in any way that the Government should not consider seeking appellate review of the merits ofthis Court’s decision That is by any standard fair game But the temptation by some to view this decision as being something more than a resolution ofthis specific case should be resisted by one and all l

The Government here has taken its best shot to block the merger based on the law and facts and within the time allowed The defendants did their best to oppose it. The Court has spoken To use a stay to accomplish indirectly what could not be done directly

- especially when it would cause certain irreparable harm to the defendants ~ simply would

be unjust l hope and trust that the Government will have the good judgment, wisdom, and courage to avoid such a manifest injustice To do otherwise l fear, would undermine the faith in our system ofjustice of not only the defendants but their millions of shareholders

and the business community at large

Thus, for all of the foregoing reasons the Government’s request to enjoin the

proposed merger is DENIED.

W»Qlwl

RICHARD J.\h-EbN United States District ludge