Antitrust expert Chris Sagers of Cleveland State University enumerates the failings of Judge Richard Leon’s dismissal last week of the Department of Justice’s attempt to block the AT&T-Time Warner merger. “I sure hope they appeal,” Sagers notes.
In a regrettable, frustrating, Homeric saga of an opinion issued on Tuesday, Judge Richard Leon of the US District Court for the District of Columbia rejected the Justice Department’s challenge to the merger of AT&T and Time Warner, Inc.
That it seems like one of the worst antitrust opinions I’ve ever read is really wholly aside from it reaching a bad result that will cause harm. That’s been a routine of reading antitrust cases for 40 years. It wasn’t even that so many key points within the rambling stream of factual reasoning didn’t make sense and several times seemed obviously wrong. It was that the opinion is obsessively, relentlessly, lopsidedly predisposed against the government. At every step, Judge Leon holds the government—and by extension the public interest, if you think antitrust does any good at all—to an extraordinary burden of proof. Meanwhile, he accepts the defendants, repeatedly finding facts in their favor for no reason except that their own executives came to court and told him so. “The sole consistency,” paraphrasing Justice Stewart ever so slightly, “is that the Government always [loses].” He did all that on no good or well-demonstrated basis at all, and emphatically not for any reason ever expressed by Congress or decisions of the Supreme Court. As we shall see, though, that an opinion this bad is more than good enough as far as our merger law is concerned is proof of just how broken the law is. And that in itself is hardly Judge Leon’s fault. The blame is with the conservative lower courts that took it upon themselves essentially to kill merger law during the 1980s and 1990s.
The opinion was surely thorough in the sense that it was long—172 pages—and it addressed a lot of individual issues. It also seemed legally pretty unremarkable, breaking no ground except for some discussion of the vertical merger cause of action (beginning on p. 55), and that all seemed probably fair enough on the basis of what little law exists. But I don’t have much else good to say about it. It may well be appeal-proof, but not because I think it’s good.1)While I think appeal will be an uphill battle—it always is, after all—I sure hope the Justice Department does appeal, and there are in fact issues of law that seem appealable to me. First, the handling of vertical efficiencies breaks important and potentially very controversial ground. Strictly speaking, Judge Leon claimed that efficiencies played no role in his decision at all, because he believed the government was required to make a prima facie case that it did not make (p. 54. n17). But his reasoning on p. 57 and again on p. 61 makes clear that he thinks the prima facie test should be a high bar in vertical cases precisely because he thinks vertical efficiencies are so likely and likely to be so big. I think that seriously undercuts his claim that he’s just applying Baker Hughes and not making efficiencies themselves legally relevant. They were hugely relevant, much more so than they would have been had he actually applied Baker Hughes and considered them only as a defense. Second, it was strange basically to say that market definition doesn’t matter, because vertical mergers don’t raise concentration. See p. 62. Theories of vertical harm do in fact usually depend on the parties’ shares within relevant markets. Finally, one potentially very significant legal idea appears very late in a footnote (p. 168 n. 61), and though it wasn’t explicitly part of a holding, it is very provocative and it may seriously have colored Judge Leon’s handling of the leverage theory. He says that even proof by a preponderance that a merger creates an “incentive” to engage in anticompetitive conduct is not itself legally sufficient to prove merger injury. If that is correct, then I guess that not only the leverage argument, but game theory and any other prospective modeling of incentives was just ruled legally irrelevant.
At every step, Judge Leon holds the government—and by extension the public interest, if you think antitrust does any good at all—to an extraordinary burden of proof. Meanwhile, he accepts the defendants’ evidence with a truly surprising, hook-line-and-sinker credulousness.
I don’t personally feel that the government’s case was weak, as many armchair pundits claim. I guess if blame has to be laid it could be on the concessions made by the government’s key economic witness Carl Shapiro, or on the government’s failure to put on more expert evidence disputing some issues. There were also apparently a lot of courtroom foibles, like government witnesses contradicting each other and seemingly powerful documentary evidence being shown on rebuttal to have been misleading. But I don’t think it would have made a difference if any of those things went differently, and I don’t believe it’s all that fair to fault them. Instead, I think the case is an object lesson in the extremely unsatisfactory state of our substantive merger law, which I don’t imagine will change in our lifetimes. Virtually the entirety of the law boils down to the nearly unreviewable fact-finding of one person in each case (the trial judge), subject to the extravagant burden of proof created from more or less whole cloth by then-Judge Clarence Thomas in the 1990 DC Circuit decision United States v. Baker Hughes. (Baker Hughes was not coincidentally the very first precedent Judge Leon cites, only a few pages into the opinion).
The Problems in Judge Leon’s Reasoning
The Big Hum-Dinger: The Leverage Theory. First, and above all, Judge Leon’s rejection of the government’s key theory of harm—the so-called leverage or bargaining theory—was simply flaccid.
The leverage theory in its details is complex, but the idea is simple. The claim is that a firm that vertically integrates might be able to raise its own prices profitably higher than when it was stand-alone, because business it loses in its original line could be compensated for by gains in its new one.
Judge Leon put a lot of emphasis on industry witnesses who just said the claim is incorrect in their experience (p. 113), though Shapiro pointed out a pretty obvious reason that they were either wrong or lying (p. 114), and though throughout the rest of the opinion he disparagingly rejected virtually every government effort to base anything on the testimony of industry participants. (Incidentally, I would like someone to explain to me how the argument on the top of p. 115, on the profit incentives of integrated firms, makes any sense at all.)
But much more importantly, he found it hugely significant that literal “blackouts” in TV distribution negotiations have been rare and short-term, and that they will still be unlikely post-merger (pp. 18, 72-73, 82, 115 & nn.34-35; cf. 96-97). That is, he found it hugely significant that content producers don’t actually literally, completely exclude particular distributors from carrying their content.
The implication was that the leverage theory cannot work unless there are in fact actual, long-term blackouts with some frequency, and unless the acquired content is so important to competitors that it is “literally must have,” apparently meaning that competitors would literally be required to exit without it (p. 75). (Note, incidentally, the credulousness as to the statements of defendants’ own executives on this point in n. 36.)
But that is obviously, foolishly false. Borrowing from a very wise friend, it’s like saying that the unlikelihood of actually using nuclear weapons means that they are irrelevant. What happens instead is that even if everybody knows there won’t be a blackout, both parties can gauge that the other wants something, and has a reservation price. Carl Shapiro did not model the likelihood of an actual blackout, he modeled the psychological experience of bilateral negotiators with market power who need one another, and must try to assess one another’s strength. Under his model the merger will cause judgements about reservation price to change in a way harmful to consumers. If he’s right, that makes it illegal.
Judge Leon literally did not get it.
Kind of Another Humdinger: Dynamic Disruption. Second, Judge Leon’s rejection of one other government theory of injury—that the merger would allow integrated oligopolists to stave off dynamic technological innovation—is breezy and credulous. It consists mostly of a few pages of attacks on offhand, in-court statements by Shapiro that I think Judge Leon misunderstood (beginning on p. 152). Here he kind of repeats the same weird obsession with the unlikelihood of actual blackouts. He says that because Turner probably will continue to license content to technologically disruptive entrants, that it therefore cannot harm them or consumers. It’s almost literally the Cellophane fallacy, and by the way it is again supported with elaborate, uncritical acceptance of the self-serving testimony of the parties’ own top executives (pp. 154-56).
Lopsided Anti-Government Bias: There Will Be No Harm Because Defendants Gave Their Scout’s Honor. Judge Leon has a long reputation as a maverick and a conservative, libertarian dissident, and a former white-collar defense lawyer given to strong stands of principle against government. Perhaps as a result he also has a very high reversal rate. It was on extravagant display.
For example, Judge Leon made certain early findings that defendants likely had pro-competitive motives. Later, he finds that those motives rebut the government’s evidence of anticompetitive intent. That’s fine, but the way he found it was not.
Specifically, he found that currently unintegrated firms like the parties needed to reduce negotiation costs by integration—à la Oliver Williamson—so that they could compete with organically integrated firms like Amazon and Netflix (pp. 19-20). Likewise, he found that the integrated firm would have “superior access to customer data,” which producers need in this new age of more efficient advertising but can’t get because distributors won’t give it to them (p. 20). And then he makes a range of findings to the effect that this dynamic industry is now very competitive and likely to stay that way, and that the defendants themselves are losing market share because of it. They have to merge or they’ll die.
However, he based these important findings heavily—in many cases almost exclusively—on the opinion testimony of the defendant’s own executives.
This is extraordinarily galling. Perhaps 30 pages or more of other parts of the opinion consist of his disparagement of the government’s third-party testimony (e.g., pp. 81-82, 91-108), and his very stingy, uncharitable rejection of any government reliance on the defendants’ internal documents and statements in prior proceedings (e.t., pp. 82-91).
Dozens more examples of this are peppered throughout the opinion. One of the truly most remarkable is Judge Leon’s rejection of a government theory that the merger would make it easier for the integrated AT&T to collude with Comcast/NBCU. He based that rejection too largely on the top executives’ own testimony (beginning on p. 160). This judge held that two integrated, horizontally competing oligopolists would not conspire because the CEO of one of them gave scout’s honor that he wouldn’t (p.161). (You might say, wait, didn’t he make a good point that oligopoly arguments aren’t so obvious here, because this is a vertical merger, and there is no increase in horizontal concentration? But I would say no! Like Judge Leon, I would use exclamation points! A central theory of harm from vertical merger has always been the integration would facilitate horizontal collusion, in various different ways!)
In other words, Judge Leon effectively threw out a big proportion of the government’s evidence in this case largely on small bits of the defendants’ own self-serving testimony. The things business executives say apparently never matter, nosiree, except when they cause the government to lose.
Couldn’t We Please Just Have Some Thought for Policy?
Finally, and most importantly, there is a fundamental, implicit policy problem running throughout all of the reasoning in this opinion. It is hardly unique to Judge Leon; it infuses all of modern merger law, and it begins with his appellate superiors going back to General Dynamics. He emphasized repeatedly that traditional cable companies are losing market share to the upstarts. He implied at some length that therefore, in various ways, video distribution is actually very competitive and is likely to stay that way. His discussion on those points included a pregnant quotation from General Dynamics on p.25 n.6.
If Netflix and Amazon Prime could build up premier, internal content-production operations, organically, why can’t AT&T do the same thing?
Now, put aside the centuries-long history of prior instances of industry transition in which there was lots of entry, and then a seemingly very competitive scrimmage for a while, which was then followed by a bunch of business failures and acquisitions and a market with four or six big firms left or whatever. And in all those cases, when someone tried to challenge one of the acquisitions by which one of the incumbents was trying to stave off the competitiveness of the changing market, the courts were quite happy to fall for the defendants’ pleas that competition was just really vigorous.
Wholly aside from that, there is a huge policy problem in letting this be relevant. A key implicit insight in all this reasoning is that these defendants need their merger to compete with Amazon and Netflix and other content producers that are already integrated internally. That begs the huge question that Judge Leon never once asks: if Netflix and Amazon Prime could do that—build up premier, internal content-production operations, organically—why can’t AT&T do the same thing? And because it’s obviously not impossible that they could, why shouldn’t the law require them to do it?
In other words, it is already nominally the law that growth by acquisition is disfavored, because Congress adopted the Clayton Act incipiency standard. Neither Congress nor the Supreme Court has said anything to qualify it—ever—unless you read General Dynamics as broadly as the lower courts did in Baker Hughes and Waste Management. Moreover, if you believe even half as much in the power of economic theory as antirust conservatives claim to do, then there seems rarely reason to believe that desirable integrations just have to be accomplished by acquisition. If some development is worthwhile, there should in all but marginal cases be enough incentive to do it through internal growth as an alternative.
And internal growth is actually competition. It’s the offering of a new thing to consumers, rather than just getting control over a thing they’ve already got.
Last Thoughts on Ambitious AAGs
And so where does that leave us? Vertical mergers probably just got even more legal than they were. For the last few decades, the agencies could at least get “conduct” or “behavioral” concessions now and again. After this fiasco, I don’t see why parties would agree to them ever again, so long as they’re more than window dressing. If an agency proposed them, I would say, “Pffft. What are you going to do, sue us? You just lost the biggest, most embarrassing case in a long while on just the same theory. Go ahead.”
For the last few decades, the agencies could at least get “conduct” or “behavioral” concessions now and again. After this fiasco, I don’t see why parties would agree to them ever again.
Does that mean it was a mistake for DoJ to bring this case? I definitely think it was not. If government enforcers are not willing to take risks in litigation, we get a one-way ratchet toward more and more conservative law. So, they rolled the dice here. Had they won (and they still might), it would be a huge shot in the arm for enforcement. But you can’t pick your judge, and their luck was about as bad here as it could be. I’m still glad they took the risk, and I think the agencies should take more risks.
Disclaimer: The ProMarket blog is dedicated to discussing how competition tends to be subverted by special interests. The posts represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty. For more information, please visit ProMarket Blog Policy.
References [ + ]
|1.||↑||While I think appeal will be an uphill battle—it always is, after all—I sure hope the Justice Department does appeal, and there are in fact issues of law that seem appealable to me. First, the handling of vertical efficiencies breaks important and potentially very controversial ground. Strictly speaking, Judge Leon claimed that efficiencies played no role in his decision at all, because he believed the government was required to make a prima facie case that it did not make (p. 54. n17). But his reasoning on p. 57 and again on p. 61 makes clear that he thinks the prima facie test should be a high bar in vertical cases precisely because he thinks vertical efficiencies are so likely and likely to be so big. I think that seriously undercuts his claim that he’s just applying Baker Hughes and not making efficiencies themselves legally relevant. They were hugely relevant, much more so than they would have been had he actually applied Baker Hughes and considered them only as a defense. Second, it was strange basically to say that market definition doesn’t matter, because vertical mergers don’t raise concentration. See p. 62. Theories of vertical harm do in fact usually depend on the parties’ shares within relevant markets. Finally, one potentially very significant legal idea appears very late in a footnote (p. 168 n. 61), and though it wasn’t explicitly part of a holding, it is very provocative and it may seriously have colored Judge Leon’s handling of the leverage theory. He says that even proof by a preponderance that a merger creates an “incentive” to engage in anticompetitive conduct is not itself legally sufficient to prove merger injury. If that is correct, then I guess that not only the leverage argument, but game theory and any other prospective modeling of incentives was just ruled legally irrelevant.|