The consumer-welfare standard that undergirds modern antitrust enforcement is under attack. Although the standard hasn’t changed since the landmark U.S. v. Microsoft litigation, there has not been a single case brought by a U.S. antitrust agency under Section 2 of the Sherman Act against a monopolist threatening innovation in the past two decades. What explains this retreat by antitrust enforcers? Some antitrust scholars suggest that courts increasingly demand tangible evidence of consumer injury to satisfy the consumer-welfare standard, and because such proof is difficult in many cases, the agencies are reluctant to pursue cases they are bound to lose.
In a new paper released by the Roosevelt Institute, economist Marshall Steinbaum and Professor Maurice Stucke of the University of Tennessee College of Law propose the “effective competition standard” as an alternative to the consumer-welfare (“CW”) standard that would revive the original aims of antitrust law—namely, to preserve competitive market structures. They explain that a “price-centric” approach that flows from the CW standard misses important metrics such as harms to quality, privacy, innovation, and input providers, including workers. The effective competition standard would call for “the preservation of competitive market structures that protect individuals, purchasers, consumers, and producers.” Courts would rely far less on the Supreme Court’s rule-of-reason framework under the new standard, and far more on simpler legal presumptions.
Are these changes necessary? And would they constitute an improvement relative to the CW standard? To answer those questions, we invited the Roosevelt’s Marshall Steinbaum, the Capitol Forum’s Sally Hubbard and the Hoover Institution’s Nicolas Petit to discuss the latest attempt to reinvigorate antitrust enforcement. The Chat was moderated by Hal Singer, editor of Washington Bytes and Senior Fellow of the George Washington Institute of Public Policy. The transcript was edited lightly for readability.
* * *
Hal Singer: To peddle any alternative standard, including the effective competition standard, one must believe that the CW standard yields too little antitrust enforcement, or what economists describe as too many “false negatives”—for example, approving too many anticompetitive mergers or condoning too much anticompetitive conduct. What specific cases should we look back on as proof that these false negatives actually occur under the CW standard? Any mergers that should have been blocked or conduct that should have been barred, which now with the benefit of hindsight, were clearly anticompetitive?
Marshall Steinbaum: First of all, I follow Jon Baker in rejecting the whole framework of error-cost analysis in antitrust. I think that framework assumes the economy works in a particular way “naturally,” and the role of antitrust is to fiddle around the edges, clearing up problems here or there, and the aim is to specify the right decision rule for inherently edge cases, which is to say, all cases. Instead, the role of antitrust is to structure how the economy works, in some sense, whether it “wants to” or not. The “Great Merger Wave” of 1895-1903 was caused by the Supreme Court’s decision in US v. E.C. Knight Co., ruling that a 98% monopoly in sugar refining wasn’t illegal under the Sherman Act because the Sherman Act couldn’t pertain to manufacturing industries since they aren’t inherently “interstate commerce.” Similarly, the 1920 Supreme Court decision against the government in its attempt to break up U.S. Steel caused another merger wave. What we have right now is the proliferation of business models that have been allowed to grow up under lax antitrust enforcement, and specifically under the policy regime that blesses anything that can be made to look beneficial to consumers. I thought Lina Khan made this point really well in “Amazon’s Antitrust Paradox,” and I would say it holds elsewhere in tech and throughout the economy,
Singer: So you are refusing to answer? Should we call the judge? Anyone else willing to give me some examples of false negatives?
Sally Hubbard: Facebook-Instagram, Facebook-Whatsapp, Google-DoubleClick, Google-AdMob. In 2016, Facebook and Google accounted for 85 to 90 percent of new digital advertising revenues in the US. These mergers are part of the problem.
Singer: Tim Wu agrees with you! He recently acknowledged that approving Facebook-Instagram was a mistake. But it’s not clear that this Obama-era error was caused by the CW standard or by regulators generally misunderstanding of tech developments.
Hubbard: The CW standard is part of the problem but not the entire problem. Several of the mergers I mentioned were deemed “vertical,” and the CW standard is part of the reason why we have underenforcement in vertical mergers.
Nicolas Petit: It’s funny how consumer welfare became the totem of all antitrust evils. Twenty years ago, we had the same discussion, but the semantics were pointed at “efficiency” as the culprit. I guess the CW standard can yield both under- and overenforcement. It may be under-enforcing when strict recoupment is required in predatory pricing cases as under Brooke Group. But it may also be overenforcing when it leads to categorically disregarding transaction-specific efficiencies on the ground that price effects categorically trump cost reductions in merger review.
Steinbaum: Amazon-Whole Foods is a prime example of underenforcement due to an excessive focus on consumer prices. We know that post-merger, Amazon has imposed a bunch of restrictions on Whole Foods’ suppliers, including their access to Whole Foods stores. This is a direct reduction in competition due to the merger, and yet, no consumer price effect. Ergo, no case.
Singer: But Marshall, the laser-like focus on proving price effects keeps an army of econs busy!
Hubbard: I studied Amazon for two years before that merger and was aghast when I heard from a decisionmaker at the FTC that they didn’t view the merger as involving data. Former Amazon employees have told me Amazon is “a data company that happens to sell stuff.” To say that any merger involving Amazon doesn’t involve data is to fundamentally misunderstand Amazon’s business. Part of the problem is the CW standard and part of the problem is enforcers not understanding that data is the currency consumers pay to tech platforms.
Steinbaum: Just astounding that the FTC didn’t consider that. Literally a jaw-dropping revelation.
Hubbard: Have y’all forked over your data to Amazon at Whole Foods yet by logging in as a prime member? Have you considered that your data may be more valuable than the discount you receive?
Petit: Would anyone here volunteer to bring hard facts that provide evidence for the claim of a harmful effect of those post-merger restrictions? Once this is done, it’d be interesting to discuss as a prelim whether this cannot be adequately addressed with ex post enforcement regimes.
Steinbaum: Antitrust is specifically about structuring how the economy works, and under the CW standard, we let companies get away with business models that are highly profitable and extractive, so long as they don’t harm consumers through higher prices. There’s a whole heck of a lot you can do within that very large box in order to prey upon other economic stakeholders, as you well know, Hal, from your program-carriage cases.
Petit: I disagree. Antitrust does not prescribe an optimal structure for the economy, it proscribes specific forms of bad conduct that create or increase market power.
Steinbaum: Antitrust doesn’t prescribe an optimal structure, but it does (or should) rule certain anticompetitive business models out of bounds and impede companies on their path to erecting them. Beyond the tech space, where acquisitions have clearly been about monopoly maintenance, you have the “killer acquisitions” in pharma designed to shut down competing research lines. And doing so right under the Hart-Scott-Rodino threshold—in which case, you’re not even in the territory of “false negative”!
Singer: Defenders of the CW standard argue that it “cabins antitrust enforcement to economic matters rather than a hodgepodge of political and social objectives.” They are quick to ask for empirical evidence of the consumer harms associated with these (alleged) false negatives: Aren’t prices lower when a branded product is displaced by Amazon’s private-label products? Doesn’t Google’s answers to math questions or driving instructions constitute an improvement for its users? How can users on Facebook’s platform be harmed when the service is free? Marshall, where would you point skeptics of a new approach for evidence of the social costs of underenforcement?
Steinbaum: Google’s or Facebook’s harvesting of third-party content and presenting it directly to users/consumers as its own has had a disastrous effect on independent journalism and other forms of content creation.
Hubbard: Only looking at one side of a consumer—the prices they pay—is like your doctor saying she is going to give you meds that will save your kidneys but destroy all of your other organs. It doesn’t make sense. You can’t slice and dice human beings this way.
Singer: Are you saying our preferences can’t be decomposed into the summation of price and non-price attributes of a product? Next thing you’ll tell me is that we have feelings too!
Hubbard: What good are lower prices if your wages are also lower because there are fewer employers competing for your labor? Or because you work for a company that sells to Amazon and is getting squeezed on its margins? What good is Google incorporating everyone else’s content and innovations into its monopoly search engine if it means you can’t make a living as a content creator or an innovator? And if entrepreneurs can’t bring new products and services to market because they can’t raise funding for their startup without showing they won’t get squashed by a tech giant, not only do the entrepreneurs lose but so do all consumers who miss out on these innovations.
Petit: Weird. There is more content around us than ever. Digital technologies have brought us an abundant supply of micro content providers. Indie publishing has gone through the roof in the 2010s. Just check ISBN numbers. Again, unless you back those intuitions with hard facts – here’s a report – we risk ending up in a “this is so because I say so” discussion. Isn’t it actually puzzling that the large corporate powers of the publishing industry are today the main complainants against big tech? Digital has decentralized content production, and empowered creative individuals.
Steinbaum: The idea that content creators should be thanking the platforms for acting as middlemen is … a bit out of touch. I was just reading the Paramount consent decree, a very clear instance in which the DOJ concluded that vertical integration threatened both artists and consumers, completely within the scope of the Sherman Act. It’s astounding to think that the DOJ is considering getting rid of it, when the evidence the DOJ cites for its irrelevance is actually evidence of its success. (This reminded me of John Roberts’ justification for gutting the Voting Rights Act.)
Petit: Is this a feeling or a fact? The overall social decrease of creative content?
Steinbaum: I would point you to the employment trends for journalism versus PR, as one factoid.
Hubbard: You can create all the content you want and shout it into the black hole, but making a living at or making it a viable business is harder when Facebook and Google get all the ad dollars.
Petit: Yes, it’s sad but true – no Metallica pun intended – that not everyone can be a professional musician or writer. Though we each have our individual perceptions, we should all agree that the antitrust process must be based on hard facts: prices, output, investment, innovation, etc. Because these are all amenable to some form of estimation.
Singer: I know I’m the neutral moderator, but I have a paper that says we can’t measure innovation. Sorry. Carry on.
Petit: This is an abuse of dominance!
Hubbard: “Not everyone” is an understatement. Really only bestsellers can be professional writers, which didn’t use to be the case. But for that we can thank Amazon mostly.
Singer: Any other examples of underenforcement attributable to the CW standard?
Hubbard: Here’s one: Under the CW standard as currently construed, Facebook is largely immune because it’s “free.” But Facebook isn’t free! We are paying with our private data and the price we are paying is too high. We are harmed when Russian disinformation agents get our personal data. After the Cambridge Analytica scandal broke, so many smart people told me they were quitting Facebook but still using Instagram without fully comprehending that Facebook owns Instagram—the next best substitute for most people. When consumers don’t have choices and can’t vote with their feet when a tech platform screws them over, well, tech platforms can get away with screwing consumers over (not because they delight at screwing consumers over, but because they can pursue profits without regards to consequences). How is that good for consumer welfare?
Petit: I don’t agree with the idea of a zero-price market being categorically immune from antitrust. This is a straw man. No decent agency or court would ever say that Facebook is immune from antitrust because the service is nominally free.
Hubbard: I hope you’re right Nicolas, but I haven’t seen any enforcement on tech platforms!
Steinbaum: But the Supreme Court implicitly argued that Facebook was immune, in Ohio v. Amex. For example, the decision states “Price increases on one side of the platform… do not suggest anticompetitive effects without some evidence that they have increased the overall cost of the platform’s services. Thus, courts must include both sides of the platform—merchants and cardholders—when defining the credit-card market.” Of course, it remains to be seen how widely courts will interpret the “two-sided platform” exception to antitrust law that the Court created in Amex, but let’s just say that the tech industry trade association was very, very pleased by that ruling.
Petit: Yes, Ohio v. Amex is not the best Supreme Court opinion. But the language you quote does not imply, neither by logic nor intuition, antitrust immunity for zero-price markets. And even less talk about Facebook, which has a very different type of multi-sided model.
Singer: Against those costs of false negatives or cases of underenforcement, one must weigh the costs of “false positives” from overenforcement. Blocking an efficient merger or condemning procompetitive conduct generates costs as well. Given the general lack of antitrust enforcement in the past two decades—at least from U.S. antitrust agencies—it’s hard to seize on natural experiments. But perhaps more aggressive enforcement in Europe provides a test ground for what can potentially go wrong on the other side of the equation. Nicolas, do you have any examples of errors from overenforcement in Europe, and what those costs might look like?
Petit: While I agree that Europe is more aggressive, I would not characterize this as overenforcement. Instead, I would say that Europe’s main feature is to have a very discretionary enforcement system. Some time ago, we ran a survey on a blog asking, “What is a restriction of competition?” Someone replied, “Whatever DG COMP thinks it is.” The European Commission has almost full discretion over antitrust in unilateral conduct cases. By this I mean that it has ability to choose among a wide range of theories of harm, both exploitative and exclusionary. It can intervene in less concentrated markets, due to the lower threshold requirements for dominance. Its decisions are binding, meaning and it does not need to try a case before a judge to bring market outcomes. Even when there is judicial review, it faces more deference than agencies in the US appeals system. And it has traditionally been more reluctant to use economics evidence in unilateral conduct cases. This cocktail of institutional, educational, and textual idiosyncrasies gives much more leeway to the EU administrative agency than its US counterparts. This also explains that Europe has a more experimental approach to enforcement… now just recall for a minute that experiments do fail, and sometimes they do not fail well.
Steinbaum: I don’t think there’s any better example of discretionary enforcement than in the United States, where we tailored our antitrust policy to the special pleading of the tech industry. The idea that contemporary U.S. antitrust enforcement isn’t discretionary is… wishful thinking. And I would just point out that ever since we in the United States supposedly started caring only about efficiency in antitrust enforcement, the profit share of national income increased from 5% to 15%. So if all we care about is efficiency, we’re not doing a very good job.
Petit: Now, to Hal’s question regarding the costs of false positives: The EU’s Google-Android decision increases the returns to closed ecosystems, and as a matter of fact penalizes ecosystems that are relatively more open. This was not carefully thought out.
Hubbard: Agree that Apple got away with its closed system, but that does make Google’s conduct legal. The Google Android case to me is U.S. v. Microsoft all over again.
Petit: I sense a possible agreement here: discretionary enforcement creates both frustration for defendants and complainants alike.
Singer: That’s enough agreement for one Bytes Chat. Disagreement and conflict are what drives clicks!
Hubbard: Google has a huge worldwide market share on the Android OS and its agreements with manufacturers excluded competition from other apps. It’s really a straightforward case actually.
Singer: Other examples of overenforcement attributable to giving enforcers too much discretion?
Petit: Many complaints are currently pending before the EU Commission, and plaintiffs do not know what to do to make them fly. More concrete examples of possible overenforcement can be seen over the cycle of cases. IBM: guilty of infringement, then complainant against Microsoft. Microsoft: guilty of infringement, then complainant against Google. Intel: guilty of infringement, then alleged victim in the Qualcomm case. Amazon: de facto plaintiff against Apple in the hub and spoke cases, now on the receiving end of antitrust proceedings. I could carry on with more examples. You see that overly discretionary antitrust creates a rent-seeking structure.
Steinbaum: I think this idea that overenforcement of antitrust laws creates opportunities for inefficient competitors to avoid their “natural” death is totally unsupported in the empirical record.
Singer: Any other costs associated with overenforcement besides encouraging rent seeking by rivals? For example, did the overenforcement lead antitrust defendants to invest less? Maybe a diversion of a defendant’s resources?
Petit: When you think of overenforcement, you must think of antitrust cases as kick starters for regulation.
Hubbard: What do you mean? I think of antitrust enforcement as a way to avoid utility-style regulation.
Petit: My point is that antitrust law untethered from CW provides an entry point for regulation. You can see that with platform regulation today in the EU. Antitrust proceedings ramp up pressure on policymakers to do something at scale, to use a tech word.
Singer: Here’s where I’m driving, and then we can move onto the next question: Those who want to preserve the status quo need to explain the parade of horribles that would happen if we altered the CW standard in a way to allow for slightly more errors in overenforcement (and fewer errors in underenforcement).
Petit: I think the burden of proof is on the other side—those who want to change it.
Singer: It’s not clear to me that either side of this debate uniquely bears the burden. If the current antitrust regime results in too many errors in one direction (underenforcement), then both sides need to marshal evidence in support of their positions to change (or maintain) the standard. This is a policy decision, not a legal one.
Steinbaum: Demsetz, Bork and Easterbrook didn’t bother checking to see if their theories were actually true before writing them into the law. Wild swings in policy based on no change in our understanding of how the economy works. There’s some great stuff in Gorsuch’s Wordperfect decision echoing this idea—whispered in his ear (I guess) by ideologically-biased economists who pretended the economy works the way they say it works.
Singer: Let’s talk about harms to innovation caused by underenforcement. What I mean here is less entrepreneurial activity taking place in the “edges” of the platforms, either because independents feel the landscape is so slanted against them, that innovation is not worth the effort, or because investors are not willing to bet on ideas that can be easily appropriated by a platform. Sally, should we be asking antitrust plaintiffs to demonstrate—in satisfaction of the consumer-welfare standard—an empirical connection between a platform provider’s discriminatory conduct and a reduction in edge innovation? Or is that a fool’s errand?
Hubbard: No, we should not be asking antitrust plaintiffs to demonstrate an empirical connection between a platform provider’s discriminatory conduct and a reduction in edge innovation. It’s really hard to prove what innovation would have existed but for the platform’s anticompetitive conduct. And in any event, that’s not what the law actually requires. In the monopolization context, plaintiffs need to show that defendant possesses monopoly power and has acquired, enhanced, or maintained that power by using exclusionary conduct. What we are seeing a lot with tech platforms is monopoly leveraging: the elements after Trinko include (1) the possession of monopoly power in one market, (2) the use of that power in an anticompetitive manner to (3) create a dangerous probability of monopolizing a second market. If we adequately enforced monopolization and monopoly leveraging, consumers would benefit from greater innovation.
Steinbaum: I agree with Sally. This is why I don’t accept the criticism that departing from the CW standard would harm consumers. We know that an antitrust policy that takes monopoly leveraging seriously and is thus skeptical of vertical integration by firms with market power is a policy that serves both consumers and would-be entrants and entrepreneurs far better than the status quo.
Petit: I agree with “3) dangerous probability of monopolizing” if by this Sally means evidence of likely anticompetitive effects.
Hubbard: The case law for 3) is about what are the odds of success. Courts require a high market share in the secondary market, which I think is a mistake. By the time the monopolist has that much market share, it’s too late.
Petit: Doctrine must also set threshold levels of unlawful competitive harm. To me, anticompetitive leveraging implies likely exclusion, not mere competitive disadvantage.
Singer: What I hear Sally arguing is that it’s not the CW standard so much as it is the court’s interpretation of the original statute that matters. Is that right, Sally? Put another way, are courts really constrained by the CW standard?
Hubbard: It didn’t constrain the court in U.S. v. Microsoft if we’re talking about monopolization cases.
Singer: Indeed! The appeals court routinely cited the decline in rival browers’ shares as evidence of anticompetitive effects. Of course, that has almost nothing to do with consumer welfare.
Steinbaum: Well, the CW standard has acted as a means of restricting what “counts” as reducing competition. So I at least don’t think you can separate the two.
Singer: Where in the case law would I go looking for a requirement that a court must consider short-run injury to a consumer before deciding antitrust liability?
Steinbaum: Ohio v. Amex.
Hubbard: Ohio v. Amex is limited to “transaction platforms,” which could be just credit card companies. Let’s not give that terrible decision more power than it has!
Steinbaum: I think there are many others, but that one makes the point so starkly I think we should start there.
Hubbard: The CW standard is certainly a huge part of merger analysis. But when it comes to monopolization cases, the precedent of U.S. v. Microsoft makes clear a court can prohibit conduct that harms competition without focusing on consumer harm. The CW standard mostly comes into play in determining whether to bring a Section 2 case at all (the answer is usually no), and also determining whether conduct is exclusionary or just robust competition. But Section 2 makes it illegal to “monopolize, or attempt to monopolize, or combine or conspire with any other person or persons, to monopolize any part of the trade or commerce…” Consumer welfare, prices, and efficiencies are not mentioned at all in the statute, and the legislative history is all about preventing concentrated power!
Steinbaum: I think you can see the influence of CW in the way merger cases are litigated. That’s the only way you get to the narrow way AT&T-Time Warner was fought.
Petit: On startup or “edge” innovation, as Hal puts it, I understand that your related and underlying claim is that VCs may be loathe to fund innovative founders when their innovation risk being subject to monopolistic exclusion or imitation by a big tech platform. Again, we’d need data here. And as anecdote, Paul Graham said the other day that only “maybe one out of 1,900” of its portfolio companies has been killed by a rival that’s tackling the same problem. Furthermore, as Nobel prize economist Jean Tirole recently pointed out, the effect could plausibly cut the other way. The current regime might actually encourage edge providers to produce mediocre inventions in the hope of being bought out, and not try hard enough to push the technology frontier. Another VC legend stated that startup founders often did not push hard enough, and sold too early. If we set too lax a threshold of unlawful discrimination, we will assist developers with marginally interesting inventions, and artificially raise their exit price. Now, whether this is bad for society is open for debate. But at the very least, it’s not clear that antitrust law as currently applied discourages rather than encourages edge – I prefer disruptive – innovation. As you guys like to stress in the US, we Europeans do not have a Google or a Facebook.
Singer: I will simply point out that per Crunchbase, venture investing in tech in the United States, as measured by the number of deals, has declined by 23 percent per year since 2015. But let’s get back to Marshall’s paper. It touches on an important and overlooked aspect of case-by-case adjudication pursuant to the CW standard—namely, timeliness. You fault this process for being “too costly and time-consuming for anyone other than a well-financed plaintiff to undertake,” which “allows market power to fester and accumulate unchallenged.” But the FCC adjudicates program carriage complaints on a case-by-case basis pursuant to a nondiscrimination standard, and those cases take roughly half the time as a typical section 2 case. Doesn’t that suggest that the real problem when it comes to timeliness is not case-by-case review per se, but instead the proper evidentiary standard?
Steinbaum: I think there’s room for both. We’ve only ever really had case-by-case review in U.S. antitrust. It once gave rise to a reasonable policy, and now it doesn’t. So that right there implies that case-by-case review isn’t per se bad. On the other hand, when it worked, that was long before the **cough cough, Hal*** economic experts *cough* got so deeply involved.
Singer: Ah, blame the economists! Moving to the nuts and bolts of your proposal, the first ask is to permit plaintiffs to prove market power directly in cases challenging so-called “vertical restraints” such as exclusive dealing, bundling, or most-favored-nations clauses. The Supreme Court in American Express appears to have eliminated this pathway for antitrust plaintiffs, and instead required plaintiffs challenging vertical restraints to show market power indirectly by defining a relevant product market. Does anyone on this panel object to such a change, which in my mind, merely puts things back to where they were pre-Amex and creates a symmetry with horizontal cases, where direct proof of market power is sufficient?
Hubbard: Again, Amex is limited to transaction platforms—that is, credit card companies. It does not apply to all antitrust plaintiffs.
Singer: Fair point. But play along. If Amex removed that pathway for all antitrust plaintiffs, should the sufficiency of direct proof of market power be recreated via legislation, as Marshall suggests? I think this is a fairly modest ask.
Steinbaum: Hal, you’re giving a relatively moderate gloss to a proposal that I would say goes further, but then again, who am I to judge?
Singer: It’s the first ask in your executive summary! I thought it was fairly reasonable.
Steinbaum: Great! We’ve taken you in.
Petit: If you don’t really look at the two sides in multi-sided markets, you raise the risks of false positives.
Hubbard: Plaintiffs should be able to prove market power directly, including in vertical cases. Why prefer the second-best proxy over direct evidence?
Petit: There are some cases where market definition (and indirect proof of market power) may not be necessary. Actually, we do this in merger analysis often.
Hubbard: The Horizontal Merger Guidelines say direct evidence of competitive effects is enough. But courts still often insist on market definition.
Petit: But in Amex if I recall, output was growing by orders of magnitude, so direct evidence of market power was inconclusive.
Steinbaum: I do not agree at all with Nicolas on the economic inference from the fact of a rising number of credit card transactions.
Petit: Then please proceed to describe the Nirvana equilibrium that is adversely impacted by the conduct under consideration 😉 I mean, I’d like you to describe your idealized state of the world where output is maximized. And give measures. I can point you out to what Demsetz called that the Nirvana fallacy.
Singer: The relevant question is whether credit-card transactions were lower than what they would have been in a world without the restraint. That output was rising over the relevant period says nothing about the marginal effect of the restraint.
Steinbaum: In fact, this point is super-important: Any antitrust economist will tell you that you cannot infer market power from concentration. But neither can you rule it out (in a completely different market!!) through a cooked up “output effect” like that. So the court’s economic reasoning from that fact is just sloppy.
Singer: So undoing the damage from Amex is fairly modest. But then Marshall and Maurice start to get a bit more radical. For example, the paper then calls for making predatory pricing a presumptive violation of Section 2. Are there particular episodes of predatory pricing occurring right now that escape scrutiny under the consumer-welfare standard? And if so, why is this a concern given the short-run benefit of lower prices? Also, my economic training suggests that some forms of price discrimination could be good for consumers. What are the textbooks missing?
Hubbard: The most obvious episode of predatory pricing is Amazon. If you’ve read Lina Khan’s Amazon’s Antitrust Paradox you’ll know that. I have personally talked to former Amazon employees who tell me Amazon loses money on a huge number of products. As a consumer, I want to have the option of several different online marketplaces. As a producer, I want to have the option to sell to several different online marketplaces. As a taxpayer, I don’t want Amazon to be so huge that it has the power to play the nation’s local governments against one another to give gigantic tax breaks that I as a taxpayer will have to pay for. How did Amazon get this huge? Predatory pricing is an incredibly successful strategy for eliminating competition! We need to enforce predatory pricing laws, otherwise no competitor – no matter how efficient, innovative, or high quality – can compete against Amazon.
Steinbaum: To Sally’s Amazon example, I would add Uber. It’s clearly a recurrent Silicon Valley business model for VCs to underwrite loss-making sales with the aim of monopolizing the consumer side and then using control of a utility-like network to squeeze suppliers.
Hubbard: Yes, sometimes predatory pricing losses are paid for by cross-subsidization, and sometimes by investors. Investors floated Amazon despite not making a profit
Petit: I am happy to discuss the recoupment test, but I don’t see why we need to change the other building blocks of the case law.
Hubbard: The recoupment test is what makes predatory pricing law unenforceable.
Petit: First, Amazon made actually few losses, and sells above costs for a very long while. Second, Amazon’s prices are financed by its own revenue. It has raised no new equity since its IPO in 1997, nor debt (with exception of Whole foods acquisition, for more subscribe to @BenedictEvans twitter account). Third, access to VC funding is open to many inventive firms, not locked in by tech platforms. Fourth, the VC market is about funding firms without sales. So unless you want to declare all VC-funded activity predatory, I am not sure where you want to go.
Hubbard: Or can we just not see the losses because Amazon cross-subsidized? Several former Amazon employees have told me about selling below cost.
Singer: Moving to merger enforcement, the paper calls for amending Section 7 of the Clayton Act to remove the burden of proof on the government to show a likelihood of lessening competition, and instead place the burden on parties seeking to merge “to prove that their proposed acquisition will not materially lessen competition, create a monopoly or monopsony, or help maintain their market power.” This would obviously be a game changer. The question is why should the burden be placed on the merging parties? How does one prove that a business initiative does not help maintain a market position, as that is ostensibly the purpose that drives most business investments by firms with market power? Does the EU do it this way? And what does economics or the law have to say about the optimal placement of burdens?
Steinbaum: I guess I can speak to what “economics says,” (i.e., what I say): that when an economy is suffering from a market power crisis, now is not the time to be super-deferential to claims made by merging parties.
Hubbard: We’ve had decades of merger mania, and I think it makes sense to make a significant course correction by putting the burden on the merging parties.
Singer: Hmm. I was looking for some reasons asides from not liking the results. Let me offer one idea from econ—that the parties in the best position to produce the information should bear the burden. Does that make sense here? Clearly, that would be the merging parties.
Hubbard: That makes sense to me. Also, the Clayton act doesn’t say anticompetitive mergers are ok as long as they are efficient.
Petit: Most merger integration yields efficiencies. The default presumption should be one of lawfulness.
Steinbaum: What empirical evidence is there for that?
Petit: Well, lots of econ papers, both formal and empirical. And the management literature. All this actually convinced the legislature.
Singer: Binders full of merger-efficiency papers! Sorry, couldn’t help myself.
Petit: Again, this is a very bold statement. Lots of benevolent scholars who accept the efficiency rationale for mergers are nonetheless ready to accept that the question of harm to competition is empirical in nature. What you want is to have a merger-control system that can take another view when specific transactions reach excessive HHI levels. But this is standard practice. Above certain thresholds, like dominance, increments in market power are scrutinized more thoroughly. Why do we need to reverse the burden for *all* reportable mergers? Sure, you may come up with a Tim Wu-esque story like the Kronos effect. But these cases are outliers. They can, and should, be dealt with ex post under a fully functional antitrust enforcement system.
Steinbaum: Well for all these supposed “efficiencies,” I would reiterate that the profit share is now 15%. It used to be 5%. Where’s the efficiency?
Singer: With respect to mergers between firms stacked vertically along the supply chain, the paper calls for Congress to prohibit such vertical mergers “when they could foster the firm’s ability and incentive to distort competition.” The Department of Justice recently challenged a vertical merger (AT&T-Time Warner) and largely focused its energy on proving price effects for rival distributors and their customers—presumably with an eye toward satisfying the consumer-welfare standard. This new instruction would permit agencies to pursue non-price harms in merger reviews. But as it’s worded, it seems like it would give agencies a blank slate, allowing political winds to sway prosecutorial discretion. Marshall, what did you have in mind? And why should we trust them with such discretion?
Steinbaum: Reading both the government’s complaint and the judge’s decision in AT&T-Time Warner was really a revelation to me. So many reasons not to want the telecoms sector to become **even more** vertically integrated: leverage over content creators, monopoly leveraging from pay-TV/content into wireless, etc. And yet here this enormously consequential economic policy decision comes down to arguing which of two observables to plug into a structural model aimed at predicting consumer price effects. When that’s where you are, your antitrust policy has gone astray. So the point of our vertical merger proposals is to resurrect the many theories of harm that once guided antitrust enforcement, because it seems to me, the economy worked pretty well back when enforcers had access to those.
Singer: Any concerns about giving enforcers more discretion in this area? I’m thinking about a DOJ that might be hostage to a hypothetical demagogue who used antitrust to punish political enemies.
Steinbaum: DOJ is already held hostage by a demagogue, I’m sorry to say: Bork (and Easterbrook).
Singer: Finally, the paper calls for certain non-price vertical restraints such as non-compete clauses and other provisions restricting workers’ rights in labor contracts to be per se illegal, rather than subjected to rule-of-reason treatment as they are currently. The only exception the paper would tolerate is in “circumstances when no party to them possesses market power and the restraints are necessary to foster innovation and competition.” What are examples of under-enforcement in this area owing to the more charitable rule-of-reason?
Steinbaum: Oh, yes, well–on labor, they’re clear. Ashenfelter and Krueger have turned up this no-poach language in franchising contracts that’s clearly flagrantly illegal and yet has gone unnoticed just because enforcers are so focused on consumers. For non-compete clauses, it’s arguably both the focus on consumers and the rule of reason that keeps enforcers away.
Singer: And why shouldn’t firms with market power be allowed to provide efficiency defenses for these restraints? Can you not conceive of any plausible efficiencies here?
Steinbaum: As for not being permitted to show efficiencies as a defense, we know from the investigations into how non-competes are used that they rarely (if ever) bring about these efficiencies, so, why should the law provide a gaping hole to claim them? This holds elsewhere in antitrust, but to me, it’s weird that you can say “oh yeah we did that but it was good for the economy” and thereby get away with violating the law. Should we institute that jurisprudence for murder? It’s not that no efficiencies are conceivable, it’s that none are empirically present.
Petit: I agree at the high level that antitrust should look more inside firms. In the tech space, that means agencies should focus on employees but also on internal divisions and business units. Firms are not black boxes. Individuals, divisions and business units operate – and compete – within them. In fact, this may also lead us to look at to internal firm competition, and find that there’s more rivalry than we just see when we look at an alleged outside structural monopoly. Take Google: there’s Waze v Maps, YouTube v Google Music. Or Facebook: there’s Messenger v WhatsApp. Or Tencent, with the QQ v WeChat fight. Instagram is another example. Its founders – who recently left the company – have for long operated with much operational independence from Facebook. When this happens, alarmist concerns of post-merger vertical or integration by the platform, as well as calls for breakups, are witch-hunts.
Singer: Final question! An alternative to amending the consumer-welfare standard is to shrink the portfolio of antitrust, and instead use regulation to protect against harms (such as edge innovation) that is not readily cognizable under antitrust. This is precisely what Congress did in 1992 with the Cable Act, where it created a venue for independent cable networks to bring complaints against vertically integrated cable operators pursuant to a nondiscrimination (as opposed to a consumer-welfare) standard. The FCC similarly embraced a different standard to police net neutrality violations in 2010 (the nondiscrimination standard) and again in 2015 (the general conduct standard). Why shouldn’t gaps in antitrust enforcement be addressed via industry-specific regulation, as opposed to changing antitrust standards for all industries?
Hubbard: Why can’t we do both? We definitely don’t need to shrink antitrust at this moment in history. Antitrust plaintiffs could have both avenues available to them.
Steinbaum: I definitely think there’s a role for sectoral regulation. You’ve convinced me of that—especially if we stick to the CW standard, but even if we don’t. For example, Lina Khan has an excellent forthcoming paper about structural separation as a regulatory principle in tech. That would say “the economics of these platforms are such that we need to draw a border around them, impose some version of common carriage on their operations within that boundary, and then break off anything outside it, because the incentives militating toward discrimination and exclusion are so great that no regulatory regime alone could handle the burden of preventing those outcomes.” Anyway, that would amount to a sector-specific regulation aimed at enhancing competition. We’ve had that numerous times at much smaller scales: phone number portability, for example.
Petit: No doubt I will read with interest. To convince me, it’ll need to meet a three-pronged test: (1) it addresses the econ literature on economies of scope and leveraging; (2) it discusses the risk of regulatory capture; (3) it is exempt of clichés like tech firms are “natural monopolies,” data is the new “oil,” and platforms enjoy “first mover advantages.” So far, my experience with this scholarship is well… it’s nicely packaged… period.
Singer: Final thoughts?
Hubbard: So I think Marshall’s idea of scrapping the CW standard is a great approach. Another idea—if we can’t get Congress to do anything because tech platforms are now the most powerful lobbyists—is simply fixing the standard itself to not obsess on prices.
Petit: I want to concede something here: in the EU, the agencies seem to experiment more with new theories of harm. See the Dow/DuPont review or the recent work of the chief economist and his team at DG Comp on attention markets. This has little to do with the CW standard. This more experimentalist attitude is mostly the byproduct of the EU’s antitrust governance system.
Singer: Well, Marshall. There’s hope for your experimentation with the standards! Thanks to everyone for joining!