What Big Tech antitrust gets wrong

Good morning. Is TSMC cheap? The company, which manufactures chips for just about everyone, jumped nearly 10 per cent yesterday after less-bad-than-feared earnings and sunny guidance. Its forward P/E is just 15. We’re still searching for good stock picks. If you’ve got a candidate, email us: robert.armstrong@ft.com and ethan.wu@ft.com. 

Friday interview: Herbert Hovenkamp

If you own shares in Apple, Alphabet, Meta or Amazon, you need to know about US antitrust law. All save Apple are facing big antitrust challenges from US authorities, and Apple could face its own as soon as March. Meta could be forced to divest Instagram and WhatsApp; Alphabet to carve up its advertising cash machine; Amazon to change its pricing practices; and Apple to open up its walled garden. 

So it pays to listen to Herbert Hovenkamp, the pre-eminent scholar of US antitrust law and a professor at Penn Law School. Hovenkamp wrote the textbook on US antitrust, and recently published a timely paper on breaking up Big Tech. Unhedged spoke to him about the problems tech platforms pose for US law, whether Google is a natural monopoly, and what he thinks would work better than break-ups.

The interview has been edited for clarity and brevity.

Unhedged: With Big Tech, what big questions is antitrust law grappling with? 

Herbert Hovenkamp: To a significant extent, they’re grappling with the wrong things. Which industries should the antitrust authorities pursue? You look for industries that are characterised by slow growth, oligopoly, rigid market shares, not very big increases in productivity — industries of poor performance. You try to make those industries perform better.

With Big Tech, we’re looking at probably the most productive part of the economy. The rate of innovation is high. They spend a lot of money on R&D. They are among the largest patent holders. There’s very little evidence of collusion. They seem to be competing with each other quite strongly. They pay their workers relatively well and have fairly educated workforces. None of this is a sign that these are industries we should be pursuing. That doesn’t mean they don’t do some anti-competitive things. But the whole idea that we should be targeting Big Tech strikes me as fundamentally wrong-headed.

Unhedged: What’s a historical example of an industry that really needed antitrust enforcement?

Hovenkamp: Chemicals in the mid-century. Even still today, things like high-fructose corn syrup. Heavy industry, early in the 20th century. Any sales industry that is prone to price fixing, such as real estate brokerages. And we’ve got tools for identifying those: they make products that are more or less the same and it is easy to monitor prices.

Unhedged: We hear a lot about the distinction between Section 1 and Section 2 of the Sherman Antitrust Act. Why is it relevant here?

Hovenkamp: Section 1 covers contracts, conspiracies and restraint of trade. Think practices like cartels, tying arrangements, exclusive dealing. What they all have in common is that they are contractual, and involve at least two firms. Section 2 covers monopolisation — single firm dominance. If you want to break up a firm unilaterally for doing bad things, you would sue them under Section 2. 

The recent spate of cases are all under Section 2, and so they’re all being presented as [illegal monopolies]. However, when you look under the hood, virtually everything there is a challenge to some kind of an agreement! 

In the Google search case, it’s Google’s payment of big bucks, principally to Apple but also to others, for being carried as a default search engine. We don’t know too much yet about the Amazon complaint, but allegations include tying Prime membership to product shipping services. Those are contracts. The cases against Meta have now really been reduced to two acquisitions: Instagram and WhatsApp. But we’ve got a statute for dealing with acquisitions. That’s the merger statute, Section 7 of the Clayton Act. But that case is also being brought under Section 2 of the Sherman Act.

I’m not trying to defend all the practices, some of which are illegal or should be illegal. But in my view, a lot of this is mischaracterisation of contract practices as some kind of big structural offence [requiring a break-up]. We don’t usually break up firms for things like price fixing.

Unhedged: When people think about structural remedies (ie, break-ups), the first case that comes to mind is AT&T. What’s the right way to understand the history of the AT&T case in thinking about today’s antitrust lawsuits against Big Tech?

Hovenkamp: That was a consent decree, not a court judgment. It was settled by agreement between the government and AT&T. The decree consists of three parts. First, breaking up AT&T into seven regional Bell systems. Second, break-ups between the local operating companies and AT&T long-distance, and the divestiture of Western Electric, which made telephones. Third, mandatory interoperability. That created lots of competition. Carriers big and small are hooked into the same system.

Now you look back at the decree and ask which of these three things had the biggest competitive effect, and by far the biggest impact was interoperability. Frequently the best thing to do in a large network industry, particularly if it involves information technologies, is to make sure that it interoperates between firms.

Unhedged: That rhymes with the Department of Justice vs Apple case that may be coming in March. Reportedly, the allegation there is that Apple prevents its own tech from interoperating with other companies’ devices. Is that the right analogy?

Hovenkamp: It’s a pretty good analogy. Apple has defended this successfully for decades now by arguing that their [closed software and hardware ecosystem] is more virus-proof. It is less prone to glitches. Just [on Tuesday] the Supreme Court denied certiorari in Epic Games versus Apple, in which one of Apple’s defences was that they need to sell apps exclusively through the App Store, because that’s the only way we can preserve the integrity of the system. And, the Ninth Circuit approved Apple’s practices under federal law, but condemned them under state law. We’ll see how that works out.

Unhedged: Your recent paper touches on the difficulty of demonstrating the antitrust harms alleged in the Amazon case. Can you talk through some of the issues?

Hovenkamp: The first difficulty is the market power question. It creates a collision course between the antitrust left, who are opposed to bigness, and the more centrist or right-leaning antitrust [theorists], who see antitrust as a problem of market power, without regard to whether a firm is small or big. One of the problems you’ve got with Amazon is that it is very, very big. However, it sells 12mn products and does not have market-dominating shares in very many of those products.

Remember a critical principle here, which is that market power attaches to products, not firms. We can speak of Microsoft as having market power in its Windows operating system. We are not speaking of Microsoft as having market power with its search engine. Why? Because Windows has a 60-70 per cent market share in desktop/laptop operating systems. Bing struggles at just above 3 per cent.

So with Amazon, you have to look product by product. And that means several things. Number one, it’s going to be a hell of an expensive antitrust case. Number two, you’re going to get very different outcomes depending on which product you’re looking at. It’s going to be very big unless the government narrows it down by identifying particular products.

Unhedged: Our sense of the Federal Trade Commission complaint is that Amazon is able to exercise undue influence on prices everywhere because of its exclusivity rules for people who sell on its site. 

Hovenkamp: Amazon does have a fairly weak “most favoured nation” clause, which basically says if you want to sell on Amazon, we don’t want you to charge a significantly lower price on a competing seller site. I think it uses the words “appreciably lower”. But most favoured nation clauses are usually legal for non-dominant firms.

I think the way to address the most favoured nation agreements in Amazon would be to go through product by product. Of course they’ll never do that; what they’ll do is try to pick some in which the market shares are high enough that they could get a court to agree that the [entire] practice is anti-competitive.

Unhedged: What would you say is the best solution available for the Amazon case? 

Hovenkamp: I would dismiss the complaint, to be honest. Amazon dropped the most favoured nation agreements in 2019. There’s an allegation of tying the buy box [ie, better placement on a product order page] to Amazon fulfilment services. Even if that’s true, which I’m sceptical about, that’s small potatoes in the grand scheme of antitrust cases.

There’s a huge literature going back 10 or 15 years on various events that occurred with ebooks. But I don’t think there’s anything going on right now that is challengeable under the antitrust laws. The thing is that Amazon is so open. Amazon has its own file format for ebooks and its own reader format, Kindle. It gives it away for free.

Another allegation is that if you want to buy Amazon Prime, you get Prime Music, Prime Video and Prime shipping of packages. So that would be attacked as a tying arrangement. But it’s just a pricing mechanism! Could they leverage that into something with a big enough market share to make it sound anti-competitive? I can’t rule that out, but it just doesn’t sound very promising to me. So I don’t see a lot coming out of the Amazon case. 

Unhedged: How about the two cases against Google?

Hovenkamp: I think the Google ad tech case is one of the cases that the government may very well win. Google’s rules and regulations tend to steer advertising to its own assets. This is the case I’m least well-versed in, but there clearly seems to be something wrong with Google’s dominant market share in ad tech.

Search, on the other hand, is a complicated question. That case has already gone to trial; Judge Amit Mehta is sitting on it right now. If I had to make a prediction, it’s that he’s going to find a way to condemn the very large payments Google makes to Apple and others.

If the payments are eliminated, that means big device makers like Apple are going to have to decide what they want to do. One option is they continue right on using Google Search as a default, except they’re not getting paid for it anymore. Another option is they put in a choice screen, which is what happens in the EU. The third one, which I don’t expect to happen, is that Apple will try to develop its own search engine.

Unhedged: There’s a lot of commentary to the effect that search is a natural monopoly and that these remedies won’t change anything. But if it is a natural monopoly, why does Google need to give Apple $20bn a year?

Hovenkamp: That’s a very good question. And the government made lots of hay on it at the trial. There are two schools of thought. Everybody agrees that bigger is better. The question is whether it levels off at a certain point. A natural monopoly would be one where bigger is continuously better all the way up to the point that the entire market is saturated. On the other hand, if bigger is better until you hit, say, 20 per cent market share, then you could have room for as many as five search engines and they would all be equally efficient. I think the debate leans towards the first one, although it’s not conclusive. I don’t think Judge Mehta is going to be able to decide that issue in the timeframe we’ve given him. He’ll just have to come up with a best guess.

Unhedged: There is an irony here. If it is a natural monopoly, and consumers are given a choice screen and keep choosing Google, all the trial has done is save Google $20bn a year.

Hovenkamp: It’s very possible. And there’s a bigger mystery: why does Microsoft continue to make Bing the default on desktops and laptops when everybody switches away?

Unhedged: You made a distinction between interoperability remedies and structural remedies. Could interoperability solutions work in the Google case?

Hovenkamp: That’s an engineering issue that’s a little bit over my pay level. But there’s talk about making the search database kind of a public commodity and then let people that want to sell search services tap into it. The structure would be like the airlines and the travel agents. Airlines determine the schedules on flights. Travel agents don’t have any control over that. What each of them does have, however, is a terminal that permits them to log in and sell flights out of that inventory. The reason this worked with AT&T was because it was a consent decree. It was negotiated in painstaking detail. That could happen.

Unhedged: Lastly, let’s talk about Facebook/Meta. 

Hovenkamp: Facebook started out as a challenge to some contract practices plus these two acquisitions, Instagram and WhatsApp, under Section 2 of the Sherman Act. There is an enormous fight going on in the litigation over market definition. Who belongs in the social media product market? For example, under the FTC’s allegations, Twitter/X is not in the market. That strikes me as wrong, particularly when you look at the number of people who switch back and forth between Facebook and Twitter. If the market is limited the way the FTC wants it limited, then Meta is somewhere north of 60 per cent market share, which is probably enough to settle the market power issue.

Assuming the government wins on the market power issue, I think the case for undoing those two mergers is a pretty good one. There’s a big record, particularly on Instagram, in which Facebook’s managers were clearly worried that Instagram was going to mature into a rival for Facebook, and they acquired them in order to shut down that threat. That’s a classic reason for condemning a merger. But the very best the government can hope for realistically right now is a decree that divests either or both Instagram and WhatsApp. 

One good read

How to fail upwards.