Should the FTC Break Up Meta?

Facts, Theory, and Evidence

i. The Facts of the Case

In December 2020, the FTC, along with 46 states, filed suit against Meta, the company formerly known as Facebook. At issue was its acquisition of competitors, in particular Instagram and Whatsapp, and its practice of preventing the interoperability of other apps with Facebook. The state complaint was dismissed as untimely, but the FTC’s complaint survived Meta’s motion to dismiss in 2022, with a trial expected in 2025.¹

It got off to a rocky start. The initial complaint claims that Facebook is characterized by very strong network effects. The value of the service is dependent upon the number of people using it. Being the only provider of personal social networking services, they are able to provide a worse quality service to consumers, in the form of lowered privacy, and charge worse prices to prospective advertisers.

Because of the network effects, Facebook need not fear direct clones, because they would have no advantage over them. What they should fear is something offering a different service later expanding into their niche, and using the network they built up to make the jump. Instagram offered something different – a picture based social networking service – and so was able to acquire users through that channel. Whatsapp offered superior messaging services, and so had the user base to perhaps challenge Facebook. 

In addition, the FTC alleged that Facebook’s API practices were anticompetitive. APIs allow other websites to connect to Facebook, and the data for their own purposes. In order to use Facebook, other developers had to promise that they would not use the data for any direct competitor, nor could they use it to assist a competitor. Suppose you wanted to offer social networking with a twist. Reconstituting your friend network is tedious and difficult; API access means you could connect the two, and quickly find your friends.

This pleading was dismissed with the opportunity to refile. The FTC simply did not do enough to assert that Facebook had monopoly power. The only contention on that score was that, within the market of their choosing, Facebook possessed 60% of the market. As Judge Boasberg wrote, “such an unsupported assertion might (barely) suffice in a Section 2 case involving a more traditional goods market, in which the Court could reasonably infer that market share was measured by revenue, units sold, or some other typical metric. But this case involves no ordinary or intuitive market. Rather, PSN services are free to use, and the exact metes and bounds of what even constitutes a PSN service … are hardly crystal clear. In this unusual context, the FTC’s inability to offer any indication of the metric(s) or method(s) it used to calculate Facebook’s market share renders its vague “60%-plus” assertion too speculative and conclusory to go forward.” 

The amended motion keeps the arguments as to why Facebook is anti-competitive, and supplements it with enough argument that there is a monopoly. The argumentation that the consumer is harmed was tenuous, but the bar to prevent summary dismissal is fairly low. That they missed it to begin with an altogether bizarre error. The claims around Facebook’s API access policies were dismissed as untimely and irrelevant. The policies were last enforced in 2013, and dropped in 2018, and the FTC has no statutory authority to pursue the matter. Nevertheless, the case will go to trial, and it seems the Trump administration will not drop the case (unlike the Google case).

ii. What It’s Really About

There are several questions at issue. The primary question we face is whether an established company buying up a new competitor tends to increase or decrease welfare, with some of the secondary questions being what the relevant market is that Facebook monopolizes, and whether (given that Facebook is a two-sided market) consumers are actually being harmed at all. In addition, since (as the first dismissal notes) many of the relevant policies had lain in abeyance for years, the FTC might lack any authority to do get an injunction at all. 

Let’s deal with the big questions first. The case against Facebook purchasing Instagram and Whatsapp is that, in the absence of competition, the large firm can extract monopoly profits and underproduce what people want. If there is perfectly free entry, then no company would be able to buy out all of its competitors, but since there are strong network effects, every competing start up would have to have a new idea sufficient to bring in a lot of people. There are not many billion dollar company ideas lying around; doubtless we could run out of competitors. As Zuckerberg wrote to CFO David Ebersman, “there are network effects around social products and a finite number of different social mechanics to invent. Once someone wins at a specific mechanic, it’s difficult for others to supplant them without doing something different.”

That Facebook would get excess profits from their market power is a dark mirror of why it would be good for them to buy it, which is that they may benefit from economies of scale. Zuckerberg and co may be better managers than the founders, or they may resolve difficulties of unspecifiable contracts by uniting ownership of the assets. Big companies have lower implementation costs, being more experienced. What’s more, the prospect of being paid induces the entry of more firms into the market. To require all ideas be carried out start to finish by their creators would be like requiring architects to hire their own construction workers. We could hardly have comparative advantage in such a world. 

Which prevails? I should note that we cannot expect there to be one value which prevails for all industries. Cunningham, Ederer, and Ma (2021) show that in the pharmaceutical market allowing the acquisition of small competitors is really bad for consumers. When a prospective drug overlaps with a drug they own under patent, the large firms simply don’t develop it at all. The point was to protect their monopoly. We can infer from this that as the monopoly becomes more sure, the harms of allowing acquisitions grow. This aligns with Fons-Rosen, Roldan-Blanco, and Schmitz, who find the key predictor for the competitive impacts is the acquisition premium, or the difference between a start-up’s best and second best offer. In other words, if the value of the asset is primarily in increasing the market power of an established firm, the next best offer will be very low. Now, some of the acquisition premium is due to some firms being especially able to improve the assets value, so we cannot infer that they are anticompetitive perfectly, but it is still a useful heuristic. When F-R, R-B, & S fit their model of endogenous growth, they find that the anti-competitive impacts are very slightly outweighed by the pro-competitive and pro-innovation effects at the scale of the whole economy, and that bans on start-up acquisitions would reduce the growth rate.

In order to estimate the acquisition premium, we need to know the value that the companies would have to other competitors. Facebook bought Instagram for $1 billion, split 70-30 between equity and cash. (A source of fuzziness is that Facebook would not have an IPO for another month. I see no reason to think the true value was substantially different to the headline figure, though). The funding round a week before Facebook’s acquisition valued Instagram at $500 million, although I do not think we can naively take that as the value. Investors into start-ups are rarely truly passive, and Instagram could have been seeking to bring on venture capitalists as consultants. The acquisition kept the management, and so some of it was in order to retain talent. Estimating the cost of retaining talent is pure vibes, but I will go with 20%, in line with the purchase of Whatsapp. I have heard that Twitter made an offer of $525 million, although my source is nothing but asking people on twitter. Yahoo offered $1.1 billion for Tumblr in 2013 (in yet another display of the sterling business sense shown by Yahoo). Allowing for the possibility that someone else would have made a higher offer, I will conservatively say that it’s next best value is $550 million, and that the value of their assets to Facebook was about 840 million dollars.

The cost to acquire Whatsapp is a bit fuzzier, as Facebook initially offered $4 billion in cash, and the equivalent of $12 billion in stock. They increased their offer by $3.6 billion in order to retain the staff, and the shares increased in value between the time the deal was signed and when they were transferred. Because the stock price price was endogenous to the news that the sale was happening, I will drop that and the staff acquisition, and say that the price paid was $16 billion. It was much harder to estimate the value of Whatsapp than Instagram. A private equity group valued it at $11 billion in November 2013, with 350 million users. I think the value would have grown in the time, as the motion to dismiss claims that they had 450 million users by early 2014. 

We can use the model presented in Fons-Rosen, Roldan-Blanco, and Schmitz to calculate whether the acquisition would have been pro-competitive or not, so long as we know the acquisition premium. Remember, they found that the pro- and anti-competitive impacts were almost exactly balanced, and that as the acquisition premium increased, the anti-competitive impacts would grow stronger. On page 30, they find that the average startup premium was 49.3 percent. Plugging in the numbers, we arrive at an acquisition premium of 52% for Instagram, and 45% for Whatsapp. I tried to be conservative in estimating the valuation of the companies bought, and so believe this to be an overestimate of the premium. 

To me, the effect of startup acquisitions on growth is the primary question. Nonetheless, some of the minor questions are important too. Facebook takes issue with the FTC’s definition of their market. Markets must not be drawn too narrow as to be tautological. Taylor Swift, for example, holds a monopoly over Taylor Swift concert performances; but she is clearly competing with other entertainers and musicians. Facebook provides a particular sort of personal networking, but is a very small part of social media as a whole. The FTC drew the boundaries of the market narrowly. It did not include Linkedin, Twitter, Youtube, Reddit, or Tiktok. The main competition which they include in their market is Snapchat. I agree with the exclusion of Youtube, which, similarly to “Spotify, Netflix or Hulu” (p. 57, amended complaint) is focused on the “passive consumption of specific media content … from and to a wide audience of typically unknown users.” It would be helpful, however, if the FTC broke down time spent on Facebook by use case. Viewing a video from your friends is something Youtube does little of, but viewing a video from a content creator on Facebook is not personal networking at all. On top of that, drawing a line between using apps for interests, compared to using apps for personal networking, seems specious. Do people not talk to their friends on Twitter? I do!

More importantly, even if the other sites do not occupy the exact same market, Facebook fears that they may. Instagram and Whatsapp did not offer the same services as Facebook, yet still competed with them. The amended complaint is shockingly light on the threat of other instant messaging apps like WeChat and Telegram, who have usage rates in the billions, entering the market. Market share can be misleading if the markets are easily contestable. The three conditions needed are that there are no entry or exit barriers, there are no sunk costs, and that all companies have access to the same level of technology. The first two seem very close to true, and the third is satisfied by virtue of the competing instant messaging apps having a very large userbase. The failure of Google to offer a competitor is not dispositive evidence of a monopoly – can we really exclude the possibility that Google bungled it on their own? 

The FTC had great difficulty showing that the consumer was harmed, although for this they may be forgiven. It is free to the consumer, so the harms must come in the form of a worse quality of service. These future innovations are, almost by definition, unknowable ex ante, but it would have been good to point to some examples of features found elsewhere which were not implemented within Facebook. Perhaps they could also find internal emails of frustrated engineers trying (and failing) to get new ideas implemented. The best analogy I can give is that we believe the FDA’s regulations lower the rate of drug discovery, even though we cannot point to which drugs were prevented. If we grant unknowable innovations in one case, we should grant them in others. 

iii. Judgement Day

My ruling? I shall abstract away from what the law says, and answer only what I believe would lead to the best outcomes.  I believe that there could have been an argument then, but it is now too late. The FTC took an unusually long time reviewing the procedure – four months – and approved it only after receiving additional documentary evidence. If it didn’t have enough then, it doesn’t have enough now. Companies need to be able to make investments without fearing that they will later be wrested away. What’s more, ex post reviews are biased against successful firms. As Bialystock and Bloom might say, who examines a flop? Companies are uncertain about the profits from a merger ex ante – examining only the cases which work out prevents many beneficial mergers. This is especially harmful when the investments are not defined. If you are going to unwind a manufacturing company, you can at least say that they spent such and such on machines and such and such on land. How are you supposed to unwind an innovation? How are you supposed to unwind tacit managerial practices?

The examination of Meta is not entirely driven by scrupulous regard for the public either. During the 2016 and 2020 elections, Facebook offended politicians of both parties, who now seek to bloody Meta’s nose. If the powers of law are to be picked up and wielded unfairly, it is better they lack that power at all. It is like when an anti-corruption drive in an authoritarian country both convicts only people genuinely guilty of corruption, and only those who belong to the opposite party. 

Had the FTC chosen to prevent the acquisition then, I believe that it would not have been correct, although it would have been very close. The acquisition premium was insufficiently high to indicate that it would be anti-competitive. I believe the FTC when it reviewed the evidence, and found that it wouldn’t be anti-competitive. 

The FTC’s present case is not entirely meritless, however. I believe they should prevail in its injunction against barring API usage for a competitor. The court rejected it on the grounds of timeliness, as the FTC lacks the authority to seek an injunction against a practice last done seven years prior. To have standing to sue, the FTC would need to challenge a particular incident; but I think this essay would be ill-served by keeping to rules of procedure. Facebook has not exercised it since 2013, and removed the policy entirely in 2018, but one might reasonably fear the “chilling effect” of a policy change. If it is irrelevant, then banning it is of little harm; and if it could be used, it would surely only be for mischief. 

Any examination of antitrust cases is humbling. I was overwhelmingly struck by Hayek’s insight – it is not possible to simply know what all the prices are. The key parameter to be identified – the acquisition premium – is adrift in a sea of hypotheticals. It is with great difficulty that we can say whether an acquisition is good or bad. I believe the totality of the evidence shows that Facebook should be allowed to retain Instagram and Whatsapp. I do not claim strong confidence in this. 

1. No such timeliness requirement exists for the FTC. Congress has explicitly provided that the FTC can challenge any acquisition, at any time, and under any provision.