Home Antitrust and Competition In FTC vs. Facebook, the Government Lost the First Round. It Could...

In FTC vs. Facebook, the Government Lost the First Round. It Could Still Win the Fight.

0
Illustration by woolzian via Getty Images

Despite some compelling allegations, a federal judge dismissed the FTC’s antitrust complaint against Facebook due to the agency’s failure to explain how it calculated Facebook’s market share. To win, the FTC would need to explain why several metrics could actually be useful.


Federal Trade Commission v. Facebook is an early contender for the antitrust case of the century. It’s a traditional break-‘em-up monopoly action, unlike NCAA v. Alston (which focused on the NCAA’s rules against paying student-athletes) or the DOJ’s ongoing antitrust suit against Google (in which a breakup seems unlikely). But despite alleging some fairly damning facts, the FTC lost the first round: Late last month, a federal judge dismissed the Commission’s blockbuster complaint.

Yet this particular fight is far from over. The FTC’s new chair, Lina Khan, has long supported a revival of antitrust enforcement. Key lawmakers in Congress like Senator Amy Klobuchar (D-MN) have joined the fray, urging the agency to “consider all available options.” Against this backdrop, the Commission is highly unlikely to throw in the towel. Judge James E. Boasberg granted thirty days to file an amended complaint. Here’s how the FTC can best position itself for a win.

The primary reason for dismissal was the agency’s failure to explain how it calculated Facebook’s market share to be “in excess of 60 percent.” Fair enough—a bit more detail is often needed in antitrust cases.

This may seem like an easy fix, because the market definition itself, “personal social networking services,” was deemed acceptable. But Judge Boasberg’s opinion also went on to criticize just about every possible way to measure shares in social-networking markets: time spent on-site, daily or monthly active users, and advertising revenues. To win, the FTC would need to explain not only which metric(s) it is using, but also why these metrics can actually be helpful and should be acceptable to a court.

Time On-Site: Parts of a Product Are Not the Whole

Let’s start with time spent on Facebook or Instagram, which is one common way to calculate social-networking market shares. Judge Boasberg opined that this could be misleading, because some features offered by social networks are also offered by other digital apps. Under that logic, any time spent watching Instagram Reels should be deducted from Facebook’s total market share, because users can also get video content from non-market participants like YouTube or Netflix.

But when it comes to relevant markets, the whole has long been treated differently from the parts. Another Sherman Act § 2 case involving mergers, U.S. v. Grinnell Corp. (1966), offers a great example. The defendants sold burglar- and fire-alarms that also alerted local law enforcement and first responders. The Supreme Court noted that alternatives like night-watchman services or non-connected alarms offered some of the same functions—but not all of them. And that was enough to make the defendants’ products unique.

More recent cases follow the same logic. In FTC v. Staples (1997), which involved a proposed merger between Staples and Office Depot, the market was “consumable office supplies sold through office superstores.” Non-superstores also sell some office supplies. But the district court did not discount all Staples/Office Depot sales of those items. That would’ve missed what makes “office superstores” unique in the first place.

Watching a Reel on Instagram is not the same as watching a video on Netflix. Reels are shared, interacted with, and viewed from within Instagram’s personal social-media network.  Instagram users can create Reels and share them with friends via a personal feed and profile. You can share other users’ Reels with your friends—adding a comment, if desired—via feed or by direct message. While viewing Reels, you still get notifications when a friend goes “live,” and you can instantly switch over to interact with them during the livestream, and so on. Trying to analyze one particular part of the social experience in a vacuum can ignore what makes personal social-networking services so unique.

“Active Users” May Not Be Perfect—But Nothing Is

Another common way to measure shares in social-networking markets—as one federal district court has already recognized—is active monthly (or daily) users. But Judge Boasberg downplayed this as potentially misleading, since it might over- or understate actual competitive significance. That’s true, but the same could be said about any method of assigning market shares. In fact, the entire market-definition exercise is just an indirect way of trying to get at what we really care about, market power. It always runs a risk of over- or understating.

Active-user statistics could help judges and enforcers understand important dynamics. Social networks are, by their nature, differentiated products. Suppose one network successfully targets a general audience, while another is designed to attract a much smaller—but highly engaged—demographic. Looking only at time on-site could suggest that the two wield equivalent power.  But in reality, the second network may be unable to truly threaten the first, if doing so would require abandoning what made it uniquely attractive to its own core users. Active-user data may not be perfect, but it could help to tease out differences like these.

“The entire market-definition exercise is just an indirect way of trying to get at what we really care about, market power. It always runs a risk of over- or understating.”

Ad Revenues and Social-Networking Services

Finally, Judge Boasberg seemed to reject the possibility of using ad revenues, because the “market for advertising” is different from the personal social-networking market. But ad revenues can also help shed light on competitive dynamics.

Social-networking users barter attention and information to social networks in exchange for services. Social networks seek our attention not for its own sake, but because it is valuable to advertisers. As with other raw materials, attention’s value can vary. Certain users’ attention, or certain types of attention, are especially prized. These differences can therefore be important for antitrust purposes. By way of analogy, we might want to know whether a defendant controls a particularly valuable oil well or gold mine, even if we also have data on how many barrels or ounces are being produced.

A Solid Start—But Not Enough

The FTC’s most immediate task is to explain how it calculated Facebook’s market share and why its methodology is appropriate. That shouldn’t be the agency’s only change in tactics, though. If possible, the FTC should also beef up its direct proof of monopoly power. The court seems ready to entertain such evidence. Did Facebook set attention-cost or information-cost levels for users without worrying about rivals’ responses? Has it raised cost levels without losing much (or any) business? Natural experiments can be extremely useful. Did Facebook fail to lower attention or information costs in response to a new rival’s entry?

Finally, the FTC should be prepared to head off in its briefing—or correct via appeal, if necessary—an erroneous ruling on Facebook’s app-developer policies. Judge Boasberg seemed to reason that only a “blanket prohibition” on third-party apps’ ability to deal with Facebook’s rivals could be a violation. To put it bluntly, that’s wrong. A monopolist can violate antitrust laws by discouraging—even without totally prohibiting—its counterparties from dealing with its rivals. (Bundled discounting offers one example.)

The case still won’t be easy, and the inevitable appeals may drag out for years. But the FTC’s allegations raise serious concerns. Its first punch may not have landed, but the Commission doesn’t deserve to get knocked out in the first round.