The Wall Street Journal reports that “The Federal Trade Commission on Tuesday ordered Amazon.com Inc., Apple Inc., Facebook Inc., Microsoft Corp., and Google owner Alphabet Inc. to provide detailed information about their acquisitions of fledgling firms over the past 10 years.” These deals, which regulators approved at the time, might be undone after the fact.
This is likely illegal. Both federal and state governments are prohibited from making ex post facto laws punishing past actions that were legal when committed. This is a complicated legal question, however. Antitrust law usually comes in the form of judicial decisions, not congressional legislation. The Sherman Act, for example, is only two pages long. Its use of key terms is so vague that judges have been defining and redefining those terms at will for more than a century. In an antitrust case, it is not enough to have truth, justice, and the merits on your side. You must also have the judge.
If regulators follow through with their ex post facto threats and judges agree, they will create enormous uncertainty in the mergers and acquisitions market. Buyers risk prosecution if a deal works out better than expected. The potential chilling effect on competitive behavior is obvious.
Moreover, many of the technologies from years-old acquisitions are so thoroughly merged with the buyer’s operations that unwinding the deals is simply unfeasible. It would be like trying to turn a book back into a tree.
The whole scheme highlights fundamental problems with antitrust law. To see why, let’s step back and take a larger four-dimensional view.
Companies have long risked prosecution for both present and future behavior. But to reach back into the past ex post facto is something new. For example, if a company’s present size is too big for regulators’ tastes, they might break it up. That in-the-moment concern motivated Standard Oil’s 1911 breakup despite its declining market share. It also ended the government’s protection of AT&T’s monopoly in 1984, when regulators decided to allow competition, although in a weird, top-down way. Those cases did not create new offenses out of years-ago actions that were legally permissible at the time.
Antitrust regulators are also concerned with the future. If a company is doing nothing wrong now but might do something bad in the future, some regulators believe they have cause to act now. This is called the incipiency doctrine. For example, if Sprint and T-Mobile merge, will the wireless market become too concentrated, leading to potential future bad behavior? Regulators asked the question. Courts said no, and my colleague Jessica Melugin agrees. Other mergers have been blocked because of possible future effects, as has happened twice with Staples and Office Depot.
Now the distant past is coming into play. Over the last decade, the bigger tech companies, such as Google, Facebook, and Amazon, have bought out as many as 400 startups that had developed promising new products, technologies, or business models. Many of the deals fell below the minimum dollar-value threshold for an antitrust investigation. Regulators approved all of the deals they did examine.
Most of these deals ended up being duds; the rule of thumb is a 90 percent failure rate. But after a decade or so, some of the acquisitions turned out to be important to the bigger companies’ success. Facebook’s 2012 acquisition of Instagram is one example. As Facebook’s primary user base gets older and grayer, Instagram is keeping the company relevant with younger people. Countless algorithms and other under-the-hood technologies that now power different parts of Google and Amazon’s operations were originally developed at acquired firms. Now regulators are mulling undoing these past deals, which were previously approved.
The Horizontal, Vertical, and Depth Dimensions
Regulators should instead use a simpler framework with fewer dimensions. To show why, it is worth asking basic questions about business organization. What if Google or Facebook had come up with the successful technologies in-house, rather than having bought them from elsewhere? Would that be an offense? If not, why should developing them via buyouts be treated differently? This is similar to the lesson from economist David Friedman’s Iowa Car Crop story. It is a distinction without a difference.
My theory is that these sorts of multi-dimensional concerns are rationalizations that distract from the main issue: size. In our extended (and imperfect, but useful) analogy, this is equivalent to the depth dimension. Arguments about ex post facto enforcement or different horizontal and vertical arrangements are, in the end, really about size, or the depth of market competition. What appears to be four-dimensional regulation actually concerns one dimension.
Distractions from the Real Issue: Size
Many members of the Neo-Brandeisian antitrust revival are open about believing, like Justice Louis Brandeis, that large size is an inherent antitrust offense. The arguments investigators are floating about different past, present, and future actions, or about different places along the horizontal and vertical dimensions, are only intended to apply to companies of a certain size or to markets with fewer than a certain number of competitors.
Even the number of competitors in a market is a problematic measure. (I earlier gave two reasons why here and here.) A third way to look at it is this: In a way, startup tech entrepreneurs eager to sell out are similar to independent contractors, similar to the way a company might outsource its payroll to an outside contractor or a family might outsource household repairs to a handyman. Sometimes doing something in-house is better. Sometimes it’s not. Every case is different. But doing something in-house means fewer, and larger, firms in the market. Outsourcing means more, and smaller, firms. One arrangement is not inherently more competitive than the other, yet antitrust regulators treat them differently. This is not a coherent position.
Circumstances also change over time. Maybe a company’s in-house R&D team loses a key person or is stuck in a rut. Sometimes a fresh perspective from an outsider might be helpful. Maybe a contractor is too far away from her customers to communicate with them effectively. Maybe a company is having trouble coordinating multiple outside contractors. In these cases, bringing the contractors in-house could make the companies more competitive, even as it reduces the number of firms in the market.
Competition Is a Spectrum and a Process, Not an On/Off Switch
Even within the outside contractor model, there are lots of places along this vertical dimension. Maybe one company contracts with a startup. Another licenses a startup’s technology and brings it in-house but doesn’t buy the company itself. Maybe the license is exclusive; maybe it isn’t. A third company hires the outside person with the bright idea but doesn’t buy her company. Maybe that person’s team and their equipment are necessary to make the most of that idea. If that’s the case, maybe a buyout is easier, and likely cheaper, than hiring away one or two key people. Maybe another company makes overtures to a horizontal competitor or certain of its employees.
Here we find there are angles between the purely horizontal and the purely vertical. Again, competition is not a binary switch, fully on or fully off. It is a spectrum with all kinds of in-betweens. Competition is a complicated, evolving process with nuances that don’t neatly fit into categories.
Every case is different. Nobody knows in advance which possible course of action is the right one—or if there even is a right one. Remember, as noted, mergers have about a 90 percent failure rate—and each and every one was entered into with confidence.
Here is another way to put it. A company with in-house counsel has essentially bought its own law firm. For antitrust purposes, how is that conceptually different from using an outside attorney? These are two different places on the spectrum of vertical integration, but they irrelevant to market competition.
There are similar concerns for the horizontal spectrum. Some cases require multiple attorneys. What if attorneys from competing firms collaborate on the same side of a case? What if some mix of in-house and outside attorneys work together? The result is the same. People or companies who need legal services buy them in the manner of their choosing. That is not a proper antitrust issue.
Same goes with the mishmash of mergers, acquisitions, and divestitures that have characterized the tech industry for decades. Regulators are only making incoherent multidimensional arguments now because the companies are bigger on the one dimension they really care about: size.
Whatever names we give to the ways big and small companies interact with each other, the end results are not that different. Someone sells something and someone else buys it. The sellers might get paid as employees, vendors, or contractors, or maybe they just take the money and move on to something else.
Why some of these arrangements are considered legitimate antitrust questions while others are not is an important question. Regulators have not given a compelling answer, nor are they likely to.
A final point worth remembering: The reason firms exist in the first place is not to enable or restrict competition. It is to reduce transaction costs. There is no magic number of firms that accomplishes that goal. And if there were, it would constantly move as tastes and technology change. It would certainly move faster than the speed of antitrust litigation. Competition is an ongoing discovery process.
Antitrust regulation fails along all four dimensions—the vertical, the horizontal, depth, and time. It should be entirely repealed. At the very least, the Justice Department should immediately stop its search for ex post facto offenses against Amazon, Apple, Facebook, Google, and Microsoft.
For more problems with antitrust regulation, see Wayne Crews’ and my paper, “The Case against Antitrust Law.”