In early 2005, I was part of an International Law Institute delegation sent to Uzbekistan to teach government officials in emerging market economies about competition policy and regulatory economics. I’m sure I learned more from them than they from me — most but not all of which is a topic for another day.

During one session, I was talking about the Microsoft antitrust case, which had been settled just a few years before. I mentioned that Microsoft was pretty much a monopoly because of enormous scale economies (the first copy of a new version of Windows might cost billions to create, while subsequent copies cost nothing), network externalities (the more users the network has, the more it is worth to each) and lock-in (reluctance to switch if applications and files aren’t portable).

One of the officials in the audience challenged my assertion that Microsoft was a monopoly. When I asked why, he responded that it can’t be a monopoly, because if it were, it would have been regulated. That Uzbek official’s question, in essence, lies behind the growing controversy regarding what to do about “big tech.” Concerns are bipartisan. Democrats in the House recently weighed in with a massive report, while on a separate track (and for very different reasons) the Trump administration tried to repeal statutory protections of Internet platforms from liability for users’ content. Meanwhile, following somewhat similar efforts in the European Union, U.S. antitrust agencies have filed separate cases against Google and Facebook.

These cases are hardly slam dunks (for critiques, see here, here and here). By the same token, it is far from obvious what policy course would best to address a host of policymakers’ concerns. The problem really goes back to the question from the Uzbek official: why do we have unregulated monopolies, and what should we do about them?

How We Deal With Monopoly

One of the generic sources of the failure of market to yield efficient outcomes, appreciated since Adam Smith wrote The Wealth of Nations nearly a quarter of a millennium ago, is inadequate competition — monopoly being the extreme case. Without the fear of being undercut by rivals, a business can profit by restricting production and raising prices to a level above costs. Economists measuring the damage done by monopoly have focused on the reduced output, while consumers focus on the “distributive” impact — the price-gouging aspect.

The U.S. has had two policies to deal with monopoly for over a century. The first, antitrust law, applies when competition is feasible. Where it is, antitrust makes it illegal to subvert it. The most egregious violations occur when rivals agree not to undercut each other’s prices or to divide up the buyers among themselves rather than competing for them. Maintaining competition is also the goal for laws preventing mergers among substantial rivals and — somewhat more controversially — preventing actions taken by individual firms that keep other firms from succeeding but that have no competitive merit on their own.

However, competition is not always feasible. The textbook exception is the utility sector — electricity, telephones (in the ancient days of phones connected by physical networks of wires), natural gas distribution and letter delivery. Those businesses are what economists call “natural monopolies,” meaning essentially that the fixed costs of providing the service are so great that stable competition among multiple providers is not viable. It would be hopelessly inefficient for two electric companies to install substations, transformers, poles and wires in the same city.

When competition is not practical, control of monopoly power focuses on limiting the prices the monopolist can charge. The traditional method for regulating prices has been to base them on a calculation of the average cost of providing the service — that is, the total cost divided by output. Any higher price reduces sales and gives utilities more revenue than they need to make it possible to sustain the needed private investment. A lower price would discourage the provision of the service.

Back to the Uzbek Official’s Question

Policy to control market power, then, seems straightforward at first glance: antitrust if competition is feasible, price regulation if it is not. In that light, the Uzbek official’s inference that Microsoft must not be a natural monopoly was reasonable. If it were, it would be regulated (or, at least, the public would be howling for regulation). But, in fact, back in 2005, Microsoft was largely insulated from competition in its core business and had considerable power to use pricing to increase profits.

What gives? The fallacy in the simple framework above is that all monopolies could be regulated to achieve more efficient outcomes. This is not so. Setting a regulated price takes time, sometimes years. Regulators need to gather data on the costs of providing the service and how much of it would be purchased at a given price. This process is never easy, but to work at all, costs and demand have to be stable and predictable while the price is being set.

Boring traditional utilities fit the category. However, rapid technological change — the defining hallmark of “big tech” — makes hash of the process. Rapid ongoing innovation is constantly redefining products, upending costs and thwarting accurate predictions of demand.

Think again about Microsoft. By the time one might figure out costs and demand for the Windows 95 operating system, Windows 7 would replace it, followed by Windows 8 and 10. On top of that, the control over profits endemic to regulation might well stifle incentives to bring about these innovations.

This brings us back to the current tech giants facing antitrust action here and abroad — Google and Facebook most notably, but perhaps Amazon and Apple as well. The market conditions for all of these differ, including the credibility of assertions that they are, in fact, natural monopolies. But if they are, the challenge remains: what to do about them. There may not be significant market competition for antitrust to protect. And the unpredictable pace of technological innovation makes rational price regulation a nightmare.

Fix Number One: Changing Antitrust

A first step to filling the gap might be to amend antitrust laws to take account of natural monopoly status in these high-tech sectors. This begins with the realization that the competition in need of protection is not competition in the market, but competition for the market. In other words, the concern is to change the law to prevent an established business from preventing the new kid on the block from succeeding. An example might be if MySpace (yes, that MySpace) had been able to stop Facebook from developing its social network.

One way MySpace might have protected its market power would have been to buy the competitor before it could threaten the established business. But objections to mergers based on the potential competition argument are normally hard to sustain as a matter of economics or law, largely because it is hard to show that the firm not yet in the market presents a unique challenge to the established firm’s competitive position. One possible fix would be to reduce the burden of proof on opponents of a merger when there is good reason to believe that the market is a natural monopoly. This wouldn’t necessarily require legislation; much of antitrust practice is defined by court decisions and prosecutorial policy.

This approach might have been applied to Facebook’s acquisitions of Instagram and WhatsApp. Perhaps the Federal Trade Commission would have been willing and able to block those deals had antitrust practice allowed a lower burden of proof to protect competition in social networks, which have inherently strong natural monopoly characteristics.

Whether this approach should be extended after the fact — to undoing completed mergers — is assessed below, but in my view, the short answer is no. If approved mergers can be found to violate the law after the merged firms become successful, merger approval is meaningless, at best, and an incentive to keep firms from becoming successful, at worst.

The second anticompetitive tactic to deter is monopolization of access to an input or complement that a new firm would need to displace the established firm. For example, the key claim in the FTC’s 2009 complaint against Intelwas that it denied AMD, Intel’s chief competitor in computer processing chips, access to placement in personal computers through contracts that limited how many AMD chips each computer manufacturer could buy. I have argued elsewhere (for example, here and here) that the fundamental concern in any antitrust case involving what is called “exclusion” or “foreclosure” is tying up such access.

Considerations of this sort are at the core of the Justice Department’s case against Google, which alleges that competition in online search is impeded by contracts requiring preinstallation and preferential display of Google’s search engine in new computers. (For the record, this argument differs from an antitrust approach to Google that I proposed in a 2019 article in the Milken Institute Review.) Along with making monopolization of an input or complement needed by rivals a key to any exclusion cases, perhaps the burden of proving harm should be lower when the target firm is the only source of potential competition.

Fix Number 2: Broadening Regulation

As already noted, price regulation in markets caught up in rapid technological change is not practical. But before seeing what else might be done to contain monopoly power in this context, it is important to understand the fundamental difference between regulation and antitrust. When I worked for the DOJ’s Antitrust Division (from 1978 to 1986), the lawyers vociferously objected to claims that they were a regulatory agency rather than enforcers of the law. To this economist, the logic behind this objection was obscure, since one could certainly view antitrust as regulation of which business practices affecting competition were permissible.

However, there is a fundamental difference between the two. Under the law enforcement model of antitrust, the government cannot restructure a firm or enjoin a practice until it shows that the offending firm has violated the law. Merger enforcement is slightly different in that it is prospective. But even here one has to provide specific evidence that the merger violates the “may tend” to be anti-competitive standard of the Clayton Act.

On the other hand, regulating the prices a firm charges does not require a finding that the prices were illegal. Rather, Congress can give a government agency the authority to oversee a particular firm’s pricing and practices. Such regulatory authority may be controversial, to be sure. One need only look at the off-again, on-again, off-again, maybe on-again struggle over net neutrality rules — rules that require internet service providers to treat all content suppliers equally — at the Federal Communications Commission.

Regulators need to base decisions on a record, but violations of the law are not necessary. In fact, to the extent there is a law prohibiting a particular type of conduct, regulation to prevent it is less necessary.

But if price regulation for high-tech natural monopolies is out of the question, what else might be done? One option might be to limit the ability of the firm to expand the services it provides. For example, some have proposed that Amazon and Google be prohibited from selling their own brands, claiming they unfairly exploit information about third-party sales on their digital platforms. Guidelines on vertical merger enforcement recently issued by the Department of Justice and the FTC briefly discuss this concern. However, diversification can provide numerous benefits to consumers, so this may not merit a presumptive ban.

One regulatory approach that may be practical is breaking up a company to improve competition, even if it has done nothing wrong. The FTC’s case against Facebook is to the point, focusing on the problematic consequences of Facebook’s earlier approved acquisitions of Instagram and WhatsApp. That esteemed source of policy commentary, the Onion, noted in its list of things to know about the Facebook case, that the FTC “condemns Facebook’s monopolistic behavior like buying Instagram with FTC approval.” If illegal acts are a prerequisite for an antitrust case, labeling an approved merger illegal after the fact flies in the face of consistency.

Under a regulatory approach, however, a finding of prior illegality would not be necessary. Instead, a regulator could ask whether breaking up a legally formed company would promote the economic welfare of buyers or, depending on wording of a relevant law, be in the public interest. This is intrinsically no different than regulating the price a power company charges for electricity without showing beforehand that the firm’s price had somehow violated a law. In effect, regulation would open the door to “no-fault” antitrust — imposing a remedy without showing that a law had been violated.

But Is a Fix Practical?

We’ve identified a yawning gap between antitrust and regulation — circumstances in which competition is for control of the market rather than for profits within the market — and proposed two fixes: reducing the burden of proof under the antitrust laws, and allowing no-fault breakups. For either of these, the premise is that one can verify that these firms face no ongoing competition. This is far from obvious, as can be seen by looking at the leading big tech enterprises.

Google. One can easily choose alternative search engines (Bing, DuckDuckGo, etc.). Moreover, Google monetizes its search performance by selling advertising. And it appears to face competition in advertising both from online outlets like Facebook along with dozens of high-traffic websites and more traditional media like TV and radio.

Amazon. One can shop online from other general outlets (Target, Walmart) as well as stores specializing in particular products. Amazon may have created a competitive advantage in bundling free shipping with its Prime Video service, but that should be regarded as benefiting rather than harming consumers.

Apple. Recent complaints against Apple focus on the commissions it charges to vendors to use its application store. Under U.S. antitrust law, a firm (even a monopoly) faces few limits on the prices it can charge. Regulation of Apple’s commission may serve the public interest, but this doesn’t solve the problem of determining what the right commission would be. The alternative would be to defuse the issue by barring Apple from offering services in which it competes with the apps it sells. But note the giant problem here: this could lead to Apple monopolies if it simply refused to make apps for competing services available at its store

Facebook. Because of network externalities — that is, the spillover of benefits as people and organizations interconnect on the same social network — Facebook may be the clearest candidate for a high-tech monopoly to regulate. Such regulation could include divestiture of social network alternatives that have developed into prominent potential competitors, such as Instagram. Another option, adopted in the EU, is “data portability,” which ideally would allow a person on one social network to interact with Facebook subscribers and vice versa. Even here, however, caution is warranted. Forced divestiture would discourage the technical and marketing investments responsible for their success, and it is not clear that data portability can be defined in a helpful way.

• • •

Antitrust is designed to preserve ongoing competition, while price regulation works only for natural monopolies where the technology, product characteristics and demand are stable. As I’ve discussed, the big tech monopolies fall through the cracks. We can patch the cracks in ways alluded to above. None of this will be easy. Making the case that any firm is a candidate for expanded antitrust or regulation will require care. But a prerequisite for making that case is understanding why current frameworks have this gap.