Good afternoon. Thank you for the kind introduction and the opportunity to speak to you today. The tech, media and telecom industries on which this conference is focused compete in dynamic markets.
Today, I would like to discuss data and the role of antitrust enforcement in promoting competition in the myriad industries which make use of it. It is important in thinking about the appropriate role of antitrust enforcement to distinguish between competing for the market and competing within the market. What do I mean by this distinction? I recall the day several years ago when a few St. Albans high school graduates met with me to discuss their idea for a different kind of company. Rather than compete within the existing market for the distribution of news, those young entrepreneurs took aim at competing for a new type of market by launching a new type of subscription service, which they named The Capitol Forum. In other words, when I talk about competition for the market, I am talking about dynamic competition—competition to innovate and to provide consumers with new products and services.
Firms use data in numerous ways. Data can allow a firm to become more efficient by better allocating its employees’ time and other resources. And, data can provide important feedback and insights that allow a firm to solve problems and improve its products and offerings.
At the same time, the aggregation and commercial use of large quantities of data may give a firm a competitive advantage over its rivals, for example, if it uses the data to become a more efficient or effective competitor. Network effects may compound this effect if, as the data becomes more comprehensive or the platform gains more users, it becomes even more attractive to even more users.
Consumers have begun to realize the benefits of collecting, analyzing, and using large quantities of data—benefits not previously thought possible. When new technologies such as the printing press and the telephone were introduced, they vastly changed the speed and way in which we communicate. Today, the same type of quantum change is happening with data as our ability to use technology to collect, aggregate and manipulate data improves. Platforms and apps that couple voluminous data with machine learning are having a significant impact on almost every aspect of our lives. Everything from real-time traffic information, to targeted product recommendations, to improved travel booking experiences are now possible with more and higher quality data.
You may have noticed that I have used the term “data” rather than “big data.” Big data is an imprecise, catch-all term that describes a broad range of ideas related to the collection and commercial use of large quantities of information. The term “big data” is not only imprecise but affirmatively unhelpful to the extent it is used to imply that data is different from other assets and carries with it special obligations.
It is important to be precise when considering antitrust enforcement principles and to avoid general terms that may mean different things to different people depending on the circumstances.
Due, in part, to the belief that data is meaningfully different from other assets, concerns about how data should be treated vary. For example, some enforcers have expressed concerns that amassing data, and not sharing it with rivals, may be anticompetitive. Based on the belief that certain critical data could be a source of market power, some have expressed concern that a refusal to allow competitors access to that data could be anticompetitive. Those subscribing to this view believe that access to data may be essential to compete within the market—as it may allow potential competitors in a downstream market to overcome what otherwise might pose a significant barrier to entry—and conclude from this that making data available to competitors may be an appropriate remedy to ensure competition in the downstream market is robust.
In the United States, the antitrust agencies have had occasion to consider the competitive significance of data in the context of merger reviews. The Antitrust Division investigates any potential lessening of competition that may result from the acquisition of important data, either because the transaction combines substitutable datasets or because it transfers control of critical data on which the acquiring firm’s competitors depend and for which there are inadequate alternatives. If such an acquisition may substantially lessen competition in violation of Section 7 of the Clayton Act, divestiture of the data is appropriate, just as it would be for any other asset or critical input. For example, Thomson and Reuters were two of the three largest providers of financial data to institutions like investment banks and trading firms when they proposed to merge in 2008. The Antitrust Division concluded that the transaction would have led to higher prices and reduced innovation in this space, and so it required Thomson to divest copies of three financial datasets in order to proceed with the acquisition.
Outside of merger reviews, U.S. antitrust law generally does not impose a unilateral duty to share one’s assets with competitors. The Supreme Court has never recognized the so-called essential facilities doctrine, and U.S. courts have consistently expressed skepticism of the notion that firms have an overarching duty to deal with competitors.
Aspen Skiing, decided by the Supreme Court in 1985, is the leading case analyzing a unilateral refusal to deal claim. In Aspen Skiing, the Court acknowledged that although firms have no general duty to cooperate with rivals, that right is not “unqualified.” At issue was the defendant Ski Co.’s refusal to continue offering a joint ski lift ticket for both its mountains and its competitor’s mountain on the same terms as when the joint ticket had been developed in a more competitive environment. Evaluating that conduct under Section 2, the Court noted that Ski Co.’s refusal to deal with its rival at full retail price suggests it “was not motivated by efficiency concerns and that it was willing to sacrifice short-run benefits and consumer goodwill in exchange for a perceived long-run impact on its smaller rival.”
Decades since Aspen Skiing, courts have moved away from Section 2 liability for unilateral refusals to deal, an evolution that culminated in the Supreme Court’s 2004 Trinko decision. Writing for the Court, Justice Scalia made clear that Aspen Skiing is “at or near the outer boundary” of a Section 2 claim. The Court declined to find a duty to deal in Trinko, explaining that it is cautious in recognizing new exceptions to the general principle that a monopolist is ordinarily free to refuse to deal with its rivals. Since Trinko, valid unilateral refusal to deal claims have been very rare, and for good reason.
There are many reasons to be skeptical of using the antitrust laws to force the sharing of data.
First, forced sharing of critical assets reduces the incentive to invest in innovation. Assistant Attorney General Delrahim recently emphasized that “[n]ew inventions do not appear out of the ether, and excessive use of the antitrust laws … can overlook and undermine the magnitude of investment and risk inventors undertake.”
Data collection, storage, and analysis is not free and not always easily accomplished. Innovative companies—whether they are large tech firms or start-ups—can invest millions of dollars in developing programs to collect data, servers to store data, and computation programs and algorithms necessary to analyze data. It can be years before the firm realizes any type of profit or revenue from its investment, if ever.
What motivates a firm to invest in the development of leap-frog technology and innovation? Often, the potential to obtain monopoly profits serves as an important incentive to create better products for consumers. As the Supreme Court observed in Trinko: “The mere possession of monopoly power, and the concomitant charging of monopoly prices, is not only not unlawful; it is an important element of the free-market system. The opportunity to charge monopoly prices—at least for a short period—is what attracts ‘business acumen’ in the first place; it induces risk taking that produces innovation and economic growth.” And, that incentive to innovate persists even when a monopoly exists: firms can seek to displace that monopolist with a newer, better product.
One argument underlying the forced sharing of data is that it has the potential to promote competition within an existing market. Granting competitors ready access to necessary data certainly eases their ability to compete. However, it does so at the cost of undermining future incentives to invest in innovation aimed at competing for the market rather than competing within the market. If a firm’s ability to recoup the cost of its investment is diminished, it has less incentive to make that investment. Moreover, if businesses know that they can easily gain access to the fruits of other firms’ investments, then they too have less incentive to innovate on their own, and instead have a greater incentive to free ride on the efforts of their competitors.
If we stretch antitrust law to create competition within the market, we risk undermining the incentive to compete for the market. It is exactly this incentive that leads firms to create newer, better products. In his book, Capitalism, Socialism, and Democracy, economist Joseph Schumpeter observed that high profits serve as “baits that lure capital on to untried trails,” thereby producing a “perennial gale of creative destruction” that results in newer, better products and services. Mandating access to data is just as (or perhaps more) likely to result in less of this type of innovation as it is to enable new competition within existing markets.
A second reason to be skeptical of forced sharing is that it is an inherently regulatory approach. The Supreme Court, sixty years ago, described the Sherman Act as “a comprehensive charter of economic liberty aimed at preserving free and unfettered competition as the rule of trade.” Assistant Attorney General Delrahim also expressed concerns about deploying antitrust as a regulatory regime. He said “Antitrust is law enforcement, it’s not regulation. At its best, it supports reducing regulation, by encouraging competitive markets that, as a result, require less government intervention.”
Justice Scalia expressed the same concern in Trinko, noting that “[e]nforced sharing also requires antitrust courts to act as central planners,” because a court that imposes a duty to deal also has to set the terms on which the parties must share. Writing for the Court, Justice Scalia recognized that setting those terms “will ordinarily require continuing [court] supervision of a highly detailed decree.” Justice Scalia had reservations about courts playing that role. As he put it, “No court should impose a duty to deal that it cannot explain or adequately and reasonably supervise. The problem should be deemed irremedia[ble] by antitrust law when compulsory access requires the court to assume the day-to-day controls characteristic of a regulatory agency.”
While there are many reasons to be skeptical of forced sharing, there may be narrow circumstances in which a unilateral refusal to deal fits within the framework articulated by the Supreme Court in Aspen Skiing and Trinko.
For example, the FTC has taken the position that manufacturers of branded prescription drugs should make samples of their drugs available to firms seeking to develop corresponding generics given that Congress created a statutory process mandating that firms test their generic formulation against the corresponding branded drug to obtain FDA approval.
The FTC argued that these cases fit within the limited circumstances requiring access in Aspen Skiing but absent in Trinko. Requiring sharing in that circumstance would not undermine incentives to invest in competition for the market, but rather would facilitate compliance with the mandates created by Congress in the Hatch-Waxman Act.
Some have said that we need new tools to address these new data issues. Advocates for new tools tend to cite network effects, and argue that the winner-take-all nature of digital markets and the existence of tipping points mean that the typical means of assessing market power are ineffective. The presence of network effects may indeed change the competitive landscape, but markets subject to network effects are not inherently less competitive: economic theory shows that—although network effects may provide firms with significant competitive advantages within the market—markets with network effects can produce intense competition among firms competing for the market. In other words, firms compete intensely to become the solution most people choose. And, it is the promise of the rewards that come with that position that creates an incentive for firms to take outsized risks to invest in the development of innovative products and services.
Existing antitrust tools have been adequate to address these issues in the past, and they are adequate now too. The antitrust agencies have been analyzing network effects and winner take all or most markets for some time. While the existence of network effects is clearly relevant to an antitrust analysis, it does not prevent the use of the existing antitrust framework. It is, rather, a factual situation to which the existing antitrust framework must be applied.
The Antitrust Division applied this analysis over twenty years ago in its case against Microsoft. More recently, in the context of reviewing mergers, the Antitrust Division succeeded in prosecuting Bazaarvoice’s acquisition of competing ratings and reviews platform PowerReviews as a violation of Section 7 of Clayton Act, in part by rebutting the defendant’s claim that entry into the market was easy. The court found that the facts showed that network effects in the syndication of ratings and reviews were an important barrier preventing rapid entry by new competitors.
The Antitrust Division reached the opposite conclusion in deciding not to challenge Expedia’s acquisition of Orbitz. When Assistant Attorney General Bill Baer announced the closing of the investigation, he cited evidence that “the online travel business is rapidly evolving,” and new competitors offering similar functionality had already entered the market.
None of these cases required a change to the existing antitrust framework.
Sound antitrust law enforcement, not regulation, is generally best suited to preserving the incentive to compete for the market. Where benefits to sharing exist, they can be best captured by the parties negotiating in a free and competitive market, not by government regulation.