Antitrust and big data

Here's a popular question over drinks after antitrust conferences these days: Does antitrust law apply to so-called "big data," and if so, how? (And yes, antitrust lawyers do enjoy the occasional beer or glass of wine.) Although there is certainly no express exemption for big data written into the Sherman Act, determining exactly how antitrust law (as opposed to, say, consumer protection law) applies has been far from trivial. We will explore some of the basic issues in this article.

What are big data?

We all have some intuitive idea of what big data are: data relating to consumers and their behavior and preferences, especially where the data sets are voluminous and valuable. The literature has generally identified four factors that indicate the presence of big data:

Volume: Big data obviously need to be big in the sense that there are a lot of data (lots of bits and bytes).

Velocity: Big data are typically thought to be generated, used and analyzed at high speed. As a result, even modestly aged data may be too stale to be useful and may not constitute big data. Think traffic congestion data for yesterday's commute.

Variety: Big data typically feature a number of data fields (e.g., consumer physical location, consumer product preferences, consumer browsing history, etc.). And big data are typically collected from a number of sources, e.g., human online interactions, social medial posts or texts, as well as from Internet-connected devices that report consumer location or other activity.

Value: Big data have substantial value to their gatherers, including in connection with machine algorithms or artificial intelligence that can use and exploit the data.

Big data, it has been said, can lead to a "virtuous circle" or "virtuous loop" whereby as a company (and particularly a network platform company) expands, it obtains more and more user data that may enable it to compete more efficiently -- which may enable it to gather and use even more data. (A network platform company is a company that provides a platform linking multiple users, typically online.)

Gating questions

So how should we begin to answer the questions of if and how to apply antitrust law to big data? We can start by going back to basics. One gating question may be whether big data are really necessary to compete or enter a market, or whether a firm can still enter or expand by offering a better product, a better algorithm, a better user experience, etc. In many cases, it may be possible for rivals to compete without access to a particular data set.

Another question is: If we impose antitrust requirements on big data "amassing" firms, will we actually help consumers? And relatedly: If we force the sharing of big data, will we reduce the incentive to innovate and thus end up harming consumers anyway?

Finally, are antitrust law and consumer privacy law in harmony? For example, to the extent that consumer privacy law moves towards requiring consent for the use of consumer data, it may make data access and sharing more difficult -- which could make competition and market entry more difficult. On the other hand, antitrust remedies focused on the sharing of big data could raise consumer privacy concerns, especially if consumers do not have an opportunity to meaningfully consent to the sharing.

Potential competition law concerns with big data

So how can big data pose a risk to competition? Perhaps the fundamental concern with big data is that rival companies may not be able to compete with a "virtuous circle" unless they gain access to the data.

In the U.S., historically -- at least since the 1980s -- antitrust law has been grounded in the consumer welfare standard. Under that standard, a merger is not anticompetitive unless it will harm or at least threatens to harm consumer welfare. Non-merger business conduct (e.g., a refusal to share data) also is not anticompetitive unless it will harm consumer welfare. Consumer welfare generally is measured by prices. The lower the prices, the higher the consumer welfare. Sometimes, however, antitrust law also considers product quality as an element or dimension of consumer welfare.

One question that is now very much under discussion is whether amassing big data can be fit within the rubric of antitrust consumer welfare. A consumer's interest (or lack thereof) in his or her data is not, strictly speaking, a price-related factor -- especially for many of today's platforms that offer services to consumers for free (and make their money elsewhere, e.g., through advertising). Alternatively, should the antitrust laws' reach be expanded to reach other concerns -- for example, consumer privacy -- as a species of consumer welfare?

A complication with the consumer welfare analysis is that it can as a practical matter be difficult to value consumers' interest in big data sets -- which presumably often does involve their interest in privacy. Although consumers often say that they are concerned about privacy, in practice they are often very lax about sharing their data or information, and often are not willing to pay for privacy. Valuing privacy as a species of consumer welfare may therefore be difficult.

The number of big data/antitrust cases to date has been limited. The Federal Trade Commission considered some issues in its 2018 CDK/Auto-Mate merger case, as did the Department of Justice in Cox Automotive/Dealertrack (2015) and Bazaarvoice/PowerReviews (2014). In the Cox Automotive merger case, for example, the DOJ required Cox to divest a business unit that provided analytics and algorithms to car dealers and required Cox to enable the continuing exchange of data and content between its websites and the divested business. And in Bazaarvoice, the DOJ successfully challenged a merger based on the market power of the merged entities' platforms/data holdings (online consumer reviews). But there is as yet no body of definitive guidance in this area (for example, Cox Automotive and Bazaarvoice did not consider a unilateral refusal to share data with rivals).

Big data sold to third parties vs big data used by an incumbent firm

Let us stipulate, however, that antitrust law applies to big data sets. When we are talking about big data and antitrust, as I already alluded, we may be talking about mergers, or about business conduct. In either scenario, we may want to distinguish between firms that sell big data to third parties and firms that amass it for use in their own businesses.

There is probably a fairly broad consensus that it is more straightforward to apply antitrust law to businesses that sell data sets. If two such businesses merge, they may have more market power in the markets for their data. Or, if a dominant firm that sells big data decides to stop supplying it to a competitor, there may be a fairly standard refusal-to-deal claim.

We are in at least somewhat more controversial waters when we start talking about applying antitrust to the unilateral conduct of firms that do not sell big data but instead amass it and use it themselves. Can such data be an "essential facility" that must be provided on some sort of fair and reasonable terms to a company's rivals? Alternatively, can a firm be guilty of providing discriminatory access to its data (charging some rivals more than others)? If so, what metrics should we apply to determine whether the price discrimination is reasonable or not? And what about a firm that ties access to the data to some other product or service that it provides but that consumers would not want to buy if they had a free choice? Could this tying harm competition in the market for the tied product? For the most part, these questions have not yet been specifically answered by the courts.

Conclusion

There is little question that big data can be considered as part of antitrust analysis. However, it will take some more cases to answer the further question of how the analysis should take into account issues such as consumer privacy, consumer welfare, and the interplay between antitrust and consumer protection law. 