Presidential Candidate Elizabeth Warren thinks Big Tech is too big and wants it—and, in particular, Amazon, Facebook and Google—broken up and their past mergers and acquisitions unwound. And the FTC recently announced it was forming a Task Force to look into the technology markets. There do seem to be issues with Big Tech. But is antitrust, as currently practiced, the best tool to address them? Ms. Warren contemplates this by suggesting a new regulatory regime should be implemented to control Big Tech. Should we treat platforms like a regulated utility? Should we pass a new antitrust law that supplements current common law and allows for more vigorous enforcement? Or are there tools available to modern antitrust that can address Big Tech and the issues Ms. Warren identifies? We ask those questions and suggest some high-level responses to further the dialogue.

Ms. Warren’s Views

Ms. Warren thinks Big Tech has too much power over our economy, society and democracy. She feels that Big Tech has used their power to crush small business and innovation, and to substitute their financial interests for the interests of the American people. In particular, she believes Big Tech has used mergers to squelch new and innovative technology that might have challenged their dominance. And she believes Big Tech is abusing their dominance in platforms to expand their power in new markets.

Her proposal is to regulate the platforms as utilities and require that they divest themselves of any interest in participants on those platforms. So, in the case of Google, for example, the law would force Google to cease to provide restaurant reviews that compete with the likes of Yelp. That would eliminate Google’s incentive to promote its reviews over Yelp’s. Under the proposed law, the utilities would be required to engage in fair, reasonable, and nondiscriminatory dealing with all participants. She says specifically:

Amazon Marketplace, Google’s ad exchange, and Google Search would be platform utilities under this law. Therefore, Amazon Marketplace and Basics, and Google’s ad exchange and businesses on the exchange would be split apart. Google Search would have to be spun off as well.

She also plans to sue to unwind consummated transactions: Amazon/Whole Foods; Amazon/Zappos; Facebook/What’sApp; Facebook/Instagram; Google/Waze; Google/Nest; Google/DoubleClick.

She believes that people need more control over their personal information, content creators need to receive more value for their products, and to limit the ability of hostiles like Russia to use the platforms to subvert our political process.

The Current Law

Monopolies gained through superior product, business acumen, or historic accident are not illegal. To be illegal, one would have to demonstrate that Amazon, Facebook and Google were in fact a monopoly and engaged in exclusionary or anticompetitive conduct. Big Tech is going to argue that they invented their modes of business, and to the extent they have any market power, it’s because of business acumen. They would also certainly argue that they have no market power. For example, Amazon’s pricing New York Times Best Sellers below cost—which some have called predatory—could be described as a consumer benefit. E-commerce displaced the inefficient independent bookstore, the “buggy whips” of the book business. Google’s algorithm does appear to prefer its products over competitors but could be described as consumer-welfare enhancing. By getting to know people better and what they mean when they use particular words (and what they buy), Google can tailor the search answers uniquely to that person making their experience of the Internet better. And Facebook keeps you in touch with friends from whom you might otherwise never see or hear.

Each of the mergers Ms. Warren describes were reviewed by the antitrust agencies. Under the law, the agencies asked whether the transaction substantially lessened competition or tended to create a monopoly. In a vertical merger, the main issue is foreclosure. Does one of the merging parties control some element of competition that a combined entity could use to disadvantage competitors. In horizontal mergers, does the combination allow the merged party to raise price or lower output. Ms. Warren appears to be arguing that the targets of Big Tech consistently play an outsized role in the competitiveness of the market by virtue of their maverick innovativeness, and that by acquiring them, Big Tech has attained or expanded their dominance. By unwinding these deals, presumably, that innovativeness and ultimately competition would be restored..

There are several issues with that approach. The first is that, barring a patent, Big Tech could have duplicated the functionality on its own site without the merger. This matters because Big Tech was already big. If the functionality is not protected by patent, Big Tech could have duplicated it. That means that the target really wasn’t that competitive in the first place, and likely, under the theory, could have been displaced by Big Tech easily. The acquisition was a “make or buy” decision which is usually afforded significant deference. The target’s technology, if unprotected, can still be duplicated by other startups. The second is that the people who came up with the idea have taken their money and left. These people might not return to the target post divestiture. There’s no way to know if a newly-liberated target would be the same competitive and innovative force as it was. Most of these targets were small at the time of acquisition. It would be difficult to know whether they would have become big had they had the opportunity. MySpace comes to mind. Courts would be reluctant to force a divestiture when the uncertainty of the value of the divestiture is so high. Lastly, courts are reluctant to unwind consummated deals, and, while there have been a few cases that discuss potential competition as part of a bigger ruling, none have been based on potential competition alone. So, in the case of Amazon/Whole Foods, Amazon was not a grocer of any merit before the deal, and Whole Foods did not have an ecommerce solution that sold trillions of products all over the world. There’s some verticality in that Amazon may favor Whole Foods products over others on Amazon’s platform, but the government would have to prove Amazon had market power in ecommerce and that the preference it might give itself would negatively impact other groceries that had no choice but to sell on Amazon’s sites.

These are tough cases to bring.

How to Accomplish These Ends

To constrain Big Tech, Ms. Warren suggests regulating them, perhaps by passing an FCC-enforced regime similar to, say, the Telecommunications Act of 1996. That legislation would be met by significant resistance by Big Tech but also more generally the free marketers who believe that regulation harms consumers by reducing incentives to innovate. She would also have to change substantive merger law and somehow get the rules to apply retroactively to these consummated transactions.

Another option would be to create a new antitrust law that forbids, for example, “undue concentrations of economic power.” “Undue concentration” would be something less than monopoly or market power, and would not exempt from challenge those concentrations achieved through superior product, business acumen or historic accident. An undue concentration could occur in a merger that takes a market from 5 to 4 participants (one that normally isn’t challenged now); that deprives a competitor of “fair, reasonable and nondiscriminatory” access to a “necessary” platform (“essential facilities,” a current antitrust doctrine of access is almost gone from the courts); or pursues other socio-economically harmful outcomes otherwise outside the reach of the current antitrust laws. The standard would contemplate more than just consumer welfare or economic efficiency. The new law could also apply to mergers that “substantially lessen innovation” to give enforcers tools to prevent the problem Ms. Warren sees in incumbent platforms buying up mavericks. This “new law” approach would leave it to the courts to interpret, however. They might ultimately limit the new law much as the modern courts have limited the Sherman and Clayton Acts. But by making a new law that expressly takes into account variables beyond just consumer welfare, for example, Congress might be able to direct the courts more than they did under the existing laws.

A New Theory of Predation

There may be avenues with existing law as well. One of the first things Amazon did when it was starting up was to price New York Times Best Sellers below cost. Most independent book stores make money off the higher priced best-sellers. They use that to fund sales of mid- and backlist books (and other services like author signings or Harry Potter nights) which have much smaller margins. Essentially this model is a form of windowing, the same as you see with movies. You satisfy demand with the most price insensitive at higher prices first. Once that demand is satisfied, you move to another format and the next group consumers who are less price insensitive. You keep repeating this process over time so that demand is satisfied at prices that are optimal for those particular consumer clusters. The temporal quality makes it unnecessary to know who is in each window, a difficult process in a media-based market like books and movies. By collapsing the windows, Amazon is satisfying demand suboptimally. People who would pay more aren’t.

This would seem like a benefit. If Amazon is willing to reduce its profit by selling books below cost, more power to them. The price insensitive consumer surely benefits. And there doesn’t appear to be any recoupment either. Amazon is still pricing below cost today. And entry barriers to bricks-and-mortar book stores are low. If Amazon did raise price, it could very well prompt meaningful entry. The market appears to be competitive.

One might argue, however, that Amazon is not discounting books to dominate the bricks-and-mortar book world (although they recently entered the bricks-and-mortar). What they are doing is drawing desirable consumers to their platform so Amazon can extract their data. In essence, the below-cost pricing draws highly profitable, price insensitive consumers to their platform so that Amazon can observe their shopping habits. These are valuable consumers because they spend more. The more things they buy over time, the more Amazon will know about them and what they like. In essence, by observing you in as many possible transactions, and other “situations,” as possible, they can create a fairly comprehensive understanding of what you want to buy, when and at what price. They observe you to perfect their understanding of the demand curves unique to you. They then sell access to those demand curves.

There are benefits beyond just knowing who you are. If one has enough observations of enough people, one should be able to predict accurately the demand of others even if those others rarely use the site. Say I’m very private about what I buy on Amazon such that they only have a couple of observed events for me, a couple sales of shoes and a book. Say my neighbor is not. He buys many things, reads on Kindle, buys at Whole Foods, listens to Prime Music and watches Prime television shows. [1] But he buys a couple of the same things I do. By virtue of my one or two events and my neighbor’s millions, Amazon now knows me. And I have no control over that. As a consequence, I cannot assess the value of my two interactions because I have no visibility into what they know about people like me. And my neighbor doesn’t know the value of his data because he doesn’t know how many others there are like him. Has he given Amazon a roadmap to a million consumers or a handful. This lack of visibility may also hide potential rents that might be occurring. Since consumers cannot value their data, they cannot judge whether the 50 percent discount they are getting on a Best Seller is more than the value of the data they are exchanging. Of course, it may be the case that those profits in data do not reflect monopoly rents at all but rather the result of an unfair bargain made with unsuspecting consumers. The high valuations, and profits, of these institutions suggest that they are reaping a considerable benefit from the data. The lack of visibility suggest that consumers are transferring more value than they can understand for their “free” or “discount” products.

This is where the predation in platform industries can occur. Platforms drive out rivals in data-feeding markets not to raise price to consumers in that market, but to acquire consumers they can observe and develop more and more comprehensive understandings of their particular demand curves and those of similar people. Platform predators recoup their losses in data-feeder markets by selling access to the demand curves they’ve discovered. This scenario also presages a strong network effect. The more users, the more data, the more desirable the platform is to sellers. The more sellers, the more products to sell, the more things to observe, the more desirable the platform is to consumers. There should therefore be a tipping point.[2]

The nature of the Internet also makes detection very difficult. In a bricks-and-mortar environment, you put a shoe on sale, everyone can see the sale. So, you have to wait until you’ve sold all the shoes you can at full price before you start discounting. Amazon can do it immediately. They know that Jack is price insensitive, so he will pay full price. They know that Bill is price sensitive, so they discount to Bill. But they can sell the same shoe to two different people at two different prices immediately, without waiting for demand to be satisfied in a particular window, because there’s no way Jack will see the discounted price. Amazon can present a unique storefront to anyone at any time. This effectively eliminates the need for time-based windowing in retail. You can completely saturate a market at varying price points immediately.[3]

And Amazon can sell this not only as a perk of their platform but as a marketing analytic tool. With this data, they can now predict an optimal price for a new product, how many products will sell, and who to sell them to.

Facebook and Google do this as well through advertising. What do you click through and buy. Facebook is valuable because they correlate those commercial events with a great deal more about you personally. Facebook might know that your grandmother just died and serve you an advertisement for flowers. They might also know you like Donald Trump, would be receptive to advertisements for fringe political candidates and be able to sell ads to those candidates. Amazon may be a little better at selling to manufacturers because they are selling the actual things themselves. Facebook may be better at selling your “psychological constellation” than your demand for a particular product since they aren’t actually selling the product. They could enter into data sharing agreements with their advertisers where the seller provides that transaction data back to Facebook.

A Potential Case Against Big Data

The antitrust agencies have not been particularly receptive to arguments that Amazon’s below-cost pricing is predatory, and seem to think that there can never be anti competitive predation in no-cost markets. The agencies may buy too readily the notion that consumers are benefiting from the lower prices or better advertising and more closely-tailored products. If consumers cannot know the value of the thing they are exchanging, they cannot know whether they’ve got a reasonable deal. The goal of “predacious” Big Tech is to keep competitors in data-feeder markets out of the market to retain access to as many consumers, and events, as possible. The recoupment occurs when they monetize the data they extract. This theory could also be used to explain why data sharing between platforms can be anti competitive. Platforms innovate their platforms to acquire more and more meaningful consumer observations. Sharing data between them reduces competition between them to draw users to the platforms and provide innovative environments in which to observe them.

It may also be a case ripe for a test of the unfairness doctrine under Section 5. Consumers cannot individually value their data in the hands of a particular platform because they don’t and cannot know what other data the platform has. An individual’s data could be worthless because they are very unique or highly valuable because there are many price insensitive consumers just like her. Such a case could expose the value of data, allow consumers some redress for their “Cambridge Analytica” problems, force Big Tech to be more forthcoming about and, with a sufficiently large fine, disincentivize them from exploiting the data informational asymmetry going forward.

The FTC might also want to initiate a 6(b) investigation into how Big Tech uses customer data to market to customers and to gain insight on how to value that data and whether consumers are being exploited. How do they use data; how much do they make from the data; how does one value the data individuals give up and in the collective; could there be a tipping point for big data; do data exchange agreements between big data harm competition. The FTC should include Bureau of Consumer Protection staff in their new Technology Task Force.

Conclusion

New regulatory legislation, a new antitrust statute, or bringing new cases using existing case law and theory, have upsides and downsides. Implementing these things can limit the ability of Big Tech to innovate or they could free innovation by limiting the ability of Big Tech to capture only those technologies that protect their dominance and to exploit consumers. It’s a conversation worth having.

The Gilded Age saw the birth of a tremendous disparity in wealth among our citizens. Antitrust was one means of restoring some economic balance to our society. Wealth disparity is now at the same levels if not greater than it was in the late 19th Century. It may be time to reconsider whether the contraction of antitrust into solely a question of economic efficiency was wise. The next President and Congress may very well do so.