A new paper from Erik Hovenkamp outlines pitfalls contained in newly proposed antitrust reform legislation that targets Big Tech companies. He proposes remedies and explains the danger of not getting it right.


Q: For readers who are unfamiliar, can you briefly describe the two bills and what they would do?

The American Innovation and Choice Online Act (AICOA) focuses on large digital platforms that are vertically integrated into adjacent product markets, which means the platform competes against other firms who rely on the platform to make sales. For example, Amazon sells its own batteries, so it competes with other battery sellers on its own storefront. The antitrust concern is that a company could use its control over a major platform to skew the playing field and impair rivals’ ability to compete. AICOA would prohibit most forms of “self-preferencing” by large platforms, which is where the platform gives more favorable treatment to its own product than to the competition. For example, Google could manipulate its Search algorithm so that Gmail shows up as the first result whenever a user searches for “email services.” AICOA’s text is vague, so “self-preferencing” could potentially apply to just about any actions that treat the platform’s own products differently than those of competitors.

The Open App Markets Act (OAMA) is limited to mobile operating systems and apps stores. Like AICOA, it prohibits self-preferencing—for example, it could get Apple or Google into hot water for listing their own apps first in search results on their app stores. But OAMA would also eliminate the “walled garden” model under which all app transactions must run through the official app store associated with the operating system on your device.

Q: What do you perceive as the chief issues with the two bills?

Right now, antitrust law makes it unreasonably difficult to challenge unilateral conduct by dominant firms, especially situations where they refuse to serve competitors for no good reason. In that sense, both bills would help to prohibit certain forms of anticompetitive conduct that are not adequately addressed by current law. That’s a good thing.

The problem is that the bills don’t do a good job of limiting antitrust scrutiny to cases that plausibly threaten competition. They create a hazy morass of vague rules under which almost anything is potentially actionable. As a result, the only clear-cut way for a platform to avoid legal scrutiny may be to stop selling its own products. But that would require us to sacrifice a lot of competition and innovation.

The problems are easiest to see when you consider the differences between antitrust and regulation. Antitrust is administered by the court system, not a specialized regulatory authority. That makes it slower and more expensive, and it also means that the ultimate decision makers (generalist judges) lack expertise in the relevant field of commerce. Because of these disadvantages, antitrust takes a more limited, passive approach. It does not attempt to micromanage firms to maximize efficiency. Instead, it just prohibits behaviors that are distinctly anticompetitive.

“If any antitrust bill passes, it will immediately become the face of antitrust reform. That’s why it’s important to get it right.”

The bills have lost sight of these differences and try to do too much. Suppose a platform’s own product appears first in the search results, but that the placement of competing products has not otherwise been degraded. This may or may not signal that the platform manipulated its algorithm to force its own product into the top spot. These are complicated algorithms that account for many different factors, so this is unlikely to be cut and dry. Under the proposed bills, a court would have to go through the algorithm in detail to determine whether it ranked the products in an unreasonable manner. Each side would introduce its own experts to argue why the algorithm is or isn’t reasonable. But the judge, lacking any expertise in search algorithms, may well have a hard time evaluating these arguments. The litigation process would likely last years and cost tens of millions of dollars.

It’s not unusual for antitrust cases to be complex and fact-intensive in this way. But we don’t usually go through these difficulties unless we have reason to believe that the defendant’s conduct might pose a serious threat to competition. However, absent specific evidence to that effect, we have little reason to assume it in cases alleging such mild acts of self-preferencing. This is not a situation where the platform is accused of banishing rivals from its storefront or burying their products where they’re hard for consumers to find; actions like that could be highly anticompetitive and the law should absolutely start taking that threat seriously. But merely listing the platform’s own product first does not make it difficult for consumers to identify or access competing products. It’s more similar to advertising than to anticompetitive exclusion. It could certainly help to capture some consumers who were initially indifferent, but it doesn’t materially restrain consumer choice or information.

One might respond that stronger forms of self-preferencing really can be anticompetitive. I agree, but the point is that this danger must be demonstrated, not merely assumed. Ordinarily in an exclusion case we ask the plaintiff to demonstrate two things to give credibility to its case: market power and some indicia of anticompetitive effects. The bills omit the first of these requirements in all cases, and they omit the second one in most cases. Specifically, they ask whether a platform is large, but not whether it has the power to exclude rivals in the specific product market at issue. And the majority of violations codified by the bills do not require plaintiffs to show that anticompetitive effects are likely. Instead, such effects are simply presumed by default.

I’m not sure whether it would be a good idea for a specialized regulatory authority to perform detailed oversight of the design of search algorithms or other platform services. But I am certain that it would be a mistake to try to use antitrust for that purpose. However, by eliminating the need to prove anticompetitive effects, these bills would do exactly that. As a result, whenever a large platform introduces a new product, it may face litigation threats for any perceived special treatment of its product, even if the potential impact on competition is likely negligible. This will discourage platforms from bringing new products to market, depriving the public of an important source of competition and innovation.

Q: In what ways would you change the bills to improve them?

I think market power analysis must be brought back into the picture as a way to screen out implausible complaints. The fact that a platform is large doesn’t imply that it has market power over every product that it sells. The power to exclude competitors does not necessarily require that the platform’s own product holds a dominant share of the relevant product market, but it does typically require that a sizeable portion of trade in that market run through the platform. However, this is not always the case. For example, if only 2% of coffee sales occur on Amazon, then it is hard to see how Amazon could hope to upend competition in the coffee market. This is an easy way to rule out weak cases without having to dig too far into the details.

Additionally, as in all other exclusion cases, plaintiffs should be required to make some showing that the defendant’s conduct will likely generate material anticompetitive effects. A foreclosure requirement would be the natural solution. This would ask about the extent to which rivals are effectively cut off from potential sales opportunities. This would allow courts to draw a line between minor forms of self-promotion that don’t threaten competition and more serious forms of discrimination that may have real anticompetitive effects. For instance, if a platform just puts its own product at the top of the page without otherwise degrading the visibility of rivals’ goods, then presumably there wouldn’t be significant foreclosure; but if it deliberately buries rivals deep down in the search results or excludes them altogether, there could easily be significant foreclosure.

I want to acknowledge that courts today are often too demanding when it comes to proof of anticompetitive effects and market power. So I’m not arguing that plaintiffs should necessarily face the same evidentiary standards that we impose under existing law. But ultimately they must have some meaningful obligation to demonstrate anticompetitive effects. This would still leave an important role for antitrust to play, and it would avoid the problems created by these bills.

Q: What are the broader implications of these bills on the antitrust landscape?

I think antitrust has become too laissez-faire and that courts are too dismissive of antitrust claims in general. So I share the popular view that antitrust would benefit from pro-enforcement reform. If any antitrust bill passes, it will immediately become the face of antitrust reform. That’s why it’s important to get it right—to create sensible reforms without committing a bunch of unforced errors. That would make folks comfortable extending the reforms to other areas beyond “Big Tech” (which accounts for only a tiny fraction of all antitrust cases). In that sense, I think these bills would be detrimental to the broader push for antitrust reform. They try to do too much and they make too many unwarranted assumptions, and consequently they would cause a lot of problems. And then everyone will say, “see, that’s why antitrust reform is a bad idea.”

Erik Hovenkamp holds fellowships in the Innovators Network Foundation and the Thurman Arnold Project. He previously held a Harvard postdoctoral fellowship sponsored in part by Qualcomm. The opinions expressed in his writing are entirely his own.

Learn more about our disclosure policy here.