Herbert Hovenkamp writes that the court presiding over the Google Ad Tech case gave the government an important win. However, by relying on the per se tying rule instead of rule of reason, the court perpetuated a flawed court precedent that can preclude serious market analysis for competitive harms.
The Google Ad Tech decision is an important antitrust victory for the government. This article focuses on the court’s important conclusion that Google’s strategy to tie its publisher ad server (DFP) to its ad exchange (AdX) was unlawful “per se.” The publisher ad server alerts users to the availability of ads and, in close to real time, facilitates “targeting” them to a publisher’s website. The ad exchange is a digital auction market in which publishers and advertisers can purchase and sell ads. Google’s design of its software effectively prohibited publishers from receiving real-time bids from AdX unless they also used DFP. The presiding court found this to violate both Sections 1 and 2 of the Sherman Act.
The antitrust law on tying addresses situations in which a firm refuses to sell one good unless the buyer also takes a second good. In the language of tying, Google’s AdX is the tying product and DFP is the tied product. Three antitrust provisions address tying. Most explicitly, Section 3 of the Clayton Act identifies them as anticompetitive agreements requiring a firm not to deal with competitors. Ties can also be a “restraint of trade” under the more general language of Section 1 of the Sherman Act, or in some cases monopolizing conduct under Section 2 of the Sherman Act.
Section 3 of the Clayton Act prohibits ties only if their effect “may be to substantially lessen competition or tend to create a monopoly.” That variation of the rule of reason requires that anticompetitive effects be assessed. Section 3, as originally proposed in 1914, would have condemned ties as illegal per se, without regard to competitive effects. But even in 1914, several members of Congress protested that tying sometimes served procompetitive purposes. Thus, the final version of the Clayton Act condemned them only if they met competitive harm requirements. Congress has also addressed tying arrangements in a provision of the Patent Act. It provides that tying of a patented product is unlawful only if the patentee has market power.
Thus the only time Congress has spoken on antitrust tying, it has required use of the rule of reason. Tying arrangements link complements and have a significant capacity to make products work better or reduce their costs. As such, they are never suitable for per se treatment. Tying law’s per se rule, which applies only to actions brought under Section 1 of the Sherman Act, is a perverse Supreme Court flip of Sherman and Clayton Act standards. Although the Clayton Act was intended to provide more aggressive standards of illegality than the Sherman Act, in this case Section 1 of the Sherman Act is more aggressive by far. Today, tying remains the only vertical agreement that can be unlawful per se, but only under the Sherman Act, which the court invoked in its decision.
While Congress has never supported a per se tying rule, the Supreme Court created its version of it in its 1947 International Salt decision, where it concluded that “it is unreasonable, per se, to foreclose competitors from any substantial market….” Never mind that the tied product was ordinary rock salt and the defendant’s salt-injecting machine accounted for a miniscule share of it. Instead of requiring an actual market impact, which a rule of reason would do, the Supreme Court required only that the “volume of business” covered by the tie not be “insignificant or insubstantial.” Two years later Justice Felix Frankfurter threw in some support by stating without citation in an exclusive dealing case that ties “serve hardly any purpose beyond the suppression of competition.” Thus was born a per se rule against ties, which the Google Ad Tech decision applied. Ironically, Google Ad Tech was a case that the government would very likely have won by showing actual competitive harm under the rule of reason. That seems clear from the court’s consideration of the same claims under Section 2 of the Sherman Act, which required consideration of offsetting benefits.
International Salt also held that if the tying product was patented, market power should be presumed. It need not be separately proven. The Supreme Court overruled that presumption in 2006 but did not change tying’s per se rule. Nevertheless, tying law represents “a most peculiar per se rule,” as one decision put it. Unlike all other per se rules (e.g., price-fixing), it requires proof of market power. However, if a firm ties separate products together and has market power in the tying product, then its tie can be condemned without a serious judicial inquiry into the tie’s impact on competition.
The per se rule for naked horizontal practices such as price fixing rests on a policy judgment that detailed inquiry into competitive harm is unnecessary because it falls within Justice Frankfurter’s category of practices that “serve hardly any purpose” beyond suppressing competition. That makes sense for a practice such as naked collusion. However, as the Supreme Court stated when it overruled previous decisions declaring vertical price maintenance unlawful per se, “it cannot be stated with any degree of confidence that resale price maintenance always or almost always tend[s] to restrict competition and decrease output.” The same thing is true of tying. The Supreme Court has made clear that sometimes tying can also be assessed under the rule of reason. In fact, all tying arrangements should be treated in that way.
Another problem with the per se rule for tying in the Google Ad Tech case is that the tying in question seems to be a unilateral act. Historically, a tying arrangement was an agreement. However, important alternative mechanisms are so-called technological ties, or “tech ties,” as well as tying arrangements that are created by digital code. The court acknowledged that in this case Google’s tie was “unilaterally imposed.” In the same opinion, when addressing the duty to deal, the court acknowledged that the refusal in this case was really a tying arrangement. This made it a “secondary” duty to deal in the tied product market, and subject to Section 2 of the Sherman Act.
Tying achieved unilaterally by product design or code raises the same competitive concerns as ties imposed by agreement. In digital commerce tying-by-code is pervasive because practically anything that can be accomplished by an agreement can also be accomplished “unilaterally” through code. For example, users of the iPhone are required to make software purchases through the App Store, not because they have signed an agreement, but because Apple’s software prevents it. But tying law’s per se rule is available only under Section 1 of the Sherman Act, which requires an agreement. Following that reasoning, the 2001 Microsoft decision addressed Microsoft’s unilateral “commingling” of operating system and browser code into a single software product as a violation of Section 2, not Section 1. By contrast, the 2023 Epic Games decision was willing to address Apple’s design of its operating system so as to exclude third-party payees as both an agreement and unilateral conduct. However, that decision refused to apply the per se rule. While a contractual tie is an agreement, a unilaterally imposed tie, or tech tie, is a presumptively unilateral design feature. Addressing these under the per se rule moves dangerously in the direction of saying that a firm’s unilateral design choice ought to be treated as unlawful per se.
In one sense, of course, every product redesign involves an “agreement”—namely, the buyer’s agreement to buy the product, which appears to be what Epic Games had in mind. But stretching that into a Section 1 violation obliterates the distinction between the two statutes. Product sales are necessarily agreements, but they do not turn every monopolization claim into a conspiracy simply because the monopolized product must be sold to a customer. In any event, neither precedent nor logic places unilaterally imposed ties under the per se rule. The best way to deal with unilateral ties is to address them under the rule of reason as it applies to dominant firm practices.
Citing the highly innovative character of software, the Microsoft decision also held that the integration of additional functions into a computer operating system should be addressed under the rule of reason, even when considered as tying. To apply the per se rule in such cases, the court held, “creates undue risks of error and of deterring welfare-enhancing innovation.” This was particularly important where “the tied good [was] physically and technologically integrated with the tying good.” Since then, this particular rule of reason for software ties has been applied by other decisions involving Epic Games and Amazon.
The rule of reason for technological integration by code enables courts to avoid the worst feature of excessive per se rules, which is that they prevent courts from clearheaded analysis of the competitive consequences of product design. In computers, and particularly in software, innovation continuously involves putting products together that had previously been independent. Many of them harm rivals precisely because they are desirable to consumers. Famously, Apple’s placement of a digital camera on the iPhone in 2007 led to an 84% decline in sales of dedicated digital cameras. Its success was explained by Chase Jarvis’ adage that “the best camera is the one you have with you.” An antitrust action breaking up the smartphone-camera “tie” would be met with consumer revolt, even though it would benefit the makers of dedicated cameras.
One important difference between the per se rule and the rule of reason is that the latter permits a fulsome investigation of the relative benefits and harms that result from product or software designs that combine multiple functions. It is not tying law’s purpose to drive computer users back to the days of MS-DOS, when computer operating systems did no more than provide user access to the processor and data, and all programs had to be added separately.
Nevertheless, the Google court explicitly applied the per se rule, finding that DFP and AdX were “separate products” that were “tied together” by virtue of product design. The court also recited evidence that the tie “reinforced the exclusionary effect” of Google’s products. Under tying law’s per se rule, however, proof of anticompetitive market exclusion is not even relevant. All the government needed to show was that a “not insubstantial” amount of commerce was covered by the tie. That was easy in this case, which involved billions of dollars in covered transactions. The International Salt case mentioned above had found $500,000 sufficient to meet that requirement. The Google opinion also cited a more recent decision concluding that $200,000 was sufficient. Even a relatively trivial tie could meet these requirements.
The per se rule is a particularly bad way to evaluate ties in high tech markets, where the combining of previously independent features is both ubiquitous and valuable. For example, Westlaw recently faced a tying claim that bundling its digital research tools with its caselaw database was an unlawful tie. The plaintiff marketed a set of AI-enabled research tools, but Westlaw subscribers did not need them to the extent that Westlaw already included them with the database subscription. A generation earlier a court rejected a tying claim by a subscriber to Broadcast Music Inc.’s blanket license that combined virtually all commercially recorded music. The plaintiff wanted to pay a lower fee for a fractional license that included only country-and-western music. Common to these cases is that any court that condemns the tie must also set a lower price at which the stripped-down product is sold. That places the court in the position of regulatory agency, particularly where product design and costs change over time.
Google noted that even the government’s own expert had acknowledged that “restricting AdX real-time bidding to DFP […] improved Google’s products for publishers.” That claim should get a full evaluation under the rule of reason, and the court addressed it and condemned it under Section 2 of the Sherman Act as well. In a product design case under Section 2 of the Sherman Act, which necessarily involves the rule of reason, the Ninth Circuit has held that if a unilateral product design change was a product improvement there was no violation, and no room for “balancing” the improved product’s value against its exclusionary power. That case, just as this, one involved a tech tie: a redesigned monitor of hospital patient vitals that was incompatible with older generic body sensors. One difference is that in the Google Ad Tech case the court was willing to consider balancing of harms and benefits in its analysis under Section 2.
The Google Ad Tech decision also acknowledged numerous times that scale economies in the ad-tech industry were very significant, so much that rivals could not match Google. If economies of scale mean that the market contains room for only one product, that could be relevant to both tying liability and the remedy.
Simply put, the per se rule’s substitution of a “not insubstantial” volume of commerce as a metric for harm in tying cases is a ridiculous way of testing for competitive harm. Such cases require a more serious analysis of the relationship between product design and overall market impact. The government very likely would have won on these issues in any event, but the per se tying rule served only as an obstacle to effective analysis.
Author Disclosure: The author reports no conflicts of interest. You can read our disclosure policy here.
Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.
