# Antitrust and Trademark Settlements

**Authors:** C. Scott Hemphill, Erik Hovenkamp
**Citation:** "Antitrust and Trademark Settlements," 85 *Antitrust Law Journal* 67 (2023) (with Erik Hovenkamp)

**Abstract:** Online search advertising plays an increasingly important role in facilitating competition. Targeted ads inform consumers about the low prices and other desirable product features offered by rivals. Accordingly, firms have a strong incentive and opportunity to place anticompetitive limits on the flow of information. They do so by reaching collusive agreements in which the competitors agree not to target one another with ads. Ordinarily, such a restraint of trade might be regarded as a straightforward antitrust violation. However, these agreements take the form of settlements of trademark litigation, raising the prospect that the restraints might be justified by trademark law. There is little case law or scholarship identifying when such settlements run afoul of the antitrust laws. This Article is an effort to fill that gap. We explain why “consent to use” settlements of trademark litigation, which merely restrain what marks a firm can attach to its own product, are an unsuitable analogy for understanding these deals. We show how the standard developed in the Supreme Court’s Actavis decision, a watershed ruling about patent settlements, can be adapted and applied to trademark cases. We articulate how courts can identify anticompetitive deals without having to evaluate the merits of the underlying trademark claims. We also consider and evaluate a number of possible procompetitive justifications for restrictive trademark settlements. Our analysis uncovers substantial errors in the first appellate decision addressing these restraints.
**Source:** https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4028144

*p. 2*
here-1800contacts, the name of the leading online retailer of contact lenses. The ads target likely customers and inform them about Walgreens' business. Search ads, and in particular ads targeting a competitor, are a potent and costeffective way to advertise low prices and other attractive features.

*p. 2*
1-800 Contacts (1-800) responded to the threat posed by competitive search advertising in two ways. First, the firm filed trademark suits against Walgreens and other competitors, alleging that their ads were confusing and thereby infringed the 1-800 CONTACTS trademark. Second, it settled the litigation, with 1-800 and Walgreens each agreeing not to run ads triggered by a search for the other firm's name. As a result, consumers who use Google to reach 1-800 are left in the dark about the existence, price, and features of competitive offerings. Similar settlements have been reached in a wide range of other industries.

*p. 2*
Under what circumstances do these settlements violate antitrust law? Antitrust enforcers and private plaintiffs have challenged these deals as unlawful restraints of trade, with mixed results. 3 Despite their importance, search advertising agreements have received little scholarly attention. 4 To answer this new-and newly urgent-question about competition on digital platforms requires a deeper examination of the nature of the intellectual property (IP) privilege and its intersection with antitrust law. This intersection has received a great deal of analysis in the context of patent law. The settlement of patent litigation, in which competing firms agree to restrain competitive activity in some manner, has been a subject of particularly intense litigation and scholarly debate. This debate culminated in the Supreme Court's landmark Actavis decision, which recognized important limits on the scope of the IP privilege. 5 The case established the anticompetitive harm and illegality, under antitrust law, of an important class of patent settle-ments. The lessons of Actavis have been elaborated in a decade's worth of lower-court cases.

*p. 3*
Settlements of trademark litigation raise analogous questions that have gone largely unexplored. One uncontroversial premise is that trademark law protects firms such as 1-800 from certain forms of competition. For example, it prohibits marketing activity that creates a risk of confusion as to the identity of the seller of a product. On the other hand, competitive advertising that creates no risk of confusion is both lawful and desirable. The problem is that firms would prefer to limit their rivals' competitive advertising, whether or not it creates a risk of confusion. Consequently, firms have an incentive to enter into agreements that restrain advertising to a greater extent than is justified by trademark law. A second unremarkable premise is that winning a trademark suit is not an antitrust violation, even if the result is a restriction on competition. However, settling a trademark suit is a very different thing. Such agreements may violate the antitrust laws, just like anticompetitive settlements involving patents. However, there is relatively little case law or scholarship shedding light on when this is the case. This article is an effort to fill that gap. We analyze the competitive effects of trademark settlements and propose a means to assess whether a particular settlement is an unlawful restraint of trade. We explain how Actavis, a decision about patent settlements, can be adapted and applied to trademark cases. We do so with a view to reconciling antitrust concerns about competitive markets with the policy interests underpinning trademark law.

*p. 3*
Our first step is to distinguish two types of trademark settlements. The first type limits a competitor's ability to choose a mark for its own goods. 6 One famous example is a deal between the makers of LYSOL kitchen disinfectant and PINE-SOL floor cleaner, which set the terms of entry for a kitchen disinfectant under the PINE-SOL name. 7 Such settlements restrict what mark the competitor can attach to its own product. The maker of PINE-SOL is free to "invent around" any trademark issue by choosing a different mark. The perceived ease of doing so results in a dismissive conclusion that, in terms of exclusionary potential, trademarks are a "weak" IP right compared to patents and hence unworthy of antitrust attention. This conclusion has been influential and applied broadly-too broadly, as we explain-in denying antitrust liability where a trademark is involved.

*p. 3*
The second type of trademark settlement limits a competitor's ability to target the incumbent for competition using the incumbent's trademarked [Vol. 85 brand name. 8 Such settlements restrain an important mode of competition, one that is particularly salient for lower-priced rivals. Antitrust has long protected competitive advertising as an important mode of competition. These restraints are different in kind from the sort of restriction placed on PINE-SOL. Where a trademarked product name is the key to effective competitive advertising, the weak-trademark intuition simply does not apply.

*p. 4*
We propose a framework for evaluating trademark settlements that draws upon the analogous evaluation of patent settlements reflected in Actavis and its progeny. 9 Our starting point is the identification of the appropriate benchmark against which a settlement's competitive effects should be compared: the expected result of trademark litigation between the parties. For example, if the alleged infringer is 50 percent likely to win the case, then consumers are entitled to half of the resulting benefit from that litigated outcome. The relevant question is whether the settlement confers less welfare than this baseline. This approach is central to the evaluation of patent settlements, 10 and the same logic supports its application in trademark cases. One implication of this approach is that the possession of a trademark does not provide an unqualified right to restrain competitive advertising. After all, only marketing practices shown to create a likelihood of consumer confusion would be enjoined by a court.

*p. 4*
Assessing antitrust liability presents a practical challenge because the result might seem to depend upon the probability that the trademark holder wins the trademark case, a parameter that may be difficult to quantify. But this is not in fact necessary in practice. Rather, it is sufficient to assess antitrust liability based on the nature of the firms' agreement. Here too we draw upon the analysis of patent agreements, which similarly can be evaluated without quantifying the merits of the patent case. Our analysis provides a means to identify anticompetitive settlements without evaluating the merits of the underlying trademark claim.

*p. 4*
Anticompetitive trademark settlements fall into one of two silos. Some settlements exceed the scope of the trademark-that is, they restrain competition to a greater extent than would be achieved by a judgment for the plaintiff in the trademark case. Other settlements are observably anticompetitive due to a collusive bargaining problem that persuades an accused infringer to accept a significantly stronger restraint than it expects to incur from a court's judgment. As we explain, settlements in both categories will usually confer less expected consumer benefit than litigation, a conclusion that can be established without any need to study the merits of the trademark claim. We also analyze various practical complications that arise when using these tools.

*p. 5*
Finally, we assess several potential justifications for otherwise problematic settlements. 11 The most obvious, at first blush, is that a settlement might vindicate the interests of trademark law by avoiding consumer confusion arising from the alleged infringement. However, ordinary settlements already protect against confusion to a degree commensurate with the likelihood that the trademark plaintiff would win its case. To justify an even more restrictive outcome would require a showing that confusion ought to be avoided to a degree greater than that already provided by trademark law, a conclusion that faces important conceptual and practical challenges.

*p. 5*
Along the way, we identify important errors in the first appellate decision on the antitrust implications of trademark settlements involving online advertising. In 1-800 Contacts, Inc. v. Federal Trade Commission, the Second Circuit vacated a decision by the FTC that imposed antitrust liability for 1-800's settlements with rivals. 12 As we explain, the 1-800 court wrongly conflated the two types of trademark settlement, incorrectly dismissed the relevance of settlement terms that exceeded the mark's scope, and failed to recognize the risk of collusive bargaining, among other significant errors. This article proceeds in three parts. Part I identifies the anticompetitive harms from search advertising settlements and explains why the conventional wisdom about weak trademarks does not apply. Part II explicates the probabilistic approach to trademarks and identifies conditions under which an unreasonable settlement may be identified. Part III considers possible procompetitive justifications and provides a framework for their evaluation.

## I. THE COMPETITIVE EFFECTS OF SETTLEMENT

*p. 5*
This Part sets out two types of trademark litigation and the settlements associated with each type. The first type, considered in Part I.A, challenges the use of trademarks to facilitate targeted competition between the mark owner and a competitor. The second type, considered in Part I.B, challenges the selection of a brand name-in the jargon of trademark law, a source identifierthat is confusingly similar to that of the mark owner. The competitive effects of these two types of settlement are quite different, a point elaborated in Part I.C.

## The Settlements

*p. 6*
Search advertising settlements arise in the context of Google's search ads, which appear above or below the ordinary ("organic" or "natural") results on the search engine results page (SERP). 13 For example, Walgreens has promoted its online contacts business by buying ads that appear when a Google user searches for 1800contacts. Google sells the ad space to Walgreens in an auction at which 1-800 and other firms might also bid. The auction system, called Google Ads, matches bidders with a particular user query. The auction winner is decided based on both the price bid and the quality of the ad, particularly the likelihood that a user will click on the ad.

*p. 6*
Search ads are a powerful way for sellers to connect with customers. A user searching for 1800contacts is likely to be interested in buying contact lenses. Buying an ad that appears on the SERP for a rival's name is particularly potent. The online setting makes it easy to view competitive offerings side by side, facilitating comparison shopping. Because of these benefits, search ads, triggered by a search for a rival's name, are commonly used in marketing products and services to consumers. In online retailing of contact lenses, retailers have found it to be an effective way to attract sales. 14 The trademark infringement question arises when the advertiser targets a user query that is a trademarked term, such as 1-800 CONTACTS. In its trademark suit, 1-800 alleged that the Walgreens ads infringed its mark and sought an injunction to prevent their appearance on the SERP for 1800contacts. 15 Instead of litigating this question to judgment, the parties settled. The settlement prohibited Walgreens ads from appearing on the SERP for 1800contacts, much as if 1-800 had won its trademark suit. 16 The agreement was also reciprocal, in that 1-800 agreed not to place ads in response to a user search for, say, walgreens. In addition to Walgreens, 1-800 entered settlements with thirteen other rivals that, together with 1-800, account for the lion's share of online contact sales. 17 The settlements have no effect on Google's organic search results.

*p. 7*
Search advertising agreements have become an important feature of online competition in many industries in which online sales play an important role, from airlines to hotels to credit cards. 18 Industry consultants report that the practice is commonplace. 19 The agreements take several forms. Some cover rival manufacturers, service providers, or retailers. Others have been reached between an upstream provider and a downstream retailer, such as a clothing manufacturer and a reseller or a hotel and an online travel agent, potentially protecting the manufacturer or hotel's own direct online sales from reseller competition. 20 Some industry commentators have urged firms to reach such deals with their competitors. For example, one article-written by lawyers for an audience of in-house counsel-advises, as a "cost-efficient solution" for protecting trademarks, "to approach your competitors and agree not to bid on each other's respective trademarks." 21 As an alternative to a written agreement, a "gentleman's agreement" is suggested, the prospect of defection is considered, and a nod to antitrust liability is made. 22 Industry consultants suggest "handshake agreements" with competitors not to advertise on the SERP for a competitor's brand name. 23 18 See Colangelo, supra note 4, at 41 ("Keyword advertising restrictions appear common and widespread.").

*p. 7*
19 Jaime Sikora, When -and How -To Bid on Competitors' Keywords, 3Q/DEPT (July 25, 2013), 3qdept.com/blog/when-and-how-to-bid-on-competitors-keywords (asserting the existence of a "gentleman's agreement" not to bid "in some industries"). This article, and the citations discussed in notes 21-23 infra, are quoted in Weinstein, supra note 4. Sikora's statement is ambiguous; the author might mean interdependent conduct without any communication. See also Chris Sheen, PPC Ads: Should You Bid on a Competitor's Name?, ECONSULTANCY (Aug. 25, 2015), econsultancy.com/ppc-ads-should-you-bid-on-a-competitor-s-name (noting a successful threat to retaliate against a competing advertiser).

*p. 7*
20 See Tichy v. Hyatt Hotels Corp., 376 F. Supp. 3d 821 (N.D. Ill. 2019) (denying motion to dismiss); TravelPass Grp. LLC v. Caesars Ent. Corp., No. 5:18-cv-153, 2019 WL 4727425 (E.D. Tex. Sept. 27, 2019) (denying motion to dismiss); Case AT.40428-Guess, Comm'n Decision (Dec. 17, 2018) (summary: 2018 O.J. (C 47) 5), eur-lex.europa.eu/legal-content/EN/TXT/ ?uri=CELEX:52019XC0206(01). In addition to litigation between rivals, mark owners (such as American Airlines) have sued Google, the provider of advertising space, in an effort to curb competitive keyword advertising. These suits have settled too, on undisclosed terms. See, e.g., Linda Rosencrance, American Airlines Drops Trademark Lawsuit Against Google, COM-PUTERWORLD (Jul. 22, 2008, 12:00 AM), computerworld.com/article/2535000/american-airlinesdrops-trademark-lawsuit-against-google.html. 22 See id. (concluding that in light of antitrust concerns, an agreement "must be carefully crafted").

*p. 7*
23 See Anne Baum, Why Is My Competitor Bidding on My Brand Terms?, LOCATION3 (Aug. 6, 2013), web.archive.org/web/20131012144518/http://www.location3.com/why-is-my-competitor-bidding-on-my-brand-terms (urging, as a solution to competitor bidding, that the target "[c]reate a handshake agreement with your competitors"). [Vol. 85

## The Harm from Settlements

*p. 8*
Agreeing with a rival to limit search advertising cuts off one mode of competition for customers. Moreover, this mode is particularly well suited to helping a price-cutter get the word out about its lower prices by targeting potential customers of its more expensive rival. Suppressing such advertising tends to limit the effectiveness of price competition. 24 The likely effect is particularly strong where, as with online sales of contacts, lesser-known rivals have lower prices and consumers are unaware of the price differential. 25 The expected result of suppression is higher prices, 26 injuring consumers.

*p. 8*
Before proceeding, it is worth asking whether we should worry about restraints on advertising in the first place. After all, most people find ads annoying. Therefore, it might be tempting to suppose that restraints on competitive advertising would leave consumers happier. However, such a view is mistaken. Such restraints are unlikely to reduce the amount of advertising to which consumers are exposed because media platforms will likely continue to rely on advertising as their primary source of profit. Instead, the restraints will merely suppress the extent to which advertising stimulates competition, particularly from disruptive or low-cost competitors seeking to challenge incumbents. This harms consumers by depriving them of the benefits of more vigorous competition. These settlements of trademark litigation raise some of the same concerns as other (non-trademark) restraints on advertising. Antitrust law has long recognized that advertising is an important part of the competitive process. The case law has emphasized two themes that apply directly to settlements of trademark litigation. First, advertising can inform consumers about their options, including the set of sellers and product varieties on offer. More generally, advertising can reduce consumer search costs. Second, advertising can directly facilitate competition by providing a spotlight for firms to showcase their offerings and contrast them with rivals' products. This is a core aspect of , supra note 4 (finding an association between hotel-imposed restraints on online resellers and higher prices). A full market analysis requires evaluation of substitute channels for purchasing contacts, including from ophthalmologists, optometrists, and stores such as PearleVision and LensCrafters. The administrative law judge considering the FTC's case concluded that 1-800's settlements likely led consumers to pay higher prices. 1-800 Contacts Decision, supra note 1, at 156, 2017 FTC LEXIS 125, at *345. the competitive process. Indeed, competitive behavior, such as cutting the price of a good, is only profitable if it brings in enough new buyers. And that requires getting the word out. Thus, absent the ability to advertise, firms would have a diminished incentive to offer better deals.

*p. 9*
Accordingly, courts have recognized that a naked agreement among competitors to abstain from advertising is illegal per se. For example, in Blackburn v. Sweeney, two lawyers agreed not to advertise in each other's territories. 27 The defendants argued that this was merely ancillary to an agreement to dissolve their partnership. 28 But the court held that the restraint, which carried an "infinite duration," was not reasonably necessary for that purpose. 29 More complicated is a narrower restraint that limits certain forms of advertising that are arguably problematic in some way. For example, what if an agreement is intended merely to restrict false or misleading advertisements? Such advertising is harmful to consumers, so private agreements to eliminate it could be procompetitive. But such agreements do not necessarily allay the antitrust concerns, particularly if the restrictions are relatively broad. It is potentially dangerous to let profit-seeking firms delineate the boundary between acceptable and unacceptable advertisements. They will often have an incentive to enact overly restrictive rules that soften competition by prohibiting even procompetitive forms of advertising. Additionally, other areas of law (such as false advertising and unfair competition) already exist for the purpose of policing misleading advertising. This casts some doubt on the necessity of private restrictions on advertising.

*p. 9*
Nevertheless, courts have occasionally upheld broad horizontal restrictions on advertising. The most notable example is the Supreme Court's California Dental decision. 30 The defendant, a powerful association of dentists, released a "code of ethics" imposing severe restrictions on price and quality advertising. In practical effect, dentists were prohibited from making any claims about the quality of their services and from advertising price discounts. The defendant association said that this was necessary to avoid misleading consumers, who allegedly lack the expertise to evaluate quality or discount claims on their own. A five-Justice majority was sympathetic to this argument and reversed the lower court's holding that the arrangement could be condemned through an abbreviated "quick look" analysis. 31 27 53 F.3d 825, 827 (7th Cir. 1995). 28 Id. at 828. 29 Id. 30 Cal. Dental Ass'n v. FTC, 526 U.S. 756 (1999). 31 Id. at 769. The majority emphasized its view that dentistry is "a profession as distinguished from a business," resulting in an information asymmetry between providers and patients. Id. at 771 n.10. [Vol. 85 California Dental has been widely criticized. 32 One problem is that legal protections already existed to protect patients from misleading advertising, making it less likely that the restraint was really aimed at this problem. 33 Moreover, the restraint proscribed many advertisements that would never face scrutiny under those laws. Indeed, its restrictions were indiscriminate, making no effort to distinguish truthful, procompetitive ads from potentially misleading ones. It is hard to imagine that misleading advertising is such a pervasive and unmanageable problem in California's dental industry that the only solution is a near-categorical ban on competitive advertising, particularly since other learned professions have not deemed this necessary.

## Antitrust Challenges

*p. 10*
Search advertising settlements have attracted the attention of antitrust enforcers in multiple jurisdictions and private plaintiffs. 34 For example, the European Commission condemned settlements, reached between a clothing manufacturer and independent online retailers, that protected the manufacturer's direct online sales. 35 A similar arrangement involving a brand of shoes was condemned in Germany. 36 New Zealand's enforcer condemned an agreement among lenders not to bid on each other's brand names. 37 National competition enforcers in Europe and elsewhere have raised concerns about settlements in airlines, credit cards, hotels, and broadband, among other industries. 38 Enforcement in the United States has been less successful. FTC's judgment that 1-800's settlements with rivals violate antitrust law. 39 The court accepted the premise that "trademark settlement agreements are not automatically immune from antitrust scrutiny." 40 However, the court believed that trademark settlements are presumptively procompetitive, and thus concluded that a quick-look analysis was inappropriate. Applying the more extensive rule of reason, the court rested its judgment on two conclusions: (1) 1-800 established a procompetitive justification, 41 and (2) the FTC failed to establish a less restrictive alternative to the restraint-that is, it failed to show that these benefits could be accomplished in a manner less harmful to competition. 42 We return to these points below.

## B. LIMITS ON A CHOICE OF BRAND NAME

*p. 11*
Antitrust scrutiny of search settlements has taken place against a background of skepticism that trademarks can ever be used as an anticompetitive tool. This skepticism is perhaps surprising, given the recognition elsewhere in antitrust doctrine that trademark licensing can be used to orchestrate a horizontal restraint. 43 It is frequently objected that the exclusionary force of a trademark is not a fit subject for antitrust inquiry. 44 The ground is that trademark is regarded as a particularly weak form of IP right, compared to (say) patent law. It is often asserted that most trademarks confer only "slight mo-39 1-800 Contacts, Inc. v. FTC, 1 F.4th 102, 111-12 (2d Cir. 2021) (per curiam) (reviewing procedural history). 40 Id. at 109. 41 Id. at 119-20 (crediting the claimed procompetitive benefits of avoiding litigation expense and "protecting [1-800's] trademark rights"). 42 Id. at 120-22. Under the court's interpretation of the rule of reason, it was unnecessary to decide whether the restraint had an anticompetitive effect, and the court declined to do so. Id. at 119 ("We need not decide whether the Commission's theory of harm is viable, however, because we conclude that Petitioner has shown a procompetitive justification and the Commission fails to carry its burden at the third step."). The presence of a procompetitive justification, combined with the lack of a demonstrated less restrictive alternative, doomed the FTC's case. Under this approach, a plaintiff loses even if the anticompetitive effects of a restraint are large and any procompetitive effects are relatively small. For a critique of the resulting gap in enforcement, see 45 The dismissive conclusion is that exclusion by a mark holder ordinarily is not worth the "heavy artillery" of antitrust law. 46 In particular, trademark settlements have been regarded as largely harmless. Consider the following real-world example, which formed the basis for the 1-800 court's conclusion that trademark settlements are presumptively procompetitive. The maker of LYSOL kitchen disinfectant faced fresh entry from the makers of PINE-SOL, already a well-known mark for floor cleaners. The entrant wished to offer a competing kitchen disinfectant under the PINE-SOL name. The incumbent filed a trademark suit alleging that this use of the PINE-SOL name raised an unacceptably great risk of confusion and sought to enjoin entry on that basis.

*p. 12*
The parties ultimately reached a settlement restricting the use of the PINE- SOL mark. 47 The settlement granted PINE-SOL some competitive running room compared to the degree of exclusion initially sought by the LYSOL mark owner. 48 The agreement amounted to a compromise in the closeness of competition permitted to the maker of PINE-SOL. Such "consent-to-use" agreements are a typical outcome when a mark owner seeks to restrict the choice of mark of an alleged infringer.

*p. 12*
In response, the competition-minded critic might shrug, so what? The worst that happens as a consequence of litigation or settlement is that the rival must choose a different mark to serve as its source identifier. In Clorox, the Second Circuit considered the real-world PINE-SOL settlement. The court found the consent-to-use agreement reached by the parties to be "common" and indeed "favored, under the law." 49 Thus, "[i]n the absence of evidence to the contrary it is reasonable to presume that such arms-length agreements are pro-competitive." 50 The court also took the opportunity to express severe skepticism that trademark settlements can be anticompetitive. It opined that "[t]rademarks are by their nature non-exclusionary" and that it is "difficult to show that an unfavorable trademark agreement raises antitrust concerns." 51 There is a risk here of being too dismissive. Even this category of trademark settlement may sometimes have anticompetitive effects. Inventing 45 Sheridan, 530 F.3d at 595. 46 Id. 47 See Clorox Co. v. Sterling Winthrop, 117 F.3d 50, 52-54 (2d Cir. 1997) (summarizing the terms of settlement). 48 Id. 49 Id. at 55. 50 Id. at 60. 51 Id. at 56-57; see also 1-800 Contacts, Inc. v. FTC, 1 F.4th 102, 119 (2d Cir. 2021) (per curiam) (quoting Clorox, 117 F.3d at 57).

*p. 13*
around can be expensive enough to limit entry. 52 One scenario, perhaps illustrated by the PINE-SOL example, arises when having a well-recognized brand is an important element of success. If building a brand is expensive, then redeploying an existing brand from floor cleaners to disinfectants is a relatively important form of entry. 53 Preventing that move is consequently a useful form of entry deterrence. 54 A second scenario is that exclusive use of a term denoting a particular product attribute raises a rival's cost of developing a differentiated mark. 55 Economic theory predicts lower investment by rivals, higher consumer search costs as to the rivals' products, decreased industry supply, and deadweight loss. 56 C. RECONSIDERING THE "WEAK TRADEMARKS" CRITIQUE Whatever the merits of the Clorox view and associated "weak trademarks" critique considered above, 57 this perspective is rooted in the particular kind of trademark claim discussed in Part I.B. In these cases, the restraint limits a rival's ability to choose its own mark. That is, it forces the rival to switch from its preferred mark to an alternative that raises a lesser risk of confusing consumers. This type of agreement, however, should not be conflated with a restraint that limits a rival's ability to target the incumbent for comparative advertising-that is, to advertise how one's own product (or price) compares to that of an established competitor. The latter is much broader, as it proscribes competitive advertising without regard to its truthfulness or its propensity to confuse consumers.

*p. 13*
The competitive harm of a restraint on truthful and non-confusing competitive advertising is different in kind from the sort of settlement at issue in Clorox. It restrains an important mode of competition-one that is especially important for competitive entry. As discussed in Part I.A, courts have long recognized that competitive advertising is critical to the competitive process. If a market is currently served by a large incumbent, a rival may seek to enter by offering consumers a better deal. Communicating this to consumers may entail comparative advertising highlighting the difference in price or product features. The effect is likely to be particularly pronounced in markets for undifferentiated products that are sold mainly online.

*p. 14*
The loss of this mode of competition may be difficult to work around. 59 To be sure, firms have many other ways to advertise, even if this avenue is cut off. However, an agreement may raise antitrust concerns even though it does not restrain all possible forms of competition or all forms of competitive advertising. It is enough to restrain any form of competitive activity that is important enough to have a material effect on consumer welfare. That is the case with advertising targeting the incumbent's mark. Such ads are the most direct and effective way to let the incumbent's customers know about what their other options are. The available alternatives do not replace the competitive benefits of comparative advertising. 60 The 1-800 court missed this fundamental difference. The court repeated uncritically the blanket statement from Clorox that "trademarks are by their nature non-exclusionary." 61 Relying upon that case's analysis of consent-touse agreements, the court opined, as noted above, that "[a]greements to protect trademarks . . . should not immediately be assumed to be anticompetitive -in fact, Clorox tells us instead to presume they are procompetitive." 62 The court concluded (again quoting Clorox) that because "agreements to protect trademark interests are 'common, and favored, under the law,' . . . 'it is difficult to show that an unfavorable trademark agreement creates antitrust concerns.'" 63 58 See also Smith v. Chanel, Inc., 402 F.2d 562, 568-69 (9th Cir. 1968) ("[T]he most effective way (and in some cases the only practical way) in which others may compete . . . is to . . . tell the public they have [produced the product], and if they could be barred from this effort [trademark owners] would have found a way to acquire a practical monopoly . . . to which they are not legally entitled."). 59 Such restraints raise the further concern, discussed below, that they are much more likely to exclude to an extent that exceeds the exclusionary force of trademark law. See infra Part II.B. 60 Beyond the suppression of downstream competition, search advertising agreements create a further upstream anticompetitive effect by excluding a rival bidder in certain advertising auctions. The immediate victim of lost upstream competition is Google, for whom search advertising is the main source of profits. Much as with an old-fashioned bidding ring, a restraint on bidding transfers wealth from the seller (Google) to the buyers (advertisers) and distorts the market for advertising. This upstream harm is not a distinctive feature of trademark settlements and is therefore not a focus of the present paper. 61 The court's reliance on Clorox was misplaced. Although it is true, as the 1-800 court suggested, that most trademark settlements do not pose a threat to competition, that is because they very rarely include broad restraints on competitive advertising. It is a category mistake to defend an unusually broad restraint by pointing to more typical settlements whose restrictions are different in nature and far narrower in scope. In short, intuitions built in choice-ofmark cases such as Clorox do not carry over to target-the-incumbent cases, where the opportunity for anticompetitive effects is different and greater. Moreover, to the extent that trademark law's typically narrow impact on competition has an important antitrust implication, it is that broad restraints on competitive advertising are unlikely to be privileged by trademark law. This makes the restraint more likely to raise antitrust concerns, not less.

## II. IDENTIFYING ANTICOMPETITIVE SETTLEMENTS

*p. 15*
As explained in Part I, prohibiting a rival's targeted advertising can result in significant competitive harm. Nevertheless, if a mark owner wins a trademark suit and thereby excludes its rival through an injunction, there is no antitrust violation. Here, trademark law displaces antitrust law, notwithstanding any harm to competition. This result is unsurprising and consistent with other limited exemptions from antitrust law to effectuate (for example) the policies behind labor law and legalized export cartels.

*p. 15*
The analysis is different if the parties settle rather than litigating to judgment. Trademark law does not confer an automatic right to exclude, simply by virtue of possessing a trademark. It merely grants a right to try to exclude by bringing a trademark suit. 64 A private agreement that reproduces the effect of an injunction-or goes beyond it-does not enjoy the same protected status. Thus, settlements that harm competition beyond the limits of trademark law are unreasonable restraints of trade. In this Part, we offer an approach to identifying those unreasonable settlements.

*p. 15*
As we explain in Part II.A, antitrust law accommodates its overlap with trademark law by prohibiting settlements that, by restraining competition, leave consumers worse off than the expected outcome of litigation. The remainder of the Part spells out two types of settlements that leave consumers worse off. The first is a settlement with a term that is overbroad-that is, a term that exceeds what the mark owner could achieve through litigation. The second is a settlement that results from collusive bargaining, in which the mark owner induces its competitor to compete less by sharing the profits from [Vol. 85 avoided competition. Such settlements are the subjects of Parts II.B and II.C, respectively.

*p. 16*
A. ORDINARY BARGAINS AND THE LITIGATION BENCHMARK Most cases settle. In this respect, trademark litigation is no different from other areas of law. Settlements of trademark litigation are not only common but also typically harmless from an antitrust perspective. Consider, for example, the PINE-SOL settlement. This agreement permitted the sale of certain types of PINE-SOL disinfectants while prohibiting others. 65 It also provided that the PINE-SOL product must be marketed primarily as a cleaner rather than a disinfectant. 66 Ordinary trademark settlements have two important features. First, the result is a compromise. 67 The parties might agree to permit some, but not all, of the particular uses sought by the alleged infringer. Such a settlement offers less competition, compared to the outcome if the alleged infringer won, and more competition, compared to the state of affairs if the mark owner won.

*p. 16*
Second, the compromise is the result of a hard-fought bargaining process between competitors. The mark owner pushes for as much restraint as possible, arguing that it is likely to win its suit. The alleged infringer pushes in the other direction, making the opposite argument. The agreed-upon set of uses is a compromise, a bargain reflecting the probability of success. The middle ground reached by the parties thus reflects the parties' expectations about the outcome of litigation if taken to conclusion. This ordinary bargain is an attractive and intuitive baseline for considering the competitive effects of a particular settlement. In other words, a relevant baseline for evaluating a settlement is the expected consumer welfare from litigation that has an uncertain outcome. For example, if a win is 50 percent likely, then the litigation baseline is one-half the difference between the effect of winning and the effect of losing. From a consumer standpoint, the relevant question is whether the settlement confers less welfare than this baseline.

*p. 16*
Two points about the litigation benchmark bear emphasis. First, it is not enough for an antitrust plaintiff to show that the settlement harms consumers compared to the fully competitive outcome. The proper approach takes account of the possibility that the mark owner might otherwise win the trademark suit and exclude competition. Thus, the expected consumer benefit is less than the benefit in the fully competitive outcome.

*p. 17*
Second, consumers may be harmed not only by the settlement of "weak" trademark claims that the alleged infringer is likely to win, but also by strong claims. The upshot is that even if the mark owner is likely to win the suit, consumers are harmed if the settlement restricts competition even more than the expected outcome of litigation, without any compensating consumer benefit. 68 The litigation baseline is drawn from the analogous assessment of patent settlements. In Actavis, the Supreme Court held that patent settlements that eliminate the "risk of competition" are properly subject to antitrust scrutiny. 69 The particular context was a "reverse payment" settlement, a subject considered in more detail below. 70 In taking this approach, the Court explicitly rejected an alternative approach, focused on the "scope of the patent," 71 that arguably had been applied in older cases. 72 The scope of the patent test granted safe harbor to settlements that were no more restrictive than what could be achieved through a litigated injunction. 73 On this view, a settlement permitting entry one day prior to patent expiration would be immune from antitrust challenge. 74 In rejecting this position, the Court recognized the fundamental difference between successful litigation and an untested assertion of infringement. 75 The litigation baseline is now a widely adopted rubric for evaluating patent settlements. 76 The same approach is appropriate for the evaluation of trademark settlements. A trademark does not afford a certain right to exclude a rival's use of an allegedly confusing mark, but rather, a right to try to exclude 68 See FTC v. Actavis, Inc., 570 U.S. 136, 157 (2013) (acknowledging the existence of harm when there is "even a small risk of invalidity"). 69 73 Actavis, 570 U.S. at 147 (acknowledging "that the agreement's 'anticompetitive effects fall within the scope of the exclusionary potential of the patent'" while rejecting the implication "that that fact . . . can immunize the agreement from antitrust attack") (quoting FTC v. Watson Pharms., Inc., 677 F.3d 1298, 1312 (11th Cir. 2012). 74 See id. at 146-47 (discussing lower court's adoption of scope-of-the-patent test). 75 See id. at 147 ("[T]o refer . . . simply to what the holder of a valid patent could do does not by itself answer the antitrust question. The patent here may or may not be valid, and may or may not be infringed."). 76 it. 77 The mark owner might or might not win the trademark infringement suit. In this sense, trademarks, like patents, are probabilistic rights. Settlements are problematic from an antitrust standpoint when, by restricting competition, they provide consumers with less benefit than litigation.

*p. 18*
However, the focus on probabilistic rights and the expected value of litigation may provoke practical concerns about administrability. How is a court supposed to figure out the IP owner's probability of winning a counterfactual lawsuit? Courts would prefer not to conduct an IP trial within the antitrust case, an unappetizing "turducken" task 78 that the Actavis Court worried would "prove time consuming, complex, and expensive." 79 However, the standard adopted in Actavis and applicable here does not ask the courts to confront such difficulties. Indeed, when evaluating patent settlements in practice, courts generally do not speculate about litigation probabilities. This is possible for two reasons. First, when a settlement term is more restrictive than any relief that could be obtained through litigation, then we need not assess the merits to know that the term is more restrictive than a court's judgment would have been. Second, the nature of the firms' settlement agreement-in particular, the collection of restraints and payment obligations it contains-is generally what determines whether the firms' settlement outcome will ultimately deviate from the expected result of litigation and thereby harm consumers. 80 This makes it possible to assess a settlement's antitrust compliance without having to speculate about probabilities. 81 The remainder of this Part discusses these points and explains how they eliminate the need to assess the merits of the trademark case.

## B. OVERBREADTH

*p. 18*
The first type of problematic settlement contains an overbroad term that restricts competition beyond the nominal scope of the mark. In other words, even if the mark owner won the trademark suit and secured an injunction, it could not thereby suppress the competitive conduct in question. In such cases, the restraint does not enjoy any IP-based safe harbor and will be enjoined if anticompetitive. For example, in the patent context, courts have acknowledged that overbroad settlement terms-such as a ban on sales of a competing 77 Cf. Lemley & Shapiro, supra note 64, at 75. 78 product that makes no conceivable use of the patented technology-may violate the antitrust laws. 82 The same principle applies to trademark settlements.

*p. 19*
As discussed above, most trademark settlements are compromises that stop short of what the mark owner could achieve through litigation. Exceptions arise when the settlement prevents competitive conduct that does not even arguably infringe the trademark. Here are two examples drawn from the 1-800 litigation, centered on trademark law's requirements that the conduct must constitute a use in commerce and must result in a likelihood of confusion.

*p. 19*
Beyond the scope of use in commerce. Trademark infringement requires that the alleged infringer make a "use in commerce" of the plaintiff's mark. 83 Courts have concluded, controversially, that bidding on a trademarked term to trigger an ad qualifies as such a use. 84 However, not all bidding strategies require the advertiser to proactively choose the trademarked term.

*p. 19*
An alternative and popular bidding strategy is to employ a generic keyword, such as contact lens, as a "broad" match. Suppose that Walgreens employs such a bidding strategy. When a user searches for 1800contacts, Google determines whether the Walgreens ad keyed to contact lens is sufficiently pertinent to the user search for 1800contacts that the Walgreens bid should be included in the auction. Broad-match advertising campaigns are often a cost-effective way of reaching customers. 85 Notably, in this broad-match campaign, Walgreens makes no proactive selection of the 1800contacts keyword. Google, not Walgreens, makes the connection between the contact lens campaign and the 1800contacts query. As a consequence, Walgreens makes no use in commerce, and without any such use, 1-800's infringement claim against Walgreens must fail. 86 Nevertheless, 1-800's settlements prohibit Walgreens from showing ads in response to a user search for 1800contacts, even when Walgreens is merely bidding on contact lens in broad match. This restriction is accomplished by 82 86 Here, we discuss settlement terms that fall outside the scope of a trademark infringement claim. Depending on the facts, a particular settlement might fall within the scope of some other legal claim asserted by the plaintiff, such as false advertising. [Vol. 85 requiring Walgreens to implement 1800contacts as a "negative keyword," which instructs Google not to display the ad for any user search for the negative keyword. 87 This term goes beyond the scope of the trademark because it restricts competitive activity in a manner that trademark litigation never would. 88 Beyond the scope of likely confusion. Trademark infringement also requires a likelihood of consumer confusion, 89 a determination that depends on the specific nature of the defendant's marketing practices. If a settlement restrains activities that pose no risk of confusion, then the restraint goes beyond the scope of trademark law. For example, suppose an incumbent's rival puts up a new billboard whose text allegedly creates a risk of consumer confusion. If liability were established, a court might enter an injunction proscribing the problematic text. However, the court would not prohibit the infringer from putting up any billboards whatsoever. Accordingly, a settlement that proscribed all billboards would go beyond the scope of trademark law. 90 1-800's settlements arguably go beyond the scope of the trademark for analogous reasons. 91 Buying a trademarked keyword, in itself, is not an infringing act. 92 The likelihood of confusion hinges on the nature of the defen-87 1-800 Contacts Decision, supra note 1, at 25, 2017 FTC LEXIS 125, at *48-49. 88 The discussion in text focuses on overbreadth arising when Walgreens bids in broad match. Overbreadth also arises from the prohibition of certain bids in phrase match. Id. at 24, 2017 FTC LEXIS 125, at *47-48. A phrase match includes the advertiser in the bidding for user queries that include a particular term within a longer query. For example, if Walgreens bids on cheap lenses in phrase match, the result might be an ad on the SERP for 1800contacts cheap lenses.

*p. 20*
In this scenario, Walgreens once again has made no trademark use of the term. Nevertheless, such an ad would be suppressed by 1-800's settlements, which prohibit user searches that merely include the negative keyword, even if they also include additional terms. Id. at 49, 2017 FTC LEXIS 125, at *99-100. Yet another example, similarly involving no trademark use, is a bid on the generic category of contacts in phrase match, resulting in an ad on the SERP for 1800 contacts (with a space).

*p. 20*
89 15 U.S.C. § 1114. 90 By way of further analogy, if a defendant's semiconductor chip design is held to infringe a patent, the court will only enjoin sales of the infringing chip. It will not issue a blanket ban under which the defendant cannot sell any chips at all. 91 See IP Professors Brief, supra note 24, at 13 (making an argument to this effect). 92 dant's advertisement, such as its text and appearance. If it is misleading-for example, if its text falsely implies that the defendant's product or website is affiliated with the trademark owner-then it could very well create an actionable risk of consumer confusion. But absent such facts, there is no reason to suspect that consumers will be confused. 93 Nevertheless, 1-800's settlements with competitors impose a blanket ban on all ads on the SERPs for 1800contacts and other searches, even those creating no risk of confusion. For example, suppose the Walgreens ad made no mention of 1-800, or expressly distinguished itself from 1-800: "We are Walgreens, not 1-800 Contacts, and we offer lower prices." 94 Such an ad would pose no risk of confusion. To be sure, some particular ads could be confusing, but a court would not on that basis prohibit the defendant from running any ads triggered by the trademark regardless of whether they infringe. Thus, the restraint is more restrictive than any relief a mark owner could obtain from litigation.

*p. 21*
The 1-800 court concluded that 1-800's trademark claims, if victorious in court, "might have permitted it to preclude competitors from bidding on its trademarked terms in search advertising auctions or running advertisements on those terms." 95 The court apparently believed that the restraints fell within the scope of 1-800's trademark rights. However, this unreasoned assertion is contradicted by the lack of any plausible trademark claim against ads triggered by the broad-match process or those that raised no risk of confusion.

*p. 21*
Whenever a restraint goes beyond the scope of the trademark right, this fact precludes any trademark-based safe harbor. In such instances, the liability question is assessed in the same manner employed in cases not involving IP. The first question is whether the agreement restrains competition in some CARTHY ON TRADEMARKS AND UNFAIR COMPETITION § 25A:7 (5th ed. 2021) ("Courts almost always find no likelihood of confusion if all that defendant has done is use another's mark as a keyword to trigger an ad for defendant in which the other's trademark does not appear."); Eric Goldman, Another Court Says Competitive Keyword Advertising Doesn't Cause Confusion, TECH. & MKTG. L. BLOG (May 1, 2018), blog.ericgoldman.org/archives/2018/05/another-courtsays-competitive-keyword-advertising-doesnt-cause-confusion.htm (discussing "the long-prevailing view that competitive keyword ads don't confuse consumers in a legally recognizable way").

*p. 21*
In its 1-800 opinion, the Commission noted that the defendant could identify only one case taking a contrary position, and it was from 2008. way. If the deal includes any restrictions on non-infringing competitive advertising, such as the restraints in 1-800, then the answer to this question is almost certainly Yes. Attention then turns to the defendant, who must show that the restraint nevertheless has a procompetitive effect. We explore the latter query in Part III.

## C. COLLUSIVE BARGAINING

*p. 22*
The second type of problematic settlement need not restrain competition beyond the nominal scope of the trademark right. Instead, the problem is that the hard-fought bargain associated with an ordinary settlement-the arms' length negotiation emphasized by Clorox 96 -is replaced by collusive bargaining. Rather than pushing to be restrained as little as possible, the infringer accepts the highly restrictive settlement outcome preferred by the IP owner, even if this result deviates significantly from the expected result of litigation.

*p. 22*
As an illustration, consider reverse payment settlements of patent litigation. 97 Here is how the Supreme Court described the situation:

*p. 22*
Company A sues Company B for patent infringement. The two companies settle under terms that require (1) Company B, the claimed infringer, not to produce the patented product until the patent's term expires, and (2) Company A, the patentee, to pay B many millions of dollars. The patentee and the challenger gain; the consumer loses. 98 Note that in this example, the resulting restraint is no more restrictive than an injunction and thus does not exceed the nominal scope of the patent.

*p. 22*
The concern is that by paying the alleged infringer, the patentee can persuade it to accept a much more restrictive outcome than either firm expected to result from litigation. In other words, the settlement delays competition much longer than litigation would have (in expectation). Reverse payments accomplish this by letting the accused infringer share in the profits generated by suppressing competition. The payment distorts bargaining between the rights holder and the infringer. It persuades the alleged infringer to accept a restraint that is excessive in the sense that it is significantly more restrictive 96 98 FTC v. Actavis, Inc., 570 U.S. 136, 140-41 (2013). The payment is "reverse" because "the settlement requires the patentee to pay the alleged infringer, rather than the other way around." Id.

*p. 23*
than the expected result of litigation, as determined by the merits of the IP claims.

*p. 23*
Absent some such profit-sharing mechanism, a profit-maximizing firm generally will refuse to accept such a restraint. Rather, the firms are ordinarily led to agree on terms that align with their expectations about litigation. An example is an "entry date only" settlement in which two competitors bargain over the timing of competitive entry without using a reverse payment. 99 In fact, most conventional types of IP settlements-notably including ordinary royalty deals-have this property. That is because most settlements lack any mechanism for persuading a firm to accept a restraint that is excessive in the sense defined above. 100 In the absence of some way to share anticompetitive profits, litigants can be expected to reach a settlement outcome that reflects a genuine compromise of their claims, for they are simply unable to agree on anything else. 101 To be sure, such settlements typically restrict competition, but only to a degree that is commensurate with the merits of the underlying IP claims. This is consistent with IP policy, for it means the agreement's impact on competition will be strong if and only if the underlying IP claim is strong. Accordingly, settlements not containing a problematic profit-sharing mechanism generally do not raise antitrust concerns, 102 even when they restrain competition substantially. Even without evaluating the merits of the IP claim, a court can infer that the settlement outcome will be commensurate with the expected result of litigation. But when firms use a reverse payment or other strategy to facilitate collusive bargaining, the opposite inference is warranted. The agreement will cause harm by inducing one or both firms to accept an excessive restraint, resulting in less consumer welfare than expected from litigation. 103 Thus, a natural focus of analysis is the nature of the firms' agree-99 For example, suppose there are ten years left in the patent term and the firms privately believe that the patent is only 40% likely to be valid. Then, in expected value, litigation will generate 40% x 10 = 4 years of monopoly. Thus, in settlement negotiations, the patentee will insist that its rival is delayed by at least four years. And because there is no reverse payment, the patentee's rival will not agree to delay its entry much longer than four years. As such, the firms are naturally led to agree on a delay period of about four years, which emulates the expected result of litigation. The Supreme Court has indicated that these agreements are lawful. See Actavis, 570 U.S. at 158 (recognizing that litigating firms "may . . . settle in other ways, for example, by allowing the generic manufacturer to enter the patentee's market prior to the patent's expiration, without the patentee paying the challenger to stay out prior to that point"). 100 Hovenkamp & Lemus, supra note 80. 101 The patentee will reject settlements that are appreciably less restrictive than the expected litigation benchmark, while the accused infringer, in the absence of a profit-sharing mechanism, will reject settlements that are appreciably more restrictive. 102 We are assuming for the moment that the settlement terms are not overbroad in the sense considered supra Part II.C. 103 Cf. Carl Shapiro, Antitrust Limits to Patent Settlements, 34 RAND J. ECON. 391 (2003) (arguing that the relevant antitrust standard should be to condemn IP settlements only when they result in lower consumer welfare than the expected result of litigation). [Vol. 85 ment, particularly on whether it subverts the accused infringer's bargaining incentives. 104 A direct payment to the infringer is not the only means to facilitate collusive bargaining. A closely related strategy is a quid pro quo involving reciprocal restraints. 105 That is, rather than one firm paying the other to accept an excessive restraint, the firms agree that they will both be restrained in parallel. Each firm is compensated for accepting an excessive restraint by its rival's agreement to also accept an excessive restraint, with the result that they will share in the resulting supracompetitive profits. This dynamic is familiar in cartels. A single firm would not agree to be the only firm whose price is raised far above the competitive level; its rivals could then steal all of its sales by undercutting it. But it will accept the restraint so long as its rivals agree to be restrained in parallel, as this leaves all firms better off.

*p. 24*
Patent settlements involving reciprocal restraints are often condemned as antitrust violations. 106 For example, in Summit Technology, two firms each controlled various patents related to specialized machines used in eye surgery. 107 They agreed to form a patent pool, which charged each firm high royalties. The two firms then split the pool's proceeds. 108 A royalty obligation is a very ordinary restraint-certainly within the nominal scope of the patent. 109 Conventional royalty settlements do not raise antitrust concerns. 110 But the Summit Technology defendants' reciprocal royalty scheme was highly problematic. The royalties naturally induced both firms to raise their prices (by raising their costs). The expenditures were ultimately recouped by virtue of the firms' ownership of the patent pool. Under this arrangement, it was in each firm's interest to adopt an excessive royalty rate in order to force the 104 Hovenkamp, supra note 80, at 450-57. 105 For an economic analysis demonstrating that reverse payments and reciprocal restraints both lead to the same kind of collusive bargaining problem, see Hovenkamp & Lemus, supra note 80; Hovenkamp, supra note 80, at 450-57.

*p. 24*
106 In some such cases, the defendants' arrangement is scarcely distinguishable from an ordinary cartel, in which case its illegality is fairly obvious. See, e.g., United States v. Nat'l Lead Co., 332 U.S. 319 (1947) (condemning a settlement in which competitors agreed to confine themselves to separate territories to avoid competition); United States v. Masonite Corp., 316 U.S. 265 (1942) (condemning a settlement in which the firms agreed to fix prices).

*p. 24*
107 Summit Tech., Inc., FTC Docket No. 9286, 127 F.T.C. 208 (Feb. 23, 1999). 108 Id. at 209-10. 109 A royalty raises the accused infringer's costs, which will push its price upward and diminish its sales. But an injunction would prohibit all sales of the allegedly infringing product, which is clearly a more restrictive result.

*p. 24*
110 See Hovenkamp & Lemus, supra note 80, at 266-69 (demonstrating, in a model of oligopoly competition, that absent a collusive inducement, a firm will not agree to pay a royalty that diminishes its competitive vitality to a materially greater extent than the expected result of litigation).

*p. 25*
price upward, provided that the other firm did the same. 111 The firms ultimately abandoned this arrangement after the FTC challenged it. 112 Another example arises in the pharmaceutical industry as a variant of the reverse payment problem. The branded drug maker may compensate the generic drug maker not with cash, but rather a promise not to launch its own competing generic drug. In exchange, the generic firm will accept a more anticompetitive result-a longer delay in market entry-than it would otherwise agree to based on the merits of the IP claim. The resulting settlement restrains generic entry by both firms, not just the generic challenger. Courts and enforcement agencies have recognized the anticompetitive effects of such settlements. 113 Trademark settlements may exploit reciprocal restraints to suppress competitive advertising. For example, 1-800 and Walgreens each agreed not to advertise using the other's keywords. The settlement prohibited each from using the other's trademark to trigger keyword advertisements. 114 This kind of reciprocity can have a major impact on the firms' bargaining irrespective of the merit of the underlying infringement claims. By default, neither firm has an incentive to agree to abstain from running the ads; each relies on such ads to take sales from the other. However, the firms earn larger profits if they simply agree to stop competing in this way, both by avoiding the expense of the ads and by reducing the potency of, and hence incentive for, price competition. This is analogous to an agreement in which two cross-town rivals agree not to put up billboards in one another's territories. It does not entirely eliminate competition, but it does eliminate an important channel of competition, which is a sufficient basis for antitrust liability. 115 The 1-800 court wrongly ignored the danger of collusive bargaining. The court saw the settlement as merely the result of "hard-nosed trademark negotiations." 116 It suggested that "what is 'reasonably necessary' [to settle] . . . is likely to be determined by competitors during settlement negotiations." 117 In other words, the court erroneously suggested that there is no need to worry [Vol. 85 that competitors might agree to improper or excessive restraints when settling an IP dispute.

*p. 26*
The 1-800 court once again relied heavily on its analysis of a consent-touse agreement in Clorox. 118 There, the court had stated that "[w]here large competitors each represent their respective trademark interests, unless one party is irrational, the result should accord with how the parties view their respective rights." 119 That is true as applied to an ordinary bargain in which the parties negotiate over a single restraint on some competitive activity that is the subject of a trademark suit. 120 But the result simply does not hold when the firms negotiate reciprocal concessions in which each firm avoids competition. Thus, the 1-800 court was wrong to conclude that "while trademark agreements limit competitors from competing as effectively as they otherwise might, we owe significant deference . . . to those agreements." 121 This perspective ignores the incentive and opportunity to engage in collusive bargaining.

*p. 26*
Various arguments might be offered in defense of settlements containing reciprocal restraints. For example, the fact that the parties have filed dueling infringement claims against one another might be thought to require or somehow justify a quid pro quo between reciprocal restraints. In Part III, we explain why this perspective is mistaken. 122 Alternatively, a defendant might argue that one of the reciprocal restraints is superfluous in the sense that it does not cause any change in the firm's behavior. 123 As we explain in Part III, if this is true, it could allay the collusive bargaining concern. Nevertheless, as we argue below, such arguments should generally be rejected when assessing a settlement's legality. 124

## III. ASSESSING POTENTIAL JUSTIFICATIONS

*p. 26*
Under the rule of reason, a defendant may escape liability for its restraints on competitive activity by establishing the existence of a procompetitive justification. In this Part, we consider some possible justifications related to trademark law and policy that defendants might raise. We also explore related considerations, such as whether the defendants could obtain the asserted procompetitive benefits by less restrictive means.

## A. WHEN IS A JUSTIFICATION NECESSARY?

*p. 27*
IP settlements between competitors routinely restrain competition in one way or another. Indeed, in many IP cases, it would be impossible to settle without some restraint. After all, litigation has some probability of resulting in an injunction. Thus, for a settlement to be acceptable to the mark owner, at least a partial restraint may be necessary.

*p. 27*
And yet the large majority of such settlements do not require any detailed analysis on this point. In particular, if a court has no reason to believe that a settlement is more restrictive than the expected result of litigation, the agreement is presumptively authorized by trademark law. In such cases, the plaintiff has failed to carry its burden of establishing a prima facie violation. Indeed, Actavis implies that a plaintiff cannot establish a violation merely by showing that a settlement is likely to restrain competition in some fashion. 125 The plaintiff must give evidence suggesting that it is likely to be more restrictive than a court's judgment on the IP question would be (in expectation).

*p. 27*
As we have emphasized, this can be assessed based on the nature of the terms being negotiated. Accordingly, analysis of the defendants' justifications is necessary only when the firms have reached a settlement that is more restrictive than the expected result of litigation. This could involve either: (1) a competition-suppressing term that goes beyond the nominal scope of the trademark; or (2) a profit-sharing mechanism that facilitates collusive bargaining, such as a reverse payment or a quid pro quo between reciprocal restraints. The reason that very few settlements require a consideration of justifications is that most settling parties avoid a settlement of this kind.

## B. AVOIDING FREE RIDING

*p. 27*
One frequently offered trademark-related argument is that proscribing otherwise lawful competitive advertising serves the asserted procompetitive purpose of curbing "free riding" by the mark owner's rivals, which could help to bolster incentives for investment in one's own brand. 126 In the context of keyword advertising, the argument would proceed as follows. A firm that does most of its business online may invest heavily in making consumers familiar with its brand. That way, consumers are more likely to seek out the firm when they shop online. For example, 1-800 pioneered online sales of contact lenses and made substantial investments in marketing to make con-125 See FTC v. Actavis, Inc., 570 U.S. 136, 141, 147, 158 (2013) 126 See Brief of Richard A. Epstein, Keith N. Hylton, Thomas A. Lambert, Geoffrey A. Manne, Hal Singer, and Washington Legal Foundation as Amici Curiae in Support of Petitioner at 17-24, 1-800 Contacts, Inc. v. FTC, 1 F.4th 102 (2d Cir. 2021) (No. 18-3848) (arguing that the prevention of free riding constitutes a procompetitive justification for 1-800's settlements). [Vol. 85 sumers aware of the lower prices available online. 127 But a competitor might attempt to free ride on 1-800's brand by targeting ads at consumers who search for 1800contacts. This might diminish the incentive of firms like 1-800 to invest in their own brands. If true, a restraint on such free riding could stimulate incentives for investment, and that would be a good thing. This defense does not rely on the proposition that free riding is necessarily actionable under trademark law. There are a few cases suggesting that free riding may provide a basis for finding trademark infringement liability, 128 but there are also reasons to doubt that trademark policy supports such a strong anti-free-riding policy. 129 And there is ample judicial support within trademark law for the proposition that free riding is essential to the competitive process. 130 Indeed, in many situations, terms like "free riding" and "copying" are just pejoratives for competitive activity that an incumbent doesn't like. 131 At its core, the argument asserts that free riding on established brands is inefficient because it chills investment, and therefore it is desirable to restrain such activities. This argument, if accepted, would potentially establish a justification for all kinds of far-reaching restraints on targeted advertising-for example, an advertisement for Universal Studios placed near the entrance to Disney World-even when they create no risk of confusion.

*p. 28*
The key problem with the free riding argument is that, even if its core premise is true, it does not constitute a cognizable antitrust defense. That is, even if one is convinced that it would be efficient to restrain free riders, this would not provide a legal justification for doing so. Antitrust does not permit firms to engage in anticompetitive self-help to shore up a perceived deficit of IP protection. By way of analogy, one might take the position that 20 years of patent protection is not enough. But that policy perspective would not author-127 1-800 Contacts Decision, supra note 1, at 77, 2017 FTC LEXIS 125, at *166. 128 See Kenner Parker Toys Inc. v. Rose Art Indus., Inc., 963 F.2d 350, 353 (Fed. Cir. 1992) ("Both the mark's fame and the consumer's trust in that symbol, however, are subject to exploitation by free riders."). For discussion and a critique of this position, see Beebe & Hemphill, supra note 52, at 1375-98. 129 Beebe & Hemphill, supra note 52. 130 See, e.g., Smith v. Chanel, Inc., 402 F.2d 562, 568-69 (9th Cir. 1968) ("Disapproval of the copyist's opportunism may be an understandable first reaction, 'but this initial response to the problem has been curbed in deference to the greater public good.' By taking his 'free ride,' the copyist, albeit unintentionally, serves an important public interest by offering comparable goods at lower prices.") (citations omitted).

*p. 28*
131 Moreover, even if the avoidance of free riding were a key element of trademark law, a settlement that heavily insulated the mark owner from free riding would still face the objection discussed supra Part III.A, as such a settlement would protect an asserted trademark policy goal to a greater extent than trademark law itself. ize a private agreement that restrains a competitor's use of an invention for even longer. 132 The leading case on this point is Fashion Originators' Guild of America, Inc. v. FTC. 133 Designers in the 1930s were afflicted by style pirates, who sold copies of the designs at much lower prices. The pirates were legally permitted to engage in free riding-copyright did not cover the designs-albeit without any special legal encouragement. The designers joined forces and arranged a boycott of retailers that did business with pirates. The Supreme Court ultimately condemned the arrangement without regard to any claimed economic benefits from the avoidance of free riding. The case is often recognized for the proposition that firms are not entitled to create a private IP system whose restrictions on competitive activity reach farther than those imposed by IP law.

*p. 29*
This same point is also visible in the analogous context of competition between branded and generic drug makers. The brand incumbent often has made a large investment in R&D to develop and secure approval of the drug. When a generic rival threatens to enter the market, "piggybacking" on the brand's investment, 134 that effort is protected by antitrust law. In Actavis, the Court rejected the idea that defendants might justify a horizontal agreement that avoids competition on the ground that reduced competition increases profits, which in turn induces or is spent on increased efforts to innovate. 135 More generally, the mere fact of investment provides no privileged basis under antitrust law to enter horizontal agreements to insulate oneself from free riding that takes the form of market competition. 136 As a general matter, alleged consumer benefits premised on reductions in competition are not cognizable. For example, engineers cannot refrain from price competition on the ground that competition will result in shoddy bridges. 137 As the Supreme Court explained, considering this argument when a group of engineers made it, "the Rule of Reason does not support a defense based on the assumption that competition itself is unreasonable." 138 [Vol. 85 To be sure, free-riding arguments are sometimes recognized as antitrust justifications. For example, antitrust law takes a lenient approach to certain vertical restraints, such as resale price maintenance and exclusive retail territories, on the ground that the restraint encourages the downstream retailer to exert effort and invest in quality enhancements, which would otherwise be undercut by other distributors of the same brand. 139 The idea is that less free riding ultimately enhances competition. However, these arguments do not contend that we should permit firms to suppress competition as a means stimulate investments whose long-run value will offset the loss of competition. Rather, they argue that eliminating a certain free-riding problem will help to enhance competition right away, such as by increasing output. In this respect, they are quite unlike the free-riding argument offered in defense of anticompetitive trademark settlements.

## C. LESS RESTRICTIVE ALTERNATIVES

*p. 30*
For a proposed justification to be an effective antitrust defense, the defendant's restraint must be reasonably necessary to achieve it. For example, if the same procompetitive benefit could be achieved without the restraint on competition, then the claimed benefit does not justify the restraint. 140 To this end, a plaintiff can overcome an asserted justification by showing the existence of a less restrictive alternative (LRA). LRAs are a standard way to establish antitrust liability. 141 The LRA inquiry asks the factfinder to compare the defendant's conduct to a hypothesized alternative that serves the same claimed goal and consider whether that alternative is less restrictive and therefore poses a lesser risk of harm to competition.

*p. 30*
To offer a basis for liability, the asserted LRA needs to be feasible and practical, not unreasonably speculative. 142 The defendants can contest feasibility, but doing so demands more than a bald assertion of some subjective difficulties that make the LRA unworkable. Otherwise, defendants could avoid liability by asserting highly subjective excuses that plaintiffs cannot realistically disprove.

*p. 30*
For example, suppose a settlement contains a problematic term such as a reverse payment. The overwhelming majority of settlements do not contain such terms. This suggests that they are generally not necessary for parties to settle. Thus, the plaintiff has a very strong argument for the existence of an LRA: the firms could just settle without using the problematic term. After all, everyone else gets along fine without them.

*p. 31*
However, the defendants may make subjective arguments about why they "just can't agree on terms" without the challenged provision. 143 This is a defense alleging that the problematic term is necessary to settle and thereby avoid the ongoing expense of litigation. 144 Such arguments should be viewed quite skeptically, as they are typically not susceptible of proof. For example, defendants sometimes argue that they need to use a reverse payment because the patentee is risk-averse and, without a large payment to the alleged infringer, settlement will fail and the patentee will experience the risk-bearing cost. 145 There are good reasons to reject this argument on substantive grounds, but the more important objection relates to administrability. A patentee's claim that it is risk-averse will likely be very difficult for an antitrust court to evaluate, because it concerns the patentee's subjective feelings toward risk. If such a defense were permitted, there is a risk of widespread abuse by defendants.

*p. 31*
The same arguments apply to trademark settlements. The defendants might assert some subjective reasons why they need an otherwise-anticompetitive term to achieve a settlement. But they properly bear the burden of proving this. Further, to avoid widespread abuse, courts should not entertain any asserted justifications that cannot be demonstrated with objective evidence.

*p. 31*
The 1-800 court erred on this point. The Commission had proposed several plausible less restrictive settlements-for example, a disclosure requirement under which a rival merely promises to clearly identify itself as the seller in an ad's text. 147 But the court summarily rejected these arguments. In effect, it held that the FTC was obliged to disprove the possibility that some subjective or context-specific difficulties might prevent 1-800 from reaching an amicable settlement without a blanket ban on keyword advertising. 148 Such a demanding proof requirement is likely to be prohibitive in virtually all cases. Unfortu-[Vol. 85 nately, while the court was deeply concerned about the practical difficulties of enforcing trademark rights, 149 it expressed no concern at all for the equally important project of maintaining administrable antitrust standards.

## D. PROTECTING TRADEMARK RIGHTS AND AVOIDING LITIGATION COSTS

*p. 32*
When assessing potential justifications for restrictive IP settlements, courts often mention the protection of IP rights and the avoidance of litigation costs as relevant efficiencies. To be sure, such effects may well be desirable, but they generally cannot justify an otherwise anticompetitive settlement. 150 The reason is that any mutually acceptable settlement agreement will achieve these benefits. Clearly, any settlement will avoid litigation costs. And a settlement cannot be acceptable to the mark owner unless it grants protection against infringement that is at least on par, in an expected value sense, with the fruits of litigation. 151 This is important because, as noted in Part II, problematic terms are normally not needed in order to reach a mutually acceptable settlement. Most parties reach an ordinary settlement-that is, a settlement that restrains the alleged infringer to some degree-without an overbroad term or a reciprocal benefit to the alleged infringer. Ordinary settlements do not raise antitrust concerns. Thus, pointing out that this settlement avoids litigation costs, or protects trademarks, provides no justification for entering an anticompetitive settlement rather than an ordinary one.

*p. 32*
A defendant may protest that their preferred settlement protects trademark interests to an even greater degree than an ordinary settlement. But to return to a familiar point, it is the ordinary settlement that better approximates whatever protections are actually offered by trademark law. The extra protection from the settlement is anticompetitive and goes beyond the force of the trademark considered on its own. To be sure, the parties' preferred settlement might result in higher firm profits, but that is a predictable consequence of an anticompetitive settlement term.

*p. 32*
The strongest case for defendants is presented if they are able to demonstrate that they could not have reached an ordinary settlement instead. Such a demonstration is necessary but not sufficient to avoid liability. Under antitrust's consumer welfare standard, consumers need not bear the costs of the 149

## Id.

*p. 32*
150 But see id. at 119 (concluding otherwise). 151 The converse is not true, however. A settlement may be mutually acceptable even if it is far more restrictive than the expected result of litigation. The anticompetitive settlements discussed in Part II all share exactly this feature. Given the IP owner's strong interest in benefiting from any settlement, there is reason to worry that a particular settlement results in excessive restraints, and little reason to be concerned that it fails to sufficiently protect the owner's IP rights. firms' inability to reach an agreement except on anticompetitive terms. Such a settlement, though "necessary" insofar as the firms are concerned, might nevertheless be condemned because it is more restrictive than the expected result of litigation. 152 And there is no reason to assume the firms' litigation cost savings will outweigh the resulting injury to consumer welfare. 153 Finally, it bears note that a failure to settle does not imply that the mark owner's IP rights will go unprotected. A court's judgment will furnish such protection to the extent that the mark owner is entitled to it.

## E. MANAGING THE SETTLEMENT OF DUELING INFRINGEMENT CLAIMS

*p. 33*
As noted earlier, in many infringement claims, a mutually beneficial settlement will require that the accused infringer is restrained to some extent. 154 One might therefore wonder whether, in a case involving dueling infringement claims between two rivals, a mutually acceptable settlement somehow requires reciprocal restraints. The answer is No. More accurately, a quid pro quo between reciprocal restraints is not necessary for the firms to settle.

*p. 33*
As an illustration, suppose that two firms sell similar products with similar branding, and that each firm has recently erected some billboards in the other's territory. Each firm now alleges that its rival's billboards infringe their trademark. A resolution may indeed require that both firms are restrained in some way, at least as to the content of their billboard advertisements. As always, we would like each restraint to be commensurate with the expected result of litigation. However, if there is a quid pro quo between the two restraints, we will get the usual collusive bargaining problem. For example, the firms might agree that they will no longer erect any billboards in each other's territory. This would be unlawful, as no trademark judgment would impose a blanket ban on billboard advertising. 155 To avoid this problem, the quid pro quo ought to be eliminated, with each restraint negotiated independently. Then, assuming there are no other problematic terms being negotiated, each settlement will impose a restraint that is commensurate with the merit of its respective infringement claim. At minimum, this requires that there is no express quid pro quo between the two restraints. [Vol. 85 However, even if there is no express quid pro quo, there is still the risk of a tacit exchange. One practical solution is to separate the negotiations in time, with one firm's claim being settled before the other. This kind of arrangement has been used in the patent context to avoid collusive bargaining problems involving reverse payments. 156

## F. SUPERFLUOUS RESTRAINTS

*p. 34*
In principle, a restraint could be superfluous in the sense that it does not cause any change in the behavior of the firm being restrained. For example, consider the settlement between 1-800 and Walgreens, which prohibits each firm from using the other's brand name to trigger online ads. 1-800 might argue that it was not planning to use WALGREENS as an advertising keyword anyway. 157 If true, the restraint imposed on 1-800 would be superfluous. That would suggest that it does not confer any benefit upon Walgreens. In that case, Walgreens would not make any extra concessions-such as accepting an excessive restriction on its own advertising practices-in exchange for the restraint on 1-800. This would allay the collusive bargaining concerns discussed previously. 158 However, such arguments should generally be rejected when evaluating a settlement's legality. There are several reasons for this. First, an antitrust plaintiff generally has no realistic way to disprove a defendant's assertion that it would have behaved the same way in any event. This is why courts do not require plaintiffs to furnish such proof in ordinary collusion cases. For example, if two defendants each agree to charge $100, provided that the other does so as well, the fact of this agreement is sufficient to find a violation. The defendant may not escape liability by arguing that it would have charged $100 in any event.

*p. 34*
Second, and relatedly, a restraint's inclusion within an agreement supports a strong presumption that it is meaningful rather than superfluous. This is particularly true if the restraint arouses antitrust concerns. Indeed, it is hard to explain why a firm would subject itself to antitrust scrutiny to defend a restraint that has no impact on anyone. Third, if a restraint really is superfluous, then there is no harm in enjoining it.

*p. 35*
For these reasons, in cases involving a quid pro quo between reciprocal restraints, there should generally be a strong presumption of illegality. Assessing the magnitude of a restraint's impact on competitive behavior is properly left to the calculation of damages.

## IV. CONCLUSION

*p. 35*
The emergence of digital platforms has made online advertising an increasingly important aspect of the competitive process. Many consumers rely on platforms as their primary sources of information. As a result, firms rely on those platforms to inform consumers about their products and how they stack up against the competition. The increased competitive significance of digital advertising, and the central role of trademarks in that advertising, means that the importance of the trademark-antitrust intersection will continue to increase.

*p. 35*
Despite its importance, the interface between antitrust and trademark law remains understudied and undertheorized. This article endeavors to help fill that gap in the context of settlements of trademark litigation. There are significant differences between trademarks and other forms of IP. For instance, trademarks are not intended to confer a right to restrain or exclude competitors in the same sense that patents and copyrights do. Nevertheless, trademark agreements between competitors do sometimes restrain activities that are relevant to the competitive process, such as competitive advertising. And these agreements may raise antitrust concerns, such as when they restrain a markedly broader set of activities than a court's injunction typically would. Some jurists have opined that trademark agreements are unlikely to harm competition because trademarks are "weaker" than other forms of IP. That is, they confer less power to exclude competitors and are usually easy to invent around. However, this conclusion does not follow. Rather, the comparative weakness of trademarks merely implies that trademark law will not normally authorize broad restraints on competitive activity. In other words, if an agreement purporting to be a trademark settlement imposes broad restraints on competition, then it is likely to raise antitrust concerns.

*p. 35*
We offer a framework for evaluating the competitive effects trademark settlements that is adapted from the analogous examination of patent settlements. The heart of the inquiry is a competitive benchmark: the expected consumer benefit from litigation. A settlement that is more anticompetitive than this is unlawful. This is the basic principle underlying the Supreme Court's Actavis decision, which we adapt to the examination of trademark settlements in a manner that is cognizant of the distinctive features of trademark law.

*p. 35*
Of course, it is unrealistic to ask courts to engage in unaided speculation about the expected value of the outcome of IP litigation that never reached a conclusion. An important benefit of our framework is that it does not require courts to make such inquiries. Instead, courts can focus on the nature of the agreement-in particular, the presence of overbroad terms or collusive bargaining. Absent such features, courts can safely presume that the firms' settlement reflects their expectations about how litigation would have played out. In that scenario, the settlement would impose a strong restraint on the accused infringer only if litigation was likely to do the same.

*p. 36*
The first category of problematic settlements-a settlement with a term that goes "beyond the scope" of the trademark right-is familiar in antitrust. Such terms are more restrictive than any judgment a court might issue. An intuitive example is an agreement that restrains both infringing and non-infringing activity. When a restraint is overbroad in this sense, it does not enjoy safe harbor based on trademark law; it should thus be evaluated under the same standards applied to horizontal agreements not involving IP.

*p. 36*
The second category of problematic settlements is one that observably reflects collusive bargaining. As we have explained, collusive bargaining is achieved by using profit-sharing terms that make it mutually beneficial (from the firms' perspective) for the settlement to restrain competition excessively-that is, to make the settlement restrain competition to a greater degree than the expected result of litigation. Such terms are acceptable to an accused infringer because they permit it to share in the profits generated by restraining competition excessively.

*p. 36*
One problematic form of profit sharing, famous within antitrust, is a reverse payment. This was the subject of the Actavis decision. However, firms can achieve similarly anticompetitive results by instead relying on a quid pro quo between reciprocal restraints. This means that the firms agree that they will each be restrained in parallel. As with an ordinary cartel, the firms can benefit by agreeing that they will each be restrained excessively. Thus, as with reverse payments, this sort of agreement induces bargaining outcomes that are more anticompetitive than the expected result of litigation. In this and other respects, the patent analysis of Actavis provides valuable guidance for the analysis of trademark settlements as well.

## Footnotes

> See Clorox Co. v. Sterling Winthrop, 117 F.3d 50, 52-54 (2d Cir. 1997).

> 1-800 Contacts, Inc. v. FTC, 1 F.4th 102, 109 (2d Cir. 2021) (per curiam).

> This account is a simplification of the actual search advertising process.

> 1-800 Contacts Decision, supra note 1, at 67, 2017 FTC LEXIS 125, at *144 ("Paid search advertising . . . is an important method for marketing contacts online . . . .").

> See id. at 44-45, 2017 FTC LEXIS 125, at *88-89 (describing 1-800's trademark suit).

> In fact, the settlement reaches beyond what 1-800 could accomplish if it had won the trademark suit. See infra Part II.B.

> 1-800 Contacts Decision, supra note 1, at 40-60, 67, 2017 FTC LEXIS 125, at *76-103, *143-44. 1-800 makes a majority of online contact sales, and the settling competitors account for another 30 percent. Id.

> Mark A. Lemley & Carl Shapiro, Probabilistic Patents, J. ECON. PERSPS., Spring 2005, at 75 (noting that "a patent does not confer upon its owner the right to exclude but rather a right to try to exclude by asserting the patent in court").

> Clorox, 117 F.3d at 53-54.

> In the extreme case of a highly clear-cut claim such as counterfeit goods, the result is likely to be highly one-sided.

> This allows the firms to emulate the results of price-fixing without having to form an explicit agreement to fix prices. See Carl Shapiro, Patent Licensing and R&D Rivalry, 75 AM. ECON. REV.25, 26 (1985) (discussing the underlying economics).

> Id. at 122 (relying on Clorox, 117 F.3d at 59-60, as a basis for granting deference to "arm's length use agreements negotiated by [the] parties").

> Clorox, 117 F.3d at 61.

> In other words, it is true in a settlement that does not contain any profit-sharing mechanism, such as reverse payments or a quid pro quo between reciprocal restraints.

> See Hovenkamp, supra note 80, at 424 n.35 (concluding that a horizontal restraint that persists after patent expiration is a transparent antitrust violation).

> See FTC v. Actavis, Inc., 570 U.S. 136, 142 (2013) (noting that the law governing generic drug approval "allow[s] the generic to piggyback on the pioneer's approval efforts").

> This argument was emphasized by the Actavis dissent but did not sway the Court. Id. at 161, 174 (Roberts, C.J., dissenting) (emphasizing the role of patents in encouraging innovation); id. at 176 (criticizing Court's opinion on the ground that it "weakens the protections afforded to innovators by patents").

> See, e.g., Smith v. Chanel, Inc., 402 F.2d 562, 568-69 (9th Cir. 1968) ("A large expenditure of money does not in itself create legally protectable rights.").

> Nat'l Soc'y of Pro. Eng'rs v. United States, 435 U.S.679, 696 (1978).

> See Hovenkamp, supra note 80, at 473.

> We discuss this and a related justification in the next subsection.

> See Hovenkamp, supra note 80, at 473; Aaron Edlin, Scott Hemphill, Herbert Hovenkamp & Carl Shapiro, Actavis and Error Costs: A Reply to Critics, ANTITRUST SOURCE (Oct. 2014), at 1.

> See id.; Hovenkamp, supra note 80, at 474.

> Opinion of the Commission at 27-30, 1-800 Contacts, Inc., FTC Docket No. 9372, 2018 WL 6201693 (Nov. 7, 2018).

> 1-800 Contacts, Inc. v. FTC, 1 F.4th 102, 121 (2d Cir. 2021) (per curiam) (holding that the Commission "failed to consider the practical reasons for the parties entering into the Challenged Agreements. . . . [It] did not consider, for example, how the parties might enforce such a requirement moving forward or give any weight to how onerous such enforcement efforts would be for private parties.").

> For example, in the reverse payment context, there is no reason to believe the firms would limit their payment to the smallest level needed to settle the case. On the contrary, unless constrained by antitrust liability, they are likely to exploit this opportunity to the maximum extent. See Edlin et al., supra note 145.

> Note also that there is no reason to think litigation cost savings will pass through to consumers. Unlike variable production costs, litigation costs do not scale with output.

> See supra Part III.A.

> See, e.g., Blackburn v. Sweeney, 53 F.3d 825 (7th Cir. 1995) (finding a per se violation where competitors agreed not to place ads in each other's territories).

> See, e.g.,[Proposed] Stipulated Revised Order for Permanent Injunction and Equitable Monetary Relief at 2, 4, FTC v. Cephalon, Inc., No. 2:08-cv-2141, 2019 WL 2111253 (Feb. 21, 2019) (prohibiting patent settlements that combine brand payments to the generic firm with delayed entry, and defining contemporaneous transfers of value as a form of payment, even if contained in a purportedly separate agreement).

> It is conceivable that 1-800 would make such an argument, given that it has repeatedly claimed that it is unlawful to use rival marks to trigger online ads.

> 38 COMPETITION & MKTS. AUTH., DIGITAL COMPARISON TOOLS MARKET STUDY: FINAL RE-PORT

> 43 See, e.g., United States v. Topco Assocs., Inc., 405 U.S. 596 (1972); United States v. Sealy, Inc., 388 U.S. 350 (1967).44 See, e.g.,

> Id.; see also Impax Labs., Inc. v. FTC, 994 F.3d 484, 492 (5th Cir. 2021). 70 See infra Part II.C. 71 Actavis, 570 U.S. at 141, 147-58. 72 See, e.g., United States v. Gen. Elec. Co., 272 U.S. 476 (1926). The GE rule has been narrowly construed in later cases.

> 81 Hovenkamp & Lemus, supra note 80.