Welcome to the Antitrust Antidote—a quarterly publication analyzing significant U.S. antitrust decisions from legal and economic perspectives. Authored by former Federal Trade Commission (FTC) enforcer Koren W. Wong-Ervin with former FTC economist co-authors Jeremy Sandford and Nathan Wilson, alternating each quarter. The title of this series, “Antitrust Antidote,” while mostly meant to be humorous (perhaps limited to those who have heard Koren’s “Let’s talk economics” as a cure for a bad day), also refers to the practical guidance we aim to provide throughout the series. We hope you enjoy it!
***
There were a number of significant developments from the end of 2023 through the beginning of 2024 covering important (and highly debated) topics, such as: (1) the role of contracting as an alternative to vertical integration when determining whether benefits such as the elimination of double marginalization (EDM) are merger specific; (2) who bears the burden of proof on pre-litigation remedies and whether their consideration is properly part of the liability (as opposed to the remedies) phase; and (3) the role and proper use of win/loss data.
Below, we consider these issues in the context of the Fifth Circuit’s decision in Illumina/Grail; the grant of a preliminary injunction in the FTC’s challenge of the IQVIA/Propel Media (DeepIntent) merger; the district court’s decision blocking the JetBlue/Spirit merger; and Amazon’s termination of its iRobot acquisition in the wake of the EC’s Statement of Objections and an apparent FTC staff recommendation that the transaction be blocked.
Illumina/Grail
While the Fifth Circuit’s decision to largely accept the FTC’s reasoning in blocking the deal may well bolster agencies’ enforcement of vertical mergers, the decision is helpful to merging parties in at least two respects. First, the court’s recognition that, while contracting can, in theory, be a way to eliminate double marginalization, plaintiffs cannot merely rely on evidence that one of the parties “considered” alternatives to vertical integration but rather must take into account whether the contracting alternatives are likely to generate the same outcomes as vertical integration. Second, pre-litigation remedies are properly taken into account in the liability phase, and merging parties need only show that the remedy “sufficiently mitigated the merger’s effect such that it was no longer likely to substantially lessen competition.” In other words, the court rejected the FTC’s “total negation” standard.
Basic Facts
Galleri is a blood test that relies on Illumina’s sequencing technology. While Galleri is the only multi-cancer test of its kind on the market as of late 2023, other tests are in the process of seeking approval. These other tests will also rely on Illumina sequencing. The FTC alleged that the combined firm would disadvantage Grail’s rivals by withholding access to Illumina’s next-generation gene sequencing (NGS) or worsening the terms offered to rivals. Under this theory, rival tests would be less available, which would be expected to drive incremental sales of the Galleri test. Prior to the FTC’s liability finding, Illumina made an “Open Offer” to continue to supply its NGS platform at the same price and with the same access as Illumina provided to Grail.
Market Definition
Interestingly, the FTC argued for a broader product market than the one offered by the parties (specifically, a market that included not only existing commercial cancer test companies but also those in the R&D stage, yet not present in the consumer market). While the agencies often allege narrow markets (or at least narrower than the parties), the FTC’s approach in this matter may be consistent with the agencies’ new 2023 Merger Guidelines’ expansive approach to potential competition, which would capture (at least in concentrated markets) “the loss of even an attenuated source of competition”.
Vertical Merger Analysis: Brown Shoe & EDM
Disappointingly, the Fifth Circuit declined to resolve the issue of whether the Brown Shoe factors for vertical harm (e.g., the trend toward concentration, potential foreclosure share, the nature and purpose of the merger) can properly be used as an alternative to the “ability-and-incentive” standard that is grounded in modern economics (an issue raised in Commissioner Wilson’s concurring opinion). While the failure of both the Commission and the Fifth Circuit to reject the Brown Shoe standard is disappointing, it’s significant that the weight of both decisions is grounded in finding both ability and incentive, which is consistent with our experience of DOJ and FTC staff focusing on ability and incentives. This said, the 2023 Merger Guidelines do appear to have adopted the Brown Shoe criteria as a possible alternative to the ability-and-incentive standard (Section 2.5.A.2).
On EDM and efficiencies, the court noted that the Fifth Circuit “has never addressed the threshold question” of whether efficiencies can be a defense in a Section 7 case. The court declined to decide the issue, instead assuming for the sake of argument that such a defense can be properly considered.
Of significance is the court’s rejection of the FTC’s determination that Illumina’s EDM claim lacked merger-specificity. The court reasoned that the FTC’s finding was “based on evidence demonstrating that Grail had considered other ways to reduce or eliminate the royalty without merging with Illumina, such as a buyout or longer-term supply agreement,” yet the FTC failed to “fairly consider evidence that Grail—in coordination with its bankers at Morgan Stanley—had determined that it lacked the leverage necessary to bring Illumina to the table on these alternative proposals, leaving merger as the only realistic option.”
This is an important holding on the issue of contracting as an alternative to vertical integration. We have discussed in prior writings (see, e.g., here) the fact that contracts such as quantity forcing and two-part tariffs do not easily generate the same outcome as a vertical merger due to factors such as demand uncertainty, risk aversion, information asymmetries, and a variety of incentive or moral hazard problems such as holdup and freeriding.
The court did, however, agree with the FTC on pass-through, reasoning that, because the merger would increase Illumina’s incentive to foreclose against Grail’s rivals, “Grail had no reason to pass [on] its royalty-reduction savings […] because, if any of Grail’s competitors actually made it to market, Grail could force those competitors to pass through extra costs to their customers.” This reasoning appears to overlook a basic insight from economics that a profit-maximizing firm will respond to lower costs (e.g., by no longer paying a royalty) with lower prices, regardless of the degree of its market power.
Finally, the court agreed with the FTC that Illumina’s claimed cost savings (e.g., EDM and “significant supply chain and operational efficiencies”) were not verifiable, faulting the parties for failing to put forth sufficient economic evidence (namely, a proposed (as opposed to “illustrative”) model for calculating EDM and any model at all for the efficiencies claim). As to Illumina’s contention that it was the FTC’s burden to model the EDM benefits, the court noted that “when it comes to efficiencies, ‘much of the information relating to efficiencies is uniquely in the possession of the merging firms’” (quoting Areeda & Hovenkamp). One takeaway here is that merging parties should be prepared to engage with a plaintiff’s economic model and to understand the combined effect of any harm within that model and the procompetitive benefits of the merger.
Remedies or “Litigating-the-Fix”
This part of the decision is mixed for merging parties: it’s troubling in that it puts the burden of proof on merging parties when it comes to pre-litigation remedies; however, it’s positive in that it rejects the FTC’s position that remedies must totally redress competitive harm (“total negation standard”). Instead, the court adopted a more lenient “substantially lessening competition” standard (at least with respect to pre-litigation remedies not conditioned on a finding of liability).
The Fifth Circuit’s decision conflicts with the Ninth Circuit’s July 2023 decision in Microsoft/Activision, in which the court required the FTC to consider Microsoft’s binding commitments not to deny access as part of its prima facie case. In contrast, the Fifth Circuit held that Illumina bore the burden of showing that the remedy rebutted the prima facie case. The court did, however, hold that such a showing was properly considered as part of the liability phase, rejecting the FTC’s position that pre-litigation fixes cannot be considered until the remedy phase. The Ninth Circuit’s approach in Microsoft/Activision makes sense, given that the goal of antitrust merger analysis is to identify mergers that are likely to harm competition and increase price (or harm innovation). It is counterproductive to this goal to fall back on structural or legal technicalities when a remedy could prevent such outcomes.
IQVIA/Propel
This decision turned on market definition and evidence on customer substitution patterns, with the parties’ own documents and data playing a significant role in the court’s decision to adopt the FTC’s narrow market for programmatic advertising healthcare professionals (HCPs). From there, the court found that the parties’ combined share triggered the 30%-share threshold for the presumption of illegality set forth in the Supreme Court’s Philadelphia National Bank decision. Win/loss data played a central role, with the court crediting the FTC’s interpretation of the data as showing that most customers substituted between different HCP-focused platforms, despite the parties explaining that the data was incomplete in ways that precluded drawing reliable inferences (e.g., limited to wins/losses in the context of RFPs and failed to take into account multihoming).
Basic Facts
IQVIA’s Lasso and Propel Media’s DeepIntent have been two of three leading healthcare-focused demand-side platforms (DSPs), which facilitate the placement, targeting, and measurement of ad campaigns across the Internet targeted at specific sets of healthcare providers. The court was ultimately swayed by the argument that healthcare DSPs offer features that may be difficult or costly to implement using other forms of advertising, including the targeting of ads to specific subsets of HCPs (including individual doctors) in a variety of online contexts and the measurement of such ads using data on physician prescriptions. In addition to healthcare DSPs, healthcare advertisers use other media to place ads targeted at HCPs, including social media, search ads, generalist DSPs like Google Ads, and direct placement on healthcare-focused websites like WebMD.
Win/loss Data and Imperfect v. Unreliable Information on Customer Substitution
The defendants argued that the win/loss data were unreliable. For instance, the defendants argued that DeepIntent’s win/loss data included only RFPs sent to DeepIntent and thus mechanically excluded opportunities where the client elected to use social media, which would not necessitate sending RFPs at all. Further, the defendants presented evidence that the FTC’s analysis of the win/loss data would characterize some losses as going entirely to IQVIA when, in fact, the advertiser split spend across various platforms. Ultimately, the court concluded that the data were reliable, pointing to evidence that healthcare DSPs rely on win/loss data and that a hypothetical monopolist healthcare DSP would profitably increase price even if substitution within healthcare DSPs were much lower than that apparently suggested by the win/loss data. Thus, the court ultimately agreed with the FTC that a healthcare DSP market satisfied the HMT while granting that the parties had made some compelling points. From there, it was direct for the FTC to convince the court that the parties had high shares of the established market, indicating harm under both structural criteria and a more sophisticated merger simulation.
While, as a matter of economics, it’s not necessary to define a relevant antitrust market to analyze the ultimate question of likely competitive effects, still, market definition can, in some cases, be a useful tool. This includes as a means to help cut through ambiguous or contradictory evidence as to customer substitution patterns. The hypothetical monopolist test can define quite narrow markets, even if there exists material substitution outside of the set. Such a rigid “in/out” approach to market definition can fail to take into consideration real-world competitive constraints, and indeed, the FTC’s modeled price increases and harms within the healthcare DSP market relied on measured shares within that market as inputs. While such an approach is common, accurately measuring substitution to products outside of a defined market is a vexing and important problem, as significant substitution to out-of-market goods can materially lower forecast price effects, both in absolute terms and relative to any expected efficiencies resulting from a merger.
JetBlue/Spirit
The district court found that Spirit represented a unique competitive constraint on the “Big Four” U.S. airlines that would be lost if Spirit were absorbed into JetBlue, despite the parties’ modest shares, the existence of other bargain carriers like Spirit, and the potential for the merger to strengthen the competition between JetBlue and the Big Four. The court was particularly concerned that industry-wide capacity constraints would limit the potential for other carriers to replace Spirit’s fares (including those slated to purchase divested airport slots) and exacerbate the effect of removing seats from Spirit planes once reconfigured to be JetBlue planes. The court also cited capacity constraints in dismissing the DOJ’s potential competition markets.
The “JetBlue Effect” v. the “Spirit Effect”
At trial, the parties presented themselves as small upstarts looking to become stronger competitors against the Big Four airlines. In contrast, the DOJ depicted Spirit’s unbundled, “no frills” product as distinct from JetBlue’s more robust product, with both airlines constraining each other’s prices and those of the Big Four. The court found both visions compelling to some extent. On the one hand, according to the court, the merger would have increased the parties’ modest nationwide shares—3% for Spirit and 7% for JetBlue—and a larger combined airline with more routes would “immediately place more pressure on its greatest competitors, the Big Four.” One mechanism for this greater competitive pressure would be the substitution of JetBlue’s relatively higher quality for Spirit’s no-frills approach, which would make Spirit’s current routes more substitutable for those of the Big Four.
On the other hand, the court found that both JetBlue, a low-cost carrier (LCC), and Spirit, an ultra-low-cost carrier (ULCC), exercise distinct constraints on other airlines and on each other. JetBlue lowers costs by flying limited routes using a homogeneous fleet, while offering many features meeting or exceeding those offered by the Big Four. Spirit also controls costs by flying a limited set of routes, while additionally unbundling most ancillary services from its base ticket, including carry-on bags and seat assignments, allowing it to offer dramatically lower base fares relative to the Big Four airlines. Both airlines offer low fares, with Spirit additionally competing for price-sensitive fliers willing to give up ancillary services.
The court cited empirical work finding that average fares on a route decrease following entry by either JetBlue (coined the “JetBlue Effect”) or Spirit (the “Spirit Effect”). The court’s overarching approach is summarized in a footnote stating that “much of this litigation, in fact, centered on whether the JetBlue Effect is as strong as the Spirit Effect.” If so, the parties’ contention that a larger JetBlue would compete more strongly with the Big Four would likely receive greater weight. If not, replacing the “Spirit Effect” with the “JetBlue Effect” would lessen competitive pressure on other airlines and thus increase prices.
Spirit’s “Unique” Business Model
In finding for the DOJ, the court found that Spirit’s business model was “unique” and “would be exceedingly difficult for another airline, or a combination of other airlines, to replicate,” and absorbing Spirit into JetBlue would “harm cost-conscious travelers who rely on Spirit’s low fares.” It also found that absorbing Spirit may lessen the “JetBlue Effect,” by “further consolidat[ing] an oligopoly,” thereby lessening JetBlue’s incentive to compete on price.
Capacity Constraints, Harm, and Potential Competition
The presence of industry-wide capacity constraints (for instance, Airbus is not allowing new orders until 2029) clearly affected the court’s thinking. As a matter of economics, while capacity constraints on merging firms can mitigate merger price effects, capacity constraints on non-merging firms can exacerbate merger effects by dampening competitors’ responses to any merger-induced changes. According to the decision, JetBlue’s planned conversion of Spirit planes to roomier JetBlue configurations would have removed 11% of Spirit’s seats from the market, and the court was skeptical that other ULCCs could make up this lost capacity given their capacity constraints, even those slated to purchase divested JetBlue airport slots (JetBlue had not proposed divesting any of its planes). Further, the court found that given capacity constraints, any competitor repositioning would necessarily come at the expense of decreased capacity on another route, which could result in higher prices on that route.
The DOJ alleged a number of potential competition markets defined around routes that Spirit had not yet entered but planned to. The court rejected all of these markets as “not relevant,” citing the unlikelihood that Spirit would actually enter those routes, given capacity constraints.
Head-To-Head Competition and Competitive Effects
The court’s comparison of the relative merits of the parties’ business models may have overshadowed the more traditional question of the competitive constraints JetBlue and Spirit exercise on each other. The parties overlap on 99 nonstop direct routes between metropolitan areas, or 30-40% of their combined routes. As an empirical matter, to the extent that Spirit competes with JetBlue, the “Spirit Effect” may (or may not) amplify the “JetBlue Effect,” and vice versa. Ultimately, the court found that the post-merger “JetBlue Effect” would decrease in magnitude, resulting in higher prices.
The DOJ alleged competitive harm on overlap routes, routes served by Spirit only, and routes on which Spirit was a potential entrant, with harm flowing both from the loss of head-to-head competition between the parties and from the removal of seats on Spirit planes converted to JetBlue planes. The court accepted markets defined around all but the potential competition routes, and found that harm would result in at least some of those markets, even after accounting for divestitures and competitor responses. Specifically, the court cited a “large category [of] consumers […] who must rely on Spirit” that would be harmed and saw it as unlikely that competitors would make up even a fraction of the 6.1M annual seats that would be lost from switching Spirit planes to JetBlue’s lower capacity configuration.
While the court referred to the Big Four as JetBlue’s “greatest competitors,” it also agreed with the DOJ that anticompetitive unilateral effects do not necessarily depend on merging parties being each other’s closest competitors.
Amazon/iRobot
In January 2024, Amazon terminated its merger agreement with robot vacuum maker iRobot, originally signed in August 2022, in the wake of a Statement of Objections issued by the European Commission (EC) and an apparent FTC staff recommendation that the transaction be blocked. The agencies were apparently concerned with the potential for vertical foreclosure, meaning that Amazon would prevent other sellers of robot vacuums from listing on amazon.com, or would worsen terms for those sellers. The EC found that “it may be economically profitable” to foreclose iRobot’s rivals, as “[t]he merged entity would likely gain more from additional sales of iRobot [robot vacuums], than it would lose from fewer sales of iRobot’s rivals” and cited “additional data gathered from iRobot’s users” as one source of gains.
The vertical arithmetic described by the EC—comparing Amazon’s retail margin with its would-be robot vacuum margin—is a standard tool used to assess the potential for vertical foreclosure. It is likely that the EC and the FTC considered the potential for Amazon to steer customers to iRobot’s models who otherwise would have purchased a rival robot vacuum. Such steering may lower Amazon’s profits (to the extent it decreases profitable sales of rival robot vacuums) or may increase Amazon’s profits (to the extent it generates incremental sales of iRobot vacuums). The EC evidently concluded the latter effect would dominate.
One problem with strict reliance on static vertical math models is that they are limited to calculating diverted sales today and don’t take into account dynamic effects, such as an acquiring firm’s plans to grow the business and how that might affect the profitability analysis.
As we have described previously, the EDM can be a procompetitive effect of vertical mergers. While EDM arises from the same economic forces as do incentives for foreclosure, the EC does not appear to have considered EDM. Amazon itself appears to be the seller of the latest iRobot model on amazon.com, which likely means that Amazon purchases iRobot’s products at a wholesale price marked up above iRobot’s cost of production and then resells the product at an additional markup. Vertical integration would allow Amazon to access iRobot’s products at the cost of production and would incentivize Amazon to lower the retail price of the iRobot product.
It is somewhat difficult to understand the thinking of the EC and the FTC without the benefit of additional information. There are many routes to market for robot vacuums, and there are many instances of store brands owned by a retailer competing with third-party brands at the same retailer. It seems unlikely that even if Amazon were to disadvantage rival robot vacuums, that such foreclosure could deny scale to those rivals, cause them to exit, or materially harm consumers, because the rivals could simply sell through other retailers. Consumers would likely benefit from EDM, and some consumers valuing Amazon’s home automation products may have benefited from the integration of iRobot’s products with other Amazon products. Finally, iRobot reported an operating margin of -14.2% in 2022, the most recent full year for which it has reported results. A negative margin should be expected to mute the incentives of Amazon to disadvantage retail sales of other robot vacuums (on which it would earn a positive retail margin) in order to sell more iRobot vacuums (on which it may well lose money at prevailing margins).
Jeremy Sandford and Koren W. Wong-Ervin
The authors thank Ben Baek for his research assistance. The views and opinions set forth herein are the personal views or opinions of the authors; they do not necessarily reflect the views or opinions of the organizations with which they are affiliated or those organizations’ management, affiliates, employees, or clients.
***
| Citation: Jeremy Sandford and Koren W. Wong-Ervin, Antitrust Antidote: December 2023-February 2024, Network Law Review, Winter 2023. |
