Innovation Competition and Innovation Effects in Horizontal Mergers: Theory and Practice in the United States and European Commission

A. Literature Review

To address innovation competition cases, Gilbert & Sunshine propose the Innovation Market Analysis (IMA): a five-step procedure focused on competition through R&D which requires an innovation-relevant market definition—an approach already applied in Roche/Genentech (FTC-1990) ⁹ —and looks into concentration in R&D. ¹⁰ The methodology adapts the Hypothetical Monopolist Test ¹¹ to define an innovation market: “In general terms, an innovation market is defined as a set of activities and a geographical area in which a hypothetical monopolist would impose at least a small but significant and nontransitory reduction in R&D effort.” ¹² This approach identifies overlapping R&D activities, alternative sources of R&D (including firms that could acquire the necessary assets for R&D in a reasonable time), downstream competition, the merger’s impact on R&D, and potential R&D efficiencies. It is worth noting that the IMA does not emphasize the competitive pressure from product market rivals but instead considers pressure from pipeline projects ¹³ and capable competitors to identify substitutes that can constrain the merged entity’s market power.

Despite its advances, IMA has been subject to criticism, as listed by Katz & Shelanski and Kerber & Kern. First, some authors argue that IMA’s approach is unnecessary, as potential competition assessment could consider innovation effects ¹⁴ and future goods market analysis. ¹⁵ The second is the IMA presumption of negative effects on innovation. ¹⁶ Third, enforcement might be less predictable. ¹⁷ Fourth, the effects on non-price variables may not have a legal basis. ¹⁸

Katz & Shelanski outlined three canonical situations for assessing mergers. First, when firms are competitors in the product market (and engage in R&D efforts to create or improve their products) or are expected to be in the future (when their R&D efforts are finished), merger analysis should focus on product market competition rather than R&D, as harm results from the loss of future product market competition. Second, when there is an innovation-based race to market dominance, the merger does not affect product market competition, and authorities should preserve R&D competition to avoid a reduction in innovation incentives that could delay the market launch or result in a product with fewer benefits. For instance, in a winner-take-all market, the first firm to complete its product development becomes a monopolist. The authors considered two cases as the opposite ends of a continuum. The first case focuses on the effects of product market competition, whereas the second only considers R&D competition. The focus of the assessment depends on how close the innovation is to the market launch. ¹⁹

The third canonical situation occurs when innovation is not protected from imitation, leading to a waiting game among firms as they anticipate their rivals to replicate their efforts. In this case, authorities face a trade-off between static and dynamic factors. Approving the merger would solve the free-riding problem but concentrate on the product market, resulting in static price effects. Blocking the merger would maintain product market competition but continue the waiting game. ²⁰

Farrell and Shapiro ²¹ and Shapiro ²² contribute to assessing innovation effects similar to the estimation of unilateral price effects (which considers variables that represent direct competition, such as product substitutability, diversion ratios, and markup of diverted sales). The innovation diversion ratio is introduced in the former paper as an index that measures the diversion of a firm’s profit when a competitor innovates. ²³ In the latter, Shapiro presents three guiding principles for studying the relationship between competition and innovation, and the first one, the Contestability Principle, suggests that innovation incentives arise from the potential to gain or protect sales through greater value, also similar to the mechanism of assessing unilateral price effects. The other two principles are the Appropriability Principle, which examines how effectively a successful innovator can capture the social benefits arising from its innovation, and the Synergy Principle, which is based on the enhancement of innovation capabilities by combining complementary assets and the consequential spur to innovation. ²⁴

Federico et al., also considering the assessment of innovation effects in a similar manner to the last two authors, define what they call business-stealing effects. They defined it as the process of gaining and protecting sales from rivals by providing value to consumers through means including not only price but also innovation. ²⁵ Thus, merging firms’ internalization of innovation-related business-stealing effects, achieved through their innovation efforts or the existence of similar capabilities and research lines, would reduce innovation incentives. This is because innovation efforts focused on one of the merging parties’ products (previously owned by independent firm A) would cannibalize the sales of the other product (previously owned by independent firm B), resulting in adverse innovation effects in the event of a merger. The closer and more substitutable the firms’ innovation efforts are, the stronger the business-stealing effects of the merger, leading to greater harm to innovation.

The authors also presented three patterns of mergers that require specific assessments based on unilateral effects theories. The first pattern is pipeline overlaps, which involve at least one pipeline project (product-to-pipeline or pipeline-to-pipeline overlaps). Similar to Katz & Shelanski, the authors suggest that the assessment should focus on product market competition if the pipeline is close to market launch, that is, no significant costs need to be spent on complete development. If the pipeline still requires development and innovation incentives, the assessment should preserve these incentives. The second pattern is overlapping capabilities, in which parties place business-stealing effects on each other related to new innovation efforts in similar areas of research. Mergers may reduce future innovation efforts and internalize these effects. The third pattern is the dominant firm acquisitions of smaller innovative firms, particularly directed to digital markets, where the acquiring firm is a platform such as Google, Facebook, Apple, or Microsoft. Potential harm may result from the loss of competitive pressure and possible discontinuation of an innovative product. ²⁶

Lyra & Pires-Alves add another category of cases to those addressed by Federico et al.: when at least one of the parties involved in product market competition engages in continuous innovation efforts (mostly incremental innovations) to gain market share by improving their own products, the merger may harm innovation by eliminating that player. The assessment focuses on firms’ innovative behavior and the reduction in incentives for new innovation efforts in the product market. ²⁷

B. Lessons from the Literature Review

Although the exact measurements of unilateral innovation effects are not feasible given the uncertain nature of innovation, lessons learned from the literature suggest three guiding principles for the proper assessment of horizontal mergers involving innovation competition: (1) business-stealing effects, (2) capabilities, and (3) dynamic effects principles.

First, the same unilateral effect theories used to assess price effects can be applied to the analysis of innovation effects. As suggested by Farrell & Shapiro, Shapiro, and Federico et al., the closer the products, innovation efforts, and/or capabilities, the more likely there will be a reduction in innovation incentives.

Second, according to Gilbert & Sunshine, analysts should identify overlapping R&D efforts of merging firms and alternative sources of R&D to define an innovation market, including firms that could acquire the necessary assets for R&D quickly, ²⁸ an effort to consider the firms’ capabilities in the assessment. Even when discussing business-stealing effects and de-emphasizing the role of relevant market definitions, analysts need to consider which firms are the parties’ competitors and the competitive significance of such firms. Sidak & Teece, Kerber, and Lyra & Pires-Alves suggest building a capabilities-based merger assessment that considers a firm’s ability to innovate to capture the competitive pressure faced by merging parties in innovation competition cases.

Finally, Federico et al. highlight that competition may not only occur between products and pipelines but also between firms with overlapping capabilities and lines of research, as they impose business-stealing effects on each other. Mergers between such firms could reduce incentives to innovate in these areas. As emphasized by Lyra & Pires-Alves, competition would be harmed in a longer and unforeseeable time horizon in such cases, compared to short-run price effects and innovation effects from pipeline project interruption, as the likelihood of new innovation efforts in that research area could decrease. ²⁹ Thus, a proper assessment of innovation competition cases must consider these dynamic effects on different time horizons.

A. The U.S. and EC Experiences

B. The Four-Layer Competitive Assessment and the Novel Approach on Innovation Competition in EU Merger Control

Dow/Dupont merger (2017) ⁵⁴ marked a significant shift in EU merger control. ⁵⁵ According to Todino et al., ⁵⁶ prior to this case, there was a traditional approach to assessing mergers with innovation concerns. Late-stage pipelines that could potentially compete with incumbents were assessed for anticompetitive effects only if they constrained the incumbent’s behavior or were likely to enter the market soon, that is, if they were developed enough to be considered potential competitors. The pipeline products would have to meet one of the following criteria: (1) they already exert constraints on the incumbent’s behavior, or (2) they are likely to enter the market in a relatively short period of time and then constrain the rival’s behavior. Insufficient competition was also needed for considering the existence of anticompetitive effects. However, as the authors argue, the approach has changed gradually, as seen in three cases: Medtronic/Covidien (2014), Novartis/GlaxoSmithKline Oncology Business (2015), and Pfizer/Hospira (2015). According to Todino et al., these cases led to the approach applied in Dow/Dupont.

When assessing Dow/Dupont, the Commission identified four types of overlaps: incumbent products in many markets, incumbent and potential competitors, early pipeline projects and lines of research, and global R&D-integrated organizations. The first two are related to price/product competition, whereas the latter two are related to innovation competition. ⁵⁷ When addressing innovation competition, the Commission used the concept of innovation spaces—discovery targets pursued by firms—to assess possible overlaps in different stages of product development, including the discovery of new active ingredients (competition in innovation spaces may occur at the discovery and development stages that precede pipeline phases). The parties had many overlaps in developing products, early pipeline projects, and lines of research, which could divert revenue from one another. The reduced innovation efforts and capabilities to innovate could result in a reduction in incentives to continue ongoing innovation efforts, possibly discontinuing, delaying, or redirecting early pipeline products and lines of research, as well as reduced incentives to undertake future innovation efforts. ⁵⁸

Dow/Dupont (2017) marks the application of a new approach to merger control and represents a major shift that has led to intense debate among academics, practitioners, and authorities. ⁵⁹ This approach to assessing the impact of mergers on innovation competition refers to whether the merger results in a Significant Impediment to Effective Innovation Competition (SIEIC), named after the Significant Impediment to Effective Competition (SIEC) test applied by the Commission to check whether a specific operation would be anticompetitive. ⁶⁰ To Chadha, the 2004 ECMR and EC’s HMG set the roots for the new approach by applying the notion of SIEC to unilateral effects and by assessing innovation effects in merger control. ⁶¹ Other authors also refer to it as the innovation theory of harm (IToH), and the procedure itself was later called the four-layer competitive assessment in Bayer/Monsanto (2018) ⁶² and AbbVie/Allergan (2020). ⁶³

The four-layer competitive assessment involves examining four types of overlap, considering both price/product competition and innovation competition. The underlying principle of the procedure is to assess product/price effects related to marketed products and late-stage pipeline products, as well as innovation effects related to pipeline products in earlier stages of development, while also defining innovation spaces. The EC checks overlap with the following:

Regarding innovation competition, in the third and fourth layers, the competitive assessment is similar to unilateral effects theories as it is concerned with business-stealing effects: the competitive pressure between the firms is significant, and they are close competitors. The theory of cannibalization developed in GSK/Novartis was changed to include diversion related to innovation efforts and future products instead of only incumbent products. ⁶⁶ In addition, the importance of diversion of sales in spurring innovation is highlighted. ⁶⁷

Second, the competitive significance of the parties and their rivals is evaluated based on their innovation capabilities. In Dow/Dupont, patent shares and new active ingredients are used as indicators of the strength of these capabilities. ⁶⁸ Therefore, the EC’s assessment focuses on the business-stealing effect mechanism as an indicator of possible harm to innovation and firms’ capabilities as indicators of the market they compete in, the innovation spaces they operate in, and their competitive significance.

Third, it expands the possibilities of overlaps to pre-pipeline stages by examining competition in innovation spaces, a step further from the 1990s innovation competition cases that considered pipeline competition. Finally, it assesses harm by examining the effects of the merger on incentives for ongoing innovation efforts and the industry’s incentives to innovate for the merging parties and their rivals.

The four-layer competitive assessment is subject to criticism. Chadha lists the following: (1) not robust enough to analyze dynamic factors; (2) potential over-reliance on patent data; (3) effectiveness of remedies such as those applied in Dow/Dupont (2017); (4) failing to balance appropriability and cannibalization; (5) difficulties in satisfying the EC high standard of proof; and (6) asymmetry in addressing positive and negative innovation effects. ⁶⁹

A. Summary and Discussion of the Propositions for Merger Assessment

Before discussing these propositions, it is useful to present the two channels of harm to innovation listed by Kokkoris & Valletti. ⁷⁰ First, there may be less incentive to continue current product development, delaying and/or interrupting these innovation efforts. Second, incentives to begin new innovation efforts may be reduced, resulting in less future innovation.

Five approaches to assessing innovation effects were discussed in the previous sections: Gilbert & Sunshine, Katz & Shelanski, Federico et al., Lyra & Pires-Alves, and the four-layer competitive assessment.

The approaches discuss three categories of innovation competition cases: overlaps in pipelines, capabilities, and the product market when there are continuous innovation efforts. All of them may be assessed under the business-stealing effects principles and are not exclusive: a merger may be assessed under more than one category, as the four-layer competitive assessment clarifies.

First, overlaps in pipelines, both product-to-pipeline and pipeline-to-pipeline, may result in the delay or discontinuation of the ongoing innovation effort, the first channel of harm to innovation. Federico et al. and Lyra & Pires-Alves suggest that three types of evidence should be considered: business-stealing effects, competitive pressure from rivals, and time to market launch. To evaluate the extent of business-stealing effects, the degree of substitutability between the parties’ products should be examined, along with evidence of the current and future profitability of the incumbent product, the expected duration of overlap between the products in the market, and the remaining time of patent protection. To assess competitive pressure from rivals and potential rivals, the history of the parties in bringing innovations to the area, patent portfolios, durable barriers to entry, the degree of cumulativeness of innovative successes, and similar core capabilities and competences should be considered. ⁷¹ The time to market launch should also be taken into account by examining the development status of the product, which varies depending on the industry’s innovation process.

A few observations regarding the assessment of these cases can be drawn from the literature. A merger involving late-stage pipeline cases should only be assessed in terms of price competition, as there is no risk of harm to innovation, and a capabilities-based assessment may be applied to identify firms’ rivals and their competitive significance through their capabilities, cases where the product market exists (which also considers price effects and the interaction between them and innovation effects), and when it does not may be distinguished.

Second, overlaps in capabilities need to be assessed, as they could harm innovation in a longer term. Harm to innovation may occur through the second channel, as there may be less incentives to begin new innovation efforts in the post-merger scenario considering that line of research. Evidence in these cases can be divided into two groups: (1) assessing the extent of business-stealing effects between parties and (2) determining whether there is effective rivalry capable of exerting competitive pressure in that line of research. For both groups, overlapping capabilities and lines of research, history of innovation efforts by the parties and competitors, product and pipeline overlaps, patent portfolios, durable barriers to entry, degree of cumulativeness of innovative successes, and similar core capabilities and competences can be considered. As in the last case, a capabilities-based assessment may be applied to better assess the case.

Third, firms that compete in the product market and engage in continuous innovation efforts to improve their products may have their mergers assessed to prevent harm to innovation. It may also occur through the second channel of harm to innovation in the form of less incentive to innovate in the post-merger scenario, although in the product market. The concern is the impact of eliminating such innovators and the business-stealing effects they may have on their competitors. Evidence of the extent of business-stealing effects can be based on the degree of substitutability between the parties’ products, the current and expected profitability of diverted sales, and the parties’ history of bringing innovation to markets. In addition, assessing merging parties’ status as frequent innovators or innovation mavericks can be useful in evaluating the competitive pressure exerted on rivals. ⁷²

B. Selected Cases and Discussion

Considering the three categories of cases and the U.S. and EC Merger Control experiences, specific cases can be analyzed to determine how they fit into these forms of innovation competition and explore the assessment methodologies used. Criteria such as (1) representing different industries, (2) not being subject to simplified procedures when assessed by the authorities, (3) presenting debates on innovation effects, and (4) representing different forms of innovation and effects were used to select three case studies for analysis.

The first two cases involve pharmaceuticals and were assessed by the EC because of innovation concerns for treatments of the same diseases. The first case was not assessed explicitly with the four-layer competitive assessment, whereas the second case was. Both cases involved concerns related to pipeline projects. In the third case, assessed by the U.S. Department of Justice, the overlap was between incumbent products without any pipelines involved, and the concern was the removal of an important innovator in the market.