Antitrust agencies can remedy horizontal mergers that reduce innovation competition. Merger guidelines in the United States and the European Union say so. One of the hardest questions is not legal, but economic: how do horizontal mergers reduce innovation competition? Guidelines say little about the mechanics of harm to innovation, and agencies generally do not describe their innovation concerns with great specificity in merger challenges.
A thorough review of the cases suggests that two theories of harm to innovation competition have been applied in merger enforcement by the European Union (and, to some degree, by the United States). The first theory of harm is structural. It insists on the post-merger reduction in the number of independent R&D projects, assets or capabilities at industry level. The lost rivalry due to the merger may hinder innovation and in turn reduce competition in future products. In the United States, this framework of analysis has previously led antitrust agencies to delineate specific R&D markets (sometimes referred to as innovation markets) for future products or classes of products and to assess the number of firms remaining on these markets post-merger to determine whether the transaction substantially lessens competition, effectively retrofitting innovation markets into the concentration matrices we use for horizontal analyses. More recently, the focus of structural analysis has shifted from firms’ R&D projects toward greater emphasis on firms’ innovation capabilities. A similar approach has been followed in the European Union. In both jurisdictions, a forward looking identification of specific future product applications or research areas has been required.
The second theory of harm is less mainstream. It deems a reduction of innovation competition possible as a result of a “cannibalization” effect. The difference from the first theory is more one of perspective than of nature. Instead of a structural analysis of the number of post-merger independent R&D capabilities at industry level, the focus is on a reduction in internal R&D rivalry between the merging parties. The theory here is that an invention will cannibalize the merged entity profits in proportions greater than would result in a more competitive market structure. In this sense, the cannibalization theory echoes in some sense a unilateral effects theory focusing upon proximity of competition, without engaging directly in a unilateral effects analysis. In Novartis-GlaxoSmithKline Oncology Business, the EU Commission came close to this unilateral effects approach when it observed that the merged entity “will internalise that investing in one of the clinical research programs [for colorectal cancer] can be expected to cannibalise future sales of its other clinical research program.”
A distinct variant of both theories of harm to innovation competition has, however, emerged on the occasion of the EU review of the Dow/DuPont merger. The Commission ordered the divestiture of DuPont’s entire global R&D business on the ground that the transaction “would be likely to significantly impede effective competition as regards innovation both in innovation spaces where the Parties’ lines of research and early pipeline products overlap and overall in innovation in the crop protection industry.”
Unstated in the Commission’s formulated theory of harm to innovation competition in Dow/DuPont are two significant variations from past practice. First, the Commission’s intervention entails a broadening of the way innovation competition harms are evaluated. While the decision looks at the loss of rivals with key innovation capabilities for competing in the future, it does not seem to regard as a necessary requirement the identification of specific future products applications or R&D areas. At one extreme, a mere showing that the industry will feature one less firm that is vertically integrated into most or all significant aspects of research and development may be an actionable loss to innovation competition under a substantial lessening of competition test (or its European equivalent, the significant impediment to effective competition).
Second, the Commission’s decision marks a conspicuous attempt to shoehorn theories of harm to innovation competition within the unilateral effects model conventionally applied in horizontal merger cases. At paragraph 2043 of the decision, innovation effects are characterized as a “standard unilateral effect from a merger.” In addition, Annex 4 to the decision, entitled “Implication of the economic theory on mergers, competition and innovation in light of the features of the Transaction,” sets out more fully how “[a] merger in innovative industries generates standard unilateral effects in innovation.”
This double expansion and simplification raises complex questions of antitrust policy. Can we confidently abstract away from specified product markets or R&D pipelines in the merger analysis of innovation harms without falling prey to uncertainties that risk being fatal to any decision in the particular case? And can we predict a reduction of innovation competition simply by substituting price for R&D decisions in the standard model of unilateral effects applied in U.S. and EU merger analysis? What other metrics might be relevant to a unilateral effects analysis of innovation competition, such as proximity of product functionality, or the funding and decision-making structures of the R&D organization?
The answers to those questions are not simple. The unilateral effects theory is quite robust. It is well fitted to test a merged entity’s ability and incentives to make adjustments to marginal variables, such as price or output, and nonprice variables, such as advertisement expenditure or even quality. But are its explanatory and predictive powers equally strong in relation to “lumpy” decisions like the shutting down of entire lines of R&D research, or even incremental decisions like funding toward R&D milestones or project phases? Doubts are permitted because exit from sunk R&D investments is not a frictionless process. Moreover, the decisional costs of R&D entry or exit—and therefore firms’ path dependence toward past R&D choices—may be even greater for public companies. Accountability to shareholders raises the costs of R&D entry or exit decisions.
Or consider welfare trade-offs. Absent efficiencies, the unilateral effects model determines that a post-merger price increase in a relevant market always reduces welfare. But does this apply equally to a post-merger decrease in R&D expenditures in a relevant product area? Again, the answer is uncertain. Because R&D exit decisions can be lumpy, in the sense that they can be go/no-go decisions, the model should consider the possibility of balancing the welfare costs of a reduction of R&D expenditure in a specific research area with the welfare benefits arising from the repurposing of fixed R&D inputs in other research areas, and from the new division of innovation labor within the merged firm.
Last, the stylized analysis conducted under the unilateral effects framework in relation to R&D pays no heed to heterogeneity at the firm level. This empirical vulnerability undermines the ability of the unilateral effects model to produce meaningful predictions when theories of post-merger loss of innovation due to delayed or deferred R&D projects are scrutinized. This problem is likely to be particularly salient when merging parties display a degree of intrafirm organizational rivalry or are active in industries where R&D is more serendipitous than goal-oriented.
These questions are hardly unimportant. Since 2004, the U.S. agencies have identified innovation concerns in approximately one-third of their merger challenges, and innovation challenges are raised in approximately 80 percent of the cases concerning high R&D industries. In the European Union, 10 of the 73 remedied cases over the period 2015–2017 involved allegations of harm to innovation.
In this article, I begin with a brief overview of the new EU theory of harm to innovation competition, and I show that it constitutes a small but significant change in merger policy. For ease of exposition, I refer to this theory as “significant impediment to effective innovation competition” (SIEIC). I then analyze the economics of SIEIC and provide a critique. To the extent that SIEIC is an application of the standard unilateral effects model where the focus is shifted from price to innovation effects, I demonstrate that this variant of the theory can only deliver sound and robust empirical predictions if three critical innovation-specific questions are addressed. I conclude with a discussion of the economic methodology of merger control policy. I argue that antitrust agencies should remain free to rely on new or adjusted versions of economic models in merger control, subject to the condition that they are able to discharge the burden of persuasion. This issue is illustrated, and discussed, on the basis of EU law doctrine.
Continue reading the full text of this article in PDF format.
The author would like to thank Jonathan Baker, Gregory Werden, David Teece, Jesse Solomon, Richard Gilbert, and Gregory Langus for comments on this article, as well as anonymous referees who gave remarks to improve a previous draft.
