This paper, available here, focuses on the impact of competition policy on innovation. Disruptive firms drive a significant amount of innovation. By making its offer to customers attractive in a new way, a disruptive firm can destroy a great deal of incumbent profit while creating a large amount of consumer surplus. The resulting churn in products and market shares, as new products enter and old ones exit, and as newer business methods and business models supplant older ones, are typical of a healthy competitive process. If that competitive process is slowed or biased by mergers or by exclusionary conduct, innovation is lessened and consumers are harmed.
Competition policy seeks to protect the competitive process by which disruptive firms challenge the status quo, despite the biggest firms being some of the most impressive innovators in many industries experiencing rapid technological change. Innovation is best promoted when market leaders are allowed to exploit their competitive advantages while also facing pressure to perform from both conventional rivals and disruptive entrants. The central theme animating the analysis in this paper is that a market leader is best motivated to innovate if it fears losing its leadership position to a disruptive rival.
Section II delineates some key economic concepts to understand horizontal merger control and innovation, beginning with reasons why mergers can be detrimental to innovation.
In a market economy, competition is best described as a dynamic process of rivalry between firms seeking to attract customers by offering them a better deal. This dynamic competitive process between direct rivals centers on what are generally known as “business-stealing effects.” Business-stealing effects arise ubiquitously because one firm’s gain in customers by offering them better value typically comes at the expense of its rivals.
At times, business-stealing effects arise when one firm undertakes risky investments to develop new and improved products or production processes. Although some innovation is driven by the prospect of serving entirely new uses or capturing sales from highly competitive industries with small price/cost margins, many of the rewards to innovation are commonly driven by the prospect of attracting customers that would otherwise purchase other products with significant price/cost margins.
The importance of business-stealing effects for innovation incentives has long been understood. One of the robust lessons from this literature is that competition from rival innovators acts as a powerful incentive for innovation. A secure incumbent would invest less on R&D than a threatened incumbent, because it does not need to fear losing its business to rivals. Likewise, in the literature on R&D joint ventures, absent spillovers, cooperation between rivals leads to lower innovative efforts because the joint venture internalises the business-stealing effects on innovation.
Unilateral innovation effects in merger control are closely analogous to unilateral price effects, with the focus on firms’ decisions to invest resources to develop new products rather than on their pricing decisions. The first step in assessing possible unilateral innovation effects is to look for innovation-related business stealing effects between the two merging firms. Anticompetitive unilateral innovation effects, just like unilateral price effects, are greatest in situations where the price/cost margins on the relevant products are large and where the business-stealing effects between the two firms are substantial. The simplest and most direct way to measure the unilateral innovation effects associated with a merger between Firm A and Firm B on Firm B’s innovation incentives is to calculate the innovation diversion ratio. This ratio is defined as the expected lost profits at Firm A caused by successful development of Product B, measured in proportion to the expected extra profits Firm B would achieve from that success. In practice, if Firm A is also undertaking risky product development efforts, the innovation diversion ratio will depend on the likelihood that Firm A’s efforts bear fruit, conditional on success by Firm B. With high correlation, the merged firm may regard cancelling one project to eliminate “duplicative” projects. The problem, from a competition perspective, is that this also eliminates the prospect that the two resulting products will compete against each other.
Section II also discusses those instances where mergers can have positive effects on competition.
The second step, if unilateral innovation effects are significant, is to look for merger-specific synergies that might offset anticompetitive effects. Merging parties often assert that the merger will generate R&D synergies and thus speed up innovation. In evaluating these claims, synergies that would likely be achieved without the merger should not be credited.
One category of synergies that is amenable to economic analysis is the internalisation of involuntary spillovers. A merger can increase innovation and ultimately benefit consumers if spillovers are sufficiently large, due to higher postmerger appropriability. However, the significance of these appropriability effects in any given case may be limited. A second category of innovation synergies arises if the merger facilitates voluntary technology transfer between the merging firms. A third category of innovation synergies arises if combining the two firms’ development teams will enable them to be more efficient in developing new products.
The significance of these R&D synergies in any given merger depends heavily on how R&D is conducted at the two merging firms and whether these firms have complementary capabilities. In practice it is quite important to distinguish postmerger reductions in the incremental cost of R&D from cost-saving resulting from the elimination of R&D projects in the same area. The latter are not efficiencies that can benefit consumers. To the contrary, evidence of a planned postmerger reduction of so-called “duplicative” R&D projects provides a direct indication that a merger may lead to an anticompetitive suppression of innovation efforts. In addition, and across the board, when merging firms claim synergies of this type, the burden rests on them to establish that the beneficial effect would not have taken place without the merger (e.g., through an ex ante research joint venture or an ex post licensing agreement).
Section III moves on to apply economic theory on horizontal mergers and innovation to practical cases.
The paper divides this analysis into three parts. First, the authors consider mergers involving competing firms with identifiable products or projects in the development pipeline. Product-to-pipeline and pipeline-to-pipeline overlaps arise if one or both of the merging firms own a specific project that is being developed or considered but has not yet reached the market, e.g. a pharmaceutical. In some sectors, notably the pharmaceutical and agrochemical sectors, the development pipeline is a well-structured process driven by regulatory requirements. In other cases, the pipeline is less well structured. A distinct feature of some cases involving pipeline products is that they are associated relatively easily and directly with an existing product market, and the analytical techniques used for the assessment of existing product market competition can often be transposed to the assessment of mergers involving those pipeline products. The basic competitive concerns related to pipeline products that arise in cases of this type are that: (i) the merger may lower the probability of successful introduction of the pipeline product; (ii) the merger may delay the launch of the pipeline product; (iii) even if the pipeline product is successfully developed despite the merger, future product market competition may be less intense because the merger has brought competing products under common ownership.
Second, the authors consider mergers involving established firms with competing innovation capabilities, including several elements required for the effective discovery, development, and commercialization of new products and processes. The need to examine overlaps in capabilities is recognised in guidelines by competition agencies in the United States and in Europe, and been used in a number of high profile mergers. The theory of harm in cases of this type is that two firms with overlapping innovation capabilities could divert profitable sales from each other by coming up with new, innovative products in similar areas, and by competing in the corresponding product market. A merger between them can lead to a reduction in the incentives to initiate new R&D efforts in the overlapping R&D areas, which would deprive consumers of some of the benefits from future product market competition in those areas. A broad perspective is required when looking at overlaps in capabilities between merging parties, especially given the inevitable uncertainty associated with R&D. Several evidentiary proxies can be used, such as the importance and frequency of past product and pipeline overlaps, evidence of current product-to-pipeline or pipeline-to-pipeline overlaps, or evidence on overlapping patent portfolios.
Third, the authors consider situations in which a large firm with a dominant position seeks to acquire a smaller firm with innovative capabilities that may ripen into a threat. These cases can involve disruptive entrants, although future product overlaps may be hard to discern. This third category of cases is particularly relevant in the digital sector, where several large incumbent platforms, including Google, Facebook, Apple and Microsoft have acquired a number of smaller firms in recent years. The clearest theory of harm in the case of the acquisition of a potential competitor by a dominant incumbent is that the acquisition will eliminate a threat to the incumbent, allowing it to protect its existing rents in the market. In practice, the main challenge with developing this theory of harm often is evidentiary. Given the absence of a specific pipeline overlap, it may be difficult to establish that the acquired firm is likely to “steal” business from the incumbent in the foreseeable future. Similarly, if there is a lack of evidence of past product and pipeline competition between the incumbent and the target, it may be difficult to find a suitable proxy for a capabilities’ overlap. This difficulty may be particularly pronounced in new and fast-moving digital markets. As such, the authors suggest focusing on the costs of over- and under-enforcement, and to lean on factors such as whether competition is in or for the market, the valuation of the target or whether the acquirer has a history of engaging in killer acquisitions.
Section IV briefly (by comparison to the discussion on mergers) discusses exclusionary conduct by dominant firms in innovative industries.
The authors are especially concerned with anticompetitive business practices used by a dominant firm in dynamic, innovative markets to exclude pesky upstarts or potential entrants—the familiar agents of disruption. Given the inherent uncertainties associated with the development of new products and their market reception, it is typically impossible to know what new and innovative products would have been developed, when they would have been introduced, or how popular they would have been, if not for the challenged conduct.
All of this implies that the quantum of evidence required to conclude that a dominant firm’s conduct has harmed innovation and thereby violated the antitrust laws is a critical element of competition policy. A more assertive antitrust regime (cough, cough, Europe) will find antitrust violations in cases where the challenged conduct disrupts the competitive process by impeding the incentive or ability of rivals to innovate. A more timid antitrust policy (cough cough, US) would require evidence showing that rivals actually reduced their R&D on particular projects as a result of the challenged conduct, and that this reduction harmed customers because certain specific products were not developed.
The authors then turn to two types of anticompetitive conduct. The first concerns the unlawful defence of a monopolistic position. There are a variety of conducts that a dominant firm can use to exclude a rival that threatens its market power. These include tying, exclusive dealing, loyalty rebates and most favored nation (MFN) provisions. In platform markets, conduct aimed at hindering multihoming on one side of the market may be a particularly effective exclusionary strategy. Exclusion of a disruptive entrant inherently harms the competitive process, even if that disruptive entrant is (currently) less efficient than the dominant firm. Indeed, that pattern tends to be the norm in industries subject to significant economies of scale (e.g., due to network effects and/or learning by doing). Some of the hardest and most important questions in this area relate to business conduct alleged to exclude nascent competitors. As a rule, the greater and more durable the incumbent’s market power is, the lower the chance of success by the entrant required for that entrant to warrant protection from exclusionary conduct.
The second type of conduct concerns the unlawful extension of a dominant position. The concern here is that a dominant firm will extend its control to adjacent markets, using the power from its dominant position to weaken or eliminate independent rivals in those markets. There are a number of economic theories of harm that can support a concern about extension of market power from a primary market to an adjacent market, e.g. tying a primary good to a complementary product that could otherwise serve as a stepping stone for entry into the newly emerging market, or leveraging of intellectual property rights from a primary market to secondary markets in ways that detract from competition and follow on innovation in these secondary markets.
This is a comprehensive, thoughtful piece, as one would expect from the authors. Nowhere is this clearer than in the two appendixes to the article – one deals with the interaction between unilateral innovation effects and unilateral price effects, while the other reviews US and EU merger decisions on innovation – which nearly double its length. The title is misleading, though – this is really a piece about horizontal mergers and innovation, which is why I discuss it here. In this respect, however, the discussion is only short of authoritative, even if it has a clear pro-enforcement bias.
This leads to my (by now) usual concern about papers written by economists ultimately arguing for a relaxation of the burden of proof when addressing innovation issues in merger control. As the authors put it: ‘the more difficult it is to discern the specifics of future competition, the wiser it may be to rely on the general economic principles described previously. Requiring the government to offer precise quantitative evidence of future competition to meet its burden of proof regarding unilateral innovation effects would be tantamount to giving up on merger enforcement relating to the development of future products that are early in the development stage or not yet discovered.’ Please note that I have sympathy for the economists’ concerns, particularly as regards the types of evidence required in the US. I recently co-wrote a paper (which I will mention in more detail in a coming email) arguing precisely that prevailing standards of proof can make it hard to address mergers which may significantly affect potential competition – i.e. the negative impact of such mergers may be large even if their likelihood is small. The paper argues that we may need to tweek some rules on evidence to ensure that enforcement is possible when such large magnitude/high uncertainty scenarios arise, which is something the authors of this paper come back to again and again. I also wholeheartedly agree with the authors’ assumption that the rules on evidence to establish a competition infringement or prohibit a merger are a critical element of competition policy.
However, I feel uneasy with claims that we can or should substitute hard evidence for economic models or presumptions of harm to innovation, particularly under existing legal standards.
- First, I think that serious consideration needs to be given to potential benefits to innovation flowing from a merger, particularly when they are as potential and uncertain as the negative effects. I see no reason under existing legal standards to subject evidence of such benefits to stricter standards than those to which we subject evidence of anticompetitive harm.
- Second, it is unclear whether adopting presumptions of harm is justified at the current stage of economic theory (and empirical evidence), even though I would ultimately leave the assessment of this to economists. However, my reading of the literature is that presumptions that mergers harm innovation may be justified only in narrow circumstances, if at all. Even if we were able to delineate those narrow circumstances (which I have seen no attempts to do), we would then still need to consider how to design such rules of inference/presumptions, which can prove controversial (e.g. see the debate on whether the benefits to innovation from a merger are part of the effects assessment or are efficiencies, which is really a discussion by proxy about who has the burden of proof of what during merger control proceedings).
- Third, if we are to make a value judgement that we want to prevent industry consolidation because we believe it will be more conducive to innovation (or for any other reason), legislative intervention setting bright(er) line rules may be more appropriate than doing this through the back door by creating evidentiary rules with the same effect.
- Fourth, it seems to me that, at least in the areas covered by the authors, competition authorities have been able to intervene against mergers that might detrimentally affect innovation. As such, I do not see the case being made as to why a relaxation of evientiary standards is needed.
- Fifth, there is quite a lot of emphasis on how all this debate is about ‘horizontal’ mergers, which are the ones that can detrimentally impact innovation. However, this raises a conceptual question about the scope of potential competition: at which point can we say that competition over a possible future product or innovation is ‘horizontal’? This can arguably be done on a case-by-case analysis, but I’m not sure we are currently able to successfully delineate ex ante a specific group of cases to which a presumption of harm would apply. Further, an undue focus on whether potential competition/innovation is horizontal or not would risk turning debates between agencies and merging parties on the effects of the merger into debates about the right categorisation of the merger, which is of doubtful usefulness given the focus of competition law on the effects of conducts or mergers.