The Dow/DuPont merger launched an economic debate about the effects of horizontal mergers on innovation. Underpinning these debates are a number of points of agreement, beginning with consensus over the debate on the relationship between competition and innovation not being directly transferable to the effect of horizontal mergers. There are also a number of shared conclusions regarding merging firms’ ability and incentive to innovate when those firms compete in developing new products (product innovation) or in reducing their costs (process innovation). Such mergers may give rise to various efficiencies and increase the merging parties’ ability to innovate, but they can also influence the parties’ incentives to engage in R&D and implement their innovations. Ultimately, whether a merger leads to more innovation will depend on the nature and relative magnitude of the positive and negative externalities that the investments made by one party generate on the other.
Where there seems to be no agreement, however, is on the implications of the previous results for merger control policy. Calls for case-by-case analysis enjoy wide consensus. However, it has been argued that such an analysis must be premised on the assumption (or presumption) that horizontal mergers do cause a loss of innovation and, therefore, its focus should be limited to carefully assessing ‘the significance’ of such a loss.
This paper, available here, assesses whether an approach is justified.
Section II looks into whether a structural presumption against horizontal mergers is justified.
The answer to this question hinges on whether horizontal mergers always reduce the merging parties’ incentives to invest in process and/or product innovation. Such a negative effect is observed as regards prices, and hence there is a structural presumption that horizontal mergers lead to significant price increases absent efficiencies.
A structural presumption would be justified when assessing the innovation effects of a horizontal merger if theory and/or evidence showed that mergers among competitors in industries with few innovators tend to reduce the parties’ incentives to innovate and, therefore, cause a loss of innovation in the absence of efficiencies. In effect, some horizontal mergers are indeed likely to reduce the merging parties’ incentives to invest in process and/or product innovation, e.g. when product innovations are stochastic; the parties’ innovations are likely to be close competitors in the same downstream market; and there are diseconomies of scale in R&D. The author conducts a survey of the literature outlining various instances where such conditions are met.
At the same time, horizontal mergers may increase the merging parties’ incentives to invest in innovation under a variety of circumstances. Some of these papers identify three effects on innovation flowing from a horizontal merger. The first is a margin expansion effect, whereby a merger reduces the merging parties’ output, and hence reduces their incentives to invest to increase their margins, which decreases innovation incentives. The second is a demand expansion effect, which operates by increasing the incentives of the parties to invest in order to increase demand. Lastly, there is an innovation diversion effect, concerning the internalisation of the externality generated by each firm’s innovation investment on the other merging firm’s demand, which is ambiguous. When the three effects are in operation, the merger can have either a positive or a negative impact on incentives to innovate.
To this may be added spillover effects: investment in R&D by one firm may not only benefit the firm but also its rivals through technological spillovers. It follows that the assessment of the impact of a horizontal merger cannot be presumed, as it requires assessing each of the four effects mentioned above.
Section III asks whether it is likely that horizontal mergers affect the merging parties’ ability to engage in process and/or product innovation.
It follows from the above that, even when a merger creates no production or innovation synergies, i.e., even in the absence of efficiencies, a horizontal merger can have a positive effect on innovation. The merger may increase innovation incentives and, therefore, be rivalry enhancing.
Yet, a horizontal merger may also cause an increase in innovation if it gives rise to efficiencies. On the production side, efficiencies may arise because of economies of scale or economies of scope. Because production synergies increase output, they are also likely to increase innovation incentives. On the innovation side, a horizontal merger may also reduce the cost of innovation due to economies of scale and scope, since R&D costs can be defrayed over a greater total volume or a larger number of products. A merger may also save R&D costs by avoiding the duplication of investment and leading to the combined investment of the merged entity exceeding investment in the no-merger counterfactual. The merger may also combine complementary R&D assets of the merging parties.
These efficiencies will of course be difficult to document, but there is no reason to be more sceptical about them than about efficiencies in industries where innovation is not an issue. It is true that some of these efficiencies may not be merger specific. The merging parties may be able to generate similar efficiencies through other means, such as research joint ventures, patent pools, etc. In this regard, it is important to note that, as a matter of economics, it would be wrong to use different counterfactuals for the assessment of competitive effects and efficiencies. In other words, it would be wrong for competition agencies to assess the potential anticompetitive effects of a horizontal merger by comparing consumer welfare in the merger scenario with a scenario in which the parties act unilaterally but then evaluate efficiencies by comparing the merger scenario with a counterfactual where the parties engage in limited, arms-length cooperation.
Section IV asks how competition authorities and courts should balance the potential anticompetitive and procompetitive effects of horizontal mergers on innovation.
The relevant question is whether competition agencies should give structure to decision making rather simply stating that “anything can happen”. Some economists believe that the presumption of a negative innovation externality between merging parties remains a useful default. There are a number of arguments in favour of this suggestion. First, in many industries where innovation effects have been investigated in recent merger control cases, the innovation diversion externality is likely to have strong effects. Secondly, these economists rightly point out that an increase in innovation need not imply that consumers are better off once price effects are considered. Finally, these economists suggest that claims about the existence of positive innovation diversion effects should be dealt with care because whether they are true or not will be difficult to verify and depend on the precise specification of demand.
The author disagrees with this approach. Firstly, whether a horizontal merger has a positive or negative effect on innovation depends on the combined effect of the margin expansion, demand expansion, innovation diversion, and spillover effects and not just of the sign of the diversion externality. It is not possible to predict the combined outcome of these factors a priori, even for the industries subject to merger control decisions that look into innovation incentives. Secondly, the issue at stake is whether competition agencies should presume that mergers reduce innovation, and not whether they can legitimately presume that the mergers’ price effects can hurt consumers. Thirdly, proving the existence of a negative diversion externality is not enough to establish that a horizontal merger would cause a loss of innovation. Whether that is the case depends on many assumptions, including the precise specification of demand and the shape of R&D costs. Fourthly, creating a structural presumption would unduly shift the burden of proof regarding the factors impacting incentives to innovate onto the merger parties. Competition authorities could and should assess the potential positive effects of a merger on the merging parties’ ability to engage in process and/or product innovation under the traditional efficiency defence approach. However, it should not treat the four factors influencing incentives to innovate as ‘traditional efficiency defences’, which they are not. Each of those effects impacts the merging parties’ incentives to innovate, and hence affect the strength of competition and rivalry in the absence of efficiencies. The assessment of the impact of the horizontal merger on the innovation incentives of the merging parties is incumbent on the competition authority in charge of the case. The competition authority must therefore demonstrate that the net incentive effect of the merger is negative to discharge its burden of proof.
The choice of legal standard must be based on a careful balancing of the expected costs of error. A structural presumption is a rebuttable presumption of illegality, and it is only justified when (a) both Type I and Type II errors are likely and costly, but (b) the expected cost of the Type II errors exceeds the expected cost of the Type I errors. This second condition is not met, since we are not in a position to claim that one type of error is more likely and/or costly than the other. If anything, there is (limited) empirical evidence that mergers are positively and significantly correlated with firm innovation; and that mergers increase innovation by helping firms to internalise positive externalities or spillovers. Of course, there is also literature indicating that some mergers are just meant to kill the innovation efforts of new entrants, and some mergers will have negative effects on innovation. But more research is needed to understand the impact of mergers on innovation, let alone before a presumption of harm is adopted.
This is a very easily digestible review of the debate on merger control and innovation. I review it here because I think it is a good example of a more accessible paper that provides a fair overview of the debate, while still making an argument for a case-by-case approach. To better understand the various effects at play one should look at other, more technical papers –such as the two I reviewed above – but this is not a bad starting point to understand the debate.
I also think that its conclusion is probably the dominant one, for the moment. As noted in a paper reviewed earlier, the Commission pursued detailed effects’ analyses in the more controversial cases in practice; and the absence of a presumption of harm to innovation does not seem to have impeded intervention against mergers on the part of competition agencies, which has removed some steam from the debate. I do have one complaint – not particularly about this paper, but about the debate as a whole. In short, all these papers completely ignore the law on the topic. As noted in the papers I reviewed a couple of weeks ago, this debate raises serious issues of burden and standard of proof. It would be nice to see those issues being acknowledged – before they suddenly become the fulcrum of the debate merely because a merger involving these matters goes before a court that will, inevitably, focus on these matters (probably framing it as a matter concerning the scope of judicial review).