This paper presents the results of a (quantitative) empirical study on how the US antitrust authorities assessed mergers in regard to their innovation effects from 1995 to 2008. It is structured as follows:
- Section 2 contains a brief overview of the theoretical and empirical literature in economics about innovation effects. The particular characteristics of innovation processes – which lead to the unpredictability of outcomes arising from innovation processes, and the possibility that important (even revolutionary) innovations can emerge entirely unexpected from anywhere – have raised the question of whether policy-makers have enough knowledge about the determinants of innovation to adopt policy instruments for promoting innovation. The paper also covers the Arrow-Schumpeter-Aghion debate on the relation between competition and innovation, which I am not going to repeat here.
More interestingly, the authors review the model-theoretic literature about the relation between competition and innovation, and distinguish between different groups of models. In the first group of modles, the innovation incentives of firms are influenced only by competition for innovation itself (e.g., in order to get a patent or to realize first mover advantages). In the second group, the incentives to innovate are also influenced by pre-innovation product market competition. A third group of models are based upon new research about the specific characteristics of high-technology industries and the “new economy”. In these (Schumpeterian) industries, markets are hard to define and market shares do not reflect the competitive positions of firms, especially because market leaders might lose their dominant position through new innovations very fast (“creative destruction”). In this respect, they look at how concept of “dynamic competition” has made its appearance in the literature as a tool for grasping innovation competition in contrast to the well-established concept of “static (price) competition” .
Despite these models, the authors conclude that no general stable relationship between market structure and innovation has been found both from a theoretical and an empirical perspective; and that a closer look at the many sector-specific studies on innovation shows that the determinants of innovation might be very different between different sectors. This leads to the conclusion that the innovation effects of mergers might be very different under different circumstances.
- At this point, the authors move to an assessment of whether the traditional approaches adopted in merger control are suitable to deal with the potential impact of mergers on innovation, and conclude that the traditional approach is focused on static competition, and hence unsuited to capture dynamic effects. They then look at an innovation-specific approach developed in 1995 by Gilbert/Sunshine – the “Innovation Market Analysis”. These authors proposed a 5-step-procedure for identifying the relevant innovation competitors, and the possible anticompetitive and efficiency effects. The focus of this approach is on “innovation markets” i.e. research and development directed at particular new or improved goods or processes, and close substitutes for that research and development.
The crucial idea of the “innovation market analysis” is the identification of the relevant competitors in innovation competition through the identification of overlapping (substitutive) R&D activities, combined with the identification of specialized assets that are necessary for this kind of R&D – which can also be interpreted as barriers to entry for innovation. The innovation market analysis concept requires an analysis of whether the merged firms would have the capabilities and incentives for reducing or slowing down their R&D activities, either through unilateral or coordinated behaviour. Such an analysis would normally be followed by an assessment of the possibility of counterbalancing any anticompetitive effects of innovation through innovation-related efficiencies.
- The paper then looks at the various criticisms that lawyers and economists have directed at this approach. It concludes that while the current concepts of potential competition and future markets can address the anticompetitive effects of mergers related to less price competition on a current or future product market, such models are unable to take into account the negative effects of the merger in regard to innovation competition itself. As a result, such approaches fail to take into account the benefits of innovation markets – i.e. welfare increases through more and faster innovation, and not through the easing of static price concerns on current or future product markets.
- Section 3 presents and analyses the results of the empirical study. They observe that during the 1990s the U.S. merger policy started to take innovation effects in merger reviews into account much more seriously. The authors find that 135 of the 399 mergers that were challenged between 1995 and 2008 raised innovation concerns, i.e. innovation concerns were raised in a third of all challenged mergers (34%).
To understand whether the agencies adopted a traditional or an innovation specific approach, the authors looked at whether the agencies considered innovation in the market definition or only in the competitive effects analysis. Two different approaches were thus identified: a first traditional approach in which an agency claimed negative innovation effects only in the competitive assessment analysis, and an innovation-specific approach according to which the agencies took innovation into account already in the market definition, and afterwards identified either (a) explicitly negative innovation effects or (b) implicit innovation effects. They found that whereas the DOJ adopted the innovation-specific approach in only 41% of all markets, the FTC relied on this approach twice as often (82%).
They also identify two groups of innovation-related arguments. The first group reasons on the basis of various theories and models which argue that a merger might reduce innovation incentives, which might in turn lead either to less investment in R&D and/or a slowing down of research and development. These arguments are mostly based upon mainstream industrial economics approaches. A second group of arguments focused instead on evolutionary arguments about possible negative effects on competition as a process of parallel experimentation with different problem solutions. This second type of argument was only used in about one-third of sampled cases. More importantly, however, is a finding that “the agencies in the majority of cases gave no specific reasonings why the mergers might lead to negative effects on innovation”. The authors explain this on the grounds that there is a large degree of insecurity on the part of the agencies about how innovation effects should be assessed, and that agencies prefer to rely on simultaneous allegations of price effects which fit much better into the traditional antitrust approach than this new kind of innovation effects argument.
In short, the study finds that while innovation is often taken into account, the agencies have not succeeded in developing a clear and consistent approach to how innovation effects should be assessed until 2008.
- Section 4 is then devoted to a brief (qualitative) analysis of important merger cases after 2008, as well as an analysis of the reform discussion leading up to the revision of the Horizontal Merger Guidelines in 2010. The important merger cases are Pfizer/Wyeth, Merck/Schering-Plough, Thoratec/Heartware, and AT&T/T-Mobile. In these cases, the FTC adopted a traditional approach to innovation. When looking at the broad discussion about the reform of the Horizontal Merger Guidelines, the authors observe that there was a broad agreement about the great importance of innovation as part of market and for the increase of wealth. There was also broad support for the assessment of innovation concerns being developed in the new Horizontal Merger Guidelines. However, how this assessment should be pursued was a much more controversial topic.
Looking specifically at the Horizontal Merger Guidelines, the authors find that it relies on the traditional approach, and ignores innovation-specific approaches. The main focus of the Guidelines is on the reduction of innovation incentives, either for continuing an already existing R&D project or for starting a new one, as regards potential competition in existing product markets. The “guidelines remain sketchy, vague, and inconsistent, and offer no clear guidance how the agencies would protect innovation competition in merger cases. However, these guidelines reflect quite well the also very vague general opinion in the U.S. discussion about this topic, which acknowledges the problem and necessity of making such assessments, but simultaneously offers few specific advice how this should be done.”
This is a very interesting paper, and a valuable overview of how innovation-related concerns are addressed in practice. There are a number of interesting nuggets arising from this paper. For example, I did not know that in more than a third of all challenged mergers innovation concerns were raised – with no differences between the agencies and over time. However, the agencies used rather different assessment approaches, the DoJ relying more the traditional product market concept and the FTC on innovation-specific approaches. Furthermore, the agencies “did not succeed to develop a clear and consistent approach” to innovation effects.
On the other hand, and in addition to the English being somewhat hard to follow (for which I must apologise in case this is reflected above, as I often copy-paste segments), I think the paper has a very ordo-liberal bias towards process-based approaches (and an Austrian school’s focus on markets as discovery mechanisms). I have nothing against this bias, but it reflects a normative preference and it should be assumed as such.
This bias then bleeds into the authors’ assessment of the different approaches to innovation by the US agencies. The result is that, while the paper is extremely informative and knowledgeable, one gets a very poor sense of what theories of harm are deployed by the agencies in the (limited) cases in which they try to argue that a merger has a negative impact on innovation and that this meets the “substantial lessening of competition” standard.