The innovation offence (by Stephen Kinsella)
[We are happy to publish a guest post from one of the most respected and interesting practitioners in the EU market, Stephen Kinsella (he’s of course best known for having been a speaker at the first Chillin’Competition conference and for being the husband of a great novelist who is currently crowdfunding her new novel. Below he gives his views on a very topical matter on which we have also commented before. As always, we will be happy to foster discussion and are open to publishing other views on the matter. Enjoy!]
Mergers tend to get more attention in press coverage than other antitrust activity in Europe. That is partly because they have compressed timetables and obvious milestones to trigger stories (announcement, filing, enquiries, third party interventions etc) but also because they can be resolved to a binary choice between approval or block, with readily understandable consequences. Not only shareholders but other financial players follow closely each twist and turn, placing bets on rumours of setbacks or “theories of harm”.
All this froth can sometimes mask a duller reality, which is that at the EU level of an average 350 or so deals notified in Brussels each year, less than one a year is ultimately prohibited. Of those that are permitted only around 5% are subject to any modifications or commitments as the price of approval.
And there is a reason for this. The system is weighted in favour of approval. It carries within it a presumption that deals will be cleared, and speedily, unless good evidence can be brought forward of some creation or strengthening of dominance causing harm to effective competition, to the detriment of consumers denied the benefits of choice.
Granted, merger control, like other aspects of antitrust, is not static. It evolves in response to evidence, to greater learning about how markets behave and to developments in legal and economic thinking. But it does so cautiously, trying to balance the risks of excessive intervention (in the jargon, Type 1 errors) against non-intervention (Type 2 errors). The Type 1 errors could include not only hampering the ability of the merged entity to innovate, but also deterring those who would invest in creating products with the aim of selling them to another who is better able to exploit them.
One area in which we are seeing calls for such an evolution relates to “big data”. Enforcers at EU and national level are asking themselves whether the mantra that “knowledge is power” literally means that the acquisition or accumulation of data, in particular about the behaviour of and relationships between large numbers of individuals, could confer the power to exploit and exclude.
This is not to be confused with concerns over privacy. We have seen numerous statements, including from Commissioner Vestager, that competition law is not to be used to try to cure possible concerns that fall more properly in the realm of consumer (or data) protection. Rather the question is a narrower one: whether a data set might be so special, valuable and non-replicable that its concentration in one undertaking would give it an overwhelming competitive advantage that could be checked only by regulatory intervention.
Such a theory is not controversial in principle if one looks at data as if it were an essential facility. But it runs up against the objection that unlike a piece of infrastructure such as a port or a pipeline, the data (or substitutable data) may well be capable of being compiled by others, or already exist in the form of other accessible compilations. Again to cite Commissioner Vestager in a recent speech, the data might quickly go out of date and need refreshing, and “we also need to ask why competitors couldn’t get hold of equally good information”. And while we sometimes see reference to the question of whether the data is “unique”, the better way of expressing it, as recognised in the Franco-German discussion paper from May this year, is whether it is really “unmatched”. This recognition has led to understandable caution. A recent consultation exercise by the Commission is beginning to explore whether the merger rules need to be adjusted – though even here the focus is more on the jurisdictional thresholds that might be appropriate to ensure deals receive proper scrutiny, rather than suggesting that data poses particularly intractable problems.
Against this backdrop, the public discussion around the Microsoft – Linked-in transaction is interesting, and rather curious [my firm has advised Microsoft on a range of antitrust issues but I am not acting on the notification of the Linked-In deal] . I only have access to what is in the public domain, but it appears to be a case where a company acquires a target with which it is not in competition and where there is no suggestion that the target will alter its commercial strategy in terms of its market behaviour or how it makes available its data to third parties. According to press reports the acquirer has already given assurances to that effect and those assurances do not seem to be seriously disputed. Therefore it is hard to see that any adverse change in the market will inevitably occur that is “deal specific”.
At the same time, though nobody disputes the value of the data held by Linked-In, there are many other players in the market and apparently many other ways of obtaining similar or competitive data sets. In fact an increasing number of companies hold substantial amounts of data regarding their customers or others with whom they interact and some use that purely for internal purposes while others develop business models around exploiting that data. Unless there is convincing evidence that a particular data set is genuinely both non-replicable and uncontestable, it would place an unreasonable burden on competition enforcers if they were always obliged to analyse the impact on some rather nebulous “data market”.
But horizontal concerns are not the end of the story. There have been claims by opponents of the deal that in the future, in some unspecified manner, the two companies could combine their data and expertise. In doing so they would come up with some new product, for which there would be strong consumer demand, and with which third parties would struggle to compete (though I have seen no suggestion that those third parties would be forced to exit the market). Such reasoning evidently includes a number of leaps and suppositions, but reduced to its essentials it seems to try to take merger theory even beyond the notion of an “efficiency offence” (always rejected by the Commission) into the realm of an “innovation offence”.
One can well understand why any regulator would be sceptical about such an approach. Merger control has to be to some extent forward looking in that it must try to identify the suppression not only of actual but also of potential competition. But when asked to go even further and tackle some speculative impact on a form of competition that absent the transaction would not anyway have taken place, combined with the fact that the transaction will in the complainants’ “worst case” scenario introduce a new element of competition through a new product, the levels of abstraction introduce too much uncertainty into the merger review process.
Moreover, it is not as if merger control is the last and only chance that the Commission has to protect competition. If following a concentration some development occurred that put the new entity into a position of unassailable market power, there remains Article 102. Indeed we saw relatively recently in the Thomson-Reuters case that the Commission, having cleared a merger, then opened a proceeding to verify the impact on the market of the merged firm’s unilateral behaviour and extracted a package of commitments that the Court subsequently ruled was sufficient to restore competition.
Those opposed to transactions will continue to innovate with theories of harm. Regulators will continue to welcome and even encourage their contribution, while maintaining a healthy scepticism regarding their agenda. But the threshold for intervention remains high.