By Helen Kean
‘Big data’ has been under the spotlight in South African competition circles as of late. Indeed, the South African Competition Commission’s recent conference was themed ‘The Future of Competition Policy’ and included in its scope or interest ‘emerging areas of enforcement’, which was in turn stated to include interest in big data. Whilst this broad topic may span many areas of competition policy, this short blog aims to raise a few questions about its relevance in mergers, following what we can learn from inter alia the May 2017 Facebook/WhatsApp 110 million Euro post-merger fine. Follow-on blogs will raise further questions about big data’s relevance in light of abuse of dominance and collusion cases, drawing relevance from inter alia the June 2017 2.42 billion Euro fine directed at Google (an abuse of dominance case).
Importantly, before turning to the cases in question and what we can learn from them, we have to ask what is ‘big data’ and why is this relevant? The term ‘big data’, although often believed to be a highly topical and current issue within competition economics, is not at all new. The precise term, which started to gain traction in the 1990s following years of information expansion, generally refers to large and information-rich datasets that render traditional data processing inadequate. Consequently, the past decade has seen a rise in computational methods – perhaps most centrally including machine learning – to better understand, work with and leverage insights from these data sets. These and related shifts have sparked significant debate within competition-related circles, largely based on the questions of whether ‘big data’ may serve as a means of market power, as well as a barrier to entry (particularly in cases where old or existent, relative or in addition to new, data are valuable). This follows on from the fact that there may be a self-reinforcing ‘network effects’ cycle between the value of datasets to firms and value of the same to the consumer: Large datasets enable superior training of machine learning systems to extract the correct insights in order to optimally innovate, and reach and serve consumers; more consumers, as a result of improved service, provide further data as they consume; and so, the cycle continues. A central concern then hinges on whether the firm may reach a critical ‘tipping point’ where competitors may not be able to keep up (this may occur in all markets, but may be exacerbated in the case of data-intensive markets). On the other hand, and depending on the market of interest, it may be argued that there are alternative sources of data (for firms wishing to compete), and low entry barriers. Alongside this, one should also consider that whilst the consumer transparency that big data often allows for may be beneficial, this may also serve to facilitate collusive behaviour. Ultimately the net effects of these points and the determination of whether ‘big data’ presents new, or somewhat different, challenges to competition frameworks may be best ascertained on a case-by-case basis.
Turning to the 2014 Facebook/WhatsApp merger, what went wrong and what can we learn? This merger was notified to the European Commission in August 2014 and ruled upon in October 2014. The value of the Transaction was set at $19 billion, and the motivation was stipulated to be related to Facebook’s strategy of focusing its business on mobile development. In outlining the case, Facebook was defined as a provider of websites and applications for mobile devices, essentially offering social networking, consumer communications and photo/video sharing functionalities via its platforms: Facebook, Facebook Messenger and Instagram. Facebook was also understood to offer advertising space. Alongside this, WhatsApp was defined as a provider of consumer communications services via its mobile applications (‘apps’). Following on from this the Commission determined that the main market of interest was that for ‘consumer communications apps for smartphones’, with the geographical market of interest being EEA-wide, if not worldwide. As ‘social networking services’ and ‘online advertising services’, the other components of Facebook’s business presented no overlap, these were set aside. Focusing then on the main market of interest, the CC found that the merging firms has 30-40% EEA-wide market share, but that “In such a dynamic context, high market shares are not necessarily indicative of market power and therefore of lasting damage to competition.” [This sentiment importantly was also held in the 2011 Microsoft/Skype merger, where it was determined that even a 90% share (which was in that case found in the market for video calls as a result of each party having roughly half of this share pre-merger) did not pose competition concerns, given the specific market dynamics regarding innovation and entry barriers]. Subsequent to this structural assessment, the CC focused on an assessment of closeness of competition; ease/cost of consumer switching; and barriers to entry and network effects. From this it was found that Facebook and WhatsApp were not close competitors (as consumers multi-home between the two), that consumers can easily switch, and that barriers to entry are low. On the latter and also relating to network effects, stakeholders did raise concerns of interoperability (i.e. of the merged entity matching Facebook IDs with WhatsApp cell phone numbers), thereby leveraging off of a larger user base, in turn raising barriers to entry and network effects. The CC however determined that this was unlikely, given the technical difficulties in doing so (as also stated by the merging parties). With the same reasoning in mind (difficulty in truly matching the databases in question), it was determined that no effects would extend to the markets for ‘social networking services’ and ‘online advertising services’.
Despite the above determination, it was announced in May 2017 that Facebook was being fined 110 million Euros by the European Union, for misleading regulators during the 2014 review of the WhatsApp takeover. This ‘misleading’ pertained specifically to the fact that Facebook and WhatsApp were able (prior to the merger) and now planning to now combine their data, after stating in the 2014 merger review that this was not technically possible. This constraint had a strong hinging on the merger approval at the time (via the stated assessments of barriers to entry and network effects in each of the markets of interest), therefore raising a hefty fine. Importantly, while less contested, the December 2016 $26.2 billion Microsoft/Linked In merger also raised the question of network effects. In that case, non-horizontal effects were of main focus, and after consideration of the various markets of interest (with detailed assessment of ability and incentive for anti-competitive merger-specific effects, as well as detrimental harm to consumers, in each instance), it was determined that negative network effects could potentially arise in two specific markets. However, in those instances, remedies were rapidly put forward by the merging parties. These were straightforward, clear-cut, and effective within a short period of time. Evidently, such concerns and hence remedies were not raised in the Facebook/WhatsApp deal.
The above highlights a few lessons and discussion points that one can consider locally. Firstly, the various cases mentioned highlight the limitations of a structural assessment of mergers. Indeed, we see that authorities generally limit this to a starting point of assessment. Secondly, a careful effects-based approach is required, taking into account all technical details of the markets in question and practical impacts on competition and consumers. Indeed, in the Facebook/ WhatsApp merger this was the approach taken by the Commission, but perhaps not fully considered by the merging parties. It is credible that the merging parties would still have wanted to merge even if there was no possibility of integrating the systems (given that there was already multi-homing and hence overlap, and also because each business was expected to be profitable on its own), but if the ability to integrate had at the time of merger been more clearly raised, this may have ‘saved’ them a fine (particularly since the Commission did include an ‘even if’ assessment in its analysis, pertaining specifically to the possibility of technical integration). A third point of discussion pertains to where competition and regulation should, in this space, meet. If well-enforced regulation correctly controls what firms can and can’t do with users’ data, this may mitigate some of the competition concerns. For example, in the case at hand, WhatsApp data could have been stipulated as not being usable beyond the purposes of what users sign up for, thereby mitigating concerns of Facebook leveraging off of this in any integration of systems that would provide unfair network effects. Of course, this would need to be alongside alternatives being available for users who may prefer to switch in the case of being unsatisfied with updated and notified legalities of data usage. Indeed, these and other points provide a starting point for discussion of the role of big data in the context of mergers requiring approval by competition authorities. This is expected to be relevant in South Africa, especially as we have already seen many mergers in sectors where data are an asset relevant to the transaction – retail, insurance, and telecommunications sectors form just a few examples. Looking forward, this is expected to become even more relevant as data increasingly becomes corporate assets. Notwithstanding the points raised, this case appears to support the idea that existing frameworks do indeed cater for big data in competition. In this case big data appears to raise evolved, rather than entirely new, questions for competition policy.
 This term is generally credited to the US-based computer scientist, John Mashey, although many others around the same time may have used it.
 We do not deal with this market definition here, as this has more to do with high technology markets. Nevertheless, this was of interest in this case, as this evolved over time from the Cisco/Tandberg (2010) merger to the Microsoft/Nokia (2013) mergers, and to the present cases. In this case the product markets were determined along the lines of functionality and network size (it was also determined that price played a role, as consumers generally expected the services to be free).
 At the time, it was found that Facebook Messenger had approximately 100-200 million EEA users and 250-350 million global users. Concurrently, WhatsApp was found to have approximately 50-150 million EEA users and 600 million global users. Competitors were listed to include iMessage, BBM, ChatOn, Google Hangouts, Skype, Join, Libon, Tuenti, LINE, Viber, Threema, Telegram, Snapchat and WeChat.
 Page 18 of hyperlink.
 In that case it was pointed out that Viber attained 1 million customers within 3 days and 15 million within 5 months.
 It was pointed out that WhatsApp itself escalated to 600 million worldwide users in less than 5 years.
 This pertained to foreclosure of competing professional social networking providers, as well as foreclosure of online recruitment service providers, both via conglomerate non-coordinated effects (tying, bundling, etc.).
Author/s: Helen Kean
Nothing in this publication should be construed as advice from any employee of Econex and should be seen as general summaries of developments or principles of interest that may not apply to specific circumstances.