This paper, available here, presents a broad retrospective evaluation of mergers and merger decisions in markets dominated by multisided digital platforms. It identifies almost 300 acquisitions carried out by three major tech companies— Amazon, Facebook, and Google—between 2008 and 2018, looks at the business logic behind these transactions, and explores the theories of harm that have been used or, alternatively, could have been formulated by authorities. The paper then retrospectively examines two important merger cases, Facebook/Instagram and Google/Waze, providing a systematic assessment of the theories of harm considered by the UK competition authority, as well as evidence on the evolution of the market after the transactions were approved.
Section II looks at the wealth of mergers and acquisitions (M&A) carried out by key digital platforms between 2008 and 2018.
Companies active in digital markets are remarkably active in M&A, constantly seeking out interesting start-ups and purchasing them. Between 2008 and 2018, Google acquired 168 companies, Facebook acquired 71 companies, and Amazon acquired 60 companies. Following desk research, the authors grouped each target company into activity clusters (‘Communication apps and tools’, ‘Tools for developers’, ‘Physical goods and services’, ‘Digital content’, ‘Remote storage and file transfer’, ‘Advertising tools and platforms’, ‘Artificial intelligence, data science and analytics’, ‘Home, well-being, and other personal needs’ and ‘Other’). Amazon and Facebook seem to focus on ‘Physical goods and services’ and ‘Communication apps and tools’, respectively, whereas Google’s acquisitions are more evenly spread out across clusters. Moreover, Amazon, Google, and Facebook have all invested in companies that have helped them with advanced data analytics techniques
Most transactions do not seem to have a clear horizontal element. Instead, acquisitions often focus on complementary economic activities to those of the acquirers. Such acquisitions may have a variety of purposes. For instance, they may be conducted to secure a technology to be incorporated into the acquirer’s product, or to secure highly skilled staff and use their expertise to develop products. Non-horizontal mergers present significant scope for efficiencies. However, these efficiencies may also enable incumbents of digital markets to preserve their leadership and prevent other market players from challenging them. Further, such acquisitions may also have the intention or effect of wiping out potential competitors.
Section III discusses features of digital markets that pose particular challenges to competition law.
Digital markets have several features that make enforcement more difficult. This section surveys a number of mergers to assess how, and to what extent, competition authorities have addressed these challenges.
A first challenge concerns network effects. Given their potential to confer significant market power to the merged entity, network effects are sometimes central to theories of harm deployed by authorities in their assessment of digital mergers – as exemplified by the European Commission’s analysis of video calls in Microsoft/Skype. Network effects can also increase the likelihood of foreclosure, exacerbating the anticompetitive effects of the merged entity’s exclusionary strategies. This was explored in Microsoft/LinkedIn, where the Commission investigated whether the merged entity could leverage its strong market position from the markets for operating systems and productivity software for PCs into the market for professional network services. The Commission explored a number of potential foreclosing strategies, which eventually led to a set of commitments being adopted.
A second difficulty relates to free pricing, which often reflects competition for attention. Mergers involving companies in competition with one another for consumer attention may increase these companies’ ability to exert market power within broad online advertising markets, even where the services they supplied to consumers were different and not substitutable to one another. However, competing to attract consumer attention does not necessarily imply that a company exploits this attention for monetisation purposes. This possibility was explored in the Facebook/WhatsApp merger, where the European Commission explored whether, post-transaction, the merged entity could analyse WhatsApp users’ data to introduce targeted (that is, more effective) advertising on WhatsApp.
A third challenge concerns the loss of potential competition. In Google/DoubleClick, both companies were active in the online advertising sector but they provided different services. In its review, the European Commission explored whether Google could become active in the segments where DoubleClick was present, and vice-versa. It found that DoubleClick had plans to enter ad intermediation, where Google was market leader. However, it also found that, even if DoubleClick could grow into an effective competitor in the market for ad intermediation services, in the event of its exit from the market a sufficient number of other competitors would exert competitive pressure on the merged entity post-merger. A related difficulty concerns killer acquisitions and loss of innovation, even though these are not specific to the digital sector.
Another difficulty relates to the role of big data as a competitive input. This was a concern in the Apple/Shazam transaction, where the combination of their respective datasets was suspected to give Apple (Music) access to commercially sensitive information about competing music streaming platforms such as Spotify. It was investigated whether the data collected by Shazam could have been used by Apple to improve existing functionalities, or to offer additional functionalities, on digital music streaming apps. Eventually, the Commission found that the relevant data could be collected from different providers and was, in any event, not an essential input for the functioning of music streaming apps.
Section IV contains an ex post evaluation of the Facebook/Instagram merger .
The Facebook/Instagram merger was cleared by the OFT in 2012. The OFT considered three main theories of harm: (i) loss of competition for the supply of photos; (ii) removal of a potentially significant competitive threat to Facebook in the market for social network services; and (iii) foreclosure of rivals by eliminating interoperability of Instagram’s photos or degrading its quality. The first theory of harm was dismissed because Instagram faced several competitors relatively stronger than Facebook in the supply of photos, and because of the limited attractiveness of photo apps to advertisers. This was assessed by app downloads and time-on-app, instead of user engagement. The second theory of harm was dismissed because the available evidence did not show that Instagram was particularly well placed to compete against Facebook in the short run, and there were other well placed ‘recognised brand’ potential competitors as social networks. The third theory of harm was dismissed because it was assumed that Instagram’s appeal was in large part due to the possibility to upload photos to other social networks, and that limiting this possibility would reduce the popularity of Instagram and diminish the merged entity’s profitability.
After the merger, Instagram evolved into an app offering social network functionalities, with Facebook contributing improved physical infrastructure as well as its expertise in social networks and advertising markets. Facebook’s number of users has remained stable, while Instagram’s had doubled by 2018; and while Facebook’s share of time spent on social networks has fallen, Instagram’s has tripled. Snapchat is the only other social network that has emerged as a significant challenger to the merged entity. On the other side of the market, Facebook’s advertising revenue increased significantly despite the drop in time spent on its social network, while Instagram’s revenues also increased significantly, with the advertising revenue per hour spent on Facebook and Instagram being significantly larger than that of their rivals.
In short, instead of having focused on the user side of the market, the OFT should have considered the two (or more) sides of the market jointly, as choices made by the platform on the various market sides are interdependent. Relatedly, the OFT might have neglected some factors that drive advertisers’ choices: chief among these are exclusivity and size of the user base, and accuracy in ad targeting. The acquisition of Instagram has provided a competitive advantage to the merged entity across all these three dimensions. Compared with a counterfactual scenario where Instagram would still have become a popular social network, the merger has increased Facebook ability to exert market power by eliminating a viable and strong competitor in the market for social networks. Compared with a counterfactual scenario where Instagram would have not been able to grow as a social network, the merger may still have provided Facebook the means to consolidate and strengthen a competitive advantage.
Section IV also evaluates the Google/Waze merger clearance.
Before clearing this transaction in 2013, the OFT considered a number of theories of harm, in particular: (i) loss of competition in the market for mobile turn-by-turn navigation applications, with concomitant reduction of the parties’ incentives to innovate and of the quality of service offered to users; (ii) loss of innovation by removing a disruptive entrant from the market . The first theory of harm was dismissed because Waze had not reached a sufficient user base in the UK to build a map with coverage and accuracy comparable with Google’s, and because there were other mobile turn-by-turn navigation applications constraining Google. The second theory of harm was dismissed because Waze’s projected growth was too uncertain.
Google and Waze both provide turn-by-turn navigation services, which are however fundamentally different in terms of their characteristics. Unlike Google Maps, the Waze map is user-generated, and the app is mainly used by heavy drivers. By exploiting their complementarities, the merger between Google and Waze allowed the merging parties to share data and information, reduce the costs of entering new geographic areas and avoid certain cost duplications. Moreover, efficiencies that resulted in the improvement of Google Maps were realised to the benefit of all Google Maps users. In the years after the merger, the number of Waze’s active users has increased, and by 2018 Waze still represented one of the main alternatives to Google Maps, together with Apple Maps.
The OFT seems to have been very—perhaps too—cautious in its assessment of the evidence when dismissing threats to potential competition due to uncertainty in future market developments. More importantly, the OFT failed to explore how these apps could be monetised. Ultimately, Waze proved to own a very innovative technology, and this may have allowed it to grow. However, while reliance on crowd sourcing enabled Waze to supply accurate and valuable real time traffic information, this approach may have had some limits regarding maps’ accuracy or coverage. Finally, the merger enabled Google Maps and Waze to exploit their complementarities and generate efficiencies.
Section V concludes and makes some recommendations.
The ex post evaluation reveals certain gaps in the way these cases were analysed, despite the depth of the analyses carried out. Merger control enforcement has so far not proved able to cope with several of the new challenges posed by digital markets. Within the existing legal framework, there are a number of steps that can be adopted. Agencies can be more interventionist in markets where network effects mean that competition is for the market, since the risk of false positives/false negatives will be different in such instances. Agencies should also seek better to understand the monetisation strategies of digital platforms, in advertising markets in particular. This is an important step in ensuring that all sides of the market are considered jointly.
In effect, business models and monetisation strategies should be taken into account when developing theories of harm, since market power has the ultimate objective of increasing profits. Given what we now know about these models and strategies, we may need to consider longer timeframes for prospective assessment than we currently do. In addition, competition authorities need to incorporate greater uncertainty into their theories of harm.
This is serious work, and an example of how to do ex post merger analysis – and of why competition agencies should pursue them. You might be surprised to see so many Italian names conducting a retrospective on UK merger practice. The reason is simple: the article is largely based on a study they conducted for the UK’s CMA. From this springs one of the main virtues of the paper: the authors were given access to a wealth of internal documents and submissions that, together with recent advances in the literature, allowed them to provide a more accurate ex post assessment than they otherwise would.
The analysis on the nature of acquisitions closely mirrors that of Axel Gautier and Joe Lamesch ‘Mergers in the Digital Economy’ (2020) Information Economics and Policy (reviewed here). While the in-depth analysis of two mergers was very interesting, I felt that the authors could have done more to explore the implications of the type of mergers they identified. I would have particularly liked to have seen a more systematic framing of the issues, and to have seen connections being drawn to the broad merger practices identified early in the paper.