The aim of this article, which can be found here, is to analyse some of the main issues that arose in the European Commission’s Google Android decision, and to place these issues in the context of hotly debated broader themes relating to antitrust enforcement in hi-tech markets. The author is head of unit at the European Commission, so his analysis may be more authoritative than other ones, at least until the full decision is published.
The piece is structured as follows:
Section II provides an overview of the Commission’s decision.
Android is an open-source smart mobile operating system. Google started providing the core version of Android commercially to smartphone and tablet manufacturers (“OEMs”) for free, but included a range of contractual requirements relating to the terms for obtaining Google’s associated proprietary apps (e.g. Google’s search app) and services. The free and open-source provision of Android was a key part of getting all major OEMs signed up, which led (by 2011) to Google having a dominant position with Android, the associated app store (Play Store), as well as of course in search.
As a first step, the Decision defines various markets, including upstream markets for smart licensable operating systems and Android app stores. It finds that these products are provided to OEMs in an upstream market where Apple is simply not present, since iOS and the Apple App Store are not provided commercially by Apple to OEMs. That is not to say that there is no competition between Android devices and Apple devices downstream, but that is an indirect constraint that must be analysed in the context of the assessment of Google’s dominance.
Against the backdrop of Google’s 90% plus market shares and the high barriers to entry in the two upstream markets, the Decision examines in detail the extent to which downstream device competition, or to be more precise, the possibility of switching from Android devices to Apple devices, is sufficient to constrain Google’s dominance upstream. It finds, on the basis of a range of factors that such a possibility does not constrain Google’s dominance. This is because: (1) an operating system is only one component among others of a smart mobile device, meaning that it is only a limited, indirect factor that is taken into account by users when considering devices; (2) empirically, there is limited switching between Android and Apple devices, not least due to consumer switching costs; and (3) Apple devices are not present on the mid to low end of the downstream device market.
The Decision identified various abuses, all of which had in common that Google used Android as a vehicle to extend and protect its search dominance in the mobile space (and therefore its main source of revenue, which comes from search advertising). The abuses that the Decision concludes took place were the following:
Abuse 1 – Tying of the Google Search app to the Play Store, and of Google Chrome to the Play Store and Google Search.
There was a significant pre-installation advantage that Google Search and Google Chrome obtained as a result of being pre-installed on virtually all Android devices through the tie-in. For the two products concerned – search and browser – the figures show that downloads are limited, and that on devices where Google was not pre-installed (e.g. Windows smartphones), the use of Google search was significantly lower than on Android devices. Market share developments in the two products were fully consistent with the finding that pre-installation.
In terms of claimed efficiencies, Google argued that Android had brought significant benefits to the mobile ecosystem by providing the market with a free and popular product that was the only effective counterweight to Apple. Google further argued that tying Search and Chrome was indispensable for Android to be brought to market. The Decision found against Google on this point. It concluded that Google did not provide any specific contemporaneous evidence in relation to the need for the specific tying restrictions. Instead, the Decision found that there were a number of other, less restrictive ways that Google was able to monetise the Android ecosystem.
Abuse 2 – Payments conditional on exclusive pre-installation of Google Search.
While the tying described above was about ensuring pre-installation of Google Search and Chrome, payments OEMs for the exclusive pre-installation of Google Search sought to ensure the exclusive pre-installation of Google Search in mobile devices.
Since this conduct relates to payments (of a share of search advertising revenues) conditional on exclusivity, the legal framework adopted by the Commission is based inter alia on that of the ECJ’s 2017 Intel judgment. While there remains a starting presumption that such conduct is abusive, the Decision analysed in detail its harmful effects. The Decision contained a quantitative as-efficient-competitor analysis which looked at how much a rival search engine with the same revenue per search and cost parameters as Google would have had to pay a device manufacturer or mobile network operator for the loss of the revenue share payments from Google and still make profits. Based on an analysis of what share of searches would be contestable across the portfolio of devices, the Decision found that a rival would have been unable to offer such compensation and still make profits.
Abuse 3 – Restrictions on OEMs selling Android forks (so called “anti-fragmentation” obligations).
As a condition of taking the Play Store, OEMs were contractually prohibited from developing or selling even a single device running on an Android fork. Such forks were a credible competitive threat to Google’s Android. and the restriction covered virtually the whole market. There was therefore direct foreclosure of rival open source operating systems – one illustration of this was the fact that a number of large manufacturers had been prevented from developing and selling devices based on Amazon’s Android fork (“Fire OS”).
Google argued that app developers could only have a predictable development platform with a uniform Android experience. With non-compatible versions of Android, there would have been a bad consumer experience since apps would crash and any such problems would be imputed to Google. However, the Commission found that Google brought no convincing evidence that any crashes would occur on devices using Android forks. Furthermore, Google had the possibility to use a variety of branding methods to ensure that consumers could differentiate between Google’s Android and Android forks.
The paper concludes that, while each of the conducts outlined above was found abusive in its own right, the Decision also concluded that they formed part of a single and continuous strategy with the same objective – to protect and strengthen Google’s dominant position in search, which remains by far Google’s main source of revenue (via search advertising).
The legal framework was similar to that adopted in earlier Commission decisions. At the same time, the Decision’s analysis of market conditions, harm to competition and objective justifications was able to effectively take into account some so-called ‘new phenomena’ such as the role of data, “free” products, or the two-sidedness of markets.
Certain academic commentators seem to require an effects-based approach setting such a high threshold for a finding of anticompetitive conduct that there would never be competition intervention against dominant companies. While it is very important for competition authorities to demonstrate that a certain conduct is capable of having harmful effects, it is also important not to drift towards a situation where there are significant risks of under-enforcement.
I cannot comment on this, but really there is not much to say; this is an authoritative description of the Google Android decision and of the theory of harm underpinning it – defensive leveraging, as in the US Microsoft / Netscape case.