Corresponding author: Natalia S. Pavlova (
Academic editor:
Digital antitrust is at the forefront of all expert discussions and is far from becoming an area of consensus among researchers. Moreover, the prescriptions for developed countries do not fit well the situation in developing countries, and namely in BRICS: where the violator of antitrust laws is based compared to national firms becomes an important factor that links competition and industrial policy. The article uses three recent cases from Russian antitrust policy in the digital sphere to illustrate typical patterns of platform conduct that lead not just to a restriction of competition that needs to be remedied by antitrust measures, but also to noteworthy distribution effects. The cases also illustrate the approach taken by the Russian competition authority to some typical problems that arise in digital markets, e.g. market definition, conduct interpretation, behavioral effects, and remedies. The analysis sheds light on the specifics of Russian antitrust policy in digital markets, as well as their interpretation in the context of competition policy in developing countries and the link between competition and industrial policies.
Russian antitrust as a direction of economic policy has a relatively short history, but a rich experience in the application of antitrust laws in areas that are considered digital. On some issues the Russian antitrust authority has even proven to be a leading force, such as prohibiting Google’s conduct of precluding the pre-installation of competing search services on Android devices almost three years before a similar decision by the European Commission (
Russia, like other BRICS countries, has been and remains a country with a developing market economy, where the agenda is historically different from that of developed countries, where the issue of balancing protective and active instruments of competition policy, as well as industrial policy, is not so acute. However, it is law enforcement in digital cases that points to a tight connection between antitrust and industrial policy, which can become a reason to consider their balance not only in the BRICS countries, but also in the countries where a developed market economy is believed to have established itself completely and irrevocably.
Competition policy in the digital sphere is, indeed, a complicated endeavor. On the one hand, digital giants have been a source of unprecedented innovations, which are closely linked to their ability to capitalize on network effects.
In terms of the platform theory, namely platforms’ main characteristics, which contribute to the formation of their behavior in digital markets as opposed to traditional firms, at least two recent works have been published that contain thorough overviews of relevant literature (
Additionally, there are at least two severe complications to this statement that currently distinguish competition policy’s agenda. One has already been mentioned: if the dominant companies are based in developed countries and the consumers or other types of customers are based in developing countries, then the abuse of dominance leads to a redistributive effect from lower-income countries to higher-income ones. But it is not just about the producers’ and the consumers’ welfare on isolated markets: the effects can have consequences in terms of global value chains in the digital sphere. As pointed out in
To clarify our approach, we stress the differences between antimonopoly policy and competition policy, since emerging market economies confront unique challenges to finding a dynamic balance between protective and active components of competition policy. While active competition policy includes reforms of regulation and a market-oriented system of public administration, as well as reforms of industries with a natural monopoly component and competition advocacy, antitrust at its core includes deterrence of dominance abuse, deterrence of anticompetitive agreements (primarily, hardcore cartels) and control of mergers. We will also discuss how antitrust policy corresponds with some active measures of competition and industrial policy.
Section 2 describes three big recent cases in Russian antitrust that illustrate this narrative and highlight the associated issues: the Yandex–Google case (2015), the Kaspersky–Microsoft case (2017) and the Kaspersky–Apple case. We describe the details of these cases, the
It might be considered the first case in Russia that dealt with platform behavior. Yet platforms and multi-sided markets were not defined in Russian competition law at this time (and haven’t been defined to this day, although the proposed “Fifth antimonopoly package” of amendments addresses that), so the interpretations used in this case gravitated to more traditional concepts. The
Initially, the behavior of Google was interpreted in terms of unfair competition (prohibited according to Article 14 of the law “On protection of competition”), although later the case was requalified in terms of abuse of dominance, prohibited by Article 10. The anticompetitive behavior, as established by
Summing up the competition authority’s approach, we can see that the violation is interpreted in terms of anticompetitive tying, which is a “traditional” type of antitrust violation. A factor that puts “a new spin” on it is that Google Play can be considered to be a platform on a multi-sided market with at least three distinct groups of users: portable device manufacturers, consumers of devices (and applications) and independent producers of applications. Consequently, even though the customers on the market are defined as device manufacturers, dominance is in fact preconditioned on the inability of end consumers (users of devices and applications) to switch to a different means of app installation.
The anticompetitive effects also appear on markets that are different from that where Google has a dominant position — on markets for services analogous to the ones in the GMS package, namely search services. This can be seen as a leveraging strategy: by tying Google Play with the GMS package Google leverages its market power on to the markets for apps. But what makes this strategy so successful is another specific trait of digital markets, and that is the exploitation of consumers’ cognitive biases. In this case, pre-installation as a strategy becomes such an efficient channel — and precluding access to pre-installation for independent developers becomes such a powerful way to restrict competition — because of the status quo bias (
Before considering the details of this particular case, it is worth looking back briefly at the history of antitrust enforcement against Microsoft.
In the U.S. case the object was the practice of tying the
The initial court decision imposed a structural remedy: the company was to be divided into separate businesses, one dealing with the operating system and the other with browsers and other software. The decision was appealed and eventually the settlement imposed only behavioral remedies without any structural requirements. The logic that allowed the requirements to be softened was based on the concept of network effects: by virtue of having a large user data base Microsoft was encouraging software developers and computer hardware markets to focus their efforts on Windows, which led to a positive feedback loop that elevated Windows against competing services.
The
Comparing the U.S. approach with the
Having the details of these cases for context, we turn to the Russian antitrust case against Microsoft.
Microsoft unilaterally deleted independent antivirus software (drivers that enabled the software to launch), having detected its incompatibility (as assessed by Microsoft) with Windows 10;
Microsoft reduced the time for antivirus developers to ensure the compatibility of their software by providing them with the newest Windows version just one week in advance;
Microsoft actively induced (utilizing the built-in visualization options available to them) users to turn on the pre-installed antivirus software (Defender) in such a manner that would lead to independent antivirus software being de-activated;
Microsoft precluded independent antivirus software from actively using visual notifications to remind customers about the need to renew their licenses in the next three days after the license expires;
if two independent antivirus programs start to function with Windows 10 simultaneously, both are de-activated (due to a restriction on the number of antivirus programs that can run at the same time) and Defender is automatically turned on instead of them.
Kaspersky Lab pointed out that this behavior could have produced anticompetitive effects on the market for antivirus software because it led to unfair advantages for the pre-installed Defender in relation to independent antiviruses, and taking into account the dominant position of Microsoft on the market for operating systems for PCs those advantages posed a serious risk to the position of Defender’s competitors. Microsoft, in return, accentuated the security benefits of providing a pre-installed antivirus. While the case was still being considered by the competition authority, Microsoft had made certain alterations to its practice and policies which mitigated part of the antitrust concerns.
Considering the possible interpretations of Microsoft’s actions, alternative theories and concepts can be implemented.
Firstly, the OS can be interpreted as a platform that provides services to different participants of a two-sided market — the sellers of applied software that want their software to be installed on PCs with a certain OS, and users that want to buy certain software.
Secondly, the OS can be thought of as a primary product, the functionality of which requires that users buy a complementary (aftermarket) product — applied software.
Thirdly, emphasis can be put on the high specificity of applied software, developed specifically to work with a given OS, which automatically gives the developer of the OS an advantage when dealing with software developers.
Fourthly, we consider it possible to examine the OS as a natural monopoly, an infrastructure industry of a sort, which users — developers and consumers of software — “connect” to, although, at first glance, the development of an OS hardly resembles traditional infrastructure industries. But from this point of view a concept that seems to explain rather accurately the behavior of Microsoft in this case is the concept of essential facilities.
Essential facilities are assets “the use of which is a necessary condition for production in a given industry and duplicating which is unfeasible for technical or economic reasons” (
Another important aspect is the ability of the OS developers to manipulate access to information, which became a major point in this case. The developers of the OS have exclusive access to an incomparably important communication channel for users — system messages on their PCs. This allows the OS developers to exploit the users’ cognitive biases stemming from their bounded rationality.
Control over essential facilities for the software markets and the ability to provoke and exploit the cognitive biases of users allow the developers of OSs to (a) leverage market power from the market for OSs (if they have it) on to the markets for applied software, and (b) implement methods of unfair competition. Can this leveraging become anticompetitive? In our opinion, it can, if the OS developer is simultaneously a participant in a market for applied software. Such was the situation in the Kaspersky–Microsoft case, as well as in the cases against Microsoft that dealt with internet browsers and media players.
In this case, as well, different theories could be applied to interpret Apple’s conduct. One of these is, again, the essential facilities concept. Although different apps or classes of apps differ in their ease of adapting to an alternative mobile OS (from iOS to Android, for example), parental control apps are among those that are difficult to adapt, as they use different approaches to utilizing each system’s capabilities to ensure their function and are often developed by different specialists, specializing only on one of the popular mobile OSs. In order to gain access to customers the parental control app, developed for a certain mobile OS, needs to be installed onto the device with this OS. In this sense, access to the OS is an essential facility: it is technically and economically impossible for parental control app developers to reach their customers without installation on the OS. For Apple devices the only approved way of installing independent apps is via the App Store, which means that access to the App Store for app developers is an essential facility as well. The combination of iOS and App Store exhibits all the necessary characteristics of essential facilities (
control of the essential facility by a monopolist: Apple owns iOS and App Store, and due to its own policies, there are no official alternative operating systems for Apple devices, or app markets for iOS. The situation for Apple differs from that of, for example, Android devices, where official alternative app stores are available (such as Amazon AppStore, Samsung Galaxy Apps, etc.);
a competitor’s inability practically or reasonably to duplicate the essential facility: once again, this is a consequence of Apple’s own policy and its orientation on a closed eco-system. Alternative stores for apps that have been introduced (Cydia, AltStore) either require a “jailbreak” or have an otherwise reduced functionality that makes them a marginal phenomenon;
the denial of the use of the facility to a competitor: use of the App Store and access to iOS has been denied to Kaspersky Lab and other parental control apps that compete with Apple’s own pre-installed Screen Time app;
the feasibility of providing the facility: it stems from the fact that previous versions of KSK that also utilized configuration profiles had earlier been approved for the App Store: the original app, 20 more updated versions before the tightening of Apple’s policy on using MDM-profiles, 5 more after the policy change and before the introduction of iOS 12, and even one full updated version after the introduction of iOS 12 (later versions were approved only with reduced functions).
If applied, the essential facilities concept leads to the logical implication that Apple does in fact have the obligation to provide non-discriminatory access to the App Store and the iOS for the apps of independent developers and its own apps. It is important to note that it is not just the App Store that the non-discriminatory access is applied to, but also the operating system itself, as, for example, Screen Time comes pre-installed on the device and does not need to be downloaded via App Store.
Regulating the rules of access to an essential facility can be seen as a means of leveraging market power, but even without addressing the concept of essential facilities Apple’s conduct can also be interpreted as leveraging market power on the market for app stores for Apple devices onto the market for parental control apps for Apple devices.
Another side to the leveraging approach is added by considering Apple’s bundling strategy: its mobile devices come with a pre-installed OS (iOS), a pre-installed app store (App Store) and pre-installed app that can be used for parental control (Screen Time). And if the OS market and app store market for Apple devices are already monopolized, Screen Time competed against other parental control apps for iOS. Bundling a product from a competitive market with a monopoly product can be considered a form of abuse of dominance.
Another alternative to the essential facilities approach would be an interpretation in terms of platforms, which, as has already been noted in the description of the Yandex–Google case, appears to be valid when dealing with any app store. App Store connects developers of mobile apps for iOS, including parental control apps, and app users that search for the apps and install them by using the App Store. The more app developers are represented on the platform, the more apps there are for the users to choose from and the higher the utility of the users; and the more users are looking for apps with the help of the platform, the more attractive the platform is for app developers. This shows us that the App Store is a platform characterized by two-way cross-side network effects. The fact that users do not easily switch between platforms (that would entail a need for a new device, as only Apple devices support iOS) coupled with Apple’s orientation on building an eco-system of devices and software means that consumers become locked in when initially choosing a platform: they are single-home rather than multi-home with their mobile devices. In this case, losing access to a platform means losing access to its clients, and as Apple’s clients are the ones spending a larger share on services in spite of owning a smaller share of devices,
Whichever interpretation of the abovementioned cases is used, the result is not only a judgement on the efficiency outcome, but also a cause to consider redistribution. In this section we will analyze the common traits of these cases in light of more general tendencies in competition policy in the digital sphere, tendencies specific for the BRICS countries, as well as traits that are characteristic of competition policy in Russia and their relation to industrial policy.
What the antitrust cases have in common is that all three of them seem to illustrate a typical strategy for platforms that deal with so-called dual distribution. Originally dual distribution means that a manufacturer simultaneously sells to distributors, who then supply end-consumers, and to end-consumers directly. This is similar to platforms, many of which perform the dual role of marketplace and online retailer or manufacturer. In all three cases, the owner of the platform (Google, Microsoft, Apple) is also the owner of a product that is distributed via that platform (Google search, Microsoft Defender, Screen Time app) and appears to use his position as a platform to manipulate its rules in order to promote his own product. One possible interpretation for this behavior is the well-known concept of leveraging: the owner uses his market power on the platform market and leverages it to increase market power in an adjacent market. This is the sort of market strategy that is becoming widely recognized throughout the world, while the need to deal with it turns into a major point for political campaigns (see, e.g., Elizabeth Warren’s arguments during her presidential campaign in the U.S.
For developing countries, such as BRICS countries, this sort of leveraging strategy has additional implications. The three cases illustrate a typical situation of the sort: we see that the platform with the market power is owned by a global corporation, while the market where the anticompetitive effects occur has among its participants national (Russian) IT-firms, and these firms suffer the consequences of the anticompetitive practice by losing profits and market share. In each case, a “laissez-faire” attitude could have had severe consequences for the national firms, contributing perhaps to raising the risk of the Vanek–Reinert effect. Consequently, by protecting competition the antimonopoly authority contributed as well to protecting a prominent national competitor — an effect in line with the goals of industrial policy in the IT sphere.
All of these cases, as has been noted, can also be interpreted in terms of denial of access to essential facilities. Of course, this approach has its own drawbacks that are commonly associated with applying the essential facilities doctrine to industries other than natural monopolies (
Beside the general characteristics of platform strategies and the BRICS-specific (or specific for developing countries) effects, the three cases also illustrate some special traits that are characteristic of antitrust in Russia. Considering the antitrust policy itself, although all three cases can be interpreted in terms of platforms, that is not the approach chosen by the antimonopoly authority.
A possible reason for not utilizing the platform concept is that platforms are not yet defined in the Russian law “On protection of competition”: introducing a working definition of platforms into the legislation is one of the main points of the “Fifth antimonopoly package” that is currently being discussed at the government level. The Yandex–Google case is older than the first version of the package that was introduced to the public and probably influenced its creation. But in later cases avoiding using the platform logic can be taken as an effect of the primate of the legal over the economic that is prominent in Russian antitrust policy and is noted by experts. Additionally, for a legal concept to be implementable, a key factor is the role of meso-institutions (
But while the platform interpretation is not explicitly used by the competition authority in these cases, neither is the alternative approach provided by the European case against Microsoft — the essential facilities doctrine. The concept of essential facilities is not defined explicitly either by Russian competition law, or its practice, even for natural monopolies. The question is, then, what approach
In the Yandex–Google case and the Kaspersky–Microsoft case the competition authority solved the market delineating issue by defining markets narrowly, including focusing on only one side of a multi-sided market. In the Yandex–Google case the market boundaries were defined as pre-installed app stores for OS Android, localized for distribution in the Russian Federation. The buyers on this market were defined to be the manufacturers of mobile devices, which pre-install app stores in order to resell them bundled with their devices to end-consumers. In the Kaspersky–Microsoft case the market was defined as RTM-versions of OS for stationary devices for adapting application software developed in the Russian Federation. In each case the multi-sided nature of the market is not addressed explicitly, but the role of consumer (end-consumer) switching patterns is taken into account (see the next point), although the role of indirect network effects is never brought up.
A similar case in this regard is the Yandex–Uber joint venture (2017). A major challenge in the case was market definition. Although consumers tend to equate aggregator services with taxi services, neither Yandex nor Uber provided the taxi services themselves. The
What these cases do recognize explicitly is the role of switching patterns in adjacent markets as a source of market power on the markets in question. In the Yandex–Google case the fact that only pre-installed app stores (for OS Android) are included in the market is not only due to the inability of users to download Google Play on their own, but their strong preference for devices with a pre-installed app store (from 63.3% to 69% of users hold the opinion that an app store is a necessary prerequisite for a device they are willing to buy
In the Kaspersky–Apple case as well the inability or at least strong reluctance of app developers to switch is not only due to additional costs they would have to incur (hiring new personnel that specializes on app development for another operating system, letting go the ones that specialize on the OS that the company is switching from or retraining them), but also the loss of profits that stems from brand-loyalty of end-consumers of mobile devices.
As for the markets where anticompetitive effects occurred (search apps, antivirus software, parental control apps), they were not defined as rigorously as the markets where the dominant position was proven, but the fact that dominance on one market was abused to suppress competition in another market indicates that the competition authority was using (if not explicitly) the leveraging concept in its decision.
For those who control and supply the platform, pre-installing and making a service the default solution on the platform is a powerful tool of promoting their own services. This tying strategy is made even more effective by the effects of bounded rationality of end-consumers, such as inertia and the status quo bias. In both the Yandex–Google and the Kaspersky–Microsoft case the competition authority used anticompetitive tying as a theory of harm. We see that all the cases recognize consumer bias as an exacerbating factor for the ability of the platforms to restrict competition.
We can also conclude that Russian antitrust incorporates the phenomenon of leveraging. In all three cases we deal with a firm that is dominant in one market, but abuses that dominance to influence competition in another, previously competitive market. We also see that the quantification of the effects of restriction of competition in this latter market is not at the center of these cases — the main market analysis pertains to the product where the violator potentially has dominance. This, perhaps, stems from the fact that in Russian competition law, and namely in Article 10 of the law “On protection of competition” that deals with abuse of dominance, definitive proof of negative effects for competition is not necessarily needed to prove a violation: conduct that leads to “infringement of the interests of other persons (economic entities) in the sphere of entrepreneurship activity or indefinite range of consumers”
Currently the
The efficiency defense based on the ground of network effects provided by platform owners and of interoperability between a platform itself and its aftermarket products and services does not seem to be an attractive argument for
In fact it means, on the one hand, a certain distribution of the burden of proof between the platform owners and the regulator, and, on the other hand, a challenge for the potential beneficiaries of the efficiency defense arguments to create demand for knowledge of this type.
Thus, one of the key points of probable agenda for an antitrust upgrade in Russia as well as in other BRICS countries in line with challenges of systemic digital transformation of national economies is to search for adequate instruments of theoretical and applied research for antimonopoly law enactment and enforcement.
Although the cases recognize some of the effects of consumer bias due to bounded rationality, these are never addressed on a theoretical level: rather, they are addressed by introducing information from consumer polls to confirm the effects of these biases (in the Yandex–Google case — how rarely consumers tend to switch away from a pre-installed option).
The issue with incorporating behavioral biases on a theoretical level is well-known: the instruments of economic analysis that are used in antitrust and the norms shaped by them rely mostly on neoclassical economics where unbounded rationality is presupposed. Behavioral economics, although more realistic in some of its predictions, does not yet offer a full alternative for the “traditional” theory and system of instruments of economic analysis for antitrust (
In this vein, the importance of status quo bias was explicitly uncovered in the Yandex–Google case that then reverberated in the Kaspersky–Microsoft and Kaspersky–Apple cases. It illustrated how pre-installation could guarantee significant competitive advantages for software, even when monetary and time switching costs (but not cognitive switching costs) for consumers are low. This ended up influencing not just Russian competition policy– we see that the effects of pre-installation are addressed as concerns in all three cases — but also the country’s industrial policy.
The
The digital cases in Russian antitrust do not go alone. The relationship between digital and “traditional” markets might be much more complex. One recent example is the Bayer–Monsanto merger (2018) that was cleared by antitrust authorities in a number of countries, including BRICS. Although all the BRICS competition authorities were interested in the merger, the risks posed to competition differed significantly depending on the sort of crops and agricultural products for which each country was dependent on the companies. There was also a digital aspect to the case: the merging firms aimed at introducing a digital platform of precise farming that could help farmers optimize their work. In Russia, the link between antitrust and agricultural policy shaped a unique remedy issued by
Finally, another digital aspect of the junction between competition and industrial policy, and for which the
All in all, the cases discussed appear to be, on the one hand, in line with existing international practice, but, on the other hand, they are clearly restricted in the use of modern economic concepts that have not yet found their way into Russian competition law. Faced with the same problems as other competition authorities in defining digital markets and interpreting the behavior of digital firms, the
Yet, looking at just these three high-profile cases, it can already be said that the competition authority is somewhat constrained in its choice of instruments of economic analysis. As their use is closely linked with evidence standards, the limitations can become critical with the inevitable increase of scale of dealing with digital companies, which, in turn, might lead to an increase in the frequency of enforcement errors.
Finally, high profile digital cases demonstrate the challenges for authorities of countries with national producers competing with global giants. This explains the pioneering character of Russian antimonopoly law enforcement vis-à-vis the EU and US, in spite of a deficiency of economic analysis. Thus, to some extent, standards of proof are the price for timeliness of industrial policy decisions, where
✩ The article was prepared as part of research for the state assignment of RANEPA.
Two concepts used in economic literature are mutually substitutable: network externalities and network effects. Yet the term “network externality” means more than the dependence of the utility of consumers on the number of consumers, since the very concept of externality relates to costs and benefits not reflected in prices or in contract clauses. So, the internalization of externalities due to the use of different institutional arrangements might mean the move to Pareto-improvement, while network effects are simply the material basis for that.
In platform economics, direct network effect takes place when a user’s utility or benefit from the platform’s service depends on the number of users on the same side of the platform (or, in other words, from the same group of users) (see, e.g., Katz and Shapiro, 1985), while indirect network effects take place when one group’s user’s benefit depends on how well the platform does in attracting users from another group/on another side of the platform (Armstrong, 2006).
The details of these cases can be found, for example, in
This section relies heavily on our previous article (Shastitko and Kurdin, 2017).
RTM (Release to Manufacturing) version — a version provided to application manufacturers in order to make possible their adaptation to changes in OS.
API (Application Programming Interface) is a mechanism providing specific OS services to applications.
Breaking the political influence of market-dominant companies.
Similarly, other BRICS countries aren’t rushing to implement platform economics in their decisions, either — see
The exception here is the Kaspersky Lab–Apple case, where the FAS Russia explicitly uses such terms as “platform” and “multi-sided market” in its recent decision, but although the concepts are named, many of their aspects remain ignored in the decision.
The legislation will presumably cover devices that will be sold in Russia starting in 2021. It is worth noting that the proposed legislation does not include the names of any specific apps that need to be installed, but rather categories of apps corresponding to three types of devices: smart devices with sensor-screens; stationary devices, PCs and laptops; smart TVs. The categories of apps include search apps and browsers, antivirus software, maps, e-mail services and others.
Ibid.