Dealing with digital dominance: insights from Germany

The German government is looking at how competition law can better address abuse of dominance in digital markets. As part of this, an academic study has recommended giving authorities powers to intervene earlier and in additional settings where large platforms or valuable datasets are involved. How can these proposed reforms improve market outcomes while limiting the scope for unintended consequences?

This article looks at Schweitzer, H., Haucap, J., Kerber, W. and Welker, R. (2018), ‘Modernisierung der Missbrauchsaufsicht für marktmächtige Unternehmen’.

Competition in the digital economy is keeping courts and authorities busy throughout Europe and beyond. Ongoing investigations by competition authorities include the use of marketplace data by Amazon,¹ advertising at Google,² and privacy terms at Facebook.³ In addition, wider debates are looking at the state of competition in online advertising more widely,⁴ markets for data,⁵ the impact of algorithms,⁶ and other issues. Many of the markets under investigation have a small number of players, which raises the question of whether existing competition law adequately captures the concepts of market power and dominance in digital markets.⁷

Of the European member states, Germany is one of the more active when it comes to applying competition law to digital issues. The German government is currently working on an upcoming amendment to the Act against Restraints of Competition (Gesetz gegen Wettbewerbsbeschränkungen, GWB). This tenth amendment, planned for 2021, aims to modernise the law to ensure that it is fit to protect competition in the digital economy. Similar debates are taking place elsewhere, through expert panels at the European Commission and in individual member states, including the UK.⁸ As part of the ninth amendment in 2017, the German government set a precedent for legal change by introducing various reforms relating to digital markets, including a value-based threshold for merger transactions of €400m, above which they become subject to merger control.

To identify the current gaps, the German Federal Ministry of Economic Affairs and Energy (Bundesministerium für Wirtschaft und Energie, BMWi) commissioned a study from a team of academics (Professor Dr Heike Schweitzer, Professor Dr Justus Haucap, Professor Dr Wolfgang Kerber and Robert Welker).⁹ The study asked how German competition law could address concerns about market power in the digital economy. While the authors do not recommend a systematically lower threshold for intervention, they do recommend changes in relation to:

• the prevention of market ‘tipping’;
• the assessment of platform dominance;
• potential anticompetitive effects of mergers;
• access to data.

Authorities elsewhere in Europe and beyond are likely to use these insights to inform their own debates on digital competition. This article summarises some of the main recommendations made by Schweitzer et al., and discusses their economic benefits and potential challenges.

Assessing dominance

The study sets out recommendations for how to prevent firms becoming dominant, as well as how authorities can establish dominance when assessing markets.

Behaviour that encourages tipping

Market tipping occurs when one firm emerges as dominant (including where it is the only player in a market), for instance as a result of strong network effects or firm decisions to restrict interoperability. For example, Facebook is often considered to be the ‘winner’ in the market for social networks.

If a market has tipped in this way, and there is no threat of dynamic competition from would-be market ‘disruptors’, this may affect market outcomes when dominant firms act with limited competitive constraints. In these cases, authorities may decide to take strong measures to regulate and/or reintroduce competition.

Schweitzer et al. suggest amending the law to prohibit actions that encourage tipping in markets with sufficient network effects before any player becomes dominant. For example, flat rates, exclusivity requirements or long cancellation periods increase the costs of multi-homing¹⁰ or switching. According to the proposals, such actions would be considered anticompetitive if undertaken by certain types of player or players, such as a large and growing platform or several similarly sized firms. The authors state that a key benefit of this would be to maintain competition before a dominant player emerges, such that a potentially more onerous intervention at a later stage would be unnecessary.

The proposed change would be more likely to increase competition and avoid side effects if authorities and courts followed a framework that balanced clear principles and flexibility when considering market-specific factors. This could address two important questions.

• First, at what point is a market sufficiently in danger of tipping to allow for intervention below the threshold of dominance? Network effects matter in terms of strength, symmetry, whether they are direct and/or indirect, and at what critical mass they begin to develop. The value and frequency of transactions also determine how likely users are to multi-home or switch platforms. These factors can serve as a starting point for assessing platform competition, but market-specific circumstances (such as behavioural biases) may also be important.
• Second, how can actions that favour tipping be distinguished from desirable (albeit aggressive) competition on aspects such as quality or price? Schweitzer et al. suggest that one way to identify an anticompetitive practice is to consider whether it is profitable only if a monopoly position is achieved in the long term (in line with the usual test for predation in proceedings against abuse of dominance). However, they acknowledge that some practices may be anticompetitive without being based on predation, and that it is difficult to draw a clear line between the two. For example, in the case of Amazon’s Prime service, membership lowers the prices that consumers pay for individual transactions on Amazon, but it also reduces the incentive to multi-home as other platforms or retailers may struggle to compete on the prices of individual transactions. In addition, the study recognises that increasing the cost of multi-homing can also be an admissible strategy for growing market entrants, or for firms (such as Apple) that invest significantly in a ‘closed’ system.¹¹

From an economic perspective, the proposed intervention would make a difference in markets where firm behaviour can be pivotal in making the market tip. As such, it is important to get the balance in enforcement right. Over-enforcement could be particularly harmful if additional rules on platforms in markets that are unlikely to tip prevented firms from competing vigorously with each other, or if strategies that primarily enhance consumer welfare were mistaken for anticompetitive behaviour.

In contrast, the proposal may enable more dynamic competition in markets that tip irrespective of firm behaviour. In these cases, additional rules will not stop markets tipping (unless the rules are too restrictive and protect inefficient competitors or promote excessive entry), but could lower the barriers to entry for future competitors (for example, by moderating switching costs).

Such a law would need to take into account the potential benefits of preserving (more) competition on the one hand, and the potential harm from legal uncertainty and over-enforcement on the other. Authorities would therefore need to assess markets and firms’ strategies on a case-by-case basis and provide appropriate ex ante guidance.

Establishing intermediation power

The GWB currently measures market power in terms of buyer and supplier power. In the digital economy, these concepts may not capture the market power of a platform if the platform acts as an intermediary rather than as a seller or buyer.

In order to provide clarity on the concept of dominance, Schweitzer et al. suggest recognising ‘intermediation power’ as a third dimension of dominance. Their recommendation focuses on intermediaries that have a relationship with users on both sides of a transaction. However, the authors reach no firm conclusion regarding information intermediaries (which have a contractual relationship with one side only—such as search engines or news aggregators). In this case, they find that additional considerations come into play, such as potentially ineffective competition if consumers struggle to compare information intermediaries (for example, due to the underlying ranking algorithms of search engines).¹² Schweitzer et al. note that this can mean that consumers need to rely on trust, which may give a platform power even below the conventional thresholds for dominance.

Intermediation power appears to be a reasonable concept in helping authorities and courts to analyse market power. However, it is important to highlight that some platforms also compete with firms that rely on other business models. A high share in intermediation would therefore need to be considered alongside other indicators, including buyer and supplier power. For example, a flight booking platform might have a high share in intermediation, but other booking channels, such as airline websites, might also exert competitive pressure.

Finding dominance without defining a market

Market definition is particularly challenging in investigations in multi-sided markets. This is because it is not easy to apply the standard ‘hypothetical monopoly test’ framework in the presence of multi-sided price setting (with feedback effects between the sides) and zero prices (where users are not charged for services).

One shortcut that Schweitzer et al. propose is to infer dominance from an observed abuse. They suggest giving courts the option to skip the market definition exercise if firm behaviour implies that the competitive constraints are insufficient.

In practice, this would need to balance the potential benefits and harm of such a change. Shorter and more focused proceedings can save legal resources and are more likely to preserve competition, because any anticompetitive behaviour is addressed earlier on. However, as Schweitzer et al. acknowledge, if it becomes easier to make a finding of dominance (and abuse), this is likely to increase the risk of ‘false positives’—i.e. firms being found to be dominant when they are not. This is least likely to happen with strategies that are profitable only for dominant players. In these cases, skipping the market definition stage could allow authorities to analyse the alleged abuse directly.

In many cases, however, strategies can be procompetitive when implemented by a non-dominant firm, but can harm competition when a dominant firm is involved. In order for the market definition exercise to be skipped, the effects of practices would need to be assessed on a case-by-case basis (which is important anyway for demonstrating an abuse of dominance). For example, increasing the cost of multi-homing can be a suitable strategy for a smaller competitor to build a loyal user base, but may reduce competition if it is the strategy of a large platform.

The box below discusses how similar concepts to those of Schweitzer et al. are already applied in the UK.

Strategic purchase of potential competitors

Recently, some large online platforms have acquired small and growing firms, including Facebook/WhatsApp, Google/DoubleClick and Microsoft/GitHub. These transactions have raised concerns about a lack of dynamic competition,¹³ as small players have been acquired before they became large enough to pose a competitive constraint on incumbents. Although a new transaction value threshold was introduced in the ninth amendment of the GWB, it is unclear whether authorities have sufficient tools to assess the potential impact of mergers on dynamic competition.

Schweitzer et al. propose to trigger an intervention by the Bundeskartellamt if a dominant firm strategically acquires potential future competitors. This could be done ex post (in the context of abuse-of-dominance proceedings) or at the point of a merger. The study leaves open exactly how such a provision might be implemented; in principle, however, it would be beneficial to give as much guidance as possible on what is allowed at the point of merger control, as breaking up a merged platform is likely to create higher costs.

Maintaining dynamic competition is essential in digital platform markets. It would therefore be important to clearly define the types of transaction that would be considered problematic. For example, if the acquirer firm were to ‘shelve’ the innovative service of a newly acquired business, this would almost certainly result in losses to consumers.

However, acquisitions of small start-ups can also have a positive effect on innovation. For example, mergers can promote competition, as some start-ups may see such a deal as a very attractive commercial strategy, and incumbents can also provide the necessary funding to quickly scale up and bring an innovation to market. The potential difference in speed at which innovation reaches the masses means that it is difficult to judge a transaction in hindsight. In the example of Facebook/Instagram, Instagram might have been a credible competing service had it remained independent, but it could also have failed to attract as many users as it did absent the integration with Facebook.

Another challenge concerns how to assess transactions in less closely related markets. As many large platforms turn into ‘ecosystems’ or ‘conglomerate’ structures, it is difficult to anticipate which firms may become competitors (even if just as part of a conglomerate incumbent). While this goes beyond the scope of the Schweitzer et al. study, one way to address this uncertainty at the time of the merger might be to more closely examine the acquiring firm’s plans for its target’s service and introduce more forward-looking remedies in merger proceedings. Other proposals include putting more focus on the shares of multi-homing users between two merging services, as a way to identify those with a potentially high welfare loss through a concentration in advertising markets.¹⁴

Data access rights

Data increasingly represents an important asset for many digital businesses. This applies not only to data on platform users that allows the platform to tailor its services and match users efficiently, but also to data generated by devices such as fitness trackers. Access to such data varies across industries, but sharing and trading between firms tend to be limited. This is aligned with privacy concerns where the data in question includes personal characteristics. However, for other types of data, more trade may be desirable to enable innovation and competition.

Schweitzer et al. find that German competition law already enables access to data held by data-dominant firms. The law states that a firm must grant access to data that cannot be replicated and is indispensable for competing in the market in question or in related or after-markets. However, the authors find that the definition of data access rights should go beyond the scope of provisions of abuse of dominance, and call for a wider legal framework that balances innovation, competition and privacy.

The study recognises that there are likely to be situations where more open access to data could enhance competition—for example, where large datasets are needed to train algorithms. Schweitzer et al. highlight that any regulatory regime would need to ensure that firms still have sufficient incentives to collect data, while enabling the data to be put to productive use more widely than it is currently.

This raises many open questions, for example about the appropriate scope of data-sharing rules (e.g. how high should the threshold of ‘replicability’ be? How would one ensure that this will not facilitate coordination in the market?) and the appropriate mechanism for compensation (e.g. a FRAND framework¹⁵ could provide a starting point to balance the incentives for companies that collect data and those that use it).

Conclusion

Many of the gaps in competition law identified by Schweitzer et al. reflect the concerns of stakeholders across Europe, and highlight how there is an impression of under-enforcement.

At first glance, the recommended changes may appear to be minor; however, their impact on what digital platforms would be allowed to do could be significant. To the extent that imposing additional constraints on large platforms could foster both competition and consumer welfare, they could give enforcers valuable tools to prevent anticompetitive behaviour. Authorities might also be able to act more effectively by intervening earlier and faster, thereby preserving the disruptive potential of promising start-ups and ensuring that data is shared on appropriate terms with those who can use it productively.

In order to achieve these objectives, lawmakers will need to make sure that any unintended side effects do not outweigh the expected benefits. These effects might include imposing additional burdens on players without market power, and increasing legal uncertainty, which could increase the reluctance of both small and large players to engage in some types of innovative and vigorous competitive behaviour. However, if German lawmakers and authorities are able to develop these proposals into well-defined legal provisions, they may also serve as an example to other authorities worldwide.

¹ Schechner, S. and Pop, V. (2018), ‘EU Starts Preliminary Probe into Amazon’s Treatment of Merchant’, Wall Street Journal, 19 September.

² European Commission, Google AdSense, Case No. 40411.

³ Bundeskartellamt (2017), ‘Preliminary assessment in Facebook proceeding: Facebook’s collection and use of data from third-party sources is abusive’, press release, 19 December.

⁴ See Australian Competition & Consumer Commission (2018), ‘Digital Platforms Inquiry’, 26 February; Autorité de la Concurrence (2018), ‘Avis n° 18-A-03 du 6 mars 2018 portant sur l’exploitation des données dans le secteur de la publicité sur internet’, 6 March; and Bundeskartellamt (2018), ‘Bundeskartellamt launches sector inquiry into market conditions in online advertising sector’, press release, 1 February.

⁵ For example, European Commission (2018), ‘Towards a common European data space’, COM(2018) 232 final, 25 April.

⁶ For example, Competition and Markets Authority (2018), ‘Pricing algorithms. Economic working paper on the use of algorithms to facilitate collusion and personalised pricing’, 8 October; OECD (2017), ‘Algorithms and Collusion: Competition Policy in the Digital Age’, background paper; and Oxera (2017), ‘When algorithms set prices—winners and losers’, Agenda, June.

⁷ See Oxera (2018), ‘It’s what you know about who you know: market power in digital platforms’, Agenda, December.

⁸ See European Commission (2018), ‘Commission appoints Professors Heike Schweitzer, Jacques Crémer and Assistant Professor Yves-Alexandre de Montjoye as Special Advisers to Commissioner Vestager on future challenges of digitisation for competition policy’, News, 28 March; HM Treasury (2018), ‘Work kicks off to examine digital competition in UK’, News Story, 19 September; Federal Ministry for Economic Affairs and Energy (2018), ‘BMWi setzt Kommission Wettbewerbsrecht 4.0 ein – Kommission soll Vorschläge zur Modernisierung des Wettbewerbsrechts erarbeiten’, press release, 10 September.

⁹ Schweitzer, H., Haucap, J., Kerber, W. and Welker, R. (2018), ‘Modernisierung der Missbrauchsaufsicht für marktmächtige Unternehmen’.

¹⁰ I.e. using more than one service for the same purpose.

¹¹ A system that is not interoperable with those of other producers.

¹² The Bundeskartellamt has launched a sector investigation into comparison platforms that addresses some of these questions—see Bundeskartellamt (2018), ‘Konsultationspapier zur Sektoruntersuchung Vergleichsportale’, 12 December.

¹³ As discussed during the conference hosted by the European Commission, ‘Shaping competition policy in the era of digitisation’, on 17 January 2019, and in The Economist (2018), ‘American tech giants are making life tough for startups’, 2 June; and Stratechery (2017), ‘Manifestos and Monopolies’, 21 February.

¹⁴ Prat, A. and Valletti, T. (2018), ‘Attention Oligopoly’.

¹⁵ FRAND stands for fair, reasonable and non-discriminatory compensation that guides licence negotiations with holders of standard-essential patents as well as with collecting societies.